davidlt/kernel-ark
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
dc594c39f7
kernel-ark/net/ceph/osd_client.c

5453 lines
140 KiB
C
Raw Normal View History

License cleanup: add SPDX GPL-2.0 license identifier to files with no license Many source files in the tree are missing licensing information, which makes it harder for compliance tools to determine the correct license. By default all files without license information are under the default license of the kernel, which is GPL version 2. Update the files which contain no license information with the 'GPL-2.0' SPDX license identifier. The SPDX identifier is a legally binding shorthand, which can be used instead of the full boiler plate text. This patch is based on work done by Thomas Gleixner and Kate Stewart and Philippe Ombredanne. How this work was done: Patches were generated and checked against linux-4.14-rc6 for a subset of the use cases: - file had no licensing information it it. - file was a */uapi/* one with no licensing information in it, - file was a */uapi/* one with existing licensing information, Further patches will be generated in subsequent months to fix up cases where non-standard license headers were used, and references to license had to be inferred by heuristics based on keywords. The analysis to determine which SPDX License Identifier to be applied to a file was done in a spreadsheet of side by side results from of the output of two independent scanners (ScanCode & Windriver) producing SPDX tag:value files created by Philippe Ombredanne. Philippe prepared the base worksheet, and did an initial spot review of a few 1000 files. The 4.13 kernel was the starting point of the analysis with 60,537 files assessed. Kate Stewart did a file by file comparison of the scanner results in the spreadsheet to determine which SPDX license identifier(s) to be applied to the file. She confirmed any determination that was not immediately clear with lawyers working with the Linux Foundation. Criteria used to select files for SPDX license identifier tagging was: - Files considered eligible had to be source code files. - Make and config files were included as candidates if they contained >5 lines of source - File already had some variant of a license header in it (even if <5 lines). All documentation files were explicitly excluded. The following heuristics were used to determine which SPDX license identifiers to apply. - when both scanners couldn't find any license traces, file was considered to have no license information in it, and the top level COPYING file license applied. For non */uapi/* files that summary was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 11139 and resulted in the first patch in this series. If that file was a */uapi/* path one, it was "GPL-2.0 WITH Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 WITH Linux-syscall-note 930 and resulted in the second patch in this series. - if a file had some form of licensing information in it, and was one of the */uapi/* ones, it was denoted with the Linux-syscall-note if any GPL family license was found in the file or had no licensing in it (per prior point). Results summary: SPDX license identifier # files ---------------------------------------------------|------ GPL-2.0 WITH Linux-syscall-note 270 GPL-2.0+ WITH Linux-syscall-note 169 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17 LGPL-2.1+ WITH Linux-syscall-note 15 GPL-1.0+ WITH Linux-syscall-note 14 ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5 LGPL-2.0+ WITH Linux-syscall-note 4 LGPL-2.1 WITH Linux-syscall-note 3 ((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3 ((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1 and that resulted in the third patch in this series. - when the two scanners agreed on the detected license(s), that became the concluded license(s). - when there was disagreement between the two scanners (one detected a license but the other didn't, or they both detected different licenses) a manual inspection of the file occurred. - In most cases a manual inspection of the information in the file resulted in a clear resolution of the license that should apply (and which scanner probably needed to revisit its heuristics). - When it was not immediately clear, the license identifier was confirmed with lawyers working with the Linux Foundation. - If there was any question as to the appropriate license identifier, the file was flagged for further research and to be revisited later in time. In total, over 70 hours of logged manual review was done on the spreadsheet to determine the SPDX license identifiers to apply to the source files by Kate, Philippe, Thomas and, in some cases, confirmation by lawyers working with the Linux Foundation. Kate also obtained a third independent scan of the 4.13 code base from FOSSology, and compared selected files where the other two scanners disagreed against that SPDX file, to see if there was new insights. The Windriver scanner is based on an older version of FOSSology in part, so they are related. Thomas did random spot checks in about 500 files from the spreadsheets for the uapi headers and agreed with SPDX license identifier in the files he inspected. For the non-uapi files Thomas did random spot checks in about 15000 files. In initial set of patches against 4.14-rc6, 3 files were found to have copy/paste license identifier errors, and have been fixed to reflect the correct identifier. Additionally Philippe spent 10 hours this week doing a detailed manual inspection and review of the 12,461 patched files from the initial patch version early this week with: - a full scancode scan run, collecting the matched texts, detected license ids and scores - reviewing anything where there was a license detected (about 500+ files) to ensure that the applied SPDX license was correct - reviewing anything where there was no detection but the patch license was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied SPDX license was correct This produced a worksheet with 20 files needing minor correction. This worksheet was then exported into 3 different .csv files for the different types of files to be modified. These .csv files were then reviewed by Greg. Thomas wrote a script to parse the csv files and add the proper SPDX tag to the file, in the format that the file expected. This script was further refined by Greg based on the output to detect more types of files automatically and to distinguish between header and source .c files (which need different comment types.) Finally Greg ran the script using the .csv files to generate the patches. Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org> Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 14:07:57 +00:00
// SPDX-License-Identifier: GPL-2.0
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:14:15 +00:00
#include <linux/ceph/ceph_debug.h>
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:14:15 +00:00
#include <linux/module.h>
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
#include <linux/err.h>
#include <linux/highmem.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
#include <linux/slab.h>
#include <linux/uaccess.h>
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
#ifdef CONFIG_BLOCK
#include <linux/bio.h>
#endif
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
#include <linux/ceph/ceph_features.h>
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:14:15 +00:00
#include <linux/ceph/libceph.h>
#include <linux/ceph/osd_client.h>
#include <linux/ceph/messenger.h>
#include <linux/ceph/decode.h>
#include <linux/ceph/auth.h>
#include <linux/ceph/pagelist.h>
libceph, ceph: move ceph_calc_file_object_mapping() to striper.c ceph_calc_file_object_mapping() has nothing to do with osdmaps. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-02-17 09:41:20 +00:00
#include <linux/ceph/striper.h>
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
ceph: use single osd op reply msg Use a single ceph_msg for the osd reply, even when we are getting multiple replies. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-01 21:02:00 +00:00
#define OSD_OPREPLY_FRONT_LEN 512
ceph: keep reserved replies on the request structure This includes treating all the data preallocation and revokation at the same place, not having to have a special case for the reserved pages. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
2010-01-14 01:03:23 +00:00
libceph: use slab cache for osd client requests Create a slab cache to manage allocation of ceph_osdc_request structures. This resolves: http://tracker.ceph.com/issues/3926 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-05-01 17:43:04 +00:00
static struct kmem_cache *ceph_osd_request_cache;
ceph: Storage class should be before const qualifier The C99 specification states in section 6.11.5: The placement of a storage-class specifier other than at the beginning of the declaration specifiers in a declaration is an obsolescent feature. Signed-off-by: Tobias Klauser <tklauser@distanz.ch> Signed-off-by: Sage Weil <sage@newdream.net>
2010-05-20 08:40:19 +00:00
static const struct ceph_connection_operations osd_con_ops;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* Implement client access to distributed object storage cluster.
*
* All data objects are stored within a cluster/cloud of OSDs, or
* "object storage devices." (Note that Ceph OSDs have _nothing_ to
* do with the T10 OSD extensions to SCSI.) Ceph OSDs are simply
* remote daemons serving up and coordinating consistent and safe
* access to storage.
*
* Cluster membership and the mapping of data objects onto storage devices
* are described by the osd map.
*
* We keep track of pending OSD requests (read, write), resubmit
* requests to different OSDs when the cluster topology/data layout
* change, or retry the affected requests when the communications
* channel with an OSD is reset.
*/
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void link_request(struct ceph_osd *osd, struct ceph_osd_request *req);
static void unlink_request(struct ceph_osd *osd, struct ceph_osd_request *req);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
static void link_linger(struct ceph_osd *osd,
struct ceph_osd_linger_request *lreq);
static void unlink_linger(struct ceph_osd *osd,
struct ceph_osd_linger_request *lreq);
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
static void clear_backoffs(struct ceph_osd *osd);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
#if 1
static inline bool rwsem_is_wrlocked(struct rw_semaphore *sem)
{
bool wrlocked = true;
if (unlikely(down_read_trylock(sem))) {
wrlocked = false;
up_read(sem);
}
return wrlocked;
}
static inline void verify_osdc_locked(struct ceph_osd_client *osdc)
{
WARN_ON(!rwsem_is_locked(&osdc->lock));
}
static inline void verify_osdc_wrlocked(struct ceph_osd_client *osdc)
{
WARN_ON(!rwsem_is_wrlocked(&osdc->lock));
}
static inline void verify_osd_locked(struct ceph_osd *osd)
{
struct ceph_osd_client *osdc = osd->o_osdc;
WARN_ON(!(mutex_is_locked(&osd->lock) &&
rwsem_is_locked(&osdc->lock)) &&
!rwsem_is_wrlocked(&osdc->lock));
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
static inline void verify_lreq_locked(struct ceph_osd_linger_request *lreq)
{
WARN_ON(!mutex_is_locked(&lreq->lock));
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
#else
static inline void verify_osdc_locked(struct ceph_osd_client *osdc) { }
static inline void verify_osdc_wrlocked(struct ceph_osd_client *osdc) { }
static inline void verify_osd_locked(struct ceph_osd *osd) { }
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
static inline void verify_lreq_locked(struct ceph_osd_linger_request *lreq) { }
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
#endif
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* calculate the mapping of a file extent onto an object, and fill out the
* request accordingly. shorten extent as necessary if it crosses an
* object boundary.
*
* fill osd op in request message.
*/
libceph: format target object name in caller Move the formatting of the object name (oid) to use for an object request into the caller of calc_layout(). This makes the "vino" parameter no longer necessary, so get rid of it. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-16 04:10:17 +00:00
static int calc_layout(struct ceph_file_layout *layout, u64 off, u64 *plen,
libceph: don't update op in calc_layout() The ceph_osdc_new_request() an array of osd operations is built up and filled in partially within that function and partially in the called function calc_layout(). Move the latter part back out to ceph_osdc_new_request() so it's all done in one place. This makes it unnecessary to pass the op pointer to calc_layout(), so get rid of that parameter. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:01 +00:00
u64 *objnum, u64 *objoff, u64 *objlen)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: kill ceph_calc_raw_layout() There is no caller of ceph_calc_raw_layout() outside of libceph, so there's no need to export from the module. Furthermore, there is only one caller, in calc_layout(), and it is not much more than a simple wrapper for that function. So get rid of ceph_calc_raw_layout() and embed it instead within calc_layout(). While touching "osd_client.c", get rid of the unnecessary forward declaration of __send_request(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-15 17:42:29 +00:00
u64 orig_len = *plen;
libceph, ceph: change ceph_calc_file_object_mapping() signature - make it void - xlen (object extent length) out parameter should be u32 because only a single stripe unit is mapped at a time Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2018-02-17 08:29:58 +00:00
u32 xlen;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: kill ceph_calc_raw_layout() There is no caller of ceph_calc_raw_layout() outside of libceph, so there's no need to export from the module. Furthermore, there is only one caller, in calc_layout(), and it is not much more than a simple wrapper for that function. So get rid of ceph_calc_raw_layout() and embed it instead within calc_layout(). While touching "osd_client.c", get rid of the unnecessary forward declaration of __send_request(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-15 17:42:29 +00:00
/* object extent? */
libceph, ceph: change ceph_calc_file_object_mapping() signature - make it void - xlen (object extent length) out parameter should be u32 because only a single stripe unit is mapped at a time Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2018-02-17 08:29:58 +00:00
ceph_calc_file_object_mapping(layout, off, orig_len, objnum,
objoff, &xlen);
*objlen = xlen;
libceph: pass offset and length out of calc_layout() The purpose of calc_layout() is to determine, given a file offset and length and a layout describing the placement of file data across objects, where in "object space" that data resides. Specifically, it determines which object should hold the first part of the specified range of file data, and the offset and length of data within that object. The length will not exceed the bounds of the object, and the caller is informed of that maximum length. Add two parameters to calc_layout() to allow the object-relative offset and length to be passed back to the caller. This is the first steps toward having ceph_osdc_new_request() build its osd op structure using osd_req_op_extent_init(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
if (*objlen < orig_len) {
*plen = *objlen;
libceph: kill ceph_calc_raw_layout() There is no caller of ceph_calc_raw_layout() outside of libceph, so there's no need to export from the module. Furthermore, there is only one caller, in calc_layout(), and it is not much more than a simple wrapper for that function. So get rid of ceph_calc_raw_layout() and embed it instead within calc_layout(). While touching "osd_client.c", get rid of the unnecessary forward declaration of __send_request(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-15 17:42:29 +00:00
dout(" skipping last %llu, final file extent %llu~%llu\n",
orig_len - *plen, off, *plen);
}
libceph: pass offset and length out of calc_layout() The purpose of calc_layout() is to determine, given a file offset and length and a layout describing the placement of file data across objects, where in "object space" that data resides. Specifically, it determines which object should hold the first part of the specified range of file data, and the offset and length of data within that object. The length will not exceed the bounds of the object, and the caller is informed of that maximum length. Add two parameters to calc_layout() to allow the object-relative offset and length to be passed back to the caller. This is the first steps toward having ceph_osdc_new_request() build its osd op structure using osd_req_op_extent_init(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
dout("calc_layout objnum=%llx %llu~%llu\n", *objnum, *objoff, *objlen);
libceph: fix a osd request memory leak If an invalid layout is provided to ceph_osdc_new_request(), its call to calc_layout() might return an error. At that point in the function we've already allocated an osd request structure, so we need to free it (drop a reference) in the event such an error occurs. The only other value calc_layout() will return is 0, so make that explicit in the successful case. This resolves: http://tracker.ceph.com/issues/4240 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-16 04:10:17 +00:00
return 0;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: define a few more helpers Define ceph_osd_data_init() and ceph_osd_data_release() to clean up a little code. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
static void ceph_osd_data_init(struct ceph_osd_data *osd_data)
{
memset(osd_data, 0, sizeof (*osd_data));
osd_data->type = CEPH_OSD_DATA_TYPE_NONE;
}
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
static void ceph_osd_data_pages_init(struct ceph_osd_data *osd_data,
libceph: define osd data initialization helpers Define and use functions that encapsulate the initializion of a ceph_osd_data structure. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
struct page **pages, u64 length, u32 alignment,
bool pages_from_pool, bool own_pages)
{
osd_data->type = CEPH_OSD_DATA_TYPE_PAGES;
osd_data->pages = pages;
osd_data->length = length;
osd_data->alignment = alignment;
osd_data->pages_from_pool = pages_from_pool;
osd_data->own_pages = own_pages;
}
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
static void ceph_osd_data_pagelist_init(struct ceph_osd_data *osd_data,
libceph: define osd data initialization helpers Define and use functions that encapsulate the initializion of a ceph_osd_data structure. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
struct ceph_pagelist *pagelist)
{
osd_data->type = CEPH_OSD_DATA_TYPE_PAGELIST;
osd_data->pagelist = pagelist;
}
#ifdef CONFIG_BLOCK
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
static void ceph_osd_data_bio_init(struct ceph_osd_data *osd_data,
libceph, rbd: new bio handling code (aka don't clone bios) The reason we clone bios is to be able to give each object request (and consequently each ceph_osd_data/ceph_msg_data item) its own pointer to a (list of) bio(s). The messenger then initializes its cursor with cloned bio's ->bi_iter, so it knows where to start reading from/writing to. That's all the cloned bios are used for: to determine each object request's starting position in the provided data buffer. Introduce ceph_bio_iter to do exactly that -- store position within bio list (i.e. pointer to bio) + position within that bio (i.e. bvec_iter). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:10 +00:00
struct ceph_bio_iter *bio_pos,
u32 bio_length)
libceph: define osd data initialization helpers Define and use functions that encapsulate the initializion of a ceph_osd_data structure. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
{
osd_data->type = CEPH_OSD_DATA_TYPE_BIO;
libceph, rbd: new bio handling code (aka don't clone bios) The reason we clone bios is to be able to give each object request (and consequently each ceph_osd_data/ceph_msg_data item) its own pointer to a (list of) bio(s). The messenger then initializes its cursor with cloned bio's ->bi_iter, so it knows where to start reading from/writing to. That's all the cloned bios are used for: to determine each object request's starting position in the provided data buffer. Introduce ceph_bio_iter to do exactly that -- store position within bio list (i.e. pointer to bio) + position within that bio (i.e. bvec_iter). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:10 +00:00
osd_data->bio_pos = *bio_pos;
libceph: define osd data initialization helpers Define and use functions that encapsulate the initializion of a ceph_osd_data structure. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
osd_data->bio_length = bio_length;
}
#endif /* CONFIG_BLOCK */
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
static void ceph_osd_data_bvecs_init(struct ceph_osd_data *osd_data,
libceph: add osd_req_op_extent_osd_data_bvecs() ... and store num_bvecs for client code's convenience. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-04 14:57:30 +00:00
struct ceph_bvec_iter *bvec_pos,
u32 num_bvecs)
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
{
osd_data->type = CEPH_OSD_DATA_TYPE_BVECS;
osd_data->bvec_pos = *bvec_pos;
libceph: add osd_req_op_extent_osd_data_bvecs() ... and store num_bvecs for client code's convenience. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-04 14:57:30 +00:00
osd_data->num_bvecs = num_bvecs;
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
}
libceph: evaluate osd_req_op_data() arguments only once This patch changes the osd_req_op_data() macro to not evaluate arguments more than once in order to follow the kernel coding style. Signed-off-by: Ioana Ciornei <ciorneiioana@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org> [idryomov@gmail.com: changelog, formatting] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2015-10-22 15:06:07 +00:00
#define osd_req_op_data(oreq, whch, typ, fld) \
({ \
struct ceph_osd_request *__oreq = (oreq); \
unsigned int __whch = (whch); \
BUG_ON(__whch >= __oreq->r_num_ops); \
&__oreq->r_ops[__whch].typ.fld; \
})
libceph: clean up osd data field access functions There are a bunch of functions defined to encapsulate getting the address of a data field for a particular op in an osd request. They're all defined the same way, so create a macro to take the place of all of them. Two of these are used outside the osd client code, so preserve them (but convert them to use the new macro internally). Stop exporting the ones that aren't used elsewhere. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
libceph: support raw data requests Allow osd request ops that aren't otherwise structured (not class, extent, or watch ops) to specify "raw" data to be used to hold incoming data for the op. Make use of this capability for the osd STAT op. Prefix the name of the private function osd_req_op_init() with "_", and expose a new function by that (earlier) name whose purpose is to initialize osd ops with (only) implied data. For now we'll just support the use of a page array for an osd op with incoming raw data. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-11 18:33:24 +00:00
static struct ceph_osd_data *
osd_req_op_raw_data_in(struct ceph_osd_request *osd_req, unsigned int which)
{
BUG_ON(which >= osd_req->r_num_ops);
return &osd_req->r_ops[which].raw_data_in;
}
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
struct ceph_osd_data *
osd_req_op_extent_osd_data(struct ceph_osd_request *osd_req,
libceph: kill off osd data write_request parameters In the incremental move toward supporting distinct data items in an osd request some of the functions had "write_request" parameters to indicate, basically, whether the data belonged to in_data or the out_data. Now that we maintain the data fields in the op structure there is no need to indicate the direction, so get rid of the "write_request" parameters. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
unsigned int which)
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
{
libceph: clean up osd data field access functions There are a bunch of functions defined to encapsulate getting the address of a data field for a particular op in an osd request. They're all defined the same way, so create a macro to take the place of all of them. Two of these are used outside the osd client code, so preserve them (but convert them to use the new macro internally). Stop exporting the ones that aren't used elsewhere. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
return osd_req_op_data(osd_req, which, extent, osd_data);
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
}
EXPORT_SYMBOL(osd_req_op_extent_osd_data);
libceph: support raw data requests Allow osd request ops that aren't otherwise structured (not class, extent, or watch ops) to specify "raw" data to be used to hold incoming data for the op. Make use of this capability for the osd STAT op. Prefix the name of the private function osd_req_op_init() with "_", and expose a new function by that (earlier) name whose purpose is to initialize osd ops with (only) implied data. For now we'll just support the use of a page array for an osd op with incoming raw data. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-11 18:33:24 +00:00
void osd_req_op_raw_data_in_pages(struct ceph_osd_request *osd_req,
unsigned int which, struct page **pages,
u64 length, u32 alignment,
bool pages_from_pool, bool own_pages)
{
struct ceph_osd_data *osd_data;
osd_data = osd_req_op_raw_data_in(osd_req, which);
ceph_osd_data_pages_init(osd_data, pages, length, alignment,
pages_from_pool, own_pages);
}
EXPORT_SYMBOL(osd_req_op_raw_data_in_pages);
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
void osd_req_op_extent_osd_data_pages(struct ceph_osd_request *osd_req,
libceph: kill off osd data write_request parameters In the incremental move toward supporting distinct data items in an osd request some of the functions had "write_request" parameters to indicate, basically, whether the data belonged to in_data or the out_data. Now that we maintain the data fields in the op structure there is no need to indicate the direction, so get rid of the "write_request" parameters. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
unsigned int which, struct page **pages,
u64 length, u32 alignment,
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
bool pages_from_pool, bool own_pages)
{
struct ceph_osd_data *osd_data;
libceph: clean up osd data field access functions There are a bunch of functions defined to encapsulate getting the address of a data field for a particular op in an osd request. They're all defined the same way, so create a macro to take the place of all of them. Two of these are used outside the osd client code, so preserve them (but convert them to use the new macro internally). Stop exporting the ones that aren't used elsewhere. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
osd_data = osd_req_op_data(osd_req, which, extent, osd_data);
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_osd_data_pages_init(osd_data, pages, length, alignment,
pages_from_pool, own_pages);
}
EXPORT_SYMBOL(osd_req_op_extent_osd_data_pages);
void osd_req_op_extent_osd_data_pagelist(struct ceph_osd_request *osd_req,
libceph: kill off osd data write_request parameters In the incremental move toward supporting distinct data items in an osd request some of the functions had "write_request" parameters to indicate, basically, whether the data belonged to in_data or the out_data. Now that we maintain the data fields in the op structure there is no need to indicate the direction, so get rid of the "write_request" parameters. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
unsigned int which, struct ceph_pagelist *pagelist)
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
{
struct ceph_osd_data *osd_data;
libceph: clean up osd data field access functions There are a bunch of functions defined to encapsulate getting the address of a data field for a particular op in an osd request. They're all defined the same way, so create a macro to take the place of all of them. Two of these are used outside the osd client code, so preserve them (but convert them to use the new macro internally). Stop exporting the ones that aren't used elsewhere. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
osd_data = osd_req_op_data(osd_req, which, extent, osd_data);
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_osd_data_pagelist_init(osd_data, pagelist);
}
EXPORT_SYMBOL(osd_req_op_extent_osd_data_pagelist);
#ifdef CONFIG_BLOCK
void osd_req_op_extent_osd_data_bio(struct ceph_osd_request *osd_req,
libceph, rbd: new bio handling code (aka don't clone bios) The reason we clone bios is to be able to give each object request (and consequently each ceph_osd_data/ceph_msg_data item) its own pointer to a (list of) bio(s). The messenger then initializes its cursor with cloned bio's ->bi_iter, so it knows where to start reading from/writing to. That's all the cloned bios are used for: to determine each object request's starting position in the provided data buffer. Introduce ceph_bio_iter to do exactly that -- store position within bio list (i.e. pointer to bio) + position within that bio (i.e. bvec_iter). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:10 +00:00
unsigned int which,
struct ceph_bio_iter *bio_pos,
u32 bio_length)
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
{
struct ceph_osd_data *osd_data;
libceph: clean up osd data field access functions There are a bunch of functions defined to encapsulate getting the address of a data field for a particular op in an osd request. They're all defined the same way, so create a macro to take the place of all of them. Two of these are used outside the osd client code, so preserve them (but convert them to use the new macro internally). Stop exporting the ones that aren't used elsewhere. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
osd_data = osd_req_op_data(osd_req, which, extent, osd_data);
libceph, rbd: new bio handling code (aka don't clone bios) The reason we clone bios is to be able to give each object request (and consequently each ceph_osd_data/ceph_msg_data item) its own pointer to a (list of) bio(s). The messenger then initializes its cursor with cloned bio's ->bi_iter, so it knows where to start reading from/writing to. That's all the cloned bios are used for: to determine each object request's starting position in the provided data buffer. Introduce ceph_bio_iter to do exactly that -- store position within bio list (i.e. pointer to bio) + position within that bio (i.e. bvec_iter). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:10 +00:00
ceph_osd_data_bio_init(osd_data, bio_pos, bio_length);
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
}
EXPORT_SYMBOL(osd_req_op_extent_osd_data_bio);
#endif /* CONFIG_BLOCK */
libceph: add osd_req_op_extent_osd_data_bvecs() ... and store num_bvecs for client code's convenience. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-04 14:57:30 +00:00
void osd_req_op_extent_osd_data_bvecs(struct ceph_osd_request *osd_req,
unsigned int which,
struct bio_vec *bvecs, u32 num_bvecs,
u32 bytes)
{
struct ceph_osd_data *osd_data;
struct ceph_bvec_iter it = {
.bvecs = bvecs,
.iter = { .bi_size = bytes },
};
osd_data = osd_req_op_data(osd_req, which, extent, osd_data);
ceph_osd_data_bvecs_init(osd_data, &it, num_bvecs);
}
EXPORT_SYMBOL(osd_req_op_extent_osd_data_bvecs);
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
void osd_req_op_extent_osd_data_bvec_pos(struct ceph_osd_request *osd_req,
unsigned int which,
struct ceph_bvec_iter *bvec_pos)
{
struct ceph_osd_data *osd_data;
osd_data = osd_req_op_data(osd_req, which, extent, osd_data);
libceph: add osd_req_op_extent_osd_data_bvecs() ... and store num_bvecs for client code's convenience. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-04 14:57:30 +00:00
ceph_osd_data_bvecs_init(osd_data, bvec_pos, 0);
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
}
EXPORT_SYMBOL(osd_req_op_extent_osd_data_bvec_pos);
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
static void osd_req_op_cls_request_info_pagelist(
struct ceph_osd_request *osd_req,
unsigned int which, struct ceph_pagelist *pagelist)
{
struct ceph_osd_data *osd_data;
libceph: clean up osd data field access functions There are a bunch of functions defined to encapsulate getting the address of a data field for a particular op in an osd request. They're all defined the same way, so create a macro to take the place of all of them. Two of these are used outside the osd client code, so preserve them (but convert them to use the new macro internally). Stop exporting the ones that aren't used elsewhere. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
osd_data = osd_req_op_data(osd_req, which, cls, request_info);
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_osd_data_pagelist_init(osd_data, pagelist);
}
libceph: make method call data be a separate data item Right now the data for a method call is specified via a pointer and length, and it's copied--along with the class and method name--into a pagelist data item to be sent to the osd. Instead, encode the data in a data item separate from the class and method names. This will allow large amounts of data to be supplied to methods without copying. Only rbd uses the class functionality right now, and when it really needs this it will probably need to use a page array rather than a page list. But this simple implementation demonstrates the functionality on the osd client, and that's enough for now. This resolves: http://tracker.ceph.com/issues/4104 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 19:46:02 +00:00
void osd_req_op_cls_request_data_pagelist(
struct ceph_osd_request *osd_req,
unsigned int which, struct ceph_pagelist *pagelist)
{
struct ceph_osd_data *osd_data;
libceph: clean up osd data field access functions There are a bunch of functions defined to encapsulate getting the address of a data field for a particular op in an osd request. They're all defined the same way, so create a macro to take the place of all of them. Two of these are used outside the osd client code, so preserve them (but convert them to use the new macro internally). Stop exporting the ones that aren't used elsewhere. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
osd_data = osd_req_op_data(osd_req, which, cls, request_data);
libceph: make method call data be a separate data item Right now the data for a method call is specified via a pointer and length, and it's copied--along with the class and method name--into a pagelist data item to be sent to the osd. Instead, encode the data in a data item separate from the class and method names. This will allow large amounts of data to be supplied to methods without copying. Only rbd uses the class functionality right now, and when it really needs this it will probably need to use a page array rather than a page list. But this simple implementation demonstrates the functionality on the osd client, and that's enough for now. This resolves: http://tracker.ceph.com/issues/4104 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 19:46:02 +00:00
ceph_osd_data_pagelist_init(osd_data, pagelist);
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
osd_req->r_ops[which].cls.indata_len += pagelist->length;
osd_req->r_ops[which].indata_len += pagelist->length;
libceph: make method call data be a separate data item Right now the data for a method call is specified via a pointer and length, and it's copied--along with the class and method name--into a pagelist data item to be sent to the osd. Instead, encode the data in a data item separate from the class and method names. This will allow large amounts of data to be supplied to methods without copying. Only rbd uses the class functionality right now, and when it really needs this it will probably need to use a page array rather than a page list. But this simple implementation demonstrates the functionality on the osd client, and that's enough for now. This resolves: http://tracker.ceph.com/issues/4104 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 19:46:02 +00:00
}
EXPORT_SYMBOL(osd_req_op_cls_request_data_pagelist);
libceph: support pages for class request data Add the ability to provide an array of pages as outbound request data for object class method calls. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-19 20:34:49 +00:00
void osd_req_op_cls_request_data_pages(struct ceph_osd_request *osd_req,
unsigned int which, struct page **pages, u64 length,
u32 alignment, bool pages_from_pool, bool own_pages)
{
struct ceph_osd_data *osd_data;
osd_data = osd_req_op_data(osd_req, which, cls, request_data);
ceph_osd_data_pages_init(osd_data, pages, length, alignment,
pages_from_pool, own_pages);
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
osd_req->r_ops[which].cls.indata_len += length;
osd_req->r_ops[which].indata_len += length;
libceph: support pages for class request data Add the ability to provide an array of pages as outbound request data for object class method calls. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-19 20:34:49 +00:00
}
EXPORT_SYMBOL(osd_req_op_cls_request_data_pages);
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
void osd_req_op_cls_request_data_bvecs(struct ceph_osd_request *osd_req,
unsigned int which,
libceph: add osd_req_op_extent_osd_data_bvecs() ... and store num_bvecs for client code's convenience. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-04 14:57:30 +00:00
struct bio_vec *bvecs, u32 num_bvecs,
u32 bytes)
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
{
struct ceph_osd_data *osd_data;
struct ceph_bvec_iter it = {
.bvecs = bvecs,
.iter = { .bi_size = bytes },
};
osd_data = osd_req_op_data(osd_req, which, cls, request_data);
libceph: add osd_req_op_extent_osd_data_bvecs() ... and store num_bvecs for client code's convenience. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-04 14:57:30 +00:00
ceph_osd_data_bvecs_init(osd_data, &it, num_bvecs);
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
osd_req->r_ops[which].cls.indata_len += bytes;
osd_req->r_ops[which].indata_len += bytes;
}
EXPORT_SYMBOL(osd_req_op_cls_request_data_bvecs);
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
void osd_req_op_cls_response_data_pages(struct ceph_osd_request *osd_req,
unsigned int which, struct page **pages, u64 length,
u32 alignment, bool pages_from_pool, bool own_pages)
{
struct ceph_osd_data *osd_data;
libceph: clean up osd data field access functions There are a bunch of functions defined to encapsulate getting the address of a data field for a particular op in an osd request. They're all defined the same way, so create a macro to take the place of all of them. Two of these are used outside the osd client code, so preserve them (but convert them to use the new macro internally). Stop exporting the ones that aren't used elsewhere. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
osd_data = osd_req_op_data(osd_req, which, cls, response_data);
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_osd_data_pages_init(osd_data, pages, length, alignment,
pages_from_pool, own_pages);
}
EXPORT_SYMBOL(osd_req_op_cls_response_data_pages);
libceph: define ceph_osd_data_length() One more osd data helper, which returns the length of the data item, regardless of its type. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:58 +00:00
static u64 ceph_osd_data_length(struct ceph_osd_data *osd_data)
{
switch (osd_data->type) {
case CEPH_OSD_DATA_TYPE_NONE:
return 0;
case CEPH_OSD_DATA_TYPE_PAGES:
return osd_data->length;
case CEPH_OSD_DATA_TYPE_PAGELIST:
return (u64)osd_data->pagelist->length;
#ifdef CONFIG_BLOCK
case CEPH_OSD_DATA_TYPE_BIO:
return (u64)osd_data->bio_length;
#endif /* CONFIG_BLOCK */
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
case CEPH_OSD_DATA_TYPE_BVECS:
return osd_data->bvec_pos.iter.bi_size;
libceph: define ceph_osd_data_length() One more osd data helper, which returns the length of the data item, regardless of its type. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:58 +00:00
default:
WARN(true, "unrecognized data type %d\n", (int)osd_data->type);
return 0;
}
}
libceph: define a few more helpers Define ceph_osd_data_init() and ceph_osd_data_release() to clean up a little code. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
static void ceph_osd_data_release(struct ceph_osd_data *osd_data)
{
libceph: kill off osd request r_data_in and r_data_out Finally! Convert the osd op data pointers into real structures, and make the switch over to using them instead of having all ops share the in and/or out data structures in the osd request. Set up a new function to traverse the set of ops and release any data associated with them (pages). This and the patches leading up to it resolve: http://tracker.ceph.com/issues/4657 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
if (osd_data->type == CEPH_OSD_DATA_TYPE_PAGES && osd_data->own_pages) {
libceph: define a few more helpers Define ceph_osd_data_init() and ceph_osd_data_release() to clean up a little code. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
int num_pages;
num_pages = calc_pages_for((u64)osd_data->alignment,
(u64)osd_data->length);
ceph_release_page_vector(osd_data->pages, num_pages);
}
libceph: kill off osd request r_data_in and r_data_out Finally! Convert the osd op data pointers into real structures, and make the switch over to using them instead of having all ops share the in and/or out data structures in the osd request. Set up a new function to traverse the set of ops and release any data associated with them (pages). This and the patches leading up to it resolve: http://tracker.ceph.com/issues/4657 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_osd_data_init(osd_data);
}
static void osd_req_op_data_release(struct ceph_osd_request *osd_req,
unsigned int which)
{
struct ceph_osd_req_op *op;
BUG_ON(which >= osd_req->r_num_ops);
op = &osd_req->r_ops[which];
switch (op->op) {
case CEPH_OSD_OP_READ:
case CEPH_OSD_OP_WRITE:
rbd: use writefull op for object size writes This covers only the simplest case - an object size sized write, but it's still useful in tiering setups when EC is used for the base tier as writefull op can be proxied, saving an object promotion. Even though updating ceph_osdc_new_request() to allow writefull should just be a matter of fixing an assert, I didn't do it because its only user is cephfs. All other sites were updated. Reflects ceph.git commit 7bfb7f9025a8ee0d2305f49bf0336d2424da5b5b. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-10-07 15:27:17 +00:00
case CEPH_OSD_OP_WRITEFULL:
libceph: kill off osd request r_data_in and r_data_out Finally! Convert the osd op data pointers into real structures, and make the switch over to using them instead of having all ops share the in and/or out data structures in the osd request. Set up a new function to traverse the set of ops and release any data associated with them (pages). This and the patches leading up to it resolve: http://tracker.ceph.com/issues/4657 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_osd_data_release(&op->extent.osd_data);
break;
case CEPH_OSD_OP_CALL:
ceph_osd_data_release(&op->cls.request_info);
libceph: make method call data be a separate data item Right now the data for a method call is specified via a pointer and length, and it's copied--along with the class and method name--into a pagelist data item to be sent to the osd. Instead, encode the data in a data item separate from the class and method names. This will allow large amounts of data to be supplied to methods without copying. Only rbd uses the class functionality right now, and when it really needs this it will probably need to use a page array rather than a page list. But this simple implementation demonstrates the functionality on the osd client, and that's enough for now. This resolves: http://tracker.ceph.com/issues/4104 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 19:46:02 +00:00
ceph_osd_data_release(&op->cls.request_data);
libceph: kill off osd request r_data_in and r_data_out Finally! Convert the osd op data pointers into real structures, and make the switch over to using them instead of having all ops share the in and/or out data structures in the osd request. Set up a new function to traverse the set of ops and release any data associated with them (pages). This and the patches leading up to it resolve: http://tracker.ceph.com/issues/4657 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_osd_data_release(&op->cls.response_data);
break;
libceph: add SETXATTR/CMPXATTR osd operations support Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-12 06:00:43 +00:00
case CEPH_OSD_OP_SETXATTR:
case CEPH_OSD_OP_CMPXATTR:
ceph_osd_data_release(&op->xattr.osd_data);
break;
libceph: properly release STAT request's raw_data_in Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:02:35 +00:00
case CEPH_OSD_OP_STAT:
ceph_osd_data_release(&op->raw_data_in);
break;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
case CEPH_OSD_OP_NOTIFY_ACK:
ceph_osd_data_release(&op->notify_ack.request_data);
break;
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
case CEPH_OSD_OP_NOTIFY:
ceph_osd_data_release(&op->notify.request_data);
ceph_osd_data_release(&op->notify.response_data);
break;
libceph: support for CEPH_OSD_OP_LIST_WATCHERS Add support for this Ceph OSD op, needed to support the RBD exclusive lock feature. Signed-off-by: Douglas Fuller <dfuller@redhat.com> [idryomov@gmail.com: refactor, misc fixes throughout] Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-07-17 20:18:07 +00:00
case CEPH_OSD_OP_LIST_WATCHERS:
ceph_osd_data_release(&op->list_watchers.response_data);
break;
libceph: kill off osd request r_data_in and r_data_out Finally! Convert the osd op data pointers into real structures, and make the switch over to using them instead of having all ops share the in and/or out data structures in the osd request. Set up a new function to traverse the set of ops and release any data associated with them (pages). This and the patches leading up to it resolve: http://tracker.ceph.com/issues/4657 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
default:
break;
}
libceph: define a few more helpers Define ceph_osd_data_init() and ceph_osd_data_release() to clean up a little code. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
}
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
/*
* Assumes @t is zero-initialized.
*/
static void target_init(struct ceph_osd_request_target *t)
{
ceph_oid_init(&t->base_oid);
ceph_oloc_init(&t->base_oloc);
ceph_oid_init(&t->target_oid);
ceph_oloc_init(&t->target_oloc);
ceph_osds_init(&t->acting);
ceph_osds_init(&t->up);
t->size = -1;
t->min_size = -1;
t->osd = CEPH_HOMELESS_OSD;
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
static void target_copy(struct ceph_osd_request_target *dest,
const struct ceph_osd_request_target *src)
{
ceph_oid_copy(&dest->base_oid, &src->base_oid);
ceph_oloc_copy(&dest->base_oloc, &src->base_oloc);
ceph_oid_copy(&dest->target_oid, &src->target_oid);
ceph_oloc_copy(&dest->target_oloc, &src->target_oloc);
dest->pgid = src->pgid; /* struct */
libceph: introduce ceph_spg, ceph_pg_to_primary_shard() Store both raw pgid and actual spgid in ceph_osd_request_target. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:53 +00:00
dest->spgid = src->spgid; /* struct */
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
dest->pg_num = src->pg_num;
dest->pg_num_mask = src->pg_num_mask;
ceph_osds_copy(&dest->acting, &src->acting);
ceph_osds_copy(&dest->up, &src->up);
dest->size = src->size;
dest->min_size = src->min_size;
dest->sort_bitwise = src->sort_bitwise;
dest->flags = src->flags;
dest->paused = src->paused;
libceph: make sure need_resend targets reflect latest map Otherwise we may miss events like PG splits, pool deletions, etc when we get multiple incremental maps at once. Because check_pool_dne() can now be fed an unlinked request, finish_request() needed to be taught to handle unlinked requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
dest->epoch = src->epoch;
libceph: fold [l]req->last_force_resend into ceph_osd_request_target Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-05 12:45:00 +00:00
dest->last_force_resend = src->last_force_resend;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
dest->osd = src->osd;
}
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
static void target_destroy(struct ceph_osd_request_target *t)
{
ceph_oid_destroy(&t->base_oid);
libceph: rados pool namespace support Add pool namesapce pointer to struct ceph_file_layout and struct ceph_object_locator. Pool namespace is used by when mapping object to PG, it's also used when composing OSD request. The namespace pointer in struct ceph_file_layout is RCU protected. So libceph can read namespace without taking lock. Signed-off-by: Yan, Zheng <zyan@redhat.com> [idryomov@gmail.com: ceph_oloc_destroy(), misc minor changes] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-14 03:24:31 +00:00
ceph_oloc_destroy(&t->base_oloc);
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
ceph_oid_destroy(&t->target_oid);
libceph: rados pool namespace support Add pool namesapce pointer to struct ceph_file_layout and struct ceph_object_locator. Pool namespace is used by when mapping object to PG, it's also used when composing OSD request. The namespace pointer in struct ceph_file_layout is RCU protected. So libceph can read namespace without taking lock. Signed-off-by: Yan, Zheng <zyan@redhat.com> [idryomov@gmail.com: ceph_oloc_destroy(), misc minor changes] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-14 03:24:31 +00:00
ceph_oloc_destroy(&t->target_oloc);
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* requests
*/
libceph: request_init() and request_release_checks() These are going to be used by request_reinit() code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void request_release_checks(struct ceph_osd_request *req)
{
WARN_ON(!RB_EMPTY_NODE(&req->r_node));
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
WARN_ON(!RB_EMPTY_NODE(&req->r_mc_node));
libceph: request_init() and request_release_checks() These are going to be used by request_reinit() code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
WARN_ON(!list_empty(&req->r_unsafe_item));
WARN_ON(req->r_osd);
}
libceph: move and add dout()s to ceph_osdc_request_{get,put}() Add dout()s to ceph_osdc_request_{get,put}(). Also move them to .c and turn kref release callback into a static function. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:42 +00:00
static void ceph_osdc_release_request(struct kref *kref)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: move and add dout()s to ceph_osdc_request_{get,put}() Add dout()s to ceph_osdc_request_{get,put}(). Also move them to .c and turn kref release callback into a static function. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:42 +00:00
struct ceph_osd_request *req = container_of(kref,
struct ceph_osd_request, r_kref);
libceph: kill off osd request r_data_in and r_data_out Finally! Convert the osd op data pointers into real structures, and make the switch over to using them instead of having all ops share the in and/or out data structures in the osd request. Set up a new function to traverse the set of ops and release any data associated with them (pages). This and the patches leading up to it resolve: http://tracker.ceph.com/issues/4657 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
unsigned int which;
ceph: use kref for ceph_osd_request Signed-off-by: Sage Weil <sage@newdream.net>
2009-12-07 21:37:03 +00:00
libceph: move and add dout()s to ceph_osdc_request_{get,put}() Add dout()s to ceph_osdc_request_{get,put}(). Also move them to .c and turn kref release callback into a static function. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:42 +00:00
dout("%s %p (r_request %p r_reply %p)\n", __func__, req,
req->r_request, req->r_reply);
libceph: request_init() and request_release_checks() These are going to be used by request_reinit() code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
request_release_checks(req);
libceph: move and add dout()s to ceph_osdc_request_{get,put}() Add dout()s to ceph_osdc_request_{get,put}(). Also move them to .c and turn kref release callback into a static function. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:42 +00:00
ceph: use kref for ceph_osd_request Signed-off-by: Sage Weil <sage@newdream.net>
2009-12-07 21:37:03 +00:00
if (req->r_request)
ceph_msg_put(req->r_request);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
if (req->r_reply)
libceph: osd_client: don't drop reply reference too early In ceph_osdc_release_request(), a reference to the r_reply message is dropped. But just after that, that same message is revoked if it was in use to receive an incoming reply. Reorder these so we are sure we hold a reference until we're actually done with the message. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-06-04 19:43:32 +00:00
ceph_msg_put(req->r_reply);
libceph: separate read and write data An osd request defines information about where data to be read should be placed as well as where data to write comes from. Currently these are represented by common fields. Keep information about data for writing separate from data to be read by splitting these into data_in and data_out fields. This is the key patch in this whole series, in that it actually identifies which osd requests generate outgoing data and which generate incoming data. It's less obvious (currently) that an osd CALL op generates both outgoing and incoming data; that's the focus of some upcoming work. This resolves: http://tracker.ceph.com/issues/4127 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-14 18:16:43 +00:00
libceph: kill off osd request r_data_in and r_data_out Finally! Convert the osd op data pointers into real structures, and make the switch over to using them instead of having all ops share the in and/or out data structures in the osd request. Set up a new function to traverse the set of ops and release any data associated with them (pages). This and the patches leading up to it resolve: http://tracker.ceph.com/issues/4657 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
for (which = 0; which < req->r_num_ops; which++)
osd_req_op_data_release(req, which);
libceph: separate read and write data An osd request defines information about where data to be read should be placed as well as where data to write comes from. Currently these are represented by common fields. Keep information about data for writing separate from data to be read by splitting these into data_in and data_out fields. This is the key patch in this whole series, in that it actually identifies which osd requests generate outgoing data and which generate incoming data. It's less obvious (currently) that an osd CALL op generates both outgoing and incoming data; that's the focus of some upcoming work. This resolves: http://tracker.ceph.com/issues/4127 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-14 18:16:43 +00:00
libceph: switch to calc_target(), part 1 Replace __calc_request_pg() and most of __map_request() with calc_target() and start using req->r_t. ceph_osdc_build_request() however still encodes base_oid, because it's called before calc_target() is and target_oid is empty at that point in time; a printf in osdc_show() also shows base_oid. This is fixed in "libceph: switch to calc_target(), part 2". Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
target_destroy(&req->r_t);
ceph: use kref for ceph_osd_request Signed-off-by: Sage Weil <sage@newdream.net>
2009-12-07 21:37:03 +00:00
ceph_put_snap_context(req->r_snapc);
libceph: variable-sized ceph_object_id Currently ceph_object_id can hold object names of up to 100 (CEPH_MAX_OID_NAME_LEN) characters. This is enough for all use cases, expect one - long rbd image names: - a format 1 header is named "<imgname>.rbd" - an object that points to a format 2 header is named "rbd_id.<imgname>" We operate on these potentially long-named objects during rbd map, and, for format 1 images, during header refresh. (A format 2 header name is a small system-generated string.) Lift this 100 character limit by making ceph_object_id be able to point to an externally-allocated string. Apart from being able to work with almost arbitrarily-long named objects, this allows us to reduce the size of ceph_object_id from >100 bytes to 64 bytes. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-29 17:54:20 +00:00
ceph: use kref for ceph_osd_request Signed-off-by: Sage Weil <sage@newdream.net>
2009-12-07 21:37:03 +00:00
if (req->r_mempool)
mempool_free(req, req->r_osdc->req_mempool);
libceph: enable large, variable-sized OSD requests Turn r_ops into a flexible array member to enable large, consisting of up to 16 ops, OSD requests. The use case is scattered writeback in cephfs and, as far as the kernel client is concerned, 16 is just a made up number. r_ops had size 3 for copyup+hint+write, but copyup is really a special case - it can only happen once. ceph_osd_request_cache is therefore stuffed with num_ops=2 requests, anything bigger than that is allocated with kmalloc(). req_mempool is backed by ceph_osd_request_cache, which means either num_ops=1 or num_ops=2 for use_mempool=true - all existing users (ceph_writepages_start(), ceph_osdc_writepages()) are fine with that. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:50:15 +00:00
else if (req->r_num_ops <= CEPH_OSD_SLAB_OPS)
libceph: use slab cache for osd client requests Create a slab cache to manage allocation of ceph_osdc_request structures. This resolves: http://tracker.ceph.com/issues/3926 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-05-01 17:43:04 +00:00
kmem_cache_free(ceph_osd_request_cache, req);
libceph: enable large, variable-sized OSD requests Turn r_ops into a flexible array member to enable large, consisting of up to 16 ops, OSD requests. The use case is scattered writeback in cephfs and, as far as the kernel client is concerned, 16 is just a made up number. r_ops had size 3 for copyup+hint+write, but copyup is really a special case - it can only happen once. ceph_osd_request_cache is therefore stuffed with num_ops=2 requests, anything bigger than that is allocated with kmalloc(). req_mempool is backed by ceph_osd_request_cache, which means either num_ops=1 or num_ops=2 for use_mempool=true - all existing users (ceph_writepages_start(), ceph_osdc_writepages()) are fine with that. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:50:15 +00:00
else
kfree(req);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: move and add dout()s to ceph_osdc_request_{get,put}() Add dout()s to ceph_osdc_request_{get,put}(). Also move them to .c and turn kref release callback into a static function. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:42 +00:00
void ceph_osdc_get_request(struct ceph_osd_request *req)
{
dout("%s %p (was %d)\n", __func__, req,
locking/atomic, kref: Add kref_read() Since we need to change the implementation, stop exposing internals. Provide kref_read() to read the current reference count; typically used for debug messages. Kills two anti-patterns: atomic_read(&kref->refcount) kref->refcount.counter Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-11-14 16:29:48 +00:00
kref_read(&req->r_kref));
libceph: move and add dout()s to ceph_osdc_request_{get,put}() Add dout()s to ceph_osdc_request_{get,put}(). Also move them to .c and turn kref release callback into a static function. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:42 +00:00
kref_get(&req->r_kref);
}
EXPORT_SYMBOL(ceph_osdc_get_request);
void ceph_osdc_put_request(struct ceph_osd_request *req)
{
libceph: make ceph_osdc_put_request() accept NULL Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-26 13:05:29 +00:00
if (req) {
dout("%s %p (was %d)\n", __func__, req,
locking/atomic, kref: Add kref_read() Since we need to change the implementation, stop exposing internals. Provide kref_read() to read the current reference count; typically used for debug messages. Kills two anti-patterns: atomic_read(&kref->refcount) kref->refcount.counter Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-11-14 16:29:48 +00:00
kref_read(&req->r_kref));
libceph: make ceph_osdc_put_request() accept NULL Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-26 13:05:29 +00:00
kref_put(&req->r_kref, ceph_osdc_release_request);
}
libceph: move and add dout()s to ceph_osdc_request_{get,put}() Add dout()s to ceph_osdc_request_{get,put}(). Also move them to .c and turn kref release callback into a static function. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:42 +00:00
}
EXPORT_SYMBOL(ceph_osdc_put_request);
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
libceph: request_init() and request_release_checks() These are going to be used by request_reinit() code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void request_init(struct ceph_osd_request *req)
{
/* req only, each op is zeroed in _osd_req_op_init() */
memset(req, 0, sizeof(*req));
kref_init(&req->r_kref);
init_completion(&req->r_completion);
RB_CLEAR_NODE(&req->r_node);
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
RB_CLEAR_NODE(&req->r_mc_node);
libceph: request_init() and request_release_checks() These are going to be used by request_reinit() code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
INIT_LIST_HEAD(&req->r_unsafe_item);
target_init(&req->r_t);
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
/*
* This is ugly, but it allows us to reuse linger registration and ping
* requests, keeping the structure of the code around send_linger{_ping}()
* reasonable. Setting up a min_nr=2 mempool for each linger request
* and dealing with copying ops (this blasts req only, watch op remains
* intact) isn't any better.
*/
static void request_reinit(struct ceph_osd_request *req)
{
struct ceph_osd_client *osdc = req->r_osdc;
bool mempool = req->r_mempool;
unsigned int num_ops = req->r_num_ops;
u64 snapid = req->r_snapid;
struct ceph_snap_context *snapc = req->r_snapc;
bool linger = req->r_linger;
struct ceph_msg *request_msg = req->r_request;
struct ceph_msg *reply_msg = req->r_reply;
dout("%s req %p\n", __func__, req);
locking/atomic, kref: Add kref_read() Since we need to change the implementation, stop exposing internals. Provide kref_read() to read the current reference count; typically used for debug messages. Kills two anti-patterns: atomic_read(&kref->refcount) kref->refcount.counter Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-11-14 16:29:48 +00:00
WARN_ON(kref_read(&req->r_kref) != 1);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
request_release_checks(req);
locking/atomic, kref: Add kref_read() Since we need to change the implementation, stop exposing internals. Provide kref_read() to read the current reference count; typically used for debug messages. Kills two anti-patterns: atomic_read(&kref->refcount) kref->refcount.counter Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-11-14 16:29:48 +00:00
WARN_ON(kref_read(&request_msg->kref) != 1);
WARN_ON(kref_read(&reply_msg->kref) != 1);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
target_destroy(&req->r_t);
request_init(req);
req->r_osdc = osdc;
req->r_mempool = mempool;
req->r_num_ops = num_ops;
req->r_snapid = snapid;
req->r_snapc = snapc;
req->r_linger = linger;
req->r_request = request_msg;
req->r_reply = reply_msg;
}
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc,
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
struct ceph_snap_context *snapc,
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
unsigned int num_ops,
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
bool use_mempool,
libceph: don't set pages or bio in ceph_osdc_alloc_request() Only one of the two callers of ceph_osdc_alloc_request() provides page or bio data for its payload. And essentially all that function was doing with those arguments was assigning them to fields in the osd request structure. Simplify ceph_osdc_alloc_request() by having the caller take care of making those assignments Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2012-11-14 03:11:15 +00:00
gfp_t gfp_flags)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
struct ceph_osd_request *req;
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
if (use_mempool) {
libceph: enable large, variable-sized OSD requests Turn r_ops into a flexible array member to enable large, consisting of up to 16 ops, OSD requests. The use case is scattered writeback in cephfs and, as far as the kernel client is concerned, 16 is just a made up number. r_ops had size 3 for copyup+hint+write, but copyup is really a special case - it can only happen once. ceph_osd_request_cache is therefore stuffed with num_ops=2 requests, anything bigger than that is allocated with kmalloc(). req_mempool is backed by ceph_osd_request_cache, which means either num_ops=1 or num_ops=2 for use_mempool=true - all existing users (ceph_writepages_start(), ceph_osdc_writepages()) are fine with that. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:50:15 +00:00
BUG_ON(num_ops > CEPH_OSD_SLAB_OPS);
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
req = mempool_alloc(osdc->req_mempool, gfp_flags);
libceph: enable large, variable-sized OSD requests Turn r_ops into a flexible array member to enable large, consisting of up to 16 ops, OSD requests. The use case is scattered writeback in cephfs and, as far as the kernel client is concerned, 16 is just a made up number. r_ops had size 3 for copyup+hint+write, but copyup is really a special case - it can only happen once. ceph_osd_request_cache is therefore stuffed with num_ops=2 requests, anything bigger than that is allocated with kmalloc(). req_mempool is backed by ceph_osd_request_cache, which means either num_ops=1 or num_ops=2 for use_mempool=true - all existing users (ceph_writepages_start(), ceph_osdc_writepages()) are fine with that. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:50:15 +00:00
} else if (num_ops <= CEPH_OSD_SLAB_OPS) {
req = kmem_cache_alloc(ceph_osd_request_cache, gfp_flags);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
} else {
libceph: enable large, variable-sized OSD requests Turn r_ops into a flexible array member to enable large, consisting of up to 16 ops, OSD requests. The use case is scattered writeback in cephfs and, as far as the kernel client is concerned, 16 is just a made up number. r_ops had size 3 for copyup+hint+write, but copyup is really a special case - it can only happen once. ceph_osd_request_cache is therefore stuffed with num_ops=2 requests, anything bigger than that is allocated with kmalloc(). req_mempool is backed by ceph_osd_request_cache, which means either num_ops=1 or num_ops=2 for use_mempool=true - all existing users (ceph_writepages_start(), ceph_osdc_writepages()) are fine with that. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:50:15 +00:00
BUG_ON(num_ops > CEPH_OSD_MAX_OPS);
treewide: Use struct_size() for kmalloc()-family One of the more common cases of allocation size calculations is finding the size of a structure that has a zero-sized array at the end, along with memory for some number of elements for that array. For example: struct foo { int stuff; void *entry[]; }; instance = kmalloc(sizeof(struct foo) + sizeof(void *) * count, GFP_KERNEL); Instead of leaving these open-coded and prone to type mistakes, we can now use the new struct_size() helper: instance = kmalloc(struct_size(instance, entry, count), GFP_KERNEL); This patch makes the changes for kmalloc()-family (and kvmalloc()-family) uses. It was done via automatic conversion with manual review for the "CHECKME" non-standard cases noted below, using the following Coccinelle script: // pkey_cache = kmalloc(sizeof *pkey_cache + tprops->pkey_tbl_len * // sizeof *pkey_cache->table, GFP_KERNEL); @@ identifier alloc =~ "kmalloc|kzalloc|kvmalloc|kvzalloc"; expression GFP; identifier VAR, ELEMENT; expression COUNT; @@ - alloc(sizeof(*VAR) + COUNT * sizeof(*VAR->ELEMENT), GFP) + alloc(struct_size(VAR, ELEMENT, COUNT), GFP) // mr = kzalloc(sizeof(*mr) + m * sizeof(mr->map[0]), GFP_KERNEL); @@ identifier alloc =~ "kmalloc|kzalloc|kvmalloc|kvzalloc"; expression GFP; identifier VAR, ELEMENT; expression COUNT; @@ - alloc(sizeof(*VAR) + COUNT * sizeof(VAR->ELEMENT[0]), GFP) + alloc(struct_size(VAR, ELEMENT, COUNT), GFP) // Same pattern, but can't trivially locate the trailing element name, // or variable name. @@ identifier alloc =~ "kmalloc|kzalloc|kvmalloc|kvzalloc"; expression GFP; expression SOMETHING, COUNT, ELEMENT; @@ - alloc(sizeof(SOMETHING) + COUNT * sizeof(ELEMENT), GFP) + alloc(CHECKME_struct_size(&SOMETHING, ELEMENT, COUNT), GFP) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-05-08 20:45:50 +00:00
req = kmalloc(struct_size(req, r_ops, num_ops), gfp_flags);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: enable large, variable-sized OSD requests Turn r_ops into a flexible array member to enable large, consisting of up to 16 ops, OSD requests. The use case is scattered writeback in cephfs and, as far as the kernel client is concerned, 16 is just a made up number. r_ops had size 3 for copyup+hint+write, but copyup is really a special case - it can only happen once. ceph_osd_request_cache is therefore stuffed with num_ops=2 requests, anything bigger than that is allocated with kmalloc(). req_mempool is backed by ceph_osd_request_cache, which means either num_ops=1 or num_ops=2 for use_mempool=true - all existing users (ceph_writepages_start(), ceph_osdc_writepages()) are fine with that. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:50:15 +00:00
if (unlikely(!req))
ceph: make ceph_msg_new return NULL on failure; clean up, fix callers Returning ERR_PTR(-ENOMEM) is useless extra work. Return NULL on failure instead, and fix up the callers (about half of which were wrong anyway). Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-01 23:06:19 +00:00
return NULL;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: request_init() and request_release_checks() These are going to be used by request_reinit() code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
request_init(req);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
req->r_osdc = osdc;
req->r_mempool = use_mempool;
libceph: keep source rather than message osd op array An osd request keeps a pointer to the osd operations (ops) array that it builds in its request message. In order to allow each op in the array to have its own distinct data, we will need to keep track of each op's data, and that information does not go over the wire. As long as we're tracking the data we might as well just track the entire (source) op definition for each of the ops. And if we're doing that, we'll have no more need to keep a pointer to the wire-encoded version. This patch makes the array of source ops be kept with the osd request structure, and uses that instead of the version encoded in the message in places where that was previously used. The array will be embedded in the request structure, and the maximum number of ops we ever actually use is currently 2. So reduce CEPH_OSD_MAX_OP to 2 to reduce the size of the structure. The result of doing this sort of ripples back up, and as a result various function parameters and local variables become unnecessary. Make r_num_ops be unsigned, and move the definition of struct ceph_osd_req_op earlier to ensure it's defined where needed. It does not yet add per-op data, that's coming soon. This resolves: http://tracker.ceph.com/issues/4656 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-04 02:32:51 +00:00
req->r_num_ops = num_ops;
libceph: grab snapc in ceph_osdc_alloc_request() ceph_osdc_build_request() is going away. Grab snapc and initialize ->r_snapid in ceph_osdc_alloc_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-26 13:39:47 +00:00
req->r_snapid = CEPH_NOSNAP;
req->r_snapc = ceph_get_snap_context(snapc);
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
dout("%s req %p\n", __func__, req);
return req;
}
EXPORT_SYMBOL(ceph_osdc_alloc_request);
libceph: enable large, variable-sized OSD requests Turn r_ops into a flexible array member to enable large, consisting of up to 16 ops, OSD requests. The use case is scattered writeback in cephfs and, as far as the kernel client is concerned, 16 is just a made up number. r_ops had size 3 for copyup+hint+write, but copyup is really a special case - it can only happen once. ceph_osd_request_cache is therefore stuffed with num_ops=2 requests, anything bigger than that is allocated with kmalloc(). req_mempool is backed by ceph_osd_request_cache, which means either num_ops=1 or num_ops=2 for use_mempool=true - all existing users (ceph_writepages_start(), ceph_osdc_writepages()) are fine with that. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:50:15 +00:00
libceph: encode_{pgid,oloc}() helpers Factor out encode_{pgid,oloc}() and use ceph_encode_string() for oid. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:53 +00:00
static int ceph_oloc_encoding_size(const struct ceph_object_locator *oloc)
libceph: rados pool namespace support Add pool namesapce pointer to struct ceph_file_layout and struct ceph_object_locator. Pool namespace is used by when mapping object to PG, it's also used when composing OSD request. The namespace pointer in struct ceph_file_layout is RCU protected. So libceph can read namespace without taking lock. Signed-off-by: Yan, Zheng <zyan@redhat.com> [idryomov@gmail.com: ceph_oloc_destroy(), misc minor changes] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-14 03:24:31 +00:00
{
return 8 + 4 + 4 + 4 + (oloc->pool_ns ? oloc->pool_ns->len : 0);
}
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
int ceph_osdc_alloc_messages(struct ceph_osd_request *req, gfp_t gfp)
{
struct ceph_osd_client *osdc = req->r_osdc;
struct ceph_msg *msg;
int msg_size;
ceph: use single osd op reply msg Use a single ceph_msg for the osd reply, even when we are getting multiple replies. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-01 21:02:00 +00:00
libceph: variable-sized ceph_object_id Currently ceph_object_id can hold object names of up to 100 (CEPH_MAX_OID_NAME_LEN) characters. This is enough for all use cases, expect one - long rbd image names: - a format 1 header is named "<imgname>.rbd" - an object that points to a format 2 header is named "rbd_id.<imgname>" We operate on these potentially long-named objects during rbd map, and, for format 1 images, during header refresh. (A format 2 header name is a small system-generated string.) Lift this 100 character limit by making ceph_object_id be able to point to an externally-allocated string. Apart from being able to work with almost arbitrarily-long named objects, this allows us to reduce the size of ceph_object_id from >100 bytes to 64 bytes. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-29 17:54:20 +00:00
WARN_ON(ceph_oid_empty(&req->r_base_oid));
libceph: rados pool namespace support Add pool namesapce pointer to struct ceph_file_layout and struct ceph_object_locator. Pool namespace is used by when mapping object to PG, it's also used when composing OSD request. The namespace pointer in struct ceph_file_layout is RCU protected. So libceph can read namespace without taking lock. Signed-off-by: Yan, Zheng <zyan@redhat.com> [idryomov@gmail.com: ceph_oloc_destroy(), misc minor changes] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-14 03:24:31 +00:00
WARN_ON(ceph_oloc_empty(&req->r_base_oloc));
libceph: variable-sized ceph_object_id Currently ceph_object_id can hold object names of up to 100 (CEPH_MAX_OID_NAME_LEN) characters. This is enough for all use cases, expect one - long rbd image names: - a format 1 header is named "<imgname>.rbd" - an object that points to a format 2 header is named "rbd_id.<imgname>" We operate on these potentially long-named objects during rbd map, and, for format 1 images, during header refresh. (A format 2 header name is a small system-generated string.) Lift this 100 character limit by making ceph_object_id be able to point to an externally-allocated string. Apart from being able to work with almost arbitrarily-long named objects, this allows us to reduce the size of ceph_object_id from >100 bytes to 64 bytes. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-29 17:54:20 +00:00
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
/* create request message */
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
msg_size = CEPH_ENCODING_START_BLK_LEN +
CEPH_PGID_ENCODING_LEN + 1; /* spgid */
msg_size += 4 + 4 + 4; /* hash, osdmap_epoch, flags */
msg_size += CEPH_ENCODING_START_BLK_LEN +
sizeof(struct ceph_osd_reqid); /* reqid */
msg_size += sizeof(struct ceph_blkin_trace_info); /* trace */
msg_size += 4 + sizeof(struct ceph_timespec); /* client_inc, mtime */
libceph: rados pool namespace support Add pool namesapce pointer to struct ceph_file_layout and struct ceph_object_locator. Pool namespace is used by when mapping object to PG, it's also used when composing OSD request. The namespace pointer in struct ceph_file_layout is RCU protected. So libceph can read namespace without taking lock. Signed-off-by: Yan, Zheng <zyan@redhat.com> [idryomov@gmail.com: ceph_oloc_destroy(), misc minor changes] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-14 03:24:31 +00:00
msg_size += CEPH_ENCODING_START_BLK_LEN +
ceph_oloc_encoding_size(&req->r_base_oloc); /* oloc */
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
msg_size += 4 + req->r_base_oid.name_len; /* oid */
msg_size += 2 + req->r_num_ops * sizeof(struct ceph_osd_op);
libceph: make r_request msg_size calculation clearer Although msg_size is calculated correctly, the terms are grouped in a misleading way - snaps appears to not have room for a u32 length. Move calculation closer to its use and regroup terms. No functional change. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-11 12:09:15 +00:00
msg_size += 8; /* snapid */
msg_size += 8; /* snap_seq */
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
msg_size += 4 + 8 * (req->r_snapc ? req->r_snapc->num_snaps : 0);
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
msg_size += 4 + 8; /* retry_attempt, features */
libceph: make r_request msg_size calculation clearer Although msg_size is calculated correctly, the terms are grouped in a misleading way - snaps appears to not have room for a u32 length. Move calculation closer to its use and regroup terms. No functional change. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-11 12:09:15 +00:00
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
if (req->r_mempool)
ceph: warn on allocation from msgpool with larger front_len Pass the front_len we need when pulling a message off a msgpool, and WARN if it is greater than the pool's size. Then try to allocate a new message (to continue without failing). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-15 00:36:07 +00:00
msg = ceph_msgpool_get(&osdc->msgpool_op, 0);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
else
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp, true);
if (!msg)
return -ENOMEM;
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
memset(msg->front.iov_base, 0, msg->front.iov_len);
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
req->r_request = msg;
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
/* create reply message */
msg_size = OSD_OPREPLY_FRONT_LEN;
libceph: change how osd_op_reply message size is calculated For a message pool message, preallocate a page, just like we do for osd_op. For a normal message, take ceph_object_id into account and don't bother subtracting CEPH_OSD_SLAB_OPS ceph_osd_ops. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 16:32:56 +00:00
msg_size += req->r_base_oid.name_len;
msg_size += req->r_num_ops * sizeof(struct ceph_osd_op);
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
if (req->r_mempool)
msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0);
else
msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, msg_size, gfp, true);
if (!msg)
return -ENOMEM;
req->r_reply = msg;
return 0;
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
}
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
EXPORT_SYMBOL(ceph_osdc_alloc_messages);
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
static bool osd_req_opcode_valid(u16 opcode)
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
{
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
switch (opcode) {
libceph: sync osd op definitions in rados.h Bring in missing osd ops and strings, use macros to eliminate multiple points of maintenance. Signed-off-by: Ilya Dryomov <idryomov@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2014-10-02 13:22:29 +00:00
#define GENERATE_CASE(op, opcode, str) case CEPH_OSD_OP_##op: return true;
__CEPH_FORALL_OSD_OPS(GENERATE_CASE)
#undef GENERATE_CASE
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
default:
return false;
}
}
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
/*
* This is an osd op init function for opcodes that have no data or
* other information associated with them. It also serves as a
* common init routine for all the other init functions, below.
*/
libceph: specify osd op by index in request An osd request now holds all of its source op structures, and every place that initializes one of these is in fact initializing one of the entries in the the osd request's array. So rather than supplying the address of the op to initialize, have caller specify the osd request and an indication of which op it would like to initialize. This better hides the details the op structure (and faciltates moving the data pointers they use). Since osd_req_op_init() is a common routine, and it's not used outside the osd client code, give it static scope. Also make it return the address of the specified op (so all the other init routines don't have to repeat that code). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:11 +00:00
static struct ceph_osd_req_op *
libceph: support raw data requests Allow osd request ops that aren't otherwise structured (not class, extent, or watch ops) to specify "raw" data to be used to hold incoming data for the op. Make use of this capability for the osd STAT op. Prefix the name of the private function osd_req_op_init() with "_", and expose a new function by that (earlier) name whose purpose is to initialize osd ops with (only) implied data. For now we'll just support the use of a page array for an osd op with incoming raw data. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-11 18:33:24 +00:00
_osd_req_op_init(struct ceph_osd_request *osd_req, unsigned int which,
libceph: allow setting osd_req_op's flags Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:09:54 +00:00
u16 opcode, u32 flags)
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
{
libceph: specify osd op by index in request An osd request now holds all of its source op structures, and every place that initializes one of these is in fact initializing one of the entries in the the osd request's array. So rather than supplying the address of the op to initialize, have caller specify the osd request and an indication of which op it would like to initialize. This better hides the details the op structure (and faciltates moving the data pointers they use). Since osd_req_op_init() is a common routine, and it's not used outside the osd client code, give it static scope. Also make it return the address of the specified op (so all the other init routines don't have to repeat that code). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:11 +00:00
struct ceph_osd_req_op *op;
BUG_ON(which >= osd_req->r_num_ops);
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
BUG_ON(!osd_req_opcode_valid(opcode));
libceph: specify osd op by index in request An osd request now holds all of its source op structures, and every place that initializes one of these is in fact initializing one of the entries in the the osd request's array. So rather than supplying the address of the op to initialize, have caller specify the osd request and an indication of which op it would like to initialize. This better hides the details the op structure (and faciltates moving the data pointers they use). Since osd_req_op_init() is a common routine, and it's not used outside the osd client code, give it static scope. Also make it return the address of the specified op (so all the other init routines don't have to repeat that code). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:11 +00:00
op = &osd_req->r_ops[which];
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
memset(op, 0, sizeof (*op));
op->op = opcode;
libceph: allow setting osd_req_op's flags Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:09:54 +00:00
op->flags = flags;
libceph: specify osd op by index in request An osd request now holds all of its source op structures, and every place that initializes one of these is in fact initializing one of the entries in the the osd request's array. So rather than supplying the address of the op to initialize, have caller specify the osd request and an indication of which op it would like to initialize. This better hides the details the op structure (and faciltates moving the data pointers they use). Since osd_req_op_init() is a common routine, and it's not used outside the osd client code, give it static scope. Also make it return the address of the specified op (so all the other init routines don't have to repeat that code). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:11 +00:00
return op;
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
}
libceph: support raw data requests Allow osd request ops that aren't otherwise structured (not class, extent, or watch ops) to specify "raw" data to be used to hold incoming data for the op. Make use of this capability for the osd STAT op. Prefix the name of the private function osd_req_op_init() with "_", and expose a new function by that (earlier) name whose purpose is to initialize osd ops with (only) implied data. For now we'll just support the use of a page array for an osd op with incoming raw data. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-11 18:33:24 +00:00
void osd_req_op_init(struct ceph_osd_request *osd_req,
libceph: allow setting osd_req_op's flags Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:09:54 +00:00
unsigned int which, u16 opcode, u32 flags)
libceph: support raw data requests Allow osd request ops that aren't otherwise structured (not class, extent, or watch ops) to specify "raw" data to be used to hold incoming data for the op. Make use of this capability for the osd STAT op. Prefix the name of the private function osd_req_op_init() with "_", and expose a new function by that (earlier) name whose purpose is to initialize osd ops with (only) implied data. For now we'll just support the use of a page array for an osd op with incoming raw data. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-11 18:33:24 +00:00
{
libceph: allow setting osd_req_op's flags Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:09:54 +00:00
(void)_osd_req_op_init(osd_req, which, opcode, flags);
libceph: support raw data requests Allow osd request ops that aren't otherwise structured (not class, extent, or watch ops) to specify "raw" data to be used to hold incoming data for the op. Make use of this capability for the osd STAT op. Prefix the name of the private function osd_req_op_init() with "_", and expose a new function by that (earlier) name whose purpose is to initialize osd ops with (only) implied data. For now we'll just support the use of a page array for an osd op with incoming raw data. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-11 18:33:24 +00:00
}
EXPORT_SYMBOL(osd_req_op_init);
libceph: specify osd op by index in request An osd request now holds all of its source op structures, and every place that initializes one of these is in fact initializing one of the entries in the the osd request's array. So rather than supplying the address of the op to initialize, have caller specify the osd request and an indication of which op it would like to initialize. This better hides the details the op structure (and faciltates moving the data pointers they use). Since osd_req_op_init() is a common routine, and it's not used outside the osd client code, give it static scope. Also make it return the address of the specified op (so all the other init routines don't have to repeat that code). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:11 +00:00
void osd_req_op_extent_init(struct ceph_osd_request *osd_req,
unsigned int which, u16 opcode,
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
u64 offset, u64 length,
u64 truncate_size, u32 truncate_seq)
{
libceph: allow setting osd_req_op's flags Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:09:54 +00:00
struct ceph_osd_req_op *op = _osd_req_op_init(osd_req, which,
opcode, 0);
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
size_t payload_len = 0;
ceph: punch hole support This patch implements fallocate and punch hole support for Ceph kernel client. Signed-off-by: Li Wang <liwang@ubuntukylin.com> Signed-off-by: Yunchuan Wen <yunchuanwen@ubuntukylin.com>
2013-08-15 03:51:44 +00:00
BUG_ON(opcode != CEPH_OSD_OP_READ && opcode != CEPH_OSD_OP_WRITE &&
rbd: use writefull op for object size writes This covers only the simplest case - an object size sized write, but it's still useful in tiering setups when EC is used for the base tier as writefull op can be proxied, saving an object promotion. Even though updating ceph_osdc_new_request() to allow writefull should just be a matter of fixing an assert, I didn't do it because its only user is cephfs. All other sites were updated. Reflects ceph.git commit 7bfb7f9025a8ee0d2305f49bf0336d2424da5b5b. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-10-07 15:27:17 +00:00
opcode != CEPH_OSD_OP_WRITEFULL && opcode != CEPH_OSD_OP_ZERO &&
opcode != CEPH_OSD_OP_TRUNCATE);
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
op->extent.offset = offset;
op->extent.length = length;
op->extent.truncate_size = truncate_size;
op->extent.truncate_seq = truncate_seq;
rbd: use writefull op for object size writes This covers only the simplest case - an object size sized write, but it's still useful in tiering setups when EC is used for the base tier as writefull op can be proxied, saving an object promotion. Even though updating ceph_osdc_new_request() to allow writefull should just be a matter of fixing an assert, I didn't do it because its only user is cephfs. All other sites were updated. Reflects ceph.git commit 7bfb7f9025a8ee0d2305f49bf0336d2424da5b5b. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-10-07 15:27:17 +00:00
if (opcode == CEPH_OSD_OP_WRITE || opcode == CEPH_OSD_OP_WRITEFULL)
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
payload_len += length;
libceph: rename ceph_osd_req_op::payload_len to indata_len Follow userspace nomenclature on this - the next commit adds outdata_len. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-08 12:39:46 +00:00
op->indata_len = payload_len;
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
}
EXPORT_SYMBOL(osd_req_op_extent_init);
libceph: specify osd op by index in request An osd request now holds all of its source op structures, and every place that initializes one of these is in fact initializing one of the entries in the the osd request's array. So rather than supplying the address of the op to initialize, have caller specify the osd request and an indication of which op it would like to initialize. This better hides the details the op structure (and faciltates moving the data pointers they use). Since osd_req_op_init() is a common routine, and it's not used outside the osd client code, give it static scope. Also make it return the address of the specified op (so all the other init routines don't have to repeat that code). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:11 +00:00
void osd_req_op_extent_update(struct ceph_osd_request *osd_req,
unsigned int which, u64 length)
ceph: build osd request message later for writepages Hold off building the osd request message in ceph_writepages_start() until just before it will be submitted to the osd client for execution. We'll still create the request and allocate the page pointer array after we learn we have at least one page to write. A local variable will be used to keep track of the allocated array of pages. Wait until just before submitting the request for assigning that page array pointer to the request message. Create ands use a new function osd_req_op_extent_update() whose purpose is to serve this one spot where the length value supplied when an osd request's op was initially formatted might need to get changed (reduced, never increased) before submitting the request. Previously, ceph_writepages_start() assigned the message header's data length because of this update. That's no longer necessary, because ceph_osdc_build_request() will recalculate the right value to use based on the content of the ops in the request. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 19:09:05 +00:00
{
libceph: specify osd op by index in request An osd request now holds all of its source op structures, and every place that initializes one of these is in fact initializing one of the entries in the the osd request's array. So rather than supplying the address of the op to initialize, have caller specify the osd request and an indication of which op it would like to initialize. This better hides the details the op structure (and faciltates moving the data pointers they use). Since osd_req_op_init() is a common routine, and it's not used outside the osd client code, give it static scope. Also make it return the address of the specified op (so all the other init routines don't have to repeat that code). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:11 +00:00
struct ceph_osd_req_op *op;
u64 previous;
BUG_ON(which >= osd_req->r_num_ops);
op = &osd_req->r_ops[which];
previous = op->extent.length;
ceph: build osd request message later for writepages Hold off building the osd request message in ceph_writepages_start() until just before it will be submitted to the osd client for execution. We'll still create the request and allocate the page pointer array after we learn we have at least one page to write. A local variable will be used to keep track of the allocated array of pages. Wait until just before submitting the request for assigning that page array pointer to the request message. Create ands use a new function osd_req_op_extent_update() whose purpose is to serve this one spot where the length value supplied when an osd request's op was initially formatted might need to get changed (reduced, never increased) before submitting the request. Previously, ceph_writepages_start() assigned the message header's data length because of this update. That's no longer necessary, because ceph_osdc_build_request() will recalculate the right value to use based on the content of the ops in the request. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 19:09:05 +00:00
if (length == previous)
return; /* Nothing to do */
BUG_ON(length > previous);
op->extent.length = length;
ceph: update readpages osd request according to size of pages add_to_page_cache_lru() can fails, so the actual pages to read can be smaller than the initial size of osd request. We need to update osd request size in that case. Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Jeff Layton <jlayton@redhat.com>
2017-01-19 03:21:29 +00:00
if (op->op == CEPH_OSD_OP_WRITE || op->op == CEPH_OSD_OP_WRITEFULL)
op->indata_len -= previous - length;
ceph: build osd request message later for writepages Hold off building the osd request message in ceph_writepages_start() until just before it will be submitted to the osd client for execution. We'll still create the request and allocate the page pointer array after we learn we have at least one page to write. A local variable will be used to keep track of the allocated array of pages. Wait until just before submitting the request for assigning that page array pointer to the request message. Create ands use a new function osd_req_op_extent_update() whose purpose is to serve this one spot where the length value supplied when an osd request's op was initially formatted might need to get changed (reduced, never increased) before submitting the request. Previously, ceph_writepages_start() assigned the message header's data length because of this update. That's no longer necessary, because ceph_osdc_build_request() will recalculate the right value to use based on the content of the ops in the request. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 19:09:05 +00:00
}
EXPORT_SYMBOL(osd_req_op_extent_update);
libceph: add helper that duplicates last extent operation This helper duplicates last extent operation in OSD request, then adjusts the new extent operation's offset and length. The helper is for scatterd page writeback, which adds nonconsecutive dirty pages to single OSD request. Signed-off-by: Yan, Zheng <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-01-07 09:32:54 +00:00
void osd_req_op_extent_dup_last(struct ceph_osd_request *osd_req,
unsigned int which, u64 offset_inc)
{
struct ceph_osd_req_op *op, *prev_op;
BUG_ON(which + 1 >= osd_req->r_num_ops);
prev_op = &osd_req->r_ops[which];
op = _osd_req_op_init(osd_req, which + 1, prev_op->op, prev_op->flags);
/* dup previous one */
op->indata_len = prev_op->indata_len;
op->outdata_len = prev_op->outdata_len;
op->extent = prev_op->extent;
/* adjust offset */
op->extent.offset += offset_inc;
op->extent.length -= offset_inc;
if (op->op == CEPH_OSD_OP_WRITE || op->op == CEPH_OSD_OP_WRITEFULL)
op->indata_len -= offset_inc;
}
EXPORT_SYMBOL(osd_req_op_extent_dup_last);
libceph, rbd: add error handling for osd_req_op_cls_init() Add proper error handling for osd_req_op_cls_init() to replace BUG_ON statement when failing from memory allocation. Signed-off-by: Chengguang Xu <cgxu519@gmx.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-04-12 04:04:55 +00:00
int osd_req_op_cls_init(struct ceph_osd_request *osd_req, unsigned int which,
libceph: make method call data be a separate data item Right now the data for a method call is specified via a pointer and length, and it's copied--along with the class and method name--into a pagelist data item to be sent to the osd. Instead, encode the data in a data item separate from the class and method names. This will allow large amounts of data to be supplied to methods without copying. Only rbd uses the class functionality right now, and when it really needs this it will probably need to use a page array rather than a page list. But this simple implementation demonstrates the functionality on the osd client, and that's enough for now. This resolves: http://tracker.ceph.com/issues/4104 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 19:46:02 +00:00
u16 opcode, const char *class, const char *method)
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
{
libceph: allow setting osd_req_op's flags Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:09:54 +00:00
struct ceph_osd_req_op *op = _osd_req_op_init(osd_req, which,
opcode, 0);
libceph: format class info at init time An object class method is formatted using a pagelist which contains the class name, the method name, and the data concatenated into an osd request's outbound data. Currently when a class op is initialized in osd_req_op_cls_init(), the lengths of and pointers to these three items are recorded. Later, when the op is getting formatted into the request message, a new pagelist is created and that is when these items get copied into the pagelist. This patch makes it so the pagelist to hold these items is created when the op is initialized instead. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
struct ceph_pagelist *pagelist;
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
size_t payload_len = 0;
size_t size;
BUG_ON(opcode != CEPH_OSD_OP_CALL);
libceph: format class info at init time An object class method is formatted using a pagelist which contains the class name, the method name, and the data concatenated into an osd request's outbound data. Currently when a class op is initialized in osd_req_op_cls_init(), the lengths of and pointers to these three items are recorded. Later, when the op is getting formatted into the request message, a new pagelist is created and that is when these items get copied into the pagelist. This patch makes it so the pagelist to hold these items is created when the op is initialized instead. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
pagelist = kmalloc(sizeof (*pagelist), GFP_NOFS);
libceph, rbd: add error handling for osd_req_op_cls_init() Add proper error handling for osd_req_op_cls_init() to replace BUG_ON statement when failing from memory allocation. Signed-off-by: Chengguang Xu <cgxu519@gmx.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-04-12 04:04:55 +00:00
if (!pagelist)
return -ENOMEM;
libceph: format class info at init time An object class method is formatted using a pagelist which contains the class name, the method name, and the data concatenated into an osd request's outbound data. Currently when a class op is initialized in osd_req_op_cls_init(), the lengths of and pointers to these three items are recorded. Later, when the op is getting formatted into the request message, a new pagelist is created and that is when these items get copied into the pagelist. This patch makes it so the pagelist to hold these items is created when the op is initialized instead. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_pagelist_init(pagelist);
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
op->cls.class_name = class;
size = strlen(class);
BUG_ON(size > (size_t) U8_MAX);
op->cls.class_len = size;
libceph: format class info at init time An object class method is formatted using a pagelist which contains the class name, the method name, and the data concatenated into an osd request's outbound data. Currently when a class op is initialized in osd_req_op_cls_init(), the lengths of and pointers to these three items are recorded. Later, when the op is getting formatted into the request message, a new pagelist is created and that is when these items get copied into the pagelist. This patch makes it so the pagelist to hold these items is created when the op is initialized instead. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_pagelist_append(pagelist, class, size);
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
payload_len += size;
op->cls.method_name = method;
size = strlen(method);
BUG_ON(size > (size_t) U8_MAX);
op->cls.method_len = size;
libceph: format class info at init time An object class method is formatted using a pagelist which contains the class name, the method name, and the data concatenated into an osd request's outbound data. Currently when a class op is initialized in osd_req_op_cls_init(), the lengths of and pointers to these three items are recorded. Later, when the op is getting formatted into the request message, a new pagelist is created and that is when these items get copied into the pagelist. This patch makes it so the pagelist to hold these items is created when the op is initialized instead. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
ceph_pagelist_append(pagelist, method, size);
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
payload_len += size;
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
osd_req_op_cls_request_info_pagelist(osd_req, which, pagelist);
libceph: format class info at init time An object class method is formatted using a pagelist which contains the class name, the method name, and the data concatenated into an osd request's outbound data. Currently when a class op is initialized in osd_req_op_cls_init(), the lengths of and pointers to these three items are recorded. Later, when the op is getting formatted into the request message, a new pagelist is created and that is when these items get copied into the pagelist. This patch makes it so the pagelist to hold these items is created when the op is initialized instead. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
libceph: rename ceph_osd_req_op::payload_len to indata_len Follow userspace nomenclature on this - the next commit adds outdata_len. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-08 12:39:46 +00:00
op->indata_len = payload_len;
libceph, rbd: add error handling for osd_req_op_cls_init() Add proper error handling for osd_req_op_cls_init() to replace BUG_ON statement when failing from memory allocation. Signed-off-by: Chengguang Xu <cgxu519@gmx.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-04-12 04:04:55 +00:00
return 0;
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
}
EXPORT_SYMBOL(osd_req_op_cls_init);
libceph: add data pointers in osd op structures An extent type osd operation currently implies that there will be corresponding data supplied in the data portion of the request (for write) or response (for read) message. Similarly, an osd class method operation implies a data item will be supplied to receive the response data from the operation. Add a ceph_osd_data pointer to each of those structures, and assign it to point to eithre the incoming or the outgoing data structure in the osd message. The data is not always available when an op is initially set up, so add two new functions to allow setting them after the op has been initialized. Begin to make use of the data item pointer available in the osd operation rather than the request data in or out structure in places where it's convenient. Add some assertions to verify pointers are always set the way they're expected to be. This is a sort of stepping stone toward really moving the data into the osd request ops, to allow for some validation before making that jump. This is the first in a series of patches that resolve: http://tracker.ceph.com/issues/4657 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:58 +00:00
libceph: add SETXATTR/CMPXATTR osd operations support Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-12 06:00:43 +00:00
int osd_req_op_xattr_init(struct ceph_osd_request *osd_req, unsigned int which,
u16 opcode, const char *name, const void *value,
size_t size, u8 cmp_op, u8 cmp_mode)
{
libceph: allow setting osd_req_op's flags Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:09:54 +00:00
struct ceph_osd_req_op *op = _osd_req_op_init(osd_req, which,
opcode, 0);
libceph: add SETXATTR/CMPXATTR osd operations support Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-12 06:00:43 +00:00
struct ceph_pagelist *pagelist;
size_t payload_len;
BUG_ON(opcode != CEPH_OSD_OP_SETXATTR && opcode != CEPH_OSD_OP_CMPXATTR);
pagelist = kmalloc(sizeof(*pagelist), GFP_NOFS);
if (!pagelist)
return -ENOMEM;
ceph_pagelist_init(pagelist);
payload_len = strlen(name);
op->xattr.name_len = payload_len;
ceph_pagelist_append(pagelist, name, payload_len);
op->xattr.value_len = size;
ceph_pagelist_append(pagelist, value, size);
payload_len += size;
op->xattr.cmp_op = cmp_op;
op->xattr.cmp_mode = cmp_mode;
ceph_osd_data_pagelist_init(&op->xattr.osd_data, pagelist);
libceph: rename ceph_osd_req_op::payload_len to indata_len Follow userspace nomenclature on this - the next commit adds outdata_len. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-08 12:39:46 +00:00
op->indata_len = payload_len;
libceph: add SETXATTR/CMPXATTR osd operations support Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-12 06:00:43 +00:00
return 0;
}
EXPORT_SYMBOL(osd_req_op_xattr_init);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
/*
* @watch_opcode: CEPH_OSD_WATCH_OP_*
*/
static void osd_req_op_watch_init(struct ceph_osd_request *req, int which,
u64 cookie, u8 watch_opcode)
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
{
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
struct ceph_osd_req_op *op;
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
op = _osd_req_op_init(req, which, CEPH_OSD_OP_WATCH, 0);
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
op->watch.cookie = cookie;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
op->watch.op = watch_opcode;
op->watch.gen = 0;
libceph: define source request op functions The rbd code has a function that allocates and populates a ceph_osd_req_op structure (the in-core version of an osd request operation). When reviewed, Josh suggested two things: that the big varargs function might be better split into type-specific functions; and that this functionality really belongs in the osd client rather than rbd. This patch implements both of Josh's suggestions. It breaks up the rbd function into separate functions and defines them in the osd client module as exported interfaces. Unlike the rbd version, however, the functions don't allocate an osd_req_op structure; they are provided the address of one and that is initialized instead. The rbd function has been eliminated and calls to it have been replaced by calls to the new routines. The rbd code now now use a stack (struct) variable to hold the op rather than allocating and freeing it each time. For now only the capabilities used by rbd are implemented. Implementing all the other osd op types, and making the rest of the code use it will be done separately, in the next few patches. Note that only the extent, cls, and watch portions of the ceph_osd_req_op structure are currently used. Delete the others (xattr, pgls, and snap) from its definition so nobody thinks it's actually implemented or needed. We can add it back again later if needed, when we know it's been tested. This (and a few follow-on patches) resolves: http://tracker.ceph.com/issues/3861 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
}
libceph: add support for CEPH_OSD_OP_SETALLOCHINT osd op This is primarily for rbd's benefit and is supposed to combat fragmentation: "... knowing that rbd images have a 4m size, librbd can pass a hint that will let the osd do the xfs allocation size ioctl on new files so that they are allocated in 1m or 4m chunks. We've seen cases where users with rbd workloads have very high levels of fragmentation in xfs and this would mitigate that and probably have a pretty nice performance benefit." SETALLOCHINT is considered advisory, so our backwards compatibility mechanism here is to set FAILOK flag for all SETALLOCHINT ops. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-02-25 14:22:27 +00:00
void osd_req_op_alloc_hint_init(struct ceph_osd_request *osd_req,
unsigned int which,
u64 expected_object_size,
u64 expected_write_size)
{
struct ceph_osd_req_op *op = _osd_req_op_init(osd_req, which,
libceph: allow setting osd_req_op's flags Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:09:54 +00:00
CEPH_OSD_OP_SETALLOCHINT,
0);
libceph: add support for CEPH_OSD_OP_SETALLOCHINT osd op This is primarily for rbd's benefit and is supposed to combat fragmentation: "... knowing that rbd images have a 4m size, librbd can pass a hint that will let the osd do the xfs allocation size ioctl on new files so that they are allocated in 1m or 4m chunks. We've seen cases where users with rbd workloads have very high levels of fragmentation in xfs and this would mitigate that and probably have a pretty nice performance benefit." SETALLOCHINT is considered advisory, so our backwards compatibility mechanism here is to set FAILOK flag for all SETALLOCHINT ops. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-02-25 14:22:27 +00:00
op->alloc_hint.expected_object_size = expected_object_size;
op->alloc_hint.expected_write_size = expected_write_size;
/*
* CEPH_OSD_OP_SETALLOCHINT op is advisory and therefore deemed
* not worth a feature bit. Set FAILOK per-op flag to make
* sure older osds don't trip over an unsupported opcode.
*/
op->flags |= CEPH_OSD_OP_FLAG_FAILOK;
}
EXPORT_SYMBOL(osd_req_op_alloc_hint_init);
libceph: add, don't set data for a message Change the names of the functions that put data on a pagelist to reflect that we're adding to whatever's already there rather than just setting it to the one thing. Currently only one data item is ever added to a message, but that's about to change. This resolves: http://tracker.ceph.com/issues/2770 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 19:46:01 +00:00
static void ceph_osdc_msg_data_add(struct ceph_msg *msg,
libceph: set the data pointers when encoding ops Still using the osd request r_data_in and r_data_out pointer, but we're basically only referring to it via the data pointers in the osd ops. And we're transferring that information to the request or reply message only when the op indicates it's needed, in osd_req_encode_op(). To avoid a forward reference, ceph_osdc_msg_data_set() was moved up in the file. Don't bother calling ceph_osd_data_init(), in ceph_osd_alloc(), because the ops array will already be zeroed anyway. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
struct ceph_osd_data *osd_data)
{
u64 length = ceph_osd_data_length(osd_data);
if (osd_data->type == CEPH_OSD_DATA_TYPE_PAGES) {
BUG_ON(length > (u64) SIZE_MAX);
if (length)
libceph: add, don't set data for a message Change the names of the functions that put data on a pagelist to reflect that we're adding to whatever's already there rather than just setting it to the one thing. Currently only one data item is ever added to a message, but that's about to change. This resolves: http://tracker.ceph.com/issues/2770 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 19:46:01 +00:00
ceph_msg_data_add_pages(msg, osd_data->pages,
libceph: set the data pointers when encoding ops Still using the osd request r_data_in and r_data_out pointer, but we're basically only referring to it via the data pointers in the osd ops. And we're transferring that information to the request or reply message only when the op indicates it's needed, in osd_req_encode_op(). To avoid a forward reference, ceph_osdc_msg_data_set() was moved up in the file. Don't bother calling ceph_osd_data_init(), in ceph_osd_alloc(), because the ops array will already be zeroed anyway. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
length, osd_data->alignment);
} else if (osd_data->type == CEPH_OSD_DATA_TYPE_PAGELIST) {
BUG_ON(!length);
libceph: add, don't set data for a message Change the names of the functions that put data on a pagelist to reflect that we're adding to whatever's already there rather than just setting it to the one thing. Currently only one data item is ever added to a message, but that's about to change. This resolves: http://tracker.ceph.com/issues/2770 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 19:46:01 +00:00
ceph_msg_data_add_pagelist(msg, osd_data->pagelist);
libceph: set the data pointers when encoding ops Still using the osd request r_data_in and r_data_out pointer, but we're basically only referring to it via the data pointers in the osd ops. And we're transferring that information to the request or reply message only when the op indicates it's needed, in osd_req_encode_op(). To avoid a forward reference, ceph_osdc_msg_data_set() was moved up in the file. Don't bother calling ceph_osd_data_init(), in ceph_osd_alloc(), because the ops array will already be zeroed anyway. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
#ifdef CONFIG_BLOCK
} else if (osd_data->type == CEPH_OSD_DATA_TYPE_BIO) {
libceph, rbd: new bio handling code (aka don't clone bios) The reason we clone bios is to be able to give each object request (and consequently each ceph_osd_data/ceph_msg_data item) its own pointer to a (list of) bio(s). The messenger then initializes its cursor with cloned bio's ->bi_iter, so it knows where to start reading from/writing to. That's all the cloned bios are used for: to determine each object request's starting position in the provided data buffer. Introduce ceph_bio_iter to do exactly that -- store position within bio list (i.e. pointer to bio) + position within that bio (i.e. bvec_iter). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:10 +00:00
ceph_msg_data_add_bio(msg, &osd_data->bio_pos, length);
libceph: set the data pointers when encoding ops Still using the osd request r_data_in and r_data_out pointer, but we're basically only referring to it via the data pointers in the osd ops. And we're transferring that information to the request or reply message only when the op indicates it's needed, in osd_req_encode_op(). To avoid a forward reference, ceph_osdc_msg_data_set() was moved up in the file. Don't bother calling ceph_osd_data_init(), in ceph_osd_alloc(), because the ops array will already be zeroed anyway. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
#endif
libceph: introduce BVECS data type In preparation for rbd "fancy" striping, introduce ceph_bvec_iter for working with bio_vec array data buffers. The wrappers are trivial, but make it look similar to ceph_bio_iter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-01-20 09:30:11 +00:00
} else if (osd_data->type == CEPH_OSD_DATA_TYPE_BVECS) {
ceph_msg_data_add_bvecs(msg, &osd_data->bvec_pos);
libceph: set the data pointers when encoding ops Still using the osd request r_data_in and r_data_out pointer, but we're basically only referring to it via the data pointers in the osd ops. And we're transferring that information to the request or reply message only when the op indicates it's needed, in osd_req_encode_op(). To avoid a forward reference, ceph_osdc_msg_data_set() was moved up in the file. Don't bother calling ceph_osd_data_init(), in ceph_osd_alloc(), because the ops array will already be zeroed anyway. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
} else {
BUG_ON(osd_data->type != CEPH_OSD_DATA_TYPE_NONE);
}
}
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
static u32 osd_req_encode_op(struct ceph_osd_op *dst,
const struct ceph_osd_req_op *src)
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
{
if (WARN_ON(!osd_req_opcode_valid(src->op))) {
pr_err("unrecognized osd opcode %d\n", src->op);
return 0;
}
switch (src->op) {
case CEPH_OSD_OP_STAT:
break;
case CEPH_OSD_OP_READ:
case CEPH_OSD_OP_WRITE:
rbd: use writefull op for object size writes This covers only the simplest case - an object size sized write, but it's still useful in tiering setups when EC is used for the base tier as writefull op can be proxied, saving an object promotion. Even though updating ceph_osdc_new_request() to allow writefull should just be a matter of fixing an assert, I didn't do it because its only user is cephfs. All other sites were updated. Reflects ceph.git commit 7bfb7f9025a8ee0d2305f49bf0336d2424da5b5b. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-10-07 15:27:17 +00:00
case CEPH_OSD_OP_WRITEFULL:
ceph: punch hole support This patch implements fallocate and punch hole support for Ceph kernel client. Signed-off-by: Li Wang <liwang@ubuntukylin.com> Signed-off-by: Yunchuan Wen <yunchuanwen@ubuntukylin.com>
2013-08-15 03:51:44 +00:00
case CEPH_OSD_OP_ZERO:
case CEPH_OSD_OP_TRUNCATE:
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
dst->extent.offset = cpu_to_le64(src->extent.offset);
dst->extent.length = cpu_to_le64(src->extent.length);
dst->extent.truncate_size =
cpu_to_le64(src->extent.truncate_size);
dst->extent.truncate_seq =
cpu_to_le32(src->extent.truncate_seq);
break;
case CEPH_OSD_OP_CALL:
dst->cls.class_len = src->cls.class_len;
dst->cls.method_len = src->cls.method_len;
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
dst->cls.indata_len = cpu_to_le32(src->cls.indata_len);
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
break;
case CEPH_OSD_OP_WATCH:
dst->watch.cookie = cpu_to_le64(src->watch.cookie);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
dst->watch.ver = cpu_to_le64(0);
dst->watch.op = src->watch.op;
dst->watch.gen = cpu_to_le32(src->watch.gen);
break;
case CEPH_OSD_OP_NOTIFY_ACK:
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
break;
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
case CEPH_OSD_OP_NOTIFY:
dst->notify.cookie = cpu_to_le64(src->notify.cookie);
break;
libceph: support for CEPH_OSD_OP_LIST_WATCHERS Add support for this Ceph OSD op, needed to support the RBD exclusive lock feature. Signed-off-by: Douglas Fuller <dfuller@redhat.com> [idryomov@gmail.com: refactor, misc fixes throughout] Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-07-17 20:18:07 +00:00
case CEPH_OSD_OP_LIST_WATCHERS:
break;
libceph: add support for CEPH_OSD_OP_SETALLOCHINT osd op This is primarily for rbd's benefit and is supposed to combat fragmentation: "... knowing that rbd images have a 4m size, librbd can pass a hint that will let the osd do the xfs allocation size ioctl on new files so that they are allocated in 1m or 4m chunks. We've seen cases where users with rbd workloads have very high levels of fragmentation in xfs and this would mitigate that and probably have a pretty nice performance benefit." SETALLOCHINT is considered advisory, so our backwards compatibility mechanism here is to set FAILOK flag for all SETALLOCHINT ops. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-02-25 14:22:27 +00:00
case CEPH_OSD_OP_SETALLOCHINT:
dst->alloc_hint.expected_object_size =
cpu_to_le64(src->alloc_hint.expected_object_size);
dst->alloc_hint.expected_write_size =
cpu_to_le64(src->alloc_hint.expected_write_size);
break;
libceph: add SETXATTR/CMPXATTR osd operations support Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-12 06:00:43 +00:00
case CEPH_OSD_OP_SETXATTR:
case CEPH_OSD_OP_CMPXATTR:
dst->xattr.name_len = cpu_to_le32(src->xattr.name_len);
dst->xattr.value_len = cpu_to_le32(src->xattr.value_len);
dst->xattr.cmp_op = src->xattr.cmp_op;
dst->xattr.cmp_mode = src->xattr.cmp_mode;
break;
libceph: add CREATE osd operation support Add CEPH_OSD_OP_CREATE support. Also change libceph to not treat CEPH_OSD_OP_DELETE as an extent op and add an assert to that end. Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 02:47:25 +00:00
case CEPH_OSD_OP_CREATE:
case CEPH_OSD_OP_DELETE:
break;
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
default:
libceph: report defined but unsupported osd ops If osd_req_encode_op() is given any opcode it doesn't recognize it reports an error. This patch fleshes out that routine to distinguish between well-defined but unsupported values and values that are simply bogus. This and the next commit are related to: http://tracker.ceph.com/issues/4126 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-15 17:42:30 +00:00
pr_err("unsupported osd opcode %s\n",
libceph: fix wrong opcode use in osd_req_encode_op() The new cases added to osd_req_encode_op() caused a new sparse error, which highlighted an existing problem that had been overlooked since it was originally checked in. When an unsupported opcode is found the destination rather than the source opcode was being used in the error message. The two differ in their byte order, and we want to be using the one in the source. Fix the problem in both spots. Reported-by: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-04 17:08:29 +00:00
ceph_osd_op_name(src->op));
libceph: report defined but unsupported osd ops If osd_req_encode_op() is given any opcode it doesn't recognize it reports an error. This patch fleshes out that routine to distinguish between well-defined but unsupported values and values that are simply bogus. This and the next commit are related to: http://tracker.ceph.com/issues/4126 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-15 17:42:30 +00:00
WARN_ON(1);
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
return 0;
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
}
libceph: encode CEPH_OSD_OP_FLAG_* op flags Encode ceph_osd_op::flags field so that it gets sent over the wire. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-02-25 14:22:26 +00:00
libceph: define osd_req_opcode_valid() Define a separate function to determine the validity of an opcode, and use it inside osd_req_encode_op() in order to unclutter that function. Don't update the destination op at all--and return zero--if an unsupported or unrecognized opcode is seen in osd_req_encode_op(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
dst->op = cpu_to_le16(src->op);
libceph: encode CEPH_OSD_OP_FLAG_* op flags Encode ceph_osd_op::flags field so that it gets sent over the wire. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-02-25 14:22:26 +00:00
dst->flags = cpu_to_le32(src->flags);
libceph: rename ceph_osd_req_op::payload_len to indata_len Follow userspace nomenclature on this - the next commit adds outdata_len. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-08 12:39:46 +00:00
dst->payload_len = cpu_to_le32(src->indata_len);
libceph: let osd ops determine request data length The length of outgoing data in an osd request is dependent on the osd ops that are embedded in that request. Each op is encoded into a request message using osd_req_encode_op(), so that should be used to determine the amount of outgoing data implied by the op as it is encoded. Have osd_req_encode_op() return the number of bytes of outgoing data implied by the op being encoded, and accumulate and use that in ceph_osdc_build_request(). As a result, ceph_osdc_build_request() no longer requires its "len" parameter, so get rid of it. Using the sum of the op lengths rather than the length provided is a valid change because: - The only callers of osd ceph_osdc_build_request() are rbd and the osd client (in ceph_osdc_new_request() on behalf of the file system). - When rbd calls it, the length provided is only non-zero for write requests, and in that case the single op has the same length value as what was passed here. - When called from ceph_osdc_new_request(), (it's not all that easy to see, but) the length passed is also always the same as the extent length encoded in its (single) write op if present. This resolves: http://tracker.ceph.com/issues/4406 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-08 19:35:36 +00:00
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
return src->indata_len;
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
}
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
/*
* build new request AND message, calculate layout, and adjust file
* extent as needed.
*
* if the file was recently truncated, we include information about its
* old and new size so that the object can be updated appropriately. (we
* avoid synchronously deleting truncated objects because it's slow.)
*/
struct ceph_osd_request *ceph_osdc_new_request(struct ceph_osd_client *osdc,
struct ceph_file_layout *layout,
struct ceph_vino vino,
libceph: specify position of extent operation allow specifying position of extent operation in multi-operations osd request. This is required for cephfs to convert inline data to normal data (compare xattr, then write object). Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 06:40:37 +00:00
u64 off, u64 *plen,
unsigned int which, int num_ops,
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
int opcode, int flags,
struct ceph_snap_context *snapc,
u32 truncate_seq,
u64 truncate_size,
libceph: don't assign page info in ceph_osdc_new_request() Currently ceph_osdc_new_request() assigns an osd request's r_num_pages and r_alignment fields. The only thing it does after that is call ceph_osdc_build_request(), and that doesn't need those fields to be assigned. Move the assignment of those fields out of ceph_osdc_new_request() and into its caller. As a result, the page_align parameter is no longer used, so get rid of it. Note that in ceph_sync_write(), the value for req->r_num_pages had already been calculated earlier (as num_pages, and fortunately it was computed the same way). So don't bother recomputing it, but because it's not needed earlier, move that calculation after the call to ceph_osdc_new_request(). Hold off making the assignment to r_alignment, doing it instead r_pages and r_num_pages are getting set. Similarly, in start_read(), nr_pages already holds the number of pages in the array (and is calculated the same way), so there's no need to recompute it. Move the assignment of the page alignment down with the others there as well. This and the next few patches are preparation work for: http://tracker.ceph.com/issues/4127 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-02 00:00:15 +00:00
bool use_mempool)
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
{
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
struct ceph_osd_request *req;
libceph: pass offset and length out of calc_layout() The purpose of calc_layout() is to determine, given a file offset and length and a layout describing the placement of file data across objects, where in "object space" that data resides. Specifically, it determines which object should hold the first part of the specified range of file data, and the offset and length of data within that object. The length will not exceed the bounds of the object, and the caller is informed of that maximum length. Add two parameters to calc_layout() to allow the object-relative offset and length to be passed back to the caller. This is the first steps toward having ceph_osdc_new_request() build its osd op structure using osd_req_op_extent_init(). Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:00 +00:00
u64 objnum = 0;
u64 objoff = 0;
u64 objlen = 0;
ceph: propagate layout error on osd request creation If we are creating an osd request and get an invalid layout, return an EINVAL to the caller. We switch up the return to have an error code instead of NULL implying -ENOMEM. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2012-09-25 04:01:02 +00:00
int r;
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
ceph: punch hole support This patch implements fallocate and punch hole support for Ceph kernel client. Signed-off-by: Li Wang <liwang@ubuntukylin.com> Signed-off-by: Yunchuan Wen <yunchuanwen@ubuntukylin.com>
2013-08-15 03:51:44 +00:00
BUG_ON(opcode != CEPH_OSD_OP_READ && opcode != CEPH_OSD_OP_WRITE &&
libceph: add CREATE osd operation support Add CEPH_OSD_OP_CREATE support. Also change libceph to not treat CEPH_OSD_OP_DELETE as an extent op and add an assert to that end. Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 02:47:25 +00:00
opcode != CEPH_OSD_OP_ZERO && opcode != CEPH_OSD_OP_TRUNCATE &&
opcode != CEPH_OSD_OP_CREATE && opcode != CEPH_OSD_OP_DELETE);
ceph: messenger and osdc changes for rbd Allow the messenger to send/receive data in a bio. This is added so that we wouldn't need to copy the data into pages or some other buffer when doing IO for an rbd block device. We can now have trailing variable sized data for osd ops. Also osd ops encoding is more modular. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:01:27 +00:00
libceph: don't build request in ceph_osdc_new_request() This patch moves the call to ceph_osdc_build_request() out of ceph_osdc_new_request() and into its caller. This is in order to defer formatting osd operation information into the request message until just before request is started. The only unusual (ab)user of ceph_osdc_build_request() is ceph_writepages_start(), where the final length of write request may change (downward) based on the current inode size or the oldest snapshot context with dirty data for the inode. The remaining callers don't change anything in the request after has been built. This means the ops array is now supplied by the caller. It also means there is no need to pass the mtime to ceph_osdc_new_request() (it gets provided to ceph_osdc_build_request()). And rather than passing a do_sync flag, have the number of ops in the ops array supplied imply adding a second STARTSYNC operation after the READ or WRITE requested. This and some of the patches that follow are related to having the messenger (only) be responsible for filling the content of the message header, as described here: http://tracker.ceph.com/issues/4589 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 19:09:05 +00:00
req = ceph_osdc_alloc_request(osdc, snapc, num_ops, use_mempool,
libceph: pass num_op with ops Both ceph_osdc_alloc_request() and ceph_osdc_build_request() are provided an array of ceph osd request operations. Rather than just passing the number of operations in the array, the caller is required append an additional zeroed operation structure to signal the end of the array. All callers know the number of operations at the time these functions are called, so drop the silly zero entry and supply that number directly. As a result, get_num_ops() is no longer needed. This also means that ceph_osdc_alloc_request() never uses its ops argument, so that can be dropped. Also rbd_create_rw_ops() no longer needs to add one to reserve room for the additional op. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2012-11-14 03:11:15 +00:00
GFP_NOFS);
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
if (!req) {
r = -ENOMEM;
goto fail;
}
libceph: keep source rather than message osd op array An osd request keeps a pointer to the osd operations (ops) array that it builds in its request message. In order to allow each op in the array to have its own distinct data, we will need to keep track of each op's data, and that information does not go over the wire. As long as we're tracking the data we might as well just track the entire (source) op definition for each of the ops. And if we're doing that, we'll have no more need to keep a pointer to the wire-encoded version. This patch makes the array of source ops be kept with the osd request structure, and uses that instead of the version encoded in the message in places where that was previously used. The array will be embedded in the request structure, and the maximum number of ops we ever actually use is currently 2. So reduce CEPH_OSD_MAX_OP to 2 to reduce the size of the structure. The result of doing this sort of ripples back up, and as a result various function parameters and local variables become unnecessary. Make r_num_ops be unsigned, and move the definition of struct ceph_osd_req_op earlier to ensure it's defined where needed. It does not yet add per-op data, that's coming soon. This resolves: http://tracker.ceph.com/issues/4656 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-04 02:32:51 +00:00
ceph: refactor osdc requests creation functions The osd requests creation are being decoupled from the vino parameter, allowing clients using the osd to use other arbitrary object names that are not necessarily vino based. Also, calc_raw_layout now takes a snap id. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 21:51:47 +00:00
/* calculate max write size */
libceph: don't update op in calc_layout() The ceph_osdc_new_request() an array of osd operations is built up and filled in partially within that function and partially in the called function calc_layout(). Move the latter part back out to ceph_osdc_new_request() so it's all done in one place. This makes it unnecessary to pass the op pointer to calc_layout(), so get rid of that parameter. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:01 +00:00
r = calc_layout(layout, off, plen, &objnum, &objoff, &objlen);
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
if (r)
goto fail;
libceph: don't update op in calc_layout() The ceph_osdc_new_request() an array of osd operations is built up and filled in partially within that function and partially in the called function calc_layout(). Move the latter part back out to ceph_osdc_new_request() so it's all done in one place. This makes it unnecessary to pass the op pointer to calc_layout(), so get rid of that parameter. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:01 +00:00
libceph: add CREATE osd operation support Add CEPH_OSD_OP_CREATE support. Also change libceph to not treat CEPH_OSD_OP_DELETE as an extent op and add an assert to that end. Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 02:47:25 +00:00
if (opcode == CEPH_OSD_OP_CREATE || opcode == CEPH_OSD_OP_DELETE) {
libceph: allow setting osd_req_op's flags Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-04-27 03:09:54 +00:00
osd_req_op_init(req, which, opcode, 0);
libceph: add CREATE osd operation support Add CEPH_OSD_OP_CREATE support. Also change libceph to not treat CEPH_OSD_OP_DELETE as an extent op and add an assert to that end. Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 02:47:25 +00:00
} else {
libceph: define new ceph_file_layout structure Define new ceph_file_layout structure and rename old ceph_file_layout to ceph_file_layout_legacy. This is preparation for adding namespace to ceph_file_layout structure. Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-02-03 13:24:49 +00:00
u32 object_size = layout->object_size;
libceph: add CREATE osd operation support Add CEPH_OSD_OP_CREATE support. Also change libceph to not treat CEPH_OSD_OP_DELETE as an extent op and add an assert to that end. Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 02:47:25 +00:00
u32 object_base = off - objoff;
if (!(truncate_seq == 1 && truncate_size == -1ULL)) {
if (truncate_size <= object_base) {
truncate_size = 0;
} else {
truncate_size -= object_base;
if (truncate_size > object_size)
truncate_size = object_size;
}
libceph: fix truncate size calculation check the "not truncated yet" case Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com> Reviewed-by: Sage Weil <sage@inktank.com>
2013-06-02 10:40:23 +00:00
}
libceph: specify position of extent operation allow specifying position of extent operation in multi-operations osd request. This is required for cephfs to convert inline data to normal data (compare xattr, then write object). Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 06:40:37 +00:00
osd_req_op_extent_init(req, which, opcode, objoff, objlen,
libceph: add CREATE osd operation support Add CEPH_OSD_OP_CREATE support. Also change libceph to not treat CEPH_OSD_OP_DELETE as an extent op and add an assert to that end. Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 02:47:25 +00:00
truncate_size, truncate_seq);
}
libceph: clean up ceph_osd_new_request() All callers of ceph_osd_new_request() pass either CEPH_OSD_OP_READ or CEPH_OSD_OP_WRITE as the opcode value. The function assumes it by filling in the extent fields in the ops array it builds. So just assert that is the case, and don't bother calling op_has_extent() before filling in the first osd operation in the array. Define some local variables to gather the information to fill into the first op, and then fill in the op array all in one place. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 01:50:01 +00:00
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
req->r_flags = flags;
libceph: define new ceph_file_layout structure Define new ceph_file_layout structure and rename old ceph_file_layout to ceph_file_layout_legacy. This is preparation for adding namespace to ceph_file_layout structure. Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-02-03 13:24:49 +00:00
req->r_base_oloc.pool = layout->pool_id;
libceph: rados pool namespace support Add pool namesapce pointer to struct ceph_file_layout and struct ceph_object_locator. Pool namespace is used by when mapping object to PG, it's also used when composing OSD request. The namespace pointer in struct ceph_file_layout is RCU protected. So libceph can read namespace without taking lock. Signed-off-by: Yan, Zheng <zyan@redhat.com> [idryomov@gmail.com: ceph_oloc_destroy(), misc minor changes] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-14 03:24:31 +00:00
req->r_base_oloc.pool_ns = ceph_try_get_string(layout->pool_ns);
libceph: variable-sized ceph_object_id Currently ceph_object_id can hold object names of up to 100 (CEPH_MAX_OID_NAME_LEN) characters. This is enough for all use cases, expect one - long rbd image names: - a format 1 header is named "<imgname>.rbd" - an object that points to a format 2 header is named "rbd_id.<imgname>" We operate on these potentially long-named objects during rbd map, and, for format 1 images, during header refresh. (A format 2 header name is a small system-generated string.) Lift this 100 character limit by making ceph_object_id be able to point to an externally-allocated string. Apart from being able to work with almost arbitrarily-long named objects, this allows us to reduce the size of ceph_object_id from >100 bytes to 64 bytes. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-29 17:54:20 +00:00
ceph_oid_printf(&req->r_base_oid, "%llx.%08llx", vino.ino, objnum);
libceph: format target object name in caller Move the formatting of the object name (oid) to use for an object request into the caller of calc_layout(). This makes the "vino" parameter no longer necessary, so get rid of it. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-16 04:10:17 +00:00
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
req->r_snapid = vino.snap;
if (flags & CEPH_OSD_FLAG_WRITE)
req->r_data_offset = off;
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
r = ceph_osdc_alloc_messages(req, GFP_NOFS);
if (r)
goto fail;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
return req;
libceph: move message allocation out of ceph_osdc_alloc_request() The size of ->r_request and ->r_reply messages depends on the size of the object name (ceph_object_id), while the size of ceph_osd_request is fixed. Move message allocation into a separate function that would have to be called after ceph_object_id and ceph_object_locator (which is also going to become variable in size with RADOS namespaces) have been filled in: req = ceph_osdc_alloc_request(...); <fill in req->r_base_oid> <fill in req->r_base_oloc> ceph_osdc_alloc_messages(req); Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 12:15:51 +00:00
fail:
ceph_osdc_put_request(req);
return ERR_PTR(r);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:14:15 +00:00
EXPORT_SYMBOL(ceph_osdc_new_request);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* We keep osd requests in an rbtree, sorted by ->r_tid.
*/
libceph: DEFINE_RB_FUNCS macro Given struct foo { u64 id; struct rb_node bar_node; }; generate insert_bar(), erase_bar() and lookup_bar() functions with DEFINE_RB_FUNCS(bar, struct foo, id, bar_node) The key is assumed to be an integer (u64, int, etc), compared with < and >. nodefld has to be initialized with RB_CLEAR_NODE(). Start using it for MDS, MON and OSD requests and OSD sessions. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
DEFINE_RB_FUNCS(request, struct ceph_osd_request, r_tid, r_node)
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
DEFINE_RB_FUNCS(request_mc, struct ceph_osd_request, r_tid, r_mc_node)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: introduce ceph_osdc_abort_requests() This will be used by the filesystem for "umount -f". Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-05-15 13:47:58 +00:00
/*
* Call @fn on each OSD request as long as @fn returns 0.
*/
static void for_each_request(struct ceph_osd_client *osdc,
int (*fn)(struct ceph_osd_request *req, void *arg),
void *arg)
{
struct rb_node *n, *p;
for (n = rb_first(&osdc->osds); n; n = rb_next(n)) {
struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node);
for (p = rb_first(&osd->o_requests); p; ) {
struct ceph_osd_request *req =
rb_entry(p, struct ceph_osd_request, r_node);
p = rb_next(p);
if (fn(req, arg))
return;
}
}
for (p = rb_first(&osdc->homeless_osd.o_requests); p; ) {
struct ceph_osd_request *req =
rb_entry(p, struct ceph_osd_request, r_node);
p = rb_next(p);
if (fn(req, arg))
return;
}
}
libceph: osd_init() and osd_cleanup() These are going to be used by homeless OSD sessions code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
static bool osd_homeless(struct ceph_osd *osd)
{
return osd->o_osd == CEPH_HOMELESS_OSD;
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static bool osd_registered(struct ceph_osd *osd)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
verify_osdc_locked(osd->o_osdc);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
return !RB_EMPTY_NODE(&osd->o_node);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: osd_init() and osd_cleanup() These are going to be used by homeless OSD sessions code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
/*
* Assumes @osd is zero-initialized.
*/
static void osd_init(struct ceph_osd *osd)
{
libceph: convert ceph_osd.o_ref from atomic_t to refcount_t refcount_t type and corresponding API should be used instead of atomic_t when the variable is used as a reference counter. This allows to avoid accidental refcounter overflows that might lead to use-after-free situations. Signed-off-by: Elena Reshetova <elena.reshetova@intel.com> Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com> Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David Windsor <dwindsor@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-17 12:10:28 +00:00
refcount_set(&osd->o_ref, 1);
libceph: osd_init() and osd_cleanup() These are going to be used by homeless OSD sessions code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
RB_CLEAR_NODE(&osd->o_node);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
osd->o_requests = RB_ROOT;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
osd->o_linger_requests = RB_ROOT;
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
osd->o_backoff_mappings = RB_ROOT;
osd->o_backoffs_by_id = RB_ROOT;
libceph: osd_init() and osd_cleanup() These are going to be used by homeless OSD sessions code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
INIT_LIST_HEAD(&osd->o_osd_lru);
INIT_LIST_HEAD(&osd->o_keepalive_item);
osd->o_incarnation = 1;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
mutex_init(&osd->lock);
libceph: osd_init() and osd_cleanup() These are going to be used by homeless OSD sessions code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
}
static void osd_cleanup(struct ceph_osd *osd)
{
WARN_ON(!RB_EMPTY_NODE(&osd->o_node));
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
WARN_ON(!RB_EMPTY_ROOT(&osd->o_requests));
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
WARN_ON(!RB_EMPTY_ROOT(&osd->o_linger_requests));
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
WARN_ON(!RB_EMPTY_ROOT(&osd->o_backoff_mappings));
WARN_ON(!RB_EMPTY_ROOT(&osd->o_backoffs_by_id));
libceph: osd_init() and osd_cleanup() These are going to be used by homeless OSD sessions code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
WARN_ON(!list_empty(&osd->o_osd_lru));
WARN_ON(!list_empty(&osd->o_keepalive_item));
if (osd->o_auth.authorizer) {
WARN_ON(osd_homeless(osd));
ceph_auth_destroy_authorizer(osd->o_auth.authorizer);
}
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* Track open sessions with osds.
*/
libceph: provide osd number when creating osd Pass the osd number to the create_osd() routine, and move the initialization of fields that depend on it therein. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-27 04:26:43 +00:00
static struct ceph_osd *create_osd(struct ceph_osd_client *osdc, int onum)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
struct ceph_osd *osd;
libceph: osd_init() and osd_cleanup() These are going to be used by homeless OSD sessions code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
WARN_ON(onum == CEPH_HOMELESS_OSD);
libceph: allocate ceph_osd with GFP_NOFAIL create_osd() is called way too deep in the stack to be able to error out in a sane way; a failing create_osd() just messes everything up. The current req_notarget list solution is broken - the list is never traversed as it's not entirely clear when to do it, I guess. If we were to start traversing it at regular intervals and retrying each request, we wouldn't be far off from what __GFP_NOFAIL is doing, so allocate OSD sessions with __GFP_NOFAIL, at least until we come up with a better fix. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
osd = kzalloc(sizeof(*osd), GFP_NOIO | __GFP_NOFAIL);
libceph: osd_init() and osd_cleanup() These are going to be used by homeless OSD sessions code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
osd_init(osd);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
osd->o_osdc = osdc;
libceph: provide osd number when creating osd Pass the osd number to the create_osd() routine, and move the initialization of fields that depend on it therein. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-27 04:26:43 +00:00
osd->o_osd = onum;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: set peer name on con_open, not init The peer name may change on each open attempt, even when the connection is reused. Signed-off-by: Sage Weil <sage@inktank.com>
2012-06-27 19:24:08 +00:00
ceph_con_init(&osd->o_con, osd, &osd_con_ops, &osdc->client->msgr);
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
return osd;
}
static struct ceph_osd *get_osd(struct ceph_osd *osd)
{
libceph: convert ceph_osd.o_ref from atomic_t to refcount_t refcount_t type and corresponding API should be used instead of atomic_t when the variable is used as a reference counter. This allows to avoid accidental refcounter overflows that might lead to use-after-free situations. Signed-off-by: Elena Reshetova <elena.reshetova@intel.com> Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com> Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David Windsor <dwindsor@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-17 12:10:28 +00:00
if (refcount_inc_not_zero(&osd->o_ref)) {
dout("get_osd %p %d -> %d\n", osd, refcount_read(&osd->o_ref)-1,
refcount_read(&osd->o_ref));
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
return osd;
} else {
dout("get_osd %p FAIL\n", osd);
return NULL;
}
}
static void put_osd(struct ceph_osd *osd)
{
libceph: convert ceph_osd.o_ref from atomic_t to refcount_t refcount_t type and corresponding API should be used instead of atomic_t when the variable is used as a reference counter. This allows to avoid accidental refcounter overflows that might lead to use-after-free situations. Signed-off-by: Elena Reshetova <elena.reshetova@intel.com> Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com> Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David Windsor <dwindsor@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-17 12:10:28 +00:00
dout("put_osd %p %d -> %d\n", osd, refcount_read(&osd->o_ref),
refcount_read(&osd->o_ref) - 1);
if (refcount_dec_and_test(&osd->o_ref)) {
libceph: osd_init() and osd_cleanup() These are going to be used by homeless OSD sessions code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
osd_cleanup(osd);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
kfree(osd);
ceph: fix leak of osd authorizer Release the ceph_authorizer when releasing osd state. Signed-off-by: Sage Weil <sage@newdream.net>
2010-05-27 21:15:49 +00:00
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: DEFINE_RB_FUNCS macro Given struct foo { u64 id; struct rb_node bar_node; }; generate insert_bar(), erase_bar() and lookup_bar() functions with DEFINE_RB_FUNCS(bar, struct foo, id, bar_node) The key is assumed to be an integer (u64, int, etc), compared with < and >. nodefld has to be initialized with RB_CLEAR_NODE(). Start using it for MDS, MON and OSD requests and OSD sessions. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
DEFINE_RB_FUNCS(osd, struct ceph_osd, o_osd, o_node)
libceph: protect osdc->osd_lru list with a spinlock OSD client is getting moved from the big per-client lock to a set of per-session locks. The big rwlock would only be held for read most of the time, so a global osdc->osd_lru needs additional protection. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void __move_osd_to_lru(struct ceph_osd *osd)
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
{
libceph: protect osdc->osd_lru list with a spinlock OSD client is getting moved from the big per-client lock to a set of per-session locks. The big rwlock would only be held for read most of the time, so a global osdc->osd_lru needs additional protection. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd_client *osdc = osd->o_osdc;
dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd);
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
BUG_ON(!list_empty(&osd->o_osd_lru));
libceph: add maybe_move_osd_to_lru() and switch to it Abstract out __move_osd_to_lru() logic from __unregister_request() and __unregister_linger_request(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:41 +00:00
libceph: protect osdc->osd_lru list with a spinlock OSD client is getting moved from the big per-client lock to a set of per-session locks. The big rwlock would only be held for read most of the time, so a global osdc->osd_lru needs additional protection. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
spin_lock(&osdc->osd_lru_lock);
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
list_add_tail(&osd->o_osd_lru, &osdc->osd_lru);
libceph: protect osdc->osd_lru list with a spinlock OSD client is getting moved from the big per-client lock to a set of per-session locks. The big rwlock would only be held for read most of the time, so a global osdc->osd_lru needs additional protection. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
spin_unlock(&osdc->osd_lru_lock);
libceph: store timeouts in jiffies, verify user input There are currently three libceph-level timeouts that the user can specify on mount: mount_timeout, osd_idle_ttl and osdkeepalive. All of these are in seconds and no checking is done on user input: negative values are accepted, we multiply them all by HZ which may or may not overflow, arbitrarily large jiffies then get added together, etc. There is also a bug in the way mount_timeout=0 is handled. It's supposed to mean "infinite timeout", but that's not how wait.h APIs treat it and so __ceph_open_session() for example will busy loop without much chance of being interrupted if none of ceph-mons are there. Fix all this by verifying user input, storing timeouts capped by msecs_to_jiffies() in jiffies and using the new ceph_timeout_jiffies() helper for all user-specified waits to handle infinite timeouts correctly. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-05-15 09:02:17 +00:00
osd->lru_ttl = jiffies + osdc->client->options->osd_idle_ttl;
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
}
libceph: protect osdc->osd_lru list with a spinlock OSD client is getting moved from the big per-client lock to a set of per-session locks. The big rwlock would only be held for read most of the time, so a global osdc->osd_lru needs additional protection. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void maybe_move_osd_to_lru(struct ceph_osd *osd)
libceph: add maybe_move_osd_to_lru() and switch to it Abstract out __move_osd_to_lru() logic from __unregister_request() and __unregister_linger_request(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:41 +00:00
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
if (RB_EMPTY_ROOT(&osd->o_requests) &&
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
RB_EMPTY_ROOT(&osd->o_linger_requests))
libceph: protect osdc->osd_lru list with a spinlock OSD client is getting moved from the big per-client lock to a set of per-session locks. The big rwlock would only be held for read most of the time, so a global osdc->osd_lru needs additional protection. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
__move_osd_to_lru(osd);
libceph: add maybe_move_osd_to_lru() and switch to it Abstract out __move_osd_to_lru() logic from __unregister_request() and __unregister_linger_request(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-20 10:14:41 +00:00
}
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
static void __remove_osd_from_lru(struct ceph_osd *osd)
{
libceph: protect osdc->osd_lru list with a spinlock OSD client is getting moved from the big per-client lock to a set of per-session locks. The big rwlock would only be held for read most of the time, so a global osdc->osd_lru needs additional protection. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd_client *osdc = osd->o_osdc;
dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd);
spin_lock(&osdc->osd_lru_lock);
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
if (!list_empty(&osd->o_osd_lru))
list_del_init(&osd->o_osd_lru);
libceph: protect osdc->osd_lru list with a spinlock OSD client is getting moved from the big per-client lock to a set of per-session locks. The big rwlock would only be held for read most of the time, so a global osdc->osd_lru needs additional protection. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
spin_unlock(&osdc->osd_lru_lock);
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
/*
* Close the connection and assign any leftover requests to the
* homeless session.
*/
static void close_osd(struct ceph_osd *osd)
{
struct ceph_osd_client *osdc = osd->o_osdc;
struct rb_node *n;
verify_osdc_wrlocked(osdc);
dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd);
ceph_con_close(&osd->o_con);
for (n = rb_first(&osd->o_requests); n; ) {
struct ceph_osd_request *req =
rb_entry(n, struct ceph_osd_request, r_node);
n = rb_next(n); /* unlink_request() */
dout(" reassigning req %p tid %llu\n", req, req->r_tid);
unlink_request(osd, req);
link_request(&osdc->homeless_osd, req);
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
for (n = rb_first(&osd->o_linger_requests); n; ) {
struct ceph_osd_linger_request *lreq =
rb_entry(n, struct ceph_osd_linger_request, node);
n = rb_next(n); /* unlink_linger() */
dout(" reassigning lreq %p linger_id %llu\n", lreq,
lreq->linger_id);
unlink_linger(osd, lreq);
link_linger(&osdc->homeless_osd, lreq);
}
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
clear_backoffs(osd);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
__remove_osd_from_lru(osd);
erase_osd(&osdc->osds, osd);
put_osd(osd);
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* reset osd connect
*/
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static int reopen_osd(struct ceph_osd *osd)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: reformat __reset_osd() Reformat __reset_osd() into three distinct blocks of code handling the three return cases. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2012-12-07 15:57:58 +00:00
struct ceph_entity_addr *peer_addr;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd);
if (RB_EMPTY_ROOT(&osd->o_requests) &&
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
RB_EMPTY_ROOT(&osd->o_linger_requests)) {
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
close_osd(osd);
libceph: reformat __reset_osd() Reformat __reset_osd() into three distinct blocks of code handling the three return cases. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2012-12-07 15:57:58 +00:00
return -ENODEV;
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
peer_addr = &osd->o_osdc->osdmap->osd_addr[osd->o_osd];
libceph: reformat __reset_osd() Reformat __reset_osd() into three distinct blocks of code handling the three return cases. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2012-12-07 15:57:58 +00:00
if (!memcmp(peer_addr, &osd->o_con.peer_addr, sizeof (*peer_addr)) &&
!ceph_con_opened(&osd->o_con)) {
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct rb_node *n;
libceph: reformat __reset_osd() Reformat __reset_osd() into three distinct blocks of code handling the three return cases. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2012-12-07 15:57:58 +00:00
libceph: dout() is missing a newline Add a missing newline to a dout() in __reset_osd(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com>
2014-01-16 17:18:27 +00:00
dout("osd addr hasn't changed and connection never opened, "
"letting msgr retry\n");
ceph: avoid reopening osd connections when address hasn't changed We get a fault callback on _every_ tcp connection fault. Normally, we want to reopen the connection when that happens. If the address we have is bad, however, and connection attempts always result in a connection refused or similar error, explicitly closing and reopening the msgr connection just prevents the messenger's backoff logic from kicking in. The result can be a console full of [ 3974.417106] ceph: osd11 10.3.14.138:6800 connection failed [ 3974.423295] ceph: osd11 10.3.14.138:6800 connection failed [ 3974.429709] ceph: osd11 10.3.14.138:6800 connection failed Instead, if we get a fault, and have outstanding requests, but the osd address hasn't changed and the connection never successfully connected in the first place, do nothing to the osd connection. The messenger layer will back off and retry periodically, because we never connected and thus the lossy bit is not set. Instead, touch each request's r_stamp so that handle_timeout can tell the request is still alive and kicking. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-22 21:51:18 +00:00
/* touch each r_stamp for handle_timeout()'s benfit */
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
for (n = rb_first(&osd->o_requests); n; n = rb_next(n)) {
struct ceph_osd_request *req =
rb_entry(n, struct ceph_osd_request, r_node);
ceph: avoid reopening osd connections when address hasn't changed We get a fault callback on _every_ tcp connection fault. Normally, we want to reopen the connection when that happens. If the address we have is bad, however, and connection attempts always result in a connection refused or similar error, explicitly closing and reopening the msgr connection just prevents the messenger's backoff logic from kicking in. The result can be a console full of [ 3974.417106] ceph: osd11 10.3.14.138:6800 connection failed [ 3974.423295] ceph: osd11 10.3.14.138:6800 connection failed [ 3974.429709] ceph: osd11 10.3.14.138:6800 connection failed Instead, if we get a fault, and have outstanding requests, but the osd address hasn't changed and the connection never successfully connected in the first place, do nothing to the osd connection. The messenger layer will back off and retry periodically, because we never connected and thus the lossy bit is not set. Instead, touch each request's r_stamp so that handle_timeout can tell the request is still alive and kicking. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-22 21:51:18 +00:00
req->r_stamp = jiffies;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
}
libceph: reformat __reset_osd() Reformat __reset_osd() into three distinct blocks of code handling the three return cases. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2012-12-07 15:57:58 +00:00
return -EAGAIN;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: reformat __reset_osd() Reformat __reset_osd() into three distinct blocks of code handling the three return cases. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2012-12-07 15:57:58 +00:00
ceph_con_close(&osd->o_con);
ceph_con_open(&osd->o_con, CEPH_ENTITY_TYPE_OSD, osd->o_osd, peer_addr);
osd->o_incarnation++;
return 0;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static struct ceph_osd *lookup_create_osd(struct ceph_osd_client *osdc, int o,
bool wrlocked)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd *osd;
libceph: avoid unregistering osd request when not registered There is a race between two __unregister_request() callers: the reply path and the ceph_osdc_wait_request(). If we get a reply *and* the timeout expires at roughly the same time, both callers will try to unregister the request, and the second one will do bad things. Simply check if the request is still already unregistered; if so, return immediately and do nothing. Fixes http://tracker.newdream.net/issues/2420 Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2012-05-16 20:16:38 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
if (wrlocked)
verify_osdc_wrlocked(osdc);
else
verify_osdc_locked(osdc);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
if (o != CEPH_HOMELESS_OSD)
osd = lookup_osd(&osdc->osds, o);
else
osd = &osdc->homeless_osd;
if (!osd) {
if (!wrlocked)
return ERR_PTR(-EAGAIN);
ceph: revoke osd request message on request completion If an osd has failed or returned and a request has been sent twice, it's possible to get a reply and unregister the request while the request message is queued for delivery. Since the message references the caller's page vector, we need to revoke it before completing. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-08 23:57:16 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
osd = create_osd(osdc, o);
insert_osd(&osdc->osds, osd);
ceph_con_open(&osd->o_con, CEPH_ENTITY_TYPE_OSD, osd->o_osd,
&osdc->osdmap->osd_addr[osd->o_osd]);
ceph: revoke osd request message on request completion If an osd has failed or returned and a request has been sent twice, it's possible to get a reply and unregister the request while the request message is queued for delivery. Since the message references the caller's page vector, we need to revoke it before completing. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-08 23:57:16 +00:00
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("%s osdc %p osd%d -> osd %p\n", __func__, osdc, o, osd);
return osd;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
/*
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
* Create request <-> OSD session relation.
*
* @req has to be assigned a tid, @osd may be homeless.
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
*/
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void link_request(struct ceph_osd *osd, struct ceph_osd_request *req)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
verify_osd_locked(osd);
WARN_ON(!req->r_tid || req->r_osd);
dout("%s osd %p osd%d req %p tid %llu\n", __func__, osd, osd->o_osd,
req, req->r_tid);
if (!osd_homeless(osd))
__remove_osd_from_lru(osd);
else
atomic_inc(&osd->o_osdc->num_homeless);
get_osd(osd);
insert_request(&osd->o_requests, req);
req->r_osd = osd;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void unlink_request(struct ceph_osd *osd, struct ceph_osd_request *req)
{
verify_osd_locked(osd);
WARN_ON(req->r_osd != osd);
dout("%s osd %p osd%d req %p tid %llu\n", __func__, osd, osd->o_osd,
req, req->r_tid);
req->r_osd = NULL;
erase_request(&osd->o_requests, req);
put_osd(osd);
if (!osd_homeless(osd))
maybe_move_osd_to_lru(osd);
else
atomic_dec(&osd->o_osdc->num_homeless);
}
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
static bool __pool_full(struct ceph_pg_pool_info *pi)
{
return pi->flags & CEPH_POOL_FLAG_FULL;
}
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
static bool have_pool_full(struct ceph_osd_client *osdc)
{
struct rb_node *n;
for (n = rb_first(&osdc->osdmap->pg_pools); n; n = rb_next(n)) {
struct ceph_pg_pool_info *pi =
rb_entry(n, struct ceph_pg_pool_info, node);
if (__pool_full(pi))
return true;
}
return false;
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static bool pool_full(struct ceph_osd_client *osdc, s64 pool_id)
{
struct ceph_pg_pool_info *pi;
pi = ceph_pg_pool_by_id(osdc->osdmap, pool_id);
if (!pi)
return false;
return __pool_full(pi);
}
libceph: block I/O when PAUSE or FULL osd map flags are set The PAUSEWR and PAUSERD flags are meant to stop the cluster from processing writes and reads, respectively. The FULL flag is set when the cluster determines that it is out of space, and will no longer process writes. PAUSEWR and PAUSERD are purely client-side settings already implemented in userspace clients. The osd does nothing special with these flags. When the FULL flag is set, however, the osd responds to all writes with -ENOSPC. For cephfs, this makes sense, but for rbd the block layer translates this into EIO. If a cluster goes from full to non-full quickly, a filesystem on top of rbd will not behave well, since some writes succeed while others get EIO. Fix this by blocking any writes when the FULL flag is set in the osd client. This is the same strategy used by userspace, so apply it by default. A follow-on patch makes this configurable. __map_request() is called to re-target osd requests in case the available osds changed. Add a paused field to a ceph_osd_request, and set it whenever an appropriate osd map flag is set. Avoid queueing paused requests in __map_request(), but force them to be resent if they become unpaused. Also subscribe to the next osd map from the monitor if any of these flags are set, so paused requests can be unblocked as soon as possible. Fixes: http://tracker.ceph.com/issues/6079 Reviewed-by: Sage Weil <sage@inktank.com> Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
2013-12-03 03:11:48 +00:00
/*
* Returns whether a request should be blocked from being sent
* based on the current osdmap and osd_client settings.
*/
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
static bool target_should_be_paused(struct ceph_osd_client *osdc,
const struct ceph_osd_request_target *t,
struct ceph_pg_pool_info *pi)
{
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
bool pauserd = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD);
bool pausewr = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR) ||
ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
__pool_full(pi);
libceph: use target pi for calc_target() calculations For luminous and beyond we are encoding the actual spgid, which requires operating with the correct pg_num, i.e. that of the target pool. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
WARN_ON(pi->id != t->target_oloc.pool);
libceph: add an epoch_barrier field to struct ceph_osd_client Cephfs can get cap update requests that contain a new epoch barrier in them. When that happens we want to pause all OSD traffic until the right map epoch arrives. Add an epoch_barrier field to ceph_osd_client that is protected by the osdc->lock rwsem. When the barrier is set, and the current OSD map epoch is below that, pause the request target when submitting the request or when revisiting it. Add a way for upper layers (cephfs) to update the epoch_barrier as well. If we get a new map, compare the new epoch against the barrier before kicking requests and request another map if the map epoch is still lower than the one we want. If we get a map with a full pool, or at quota condition, then set the barrier to the current epoch value. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-18 13:21:16 +00:00
return ((t->flags & CEPH_OSD_FLAG_READ) && pauserd) ||
((t->flags & CEPH_OSD_FLAG_WRITE) && pausewr) ||
(osdc->osdmap->epoch < osdc->epoch_barrier);
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
}
enum calc_target_result {
CALC_TARGET_NO_ACTION = 0,
CALC_TARGET_NEED_RESEND,
CALC_TARGET_POOL_DNE,
};
static enum calc_target_result calc_target(struct ceph_osd_client *osdc,
struct ceph_osd_request_target *t,
libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
struct ceph_connection *con,
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
bool any_change)
{
struct ceph_pg_pool_info *pi;
struct ceph_pg pgid, last_pgid;
struct ceph_osds up, acting;
bool force_resend = false;
libceph: drop need_resend from calc_target() Replace it with more fine-grained bools to separate updating ceph_osd_request_target fields and the decision to resend. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
bool unpaused = false;
bool legacy_change;
libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
bool split = false;
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
bool sort_bitwise = ceph_osdmap_flag(osdc, CEPH_OSDMAP_SORTBITWISE);
libceph: make RECOVERY_DELETES feature create a new interval This is needed so that the OSDs can regenerate the missing set at the start of a new interval where support for recovery deletes changed. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-07-27 15:59:14 +00:00
bool recovery_deletes = ceph_osdmap_flag(osdc,
CEPH_OSDMAP_RECOVERY_DELETES);
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
enum calc_target_result ct_res;
libceph: make sure need_resend targets reflect latest map Otherwise we may miss events like PG splits, pool deletions, etc when we get multiple incremental maps at once. Because check_pool_dne() can now be fed an unlinked request, finish_request() needed to be taught to handle unlinked requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
t->epoch = osdc->osdmap->epoch;
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
pi = ceph_pg_pool_by_id(osdc->osdmap, t->base_oloc.pool);
if (!pi) {
t->osd = CEPH_HOMELESS_OSD;
ct_res = CALC_TARGET_POOL_DNE;
goto out;
}
if (osdc->osdmap->epoch == pi->last_force_request_resend) {
libceph: fold [l]req->last_force_resend into ceph_osd_request_target Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-05 12:45:00 +00:00
if (t->last_force_resend < pi->last_force_request_resend) {
t->last_force_resend = pi->last_force_request_resend;
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
force_resend = true;
libceph: fold [l]req->last_force_resend into ceph_osd_request_target Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-05 12:45:00 +00:00
} else if (t->last_force_resend == 0) {
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
force_resend = true;
}
}
libceph: always populate t->target_{oid,oloc} in calc_target() need_check_tiering logic doesn't make a whole lot of sense. Drop it and apply tiering unconditionally on every calc_target() call instead. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
/* apply tiering */
ceph_oid_copy(&t->target_oid, &t->base_oid);
ceph_oloc_copy(&t->target_oloc, &t->base_oloc);
if ((t->flags & CEPH_OSD_FLAG_IGNORE_OVERLAY) == 0) {
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
if (t->flags & CEPH_OSD_FLAG_READ && pi->read_tier >= 0)
t->target_oloc.pool = pi->read_tier;
if (t->flags & CEPH_OSD_FLAG_WRITE && pi->write_tier >= 0)
t->target_oloc.pool = pi->write_tier;
libceph: use target pi for calc_target() calculations For luminous and beyond we are encoding the actual spgid, which requires operating with the correct pg_num, i.e. that of the target pool. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
pi = ceph_pg_pool_by_id(osdc->osdmap, t->target_oloc.pool);
if (!pi) {
t->osd = CEPH_HOMELESS_OSD;
ct_res = CALC_TARGET_POOL_DNE;
goto out;
}
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
}
libceph: allocate the locator string with GFP_NOFAIL calc_target() isn't supposed to fail with anything but POOL_DNE, in which case we report that the pool doesn't exist and fail the request with -ENOENT. Doing this for -ENOMEM is at the very least confusing and also harmful -- as the preceding requests complete, a short-lived locator string allocation is likely to succeed after a wait. (We used to call ceph_object_locator_to_pg() for a pi lookup. In theory that could fail with -ENOENT, hence the "ret != -ENOENT" warning being removed.) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-05-23 12:46:53 +00:00
__ceph_object_locator_to_pg(pi, &t->target_oid, &t->target_oloc, &pgid);
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
last_pgid.pool = pgid.pool;
last_pgid.seed = ceph_stable_mod(pgid.seed, t->pg_num, t->pg_num_mask);
libceph: avoid unnecessary pi lookups in calc_target() Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:56 +00:00
ceph_pg_to_up_acting_osds(osdc->osdmap, pi, &pgid, &up, &acting);
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
if (any_change &&
ceph_is_new_interval(&t->acting,
&acting,
&t->up,
&up,
t->size,
pi->size,
t->min_size,
pi->min_size,
t->pg_num,
pi->pg_num,
t->sort_bitwise,
sort_bitwise,
libceph: make RECOVERY_DELETES feature create a new interval This is needed so that the OSDs can regenerate the missing set at the start of a new interval where support for recovery deletes changed. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-07-27 15:59:14 +00:00
t->recovery_deletes,
recovery_deletes,
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
&last_pgid))
force_resend = true;
if (t->paused && !target_should_be_paused(osdc, t, pi)) {
t->paused = false;
libceph: drop need_resend from calc_target() Replace it with more fine-grained bools to separate updating ceph_osd_request_target fields and the decision to resend. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
unpaused = true;
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
}
libceph: drop need_resend from calc_target() Replace it with more fine-grained bools to separate updating ceph_osd_request_target fields and the decision to resend. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
legacy_change = ceph_pg_compare(&t->pgid, &pgid) ||
ceph_osds_changed(&t->acting, &acting, any_change);
libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
if (t->pg_num)
split = ceph_pg_is_split(&last_pgid, t->pg_num, pi->pg_num);
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
if (legacy_change || force_resend || split) {
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
t->pgid = pgid; /* struct */
libceph: avoid unnecessary pi lookups in calc_target() Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:56 +00:00
ceph_pg_to_primary_shard(osdc->osdmap, pi, &pgid, &t->spgid);
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
ceph_osds_copy(&t->acting, &acting);
ceph_osds_copy(&t->up, &up);
t->size = pi->size;
t->min_size = pi->min_size;
t->pg_num = pi->pg_num;
t->pg_num_mask = pi->pg_num_mask;
t->sort_bitwise = sort_bitwise;
libceph: make RECOVERY_DELETES feature create a new interval This is needed so that the OSDs can regenerate the missing set at the start of a new interval where support for recovery deletes changed. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-07-27 15:59:14 +00:00
t->recovery_deletes = recovery_deletes;
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
t->osd = acting.primary;
}
libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
if (unpaused || legacy_change || force_resend ||
(split && con && CEPH_HAVE_FEATURE(con->peer_features,
RESEND_ON_SPLIT)))
libceph: drop need_resend from calc_target() Replace it with more fine-grained bools to separate updating ceph_osd_request_target fields and the decision to resend. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
ct_res = CALC_TARGET_NEED_RESEND;
else
ct_res = CALC_TARGET_NO_ACTION;
libceph: introduce ceph_osd_request_target, calc_target() Introduce ceph_osd_request_target, containing all mapping-related fields of ceph_osd_request and calc_target() for calculating mappings and populating it. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:23 +00:00
out:
dout("%s t %p -> ct_res %d osd %d\n", __func__, t, ct_res, t->osd);
return ct_res;
}
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
static struct ceph_spg_mapping *alloc_spg_mapping(void)
{
struct ceph_spg_mapping *spg;
spg = kmalloc(sizeof(*spg), GFP_NOIO);
if (!spg)
return NULL;
RB_CLEAR_NODE(&spg->node);
spg->backoffs = RB_ROOT;
return spg;
}
static void free_spg_mapping(struct ceph_spg_mapping *spg)
{
WARN_ON(!RB_EMPTY_NODE(&spg->node));
WARN_ON(!RB_EMPTY_ROOT(&spg->backoffs));
kfree(spg);
}
/*
* rbtree of ceph_spg_mapping for handling map<spg_t, ...>, similar to
* ceph_pg_mapping. Used to track OSD backoffs -- a backoff [range] is
* defined only within a specific spgid; it does not pass anything to
* children on split, or to another primary.
*/
DEFINE_RB_FUNCS2(spg_mapping, struct ceph_spg_mapping, spgid, ceph_spg_compare,
RB_BYPTR, const struct ceph_spg *, node)
static u64 hoid_get_bitwise_key(const struct ceph_hobject_id *hoid)
{
return hoid->is_max ? 0x100000000ull : hoid->hash_reverse_bits;
}
static void hoid_get_effective_key(const struct ceph_hobject_id *hoid,
void **pkey, size_t *pkey_len)
{
if (hoid->key_len) {
*pkey = hoid->key;
*pkey_len = hoid->key_len;
} else {
*pkey = hoid->oid;
*pkey_len = hoid->oid_len;
}
}
static int compare_names(const void *name1, size_t name1_len,
const void *name2, size_t name2_len)
{
int ret;
ret = memcmp(name1, name2, min(name1_len, name2_len));
if (!ret) {
if (name1_len < name2_len)
ret = -1;
else if (name1_len > name2_len)
ret = 1;
}
return ret;
}
static int hoid_compare(const struct ceph_hobject_id *lhs,
const struct ceph_hobject_id *rhs)
{
void *effective_key1, *effective_key2;
size_t effective_key1_len, effective_key2_len;
int ret;
if (lhs->is_max < rhs->is_max)
return -1;
if (lhs->is_max > rhs->is_max)
return 1;
if (lhs->pool < rhs->pool)
return -1;
if (lhs->pool > rhs->pool)
return 1;
if (hoid_get_bitwise_key(lhs) < hoid_get_bitwise_key(rhs))
return -1;
if (hoid_get_bitwise_key(lhs) > hoid_get_bitwise_key(rhs))
return 1;
ret = compare_names(lhs->nspace, lhs->nspace_len,
rhs->nspace, rhs->nspace_len);
if (ret)
return ret;
hoid_get_effective_key(lhs, &effective_key1, &effective_key1_len);
hoid_get_effective_key(rhs, &effective_key2, &effective_key2_len);
ret = compare_names(effective_key1, effective_key1_len,
effective_key2, effective_key2_len);
if (ret)
return ret;
ret = compare_names(lhs->oid, lhs->oid_len, rhs->oid, rhs->oid_len);
if (ret)
return ret;
if (lhs->snapid < rhs->snapid)
return -1;
if (lhs->snapid > rhs->snapid)
return 1;
return 0;
}
/*
* For decoding ->begin and ->end of MOSDBackoff only -- no MIN/MAX
* compat stuff here.
*
* Assumes @hoid is zero-initialized.
*/
static int decode_hoid(void **p, void *end, struct ceph_hobject_id *hoid)
{
u8 struct_v;
u32 struct_len;
int ret;
ret = ceph_start_decoding(p, end, 4, "hobject_t", &struct_v,
&struct_len);
if (ret)
return ret;
if (struct_v < 4) {
pr_err("got struct_v %d < 4 of hobject_t\n", struct_v);
goto e_inval;
}
hoid->key = ceph_extract_encoded_string(p, end, &hoid->key_len,
GFP_NOIO);
if (IS_ERR(hoid->key)) {
ret = PTR_ERR(hoid->key);
hoid->key = NULL;
return ret;
}
hoid->oid = ceph_extract_encoded_string(p, end, &hoid->oid_len,
GFP_NOIO);
if (IS_ERR(hoid->oid)) {
ret = PTR_ERR(hoid->oid);
hoid->oid = NULL;
return ret;
}
ceph_decode_64_safe(p, end, hoid->snapid, e_inval);
ceph_decode_32_safe(p, end, hoid->hash, e_inval);
ceph_decode_8_safe(p, end, hoid->is_max, e_inval);
hoid->nspace = ceph_extract_encoded_string(p, end, &hoid->nspace_len,
GFP_NOIO);
if (IS_ERR(hoid->nspace)) {
ret = PTR_ERR(hoid->nspace);
hoid->nspace = NULL;
return ret;
}
ceph_decode_64_safe(p, end, hoid->pool, e_inval);
ceph_hoid_build_hash_cache(hoid);
return 0;
e_inval:
return -EINVAL;
}
static int hoid_encoding_size(const struct ceph_hobject_id *hoid)
{
return 8 + 4 + 1 + 8 + /* snapid, hash, is_max, pool */
4 + hoid->key_len + 4 + hoid->oid_len + 4 + hoid->nspace_len;
}
static void encode_hoid(void **p, void *end, const struct ceph_hobject_id *hoid)
{
ceph_start_encoding(p, 4, 3, hoid_encoding_size(hoid));
ceph_encode_string(p, end, hoid->key, hoid->key_len);
ceph_encode_string(p, end, hoid->oid, hoid->oid_len);
ceph_encode_64(p, hoid->snapid);
ceph_encode_32(p, hoid->hash);
ceph_encode_8(p, hoid->is_max);
ceph_encode_string(p, end, hoid->nspace, hoid->nspace_len);
ceph_encode_64(p, hoid->pool);
}
static void free_hoid(struct ceph_hobject_id *hoid)
{
if (hoid) {
kfree(hoid->key);
kfree(hoid->oid);
kfree(hoid->nspace);
kfree(hoid);
}
}
static struct ceph_osd_backoff *alloc_backoff(void)
{
struct ceph_osd_backoff *backoff;
backoff = kzalloc(sizeof(*backoff), GFP_NOIO);
if (!backoff)
return NULL;
RB_CLEAR_NODE(&backoff->spg_node);
RB_CLEAR_NODE(&backoff->id_node);
return backoff;
}
static void free_backoff(struct ceph_osd_backoff *backoff)
{
WARN_ON(!RB_EMPTY_NODE(&backoff->spg_node));
WARN_ON(!RB_EMPTY_NODE(&backoff->id_node));
free_hoid(backoff->begin);
free_hoid(backoff->end);
kfree(backoff);
}
/*
* Within a specific spgid, backoffs are managed by ->begin hoid.
*/
DEFINE_RB_INSDEL_FUNCS2(backoff, struct ceph_osd_backoff, begin, hoid_compare,
RB_BYVAL, spg_node);
static struct ceph_osd_backoff *lookup_containing_backoff(struct rb_root *root,
const struct ceph_hobject_id *hoid)
{
struct rb_node *n = root->rb_node;
while (n) {
struct ceph_osd_backoff *cur =
rb_entry(n, struct ceph_osd_backoff, spg_node);
int cmp;
cmp = hoid_compare(hoid, cur->begin);
if (cmp < 0) {
n = n->rb_left;
} else if (cmp > 0) {
if (hoid_compare(hoid, cur->end) < 0)
return cur;
n = n->rb_right;
} else {
return cur;
}
}
return NULL;
}
/*
* Each backoff has a unique id within its OSD session.
*/
DEFINE_RB_FUNCS(backoff_by_id, struct ceph_osd_backoff, id, id_node)
static void clear_backoffs(struct ceph_osd *osd)
{
while (!RB_EMPTY_ROOT(&osd->o_backoff_mappings)) {
struct ceph_spg_mapping *spg =
rb_entry(rb_first(&osd->o_backoff_mappings),
struct ceph_spg_mapping, node);
while (!RB_EMPTY_ROOT(&spg->backoffs)) {
struct ceph_osd_backoff *backoff =
rb_entry(rb_first(&spg->backoffs),
struct ceph_osd_backoff, spg_node);
erase_backoff(&spg->backoffs, backoff);
erase_backoff_by_id(&osd->o_backoffs_by_id, backoff);
free_backoff(backoff);
}
erase_spg_mapping(&osd->o_backoff_mappings, spg);
free_spg_mapping(spg);
}
}
/*
* Set up a temporary, non-owning view into @t.
*/
static void hoid_fill_from_target(struct ceph_hobject_id *hoid,
const struct ceph_osd_request_target *t)
{
hoid->key = NULL;
hoid->key_len = 0;
hoid->oid = t->target_oid.name;
hoid->oid_len = t->target_oid.name_len;
hoid->snapid = CEPH_NOSNAP;
hoid->hash = t->pgid.seed;
hoid->is_max = false;
if (t->target_oloc.pool_ns) {
hoid->nspace = t->target_oloc.pool_ns->str;
hoid->nspace_len = t->target_oloc.pool_ns->len;
} else {
hoid->nspace = NULL;
hoid->nspace_len = 0;
}
hoid->pool = t->target_oloc.pool;
ceph_hoid_build_hash_cache(hoid);
}
static bool should_plug_request(struct ceph_osd_request *req)
{
struct ceph_osd *osd = req->r_osd;
struct ceph_spg_mapping *spg;
struct ceph_osd_backoff *backoff;
struct ceph_hobject_id hoid;
spg = lookup_spg_mapping(&osd->o_backoff_mappings, &req->r_t.spgid);
if (!spg)
return false;
hoid_fill_from_target(&hoid, &req->r_t);
backoff = lookup_containing_backoff(&spg->backoffs, &hoid);
if (!backoff)
return false;
dout("%s req %p tid %llu backoff osd%d spgid %llu.%xs%d id %llu\n",
__func__, req, req->r_tid, osd->o_osd, backoff->spgid.pgid.pool,
backoff->spgid.pgid.seed, backoff->spgid.shard, backoff->id);
return true;
}
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
static void setup_request_data(struct ceph_osd_request *req,
struct ceph_msg *msg)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
u32 data_len = 0;
int i;
if (!list_empty(&msg->data))
return;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
WARN_ON(msg->data_length);
for (i = 0; i < req->r_num_ops; i++) {
struct ceph_osd_req_op *op = &req->r_ops[i];
switch (op->op) {
/* request */
case CEPH_OSD_OP_WRITE:
case CEPH_OSD_OP_WRITEFULL:
WARN_ON(op->indata_len != op->extent.length);
ceph_osdc_msg_data_add(msg, &op->extent.osd_data);
break;
case CEPH_OSD_OP_SETXATTR:
case CEPH_OSD_OP_CMPXATTR:
WARN_ON(op->indata_len != op->xattr.name_len +
op->xattr.value_len);
ceph_osdc_msg_data_add(msg, &op->xattr.osd_data);
break;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
case CEPH_OSD_OP_NOTIFY_ACK:
ceph_osdc_msg_data_add(msg,
&op->notify_ack.request_data);
break;
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
/* reply */
case CEPH_OSD_OP_STAT:
ceph_osdc_msg_data_add(req->r_reply,
&op->raw_data_in);
break;
case CEPH_OSD_OP_READ:
ceph_osdc_msg_data_add(req->r_reply,
&op->extent.osd_data);
break;
libceph: support for CEPH_OSD_OP_LIST_WATCHERS Add support for this Ceph OSD op, needed to support the RBD exclusive lock feature. Signed-off-by: Douglas Fuller <dfuller@redhat.com> [idryomov@gmail.com: refactor, misc fixes throughout] Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-07-17 20:18:07 +00:00
case CEPH_OSD_OP_LIST_WATCHERS:
ceph_osdc_msg_data_add(req->r_reply,
&op->list_watchers.response_data);
break;
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
/* both */
case CEPH_OSD_OP_CALL:
WARN_ON(op->indata_len != op->cls.class_len +
op->cls.method_len +
op->cls.indata_len);
ceph_osdc_msg_data_add(msg, &op->cls.request_info);
/* optional, can be NONE */
ceph_osdc_msg_data_add(msg, &op->cls.request_data);
/* optional, can be NONE */
ceph_osdc_msg_data_add(req->r_reply,
&op->cls.response_data);
break;
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
case CEPH_OSD_OP_NOTIFY:
ceph_osdc_msg_data_add(msg,
&op->notify.request_data);
ceph_osdc_msg_data_add(req->r_reply,
&op->notify.response_data);
break;
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
}
data_len += op->indata_len;
}
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
WARN_ON(data_len != msg->data_length);
}
libceph: encode_{pgid,oloc}() helpers Factor out encode_{pgid,oloc}() and use ceph_encode_string() for oid. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:53 +00:00
static void encode_pgid(void **p, const struct ceph_pg *pgid)
{
ceph_encode_8(p, 1);
ceph_encode_64(p, pgid->pool);
ceph_encode_32(p, pgid->seed);
ceph_encode_32(p, -1); /* preferred */
}
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
static void encode_spgid(void **p, const struct ceph_spg *spgid)
{
ceph_start_encoding(p, 1, 1, CEPH_PGID_ENCODING_LEN + 1);
encode_pgid(p, &spgid->pgid);
ceph_encode_8(p, spgid->shard);
}
libceph: encode_{pgid,oloc}() helpers Factor out encode_{pgid,oloc}() and use ceph_encode_string() for oid. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:53 +00:00
static void encode_oloc(void **p, void *end,
const struct ceph_object_locator *oloc)
{
ceph_start_encoding(p, 5, 4, ceph_oloc_encoding_size(oloc));
ceph_encode_64(p, oloc->pool);
ceph_encode_32(p, -1); /* preferred */
ceph_encode_32(p, 0); /* key len */
if (oloc->pool_ns)
ceph_encode_string(p, end, oloc->pool_ns->str,
oloc->pool_ns->len);
else
ceph_encode_32(p, 0);
}
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
static void encode_request_partial(struct ceph_osd_request *req,
struct ceph_msg *msg)
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
{
void *p = msg->front.iov_base;
void *const end = p + msg->front_alloc_len;
u32 data_len = 0;
int i;
if (req->r_flags & CEPH_OSD_FLAG_WRITE) {
/* snapshots aren't writeable */
WARN_ON(req->r_snapid != CEPH_NOSNAP);
} else {
WARN_ON(req->r_mtime.tv_sec || req->r_mtime.tv_nsec ||
req->r_data_offset || req->r_snapc);
}
setup_request_data(req, msg);
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
encode_spgid(&p, &req->r_t.spgid); /* actual spg */
ceph_encode_32(&p, req->r_t.pgid.seed); /* raw hash */
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
ceph_encode_32(&p, req->r_osdc->osdmap->epoch);
ceph_encode_32(&p, req->r_flags);
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
/* reqid */
ceph_start_encoding(&p, 2, 2, sizeof(struct ceph_osd_reqid));
memset(p, 0, sizeof(struct ceph_osd_reqid));
p += sizeof(struct ceph_osd_reqid);
/* trace */
memset(p, 0, sizeof(struct ceph_blkin_trace_info));
p += sizeof(struct ceph_blkin_trace_info);
ceph_encode_32(&p, 0); /* client_inc, always 0 */
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
ceph_encode_timespec(p, &req->r_mtime);
p += sizeof(struct ceph_timespec);
libceph: remove req->r_replay_version Nothing uses this anymore with the removal of the ack vs. commit code. Remove the field and just encode zeroes into place in the request encoding. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-04 12:39:36 +00:00
libceph: encode_{pgid,oloc}() helpers Factor out encode_{pgid,oloc}() and use ceph_encode_string() for oid. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:53 +00:00
encode_oloc(&p, end, &req->r_t.target_oloc);
ceph_encode_string(&p, end, req->r_t.target_oid.name,
req->r_t.target_oid.name_len);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
/* ops, can imply data */
ceph_encode_16(&p, req->r_num_ops);
for (i = 0; i < req->r_num_ops; i++) {
data_len += osd_req_encode_op(p, &req->r_ops[i]);
p += sizeof(struct ceph_osd_op);
}
libceph: change how "safe" callback is used An osd request currently has two callbacks. They inform the initiator of the request when we've received confirmation for the target osd that a request was received, and when the osd indicates all changes described by the request are durable. The only time the second callback is used is in the ceph file system for a synchronous write. There's a race that makes some handling of this case unsafe. This patch addresses this problem. The error handling for this callback is also kind of gross, and this patch changes that as well. In ceph_sync_write(), if a safe callback is requested we want to add the request on the ceph inode's unsafe items list. Because items on this list must have their tid set (by ceph_osd_start_request()), the request added *after* the call to that function returns. The problem with this is that there's a race between starting the request and adding it to the unsafe items list; the request may already be complete before ceph_sync_write() even begins to put it on the list. To address this, we change the way the "safe" callback is used. Rather than just calling it when the request is "safe", we use it to notify the initiator the bounds (start and end) of the period during which the request is *unsafe*. So the initiator gets notified just before the request gets sent to the osd (when it is "unsafe"), and again when it's known the results are durable (it's no longer unsafe). The first call will get made in __send_request(), just before the request message gets sent to the messenger for the first time. That function is only called by __send_queued(), which is always called with the osd client's request mutex held. We then have this callback function insert the request on the ceph inode's unsafe list when we're told the request is unsafe. This will avoid the race because this call will be made under protection of the osd client's request mutex. It also nicely groups the setup and cleanup of the state associated with managing unsafe requests. The name of the "safe" callback field is changed to "unsafe" to better reflect its new purpose. It has a Boolean "unsafe" parameter to indicate whether the request is becoming unsafe or is now safe. Because the "msg" parameter wasn't used, we drop that. This resolves the original problem reportedin: http://tracker.ceph.com/issues/4706 Reported-by: Yan, Zheng <zheng.z.yan@intel.com> Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Yan, Zheng <zheng.z.yan@intel.com> Reviewed-by: Sage Weil <sage@inktank.com>
2013-04-15 16:20:42 +00:00
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
ceph_encode_64(&p, req->r_snapid); /* snapid */
if (req->r_snapc) {
ceph_encode_64(&p, req->r_snapc->seq);
ceph_encode_32(&p, req->r_snapc->num_snaps);
for (i = 0; i < req->r_snapc->num_snaps; i++)
ceph_encode_64(&p, req->r_snapc->snaps[i]);
} else {
ceph_encode_64(&p, 0); /* snap_seq */
ceph_encode_32(&p, 0); /* snaps len */
}
ceph_encode_32(&p, req->r_attempts); /* retry_attempt */
libceph: make encode_request_*() work with r_mempool requests Messages allocated out of ceph_msgpool have a fixed front length (pool->front_len). Asserting that the entire front has been filled while encoding is thus wrong. Fixes: 8cb441c0545d ("libceph: MOSDOp v8 encoding (actual spgid + full hash)") Reported-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2017-07-25 12:40:03 +00:00
BUG_ON(p > end - 8); /* space for features */
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
msg->hdr.version = cpu_to_le16(8); /* MOSDOp v8 */
/* front_len is finalized in encode_request_finish() */
libceph: make encode_request_*() work with r_mempool requests Messages allocated out of ceph_msgpool have a fixed front length (pool->front_len). Asserting that the entire front has been filled while encoding is thus wrong. Fixes: 8cb441c0545d ("libceph: MOSDOp v8 encoding (actual spgid + full hash)") Reported-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2017-07-25 12:40:03 +00:00
msg->front.iov_len = p - msg->front.iov_base;
msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
msg->hdr.data_len = cpu_to_le32(data_len);
/*
* The header "data_off" is a hint to the receiver allowing it
* to align received data into its buffers such that there's no
* need to re-copy it before writing it to disk (direct I/O).
*/
msg->hdr.data_off = cpu_to_le16(req->r_data_offset);
libceph: change how "safe" callback is used An osd request currently has two callbacks. They inform the initiator of the request when we've received confirmation for the target osd that a request was received, and when the osd indicates all changes described by the request are durable. The only time the second callback is used is in the ceph file system for a synchronous write. There's a race that makes some handling of this case unsafe. This patch addresses this problem. The error handling for this callback is also kind of gross, and this patch changes that as well. In ceph_sync_write(), if a safe callback is requested we want to add the request on the ceph inode's unsafe items list. Because items on this list must have their tid set (by ceph_osd_start_request()), the request added *after* the call to that function returns. The problem with this is that there's a race between starting the request and adding it to the unsafe items list; the request may already be complete before ceph_sync_write() even begins to put it on the list. To address this, we change the way the "safe" callback is used. Rather than just calling it when the request is "safe", we use it to notify the initiator the bounds (start and end) of the period during which the request is *unsafe*. So the initiator gets notified just before the request gets sent to the osd (when it is "unsafe"), and again when it's known the results are durable (it's no longer unsafe). The first call will get made in __send_request(), just before the request message gets sent to the messenger for the first time. That function is only called by __send_queued(), which is always called with the osd client's request mutex held. We then have this callback function insert the request on the ceph inode's unsafe list when we're told the request is unsafe. This will avoid the race because this call will be made under protection of the osd client's request mutex. It also nicely groups the setup and cleanup of the state associated with managing unsafe requests. The name of the "safe" callback field is changed to "unsafe" to better reflect its new purpose. It has a Boolean "unsafe" parameter to indicate whether the request is becoming unsafe or is now safe. Because the "msg" parameter wasn't used, we drop that. This resolves the original problem reportedin: http://tracker.ceph.com/issues/4706 Reported-by: Yan, Zheng <zheng.z.yan@intel.com> Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Yan, Zheng <zheng.z.yan@intel.com> Reviewed-by: Sage Weil <sage@inktank.com>
2013-04-15 16:20:42 +00:00
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
dout("%s req %p msg %p oid %s oid_len %d\n", __func__, req, msg,
req->r_t.target_oid.name, req->r_t.target_oid.name_len);
}
static void encode_request_finish(struct ceph_msg *msg)
{
void *p = msg->front.iov_base;
libceph: make encode_request_*() work with r_mempool requests Messages allocated out of ceph_msgpool have a fixed front length (pool->front_len). Asserting that the entire front has been filled while encoding is thus wrong. Fixes: 8cb441c0545d ("libceph: MOSDOp v8 encoding (actual spgid + full hash)") Reported-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2017-07-25 12:40:03 +00:00
void *const partial_end = p + msg->front.iov_len;
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
void *const end = p + msg->front_alloc_len;
if (CEPH_HAVE_FEATURE(msg->con->peer_features, RESEND_ON_SPLIT)) {
/* luminous OSD -- encode features and be done */
libceph: make encode_request_*() work with r_mempool requests Messages allocated out of ceph_msgpool have a fixed front length (pool->front_len). Asserting that the entire front has been filled while encoding is thus wrong. Fixes: 8cb441c0545d ("libceph: MOSDOp v8 encoding (actual spgid + full hash)") Reported-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2017-07-25 12:40:03 +00:00
p = partial_end;
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
ceph_encode_64(&p, msg->con->peer_features);
} else {
struct {
char spgid[CEPH_ENCODING_START_BLK_LEN +
CEPH_PGID_ENCODING_LEN + 1];
__le32 hash;
__le32 epoch;
__le32 flags;
char reqid[CEPH_ENCODING_START_BLK_LEN +
sizeof(struct ceph_osd_reqid)];
char trace[sizeof(struct ceph_blkin_trace_info)];
__le32 client_inc;
struct ceph_timespec mtime;
} __packed head;
struct ceph_pg pgid;
void *oloc, *oid, *tail;
int oloc_len, oid_len, tail_len;
int len;
/*
* Pre-luminous OSD -- reencode v8 into v4 using @head
* as a temporary buffer. Encode the raw PG; the rest
* is just a matter of moving oloc, oid and tail blobs
* around.
*/
memcpy(&head, p, sizeof(head));
p += sizeof(head);
oloc = p;
p += CEPH_ENCODING_START_BLK_LEN;
pgid.pool = ceph_decode_64(&p);
p += 4 + 4; /* preferred, key len */
len = ceph_decode_32(&p);
p += len; /* nspace */
oloc_len = p - oloc;
oid = p;
len = ceph_decode_32(&p);
p += len;
oid_len = p - oid;
tail = p;
libceph: make encode_request_*() work with r_mempool requests Messages allocated out of ceph_msgpool have a fixed front length (pool->front_len). Asserting that the entire front has been filled while encoding is thus wrong. Fixes: 8cb441c0545d ("libceph: MOSDOp v8 encoding (actual spgid + full hash)") Reported-by: "Yan, Zheng" <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2017-07-25 12:40:03 +00:00
tail_len = partial_end - p;
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
p = msg->front.iov_base;
ceph_encode_copy(&p, &head.client_inc, sizeof(head.client_inc));
ceph_encode_copy(&p, &head.epoch, sizeof(head.epoch));
ceph_encode_copy(&p, &head.flags, sizeof(head.flags));
ceph_encode_copy(&p, &head.mtime, sizeof(head.mtime));
/* reassert_version */
memset(p, 0, sizeof(struct ceph_eversion));
p += sizeof(struct ceph_eversion);
BUG_ON(p >= oloc);
memmove(p, oloc, oloc_len);
p += oloc_len;
pgid.seed = le32_to_cpu(head.hash);
encode_pgid(&p, &pgid); /* raw pg */
BUG_ON(p >= oid);
memmove(p, oid, oid_len);
p += oid_len;
/* tail -- ops, snapid, snapc, retry_attempt */
BUG_ON(p >= tail);
memmove(p, tail, tail_len);
p += tail_len;
msg->hdr.version = cpu_to_le16(4); /* MOSDOp v4 */
}
BUG_ON(p > end);
msg->front.iov_len = p - msg->front.iov_base;
msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
dout("%s msg %p tid %llu %u+%u+%u v%d\n", __func__, msg,
le64_to_cpu(msg->hdr.tid), le32_to_cpu(msg->hdr.front_len),
le32_to_cpu(msg->hdr.middle_len), le32_to_cpu(msg->hdr.data_len),
le16_to_cpu(msg->hdr.version));
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
}
/*
* @req has to be assigned a tid and registered.
*/
static void send_request(struct ceph_osd_request *req)
{
struct ceph_osd *osd = req->r_osd;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
verify_osd_locked(osd);
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
WARN_ON(osd->o_osd != req->r_t.osd);
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
/* backoff? */
if (should_plug_request(req))
return;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
/*
* We may have a previously queued request message hanging
* around. Cancel it to avoid corrupting the msgr.
*/
if (req->r_sent)
ceph_msg_revoke(req->r_request);
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
req->r_flags |= CEPH_OSD_FLAG_KNOWN_REDIR;
if (req->r_attempts)
req->r_flags |= CEPH_OSD_FLAG_RETRY;
else
WARN_ON(req->r_flags & CEPH_OSD_FLAG_RETRY);
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
encode_request_partial(req, req->r_request);
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
libceph: make sure need_resend targets reflect latest map Otherwise we may miss events like PG splits, pool deletions, etc when we get multiple incremental maps at once. Because check_pool_dne() can now be fed an unlinked request, finish_request() needed to be taught to handle unlinked requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
dout("%s req %p tid %llu to pgid %llu.%x spgid %llu.%xs%d osd%d e%u flags 0x%x attempt %d\n",
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
__func__, req, req->r_tid, req->r_t.pgid.pool, req->r_t.pgid.seed,
libceph: introduce ceph_spg, ceph_pg_to_primary_shard() Store both raw pgid and actual spgid in ceph_osd_request_target. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:53 +00:00
req->r_t.spgid.pgid.pool, req->r_t.spgid.pgid.seed,
libceph: make sure need_resend targets reflect latest map Otherwise we may miss events like PG splits, pool deletions, etc when we get multiple incremental maps at once. Because check_pool_dne() can now be fed an unlinked request, finish_request() needed to be taught to handle unlinked requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
req->r_t.spgid.shard, osd->o_osd, req->r_t.epoch, req->r_flags,
req->r_attempts);
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
req->r_t.paused = false;
req->r_stamp = jiffies;
req->r_attempts++;
req->r_sent = osd->o_incarnation;
req->r_request->hdr.tid = cpu_to_le64(req->r_tid);
ceph_con_send(&osd->o_con, ceph_msg_get(req->r_request));
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
static void maybe_request_map(struct ceph_osd_client *osdc)
{
bool continuous = false;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
verify_osdc_locked(osdc);
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
WARN_ON(!osdc->osdmap->epoch);
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD) ||
ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR)) {
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
dout("%s osdc %p continuous\n", __func__, osdc);
continuous = true;
} else {
dout("%s osdc %p onetime\n", __func__, osdc);
}
if (ceph_monc_want_map(&osdc->client->monc, CEPH_SUB_OSDMAP,
osdc->osdmap->epoch + 1, continuous))
ceph_monc_renew_subs(&osdc->client->monc);
}
libceph: allow requests to return immediately on full conditions if caller wishes Usually, when the osd map is flagged as full or the pool is at quota, write requests just hang. This is not what we want for cephfs, where it would be better to simply report -ENOSPC back to userland instead of stalling. If the caller knows that it will want an immediate error return instead of blocking on a full or at-quota error condition then allow it to set a flag to request that behavior. Set that flag in ceph_osdc_new_request (since ceph.ko is the only caller), and on any other write request from ceph.ko. A later patch will deal with requests that were submitted before the new map showing the full condition came in. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-04 12:39:37 +00:00
static void complete_request(struct ceph_osd_request *req, int err);
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
static void send_map_check(struct ceph_osd_request *req);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void __submit_request(struct ceph_osd_request *req, bool wrlocked)
libceph: factor out logic from ceph_osdc_start_request() Factor out logic from ceph_osdc_start_request() into a new helper, __ceph_osdc_start_request(). ceph_osdc_start_request() now amounts to taking locks and calling __ceph_osdc_start_request(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-31 17:33:39 +00:00
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd_client *osdc = req->r_osdc;
struct ceph_osd *osd;
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
enum calc_target_result ct_res;
libceph: introduce ceph_osdc_abort_requests() This will be used by the filesystem for "umount -f". Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-05-15 13:47:58 +00:00
int err = 0;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
bool need_send = false;
bool promoted = false;
libceph: factor out logic from ceph_osdc_start_request() Factor out logic from ceph_osdc_start_request() into a new helper, __ceph_osdc_start_request(). ceph_osdc_start_request() now amounts to taking locks and calling __ceph_osdc_start_request(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-31 17:33:39 +00:00
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
WARN_ON(req->r_tid);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("%s req %p wrlocked %d\n", __func__, req, wrlocked);
again:
libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
ct_res = calc_target(osdc, &req->r_t, NULL, false);
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
if (ct_res == CALC_TARGET_POOL_DNE && !wrlocked)
goto promote;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
osd = lookup_create_osd(osdc, req->r_t.osd, wrlocked);
if (IS_ERR(osd)) {
WARN_ON(PTR_ERR(osd) != -EAGAIN || wrlocked);
goto promote;
libceph: factor out logic from ceph_osdc_start_request() Factor out logic from ceph_osdc_start_request() into a new helper, __ceph_osdc_start_request(). ceph_osdc_start_request() now amounts to taking locks and calling __ceph_osdc_start_request(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-31 17:33:39 +00:00
}
libceph: introduce ceph_osdc_abort_requests() This will be used by the filesystem for "umount -f". Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-05-15 13:47:58 +00:00
if (osdc->abort_err) {
dout("req %p abort_err %d\n", req, osdc->abort_err);
err = osdc->abort_err;
} else if (osdc->osdmap->epoch < osdc->epoch_barrier) {
libceph: add an epoch_barrier field to struct ceph_osd_client Cephfs can get cap update requests that contain a new epoch barrier in them. When that happens we want to pause all OSD traffic until the right map epoch arrives. Add an epoch_barrier field to ceph_osd_client that is protected by the osdc->lock rwsem. When the barrier is set, and the current OSD map epoch is below that, pause the request target when submitting the request or when revisiting it. Add a way for upper layers (cephfs) to update the epoch_barrier as well. If we get a new map, compare the new epoch against the barrier before kicking requests and request another map if the map epoch is still lower than the one we want. If we get a map with a full pool, or at quota condition, then set the barrier to the current epoch value. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-18 13:21:16 +00:00
dout("req %p epoch %u barrier %u\n", req, osdc->osdmap->epoch,
osdc->epoch_barrier);
req->r_t.paused = true;
maybe_request_map(osdc);
} else if ((req->r_flags & CEPH_OSD_FLAG_WRITE) &&
ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR)) {
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("req %p pausewr\n", req);
req->r_t.paused = true;
maybe_request_map(osdc);
} else if ((req->r_flags & CEPH_OSD_FLAG_READ) &&
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD)) {
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("req %p pauserd\n", req);
req->r_t.paused = true;
maybe_request_map(osdc);
} else if ((req->r_flags & CEPH_OSD_FLAG_WRITE) &&
!(req->r_flags & (CEPH_OSD_FLAG_FULL_TRY |
CEPH_OSD_FLAG_FULL_FORCE)) &&
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
(ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
pool_full(osdc, req->r_t.base_oloc.pool))) {
dout("req %p full/pool_full\n", req);
libceph: make abort_on_full a per-osdc setting The intent behind making it a per-request setting was that it would be set for writes, but not for reads. As it is, the flag is set for all fs/ceph requests except for pool perm check stat request (technically a read). ceph_osdc_abort_on_full() skips reads since the previous commit and I don't see a use case for marking individual requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-30 14:29:14 +00:00
if (osdc->abort_on_full) {
libceph: introduce ceph_osdc_abort_requests() This will be used by the filesystem for "umount -f". Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-05-15 13:47:58 +00:00
err = -ENOSPC;
libceph: don't warn if req->r_abort_on_full is set The "FULL or reached pool quota" warning is there to explain paused requests. No need to emit it if pausing isn't going to occur. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-17 14:13:07 +00:00
} else {
pr_warn_ratelimited("FULL or reached pool quota\n");
req->r_t.paused = true;
maybe_request_map(osdc);
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
} else if (!osd_homeless(osd)) {
need_send = true;
libceph: factor out logic from ceph_osdc_start_request() Factor out logic from ceph_osdc_start_request() into a new helper, __ceph_osdc_start_request(). ceph_osdc_start_request() now amounts to taking locks and calling __ceph_osdc_start_request(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-31 17:33:39 +00:00
} else {
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
maybe_request_map(osdc);
libceph: factor out logic from ceph_osdc_start_request() Factor out logic from ceph_osdc_start_request() into a new helper, __ceph_osdc_start_request(). ceph_osdc_start_request() now amounts to taking locks and calling __ceph_osdc_start_request(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-31 17:33:39 +00:00
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
mutex_lock(&osd->lock);
/*
* Assign the tid atomically with send_request() to protect
* multiple writes to the same object from racing with each
* other, resulting in out of order ops on the OSDs.
*/
req->r_tid = atomic64_inc_return(&osdc->last_tid);
link_request(osd, req);
if (need_send)
send_request(req);
libceph: introduce ceph_osdc_abort_requests() This will be used by the filesystem for "umount -f". Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-05-15 13:47:58 +00:00
else if (err)
complete_request(req, err);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
mutex_unlock(&osd->lock);
libceph: avoid a use-after-free during map check Sending map check after complete_request() was called is not only useless, but can lead to a use-after-free as req->r_kref decrement in __complete_request() races with map check code. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-22 14:26:51 +00:00
if (!err && ct_res == CALC_TARGET_POOL_DNE)
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
send_map_check(req);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
if (promoted)
downgrade_write(&osdc->lock);
return;
promote:
up_read(&osdc->lock);
down_write(&osdc->lock);
wrlocked = true;
promoted = true;
goto again;
}
static void account_request(struct ceph_osd_request *req)
{
libceph, rbd, ceph: WRITE | ONDISK -> WRITE CEPH_OSD_FLAG_ONDISK is set in account_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:48:41 +00:00
WARN_ON(req->r_flags & (CEPH_OSD_FLAG_ACK | CEPH_OSD_FLAG_ONDISK));
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
WARN_ON(!(req->r_flags & (CEPH_OSD_FLAG_READ | CEPH_OSD_FLAG_WRITE)));
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
req->r_flags |= CEPH_OSD_FLAG_ONDISK;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
atomic_inc(&req->r_osdc->num_requests);
libceph: osd_request_timeout option osd_request_timeout specifies how many seconds to wait for a response from OSDs before returning -ETIMEDOUT from an OSD request. 0 (default) means no limit. osd_request_timeout is osdkeepalive-precise -- in-flight requests are swept through every osdkeepalive seconds. With ack vs commit behaviour gone, abort_request() is really simple. This is based on a patch from Artur Molchanov <artur.molchanov@synesis.ru>. Tested-by: Artur Molchanov <artur.molchanov@synesis.ru> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-12 16:11:07 +00:00
req->r_start_stamp = jiffies;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
}
static void submit_request(struct ceph_osd_request *req, bool wrlocked)
{
ceph_osdc_get_request(req);
account_request(req);
__submit_request(req, wrlocked);
}
libceph: remove now unused finish_request() wrapper Kill the wrapper and rename __finish_request() to finish_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-12-02 13:01:55 +00:00
static void finish_request(struct ceph_osd_request *req)
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
{
struct ceph_osd_client *osdc = req->r_osdc;
libceph: make sure need_resend targets reflect latest map Otherwise we may miss events like PG splits, pool deletions, etc when we get multiple incremental maps at once. Because check_pool_dne() can now be fed an unlinked request, finish_request() needed to be taught to handle unlinked requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
WARN_ON(lookup_request_mc(&osdc->map_checks, req->r_tid));
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("%s req %p tid %llu\n", __func__, req, req->r_tid);
libceph: make sure need_resend targets reflect latest map Otherwise we may miss events like PG splits, pool deletions, etc when we get multiple incremental maps at once. Because check_pool_dne() can now be fed an unlinked request, finish_request() needed to be taught to handle unlinked requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
if (req->r_osd)
unlink_request(req->r_osd, req);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
atomic_dec(&osdc->num_requests);
/*
* If an OSD has failed or returned and a request has been sent
* twice, it's possible to get a reply and end up here while the
* request message is queued for delivery. We will ignore the
* reply, so not a big deal, but better to try and catch it.
*/
ceph_msg_revoke(req->r_request);
ceph_msg_revoke_incoming(req->r_reply);
}
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
static void __complete_request(struct ceph_osd_request *req)
{
libceph: move more code into __complete_request() Move req->r_completion wake up and req->r_kref decrement into __complete_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-21 13:33:48 +00:00
dout("%s req %p tid %llu cb %pf result %d\n", __func__, req,
req->r_tid, req->r_callback, req->r_result);
if (req->r_callback)
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
req->r_callback(req);
libceph: move more code into __complete_request() Move req->r_completion wake up and req->r_kref decrement into __complete_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-21 13:33:48 +00:00
complete_all(&req->r_completion);
ceph_osdc_put_request(req);
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
}
libceph: defer __complete_request() to a workqueue In the common case, req->r_callback is called by handle_reply() on the ceph-msgr worker thread without any locks. If handle_reply() fails, it is called with both osd->lock and osdc->lock. In the map check case, it is called with just osdc->lock but held for write. Finally, if the request is aborted because of -ENOSPC or by ceph_osdc_abort_requests(), it is called directly on the submitter's thread, again with both locks. req->r_callback on the submitter's thread is relatively new (introduced in 4.12) and ripe for deadlocks -- e.g. writeback worker thread waiting on itself: inode_wait_for_writeback+0x26/0x40 evict+0xb5/0x1a0 iput+0x1d2/0x220 ceph_put_wrbuffer_cap_refs+0xe0/0x2c0 [ceph] writepages_finish+0x2d3/0x410 [ceph] __complete_request+0x26/0x60 [libceph] complete_request+0x2e/0x70 [libceph] __submit_request+0x256/0x330 [libceph] submit_request+0x2b/0x30 [libceph] ceph_osdc_start_request+0x25/0x40 [libceph] ceph_writepages_start+0xdfe/0x1320 [ceph] do_writepages+0x1f/0x70 __writeback_single_inode+0x45/0x330 writeback_sb_inodes+0x26a/0x600 __writeback_inodes_wb+0x92/0xc0 wb_writeback+0x274/0x330 wb_workfn+0x2d5/0x3b0 Defer __complete_request() to a workqueue in all failure cases so it's never on the same thread as ceph_osdc_start_request() and always called with no locks held. Link: http://tracker.ceph.com/issues/23978 Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-21 14:00:29 +00:00
static void complete_request_workfn(struct work_struct *work)
{
struct ceph_osd_request *req =
container_of(work, struct ceph_osd_request, r_complete_work);
__complete_request(req);
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
}
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
/*
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
* This is open-coded in handle_reply().
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
*/
static void complete_request(struct ceph_osd_request *req, int err)
{
dout("%s req %p tid %llu err %d\n", __func__, req, req->r_tid, err);
req->r_result = err;
libceph: remove now unused finish_request() wrapper Kill the wrapper and rename __finish_request() to finish_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-12-02 13:01:55 +00:00
finish_request(req);
libceph: defer __complete_request() to a workqueue In the common case, req->r_callback is called by handle_reply() on the ceph-msgr worker thread without any locks. If handle_reply() fails, it is called with both osd->lock and osdc->lock. In the map check case, it is called with just osdc->lock but held for write. Finally, if the request is aborted because of -ENOSPC or by ceph_osdc_abort_requests(), it is called directly on the submitter's thread, again with both locks. req->r_callback on the submitter's thread is relatively new (introduced in 4.12) and ripe for deadlocks -- e.g. writeback worker thread waiting on itself: inode_wait_for_writeback+0x26/0x40 evict+0xb5/0x1a0 iput+0x1d2/0x220 ceph_put_wrbuffer_cap_refs+0xe0/0x2c0 [ceph] writepages_finish+0x2d3/0x410 [ceph] __complete_request+0x26/0x60 [libceph] complete_request+0x2e/0x70 [libceph] __submit_request+0x256/0x330 [libceph] submit_request+0x2b/0x30 [libceph] ceph_osdc_start_request+0x25/0x40 [libceph] ceph_writepages_start+0xdfe/0x1320 [ceph] do_writepages+0x1f/0x70 __writeback_single_inode+0x45/0x330 writeback_sb_inodes+0x26a/0x600 __writeback_inodes_wb+0x92/0xc0 wb_writeback+0x274/0x330 wb_workfn+0x2d5/0x3b0 Defer __complete_request() to a workqueue in all failure cases so it's never on the same thread as ceph_osdc_start_request() and always called with no locks held. Link: http://tracker.ceph.com/issues/23978 Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-21 14:00:29 +00:00
INIT_WORK(&req->r_complete_work, complete_request_workfn);
queue_work(req->r_osdc->completion_wq, &req->r_complete_work);
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
}
static void cancel_map_check(struct ceph_osd_request *req)
{
struct ceph_osd_client *osdc = req->r_osdc;
struct ceph_osd_request *lookup_req;
verify_osdc_wrlocked(osdc);
lookup_req = lookup_request_mc(&osdc->map_checks, req->r_tid);
if (!lookup_req)
return;
WARN_ON(lookup_req != req);
erase_request_mc(&osdc->map_checks, req);
ceph_osdc_put_request(req);
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void cancel_request(struct ceph_osd_request *req)
{
dout("%s req %p tid %llu\n", __func__, req, req->r_tid);
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
cancel_map_check(req);
libceph: remove now unused finish_request() wrapper Kill the wrapper and rename __finish_request() to finish_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-12-02 13:01:55 +00:00
finish_request(req);
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
complete_all(&req->r_completion);
libceph: always signal completion when done r_safe_completion is currently, and has always been, signaled only if on-disk ack was requested. It's there for fsync and syncfs, which wait for in-flight writes to flush - all data write requests set ONDISK. However, the pool perm check code introduced in 4.2 sends a write request with only ACK set. An unfortunately timed syncfs can then hang forever: r_safe_completion won't be signaled because only an unsafe reply was requested. We could patch ceph_osdc_sync() to skip !ONDISK write requests, but that is somewhat incomplete and yet another special case. Instead, rename this completion to r_done_completion and always signal it when the OSD client is done with the request, whether unsafe, safe, or error. This is a bit cleaner and helps with the cancellation code. Reported-by: Yan, Zheng <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-12-02 13:01:55 +00:00
ceph_osdc_put_request(req);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
}
libceph: osd_request_timeout option osd_request_timeout specifies how many seconds to wait for a response from OSDs before returning -ETIMEDOUT from an OSD request. 0 (default) means no limit. osd_request_timeout is osdkeepalive-precise -- in-flight requests are swept through every osdkeepalive seconds. With ack vs commit behaviour gone, abort_request() is really simple. This is based on a patch from Artur Molchanov <artur.molchanov@synesis.ru>. Tested-by: Artur Molchanov <artur.molchanov@synesis.ru> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-12 16:11:07 +00:00
static void abort_request(struct ceph_osd_request *req, int err)
{
dout("%s req %p tid %llu err %d\n", __func__, req, req->r_tid, err);
cancel_map_check(req);
complete_request(req, err);
}
libceph: introduce ceph_osdc_abort_requests() This will be used by the filesystem for "umount -f". Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-05-15 13:47:58 +00:00
static int abort_fn(struct ceph_osd_request *req, void *arg)
{
int err = *(int *)arg;
abort_request(req, err);
return 0; /* continue iteration */
}
/*
* Abort all in-flight requests with @err and arrange for all future
* requests to be failed immediately.
*/
void ceph_osdc_abort_requests(struct ceph_osd_client *osdc, int err)
{
dout("%s osdc %p err %d\n", __func__, osdc, err);
down_write(&osdc->lock);
for_each_request(osdc, abort_fn, &err);
osdc->abort_err = err;
up_write(&osdc->lock);
}
EXPORT_SYMBOL(ceph_osdc_abort_requests);
libceph: add an epoch_barrier field to struct ceph_osd_client Cephfs can get cap update requests that contain a new epoch barrier in them. When that happens we want to pause all OSD traffic until the right map epoch arrives. Add an epoch_barrier field to ceph_osd_client that is protected by the osdc->lock rwsem. When the barrier is set, and the current OSD map epoch is below that, pause the request target when submitting the request or when revisiting it. Add a way for upper layers (cephfs) to update the epoch_barrier as well. If we get a new map, compare the new epoch against the barrier before kicking requests and request another map if the map epoch is still lower than the one we want. If we get a map with a full pool, or at quota condition, then set the barrier to the current epoch value. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-18 13:21:16 +00:00
static void update_epoch_barrier(struct ceph_osd_client *osdc, u32 eb)
{
if (likely(eb > osdc->epoch_barrier)) {
dout("updating epoch_barrier from %u to %u\n",
osdc->epoch_barrier, eb);
osdc->epoch_barrier = eb;
/* Request map if we're not to the barrier yet */
if (eb > osdc->osdmap->epoch)
maybe_request_map(osdc);
}
}
void ceph_osdc_update_epoch_barrier(struct ceph_osd_client *osdc, u32 eb)
{
down_read(&osdc->lock);
if (unlikely(eb > osdc->epoch_barrier)) {
up_read(&osdc->lock);
down_write(&osdc->lock);
update_epoch_barrier(osdc, eb);
up_write(&osdc->lock);
} else {
up_read(&osdc->lock);
}
}
EXPORT_SYMBOL(ceph_osdc_update_epoch_barrier);
libceph: use for_each_request() in ceph_osdc_abort_on_full() Scanning the trees just to see if there is anything to abort is unnecessary -- all that is needed here is to update the epoch barrier first, before we start aborting. Simplify and do the update inside the loop before calling abort_request() for the first time. The switch to for_each_request() also fixes a bug: homeless requests weren't even considered for aborting. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-16 16:21:34 +00:00
/*
* We can end up releasing caps as a result of abort_request().
* In that case, we probably want to ensure that the cap release message
* has an updated epoch barrier in it, so set the epoch barrier prior to
* aborting the first request.
*/
static int abort_on_full_fn(struct ceph_osd_request *req, void *arg)
{
struct ceph_osd_client *osdc = req->r_osdc;
bool *victims = arg;
libceph: make abort_on_full a per-osdc setting The intent behind making it a per-request setting was that it would be set for writes, but not for reads. As it is, the flag is set for all fs/ceph requests except for pool perm check stat request (technically a read). ceph_osdc_abort_on_full() skips reads since the previous commit and I don't see a use case for marking individual requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-30 14:29:14 +00:00
if ((req->r_flags & CEPH_OSD_FLAG_WRITE) &&
libceph: use for_each_request() in ceph_osdc_abort_on_full() Scanning the trees just to see if there is anything to abort is unnecessary -- all that is needed here is to update the epoch barrier first, before we start aborting. Simplify and do the update inside the loop before calling abort_request() for the first time. The switch to for_each_request() also fixes a bug: homeless requests weren't even considered for aborting. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-16 16:21:34 +00:00
(ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
libceph: don't abort reads in ceph_osdc_abort_on_full() Don't consider reads for aborting and use ->base_oloc instead of ->target_oloc, as done in __submit_request(). Strictly speaking, we shouldn't be aborting FULL_TRY/FULL_FORCE writes either. But, there is an inconsistency in FULL_TRY/FULL_FORCE handling on the OSD side [1], so given that neither of these is used in the kernel client, leave it for when the OSD behaviour is sorted out. [1] http://tracker.ceph.com/issues/24339 Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-30 12:58:25 +00:00
pool_full(osdc, req->r_t.base_oloc.pool))) {
libceph: use for_each_request() in ceph_osdc_abort_on_full() Scanning the trees just to see if there is anything to abort is unnecessary -- all that is needed here is to update the epoch barrier first, before we start aborting. Simplify and do the update inside the loop before calling abort_request() for the first time. The switch to for_each_request() also fixes a bug: homeless requests weren't even considered for aborting. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-16 16:21:34 +00:00
if (!*victims) {
update_epoch_barrier(osdc, osdc->osdmap->epoch);
*victims = true;
}
abort_request(req, -ENOSPC);
}
return 0; /* continue iteration */
}
libceph: abort already submitted but abortable requests when map or pool goes full When a Ceph volume hits capacity, a flag is set in the OSD map to indicate that, and a new map is sprayed around the cluster. With cephfs we want it to shut down any abortable requests that are in progress with an -ENOSPC error as they'd just hang otherwise. Add a new ceph_osdc_abort_on_full helper function to handle this. It will first check whether there is an out-of-space condition in the cluster and then walk the tree and abort any request that has r_abort_on_full set with a -ENOSPC error. Call this new function directly whenever we get a new OSD map. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-04 12:39:39 +00:00
/*
* Drop all pending requests that are stalled waiting on a full condition to
libceph: add an epoch_barrier field to struct ceph_osd_client Cephfs can get cap update requests that contain a new epoch barrier in them. When that happens we want to pause all OSD traffic until the right map epoch arrives. Add an epoch_barrier field to ceph_osd_client that is protected by the osdc->lock rwsem. When the barrier is set, and the current OSD map epoch is below that, pause the request target when submitting the request or when revisiting it. Add a way for upper layers (cephfs) to update the epoch_barrier as well. If we get a new map, compare the new epoch against the barrier before kicking requests and request another map if the map epoch is still lower than the one we want. If we get a map with a full pool, or at quota condition, then set the barrier to the current epoch value. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-18 13:21:16 +00:00
* clear, and complete them with ENOSPC as the return code. Set the
* osdc->epoch_barrier to the latest map epoch that we've seen if any were
* cancelled.
libceph: abort already submitted but abortable requests when map or pool goes full When a Ceph volume hits capacity, a flag is set in the OSD map to indicate that, and a new map is sprayed around the cluster. With cephfs we want it to shut down any abortable requests that are in progress with an -ENOSPC error as they'd just hang otherwise. Add a new ceph_osdc_abort_on_full helper function to handle this. It will first check whether there is an out-of-space condition in the cluster and then walk the tree and abort any request that has r_abort_on_full set with a -ENOSPC error. Call this new function directly whenever we get a new OSD map. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-04 12:39:39 +00:00
*/
static void ceph_osdc_abort_on_full(struct ceph_osd_client *osdc)
{
libceph: add an epoch_barrier field to struct ceph_osd_client Cephfs can get cap update requests that contain a new epoch barrier in them. When that happens we want to pause all OSD traffic until the right map epoch arrives. Add an epoch_barrier field to ceph_osd_client that is protected by the osdc->lock rwsem. When the barrier is set, and the current OSD map epoch is below that, pause the request target when submitting the request or when revisiting it. Add a way for upper layers (cephfs) to update the epoch_barrier as well. If we get a new map, compare the new epoch against the barrier before kicking requests and request another map if the map epoch is still lower than the one we want. If we get a map with a full pool, or at quota condition, then set the barrier to the current epoch value. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-18 13:21:16 +00:00
bool victims = false;
libceph: abort already submitted but abortable requests when map or pool goes full When a Ceph volume hits capacity, a flag is set in the OSD map to indicate that, and a new map is sprayed around the cluster. With cephfs we want it to shut down any abortable requests that are in progress with an -ENOSPC error as they'd just hang otherwise. Add a new ceph_osdc_abort_on_full helper function to handle this. It will first check whether there is an out-of-space condition in the cluster and then walk the tree and abort any request that has r_abort_on_full set with a -ENOSPC error. Call this new function directly whenever we get a new OSD map. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-04 12:39:39 +00:00
libceph: make abort_on_full a per-osdc setting The intent behind making it a per-request setting was that it would be set for writes, but not for reads. As it is, the flag is set for all fs/ceph requests except for pool perm check stat request (technically a read). ceph_osdc_abort_on_full() skips reads since the previous commit and I don't see a use case for marking individual requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-30 14:29:14 +00:00
if (osdc->abort_on_full &&
(ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) || have_pool_full(osdc)))
libceph: use for_each_request() in ceph_osdc_abort_on_full() Scanning the trees just to see if there is anything to abort is unnecessary -- all that is needed here is to update the epoch barrier first, before we start aborting. Simplify and do the update inside the loop before calling abort_request() for the first time. The switch to for_each_request() also fixes a bug: homeless requests weren't even considered for aborting. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-16 16:21:34 +00:00
for_each_request(osdc, abort_on_full_fn, &victims);
libceph: abort already submitted but abortable requests when map or pool goes full When a Ceph volume hits capacity, a flag is set in the OSD map to indicate that, and a new map is sprayed around the cluster. With cephfs we want it to shut down any abortable requests that are in progress with an -ENOSPC error as they'd just hang otherwise. Add a new ceph_osdc_abort_on_full helper function to handle this. It will first check whether there is an out-of-space condition in the cluster and then walk the tree and abort any request that has r_abort_on_full set with a -ENOSPC error. Call this new function directly whenever we get a new OSD map. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-04 12:39:39 +00:00
}
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
static void check_pool_dne(struct ceph_osd_request *req)
{
struct ceph_osd_client *osdc = req->r_osdc;
struct ceph_osdmap *map = osdc->osdmap;
verify_osdc_wrlocked(osdc);
WARN_ON(!map->epoch);
if (req->r_attempts) {
/*
* We sent a request earlier, which means that
* previously the pool existed, and now it does not
* (i.e., it was deleted).
*/
req->r_map_dne_bound = map->epoch;
dout("%s req %p tid %llu pool disappeared\n", __func__, req,
req->r_tid);
} else {
dout("%s req %p tid %llu map_dne_bound %u have %u\n", __func__,
req, req->r_tid, req->r_map_dne_bound, map->epoch);
}
if (req->r_map_dne_bound) {
if (map->epoch >= req->r_map_dne_bound) {
/* we had a new enough map */
pr_info_ratelimited("tid %llu pool does not exist\n",
req->r_tid);
complete_request(req, -ENOENT);
}
} else {
send_map_check(req);
}
}
static void map_check_cb(struct ceph_mon_generic_request *greq)
{
struct ceph_osd_client *osdc = &greq->monc->client->osdc;
struct ceph_osd_request *req;
u64 tid = greq->private_data;
WARN_ON(greq->result || !greq->u.newest);
down_write(&osdc->lock);
req = lookup_request_mc(&osdc->map_checks, tid);
if (!req) {
dout("%s tid %llu dne\n", __func__, tid);
goto out_unlock;
}
dout("%s req %p tid %llu map_dne_bound %u newest %llu\n", __func__,
req, req->r_tid, req->r_map_dne_bound, greq->u.newest);
if (!req->r_map_dne_bound)
req->r_map_dne_bound = greq->u.newest;
erase_request_mc(&osdc->map_checks, req);
check_pool_dne(req);
ceph_osdc_put_request(req);
out_unlock:
up_write(&osdc->lock);
}
static void send_map_check(struct ceph_osd_request *req)
{
struct ceph_osd_client *osdc = req->r_osdc;
struct ceph_osd_request *lookup_req;
int ret;
verify_osdc_wrlocked(osdc);
lookup_req = lookup_request_mc(&osdc->map_checks, req->r_tid);
if (lookup_req) {
WARN_ON(lookup_req != req);
return;
}
ceph_osdc_get_request(req);
insert_request_mc(&osdc->map_checks, req);
ret = ceph_monc_get_version_async(&osdc->client->monc, "osdmap",
map_check_cb, req->r_tid);
WARN_ON(ret);
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
/*
* lingering requests, watch/notify v2 infrastructure
*/
static void linger_release(struct kref *kref)
{
struct ceph_osd_linger_request *lreq =
container_of(kref, struct ceph_osd_linger_request, kref);
dout("%s lreq %p reg_req %p ping_req %p\n", __func__, lreq,
lreq->reg_req, lreq->ping_req);
WARN_ON(!RB_EMPTY_NODE(&lreq->node));
WARN_ON(!RB_EMPTY_NODE(&lreq->osdc_node));
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
WARN_ON(!RB_EMPTY_NODE(&lreq->mc_node));
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
WARN_ON(!list_empty(&lreq->scan_item));
libceph: support for checking on status of watch Implement ceph_osdc_watch_check() to be able to check on status of watch. Note that the time it takes for a watch/notify event to get delivered through the notify_wq is taken into account. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
WARN_ON(!list_empty(&lreq->pending_lworks));
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
WARN_ON(lreq->osd);
if (lreq->reg_req)
ceph_osdc_put_request(lreq->reg_req);
if (lreq->ping_req)
ceph_osdc_put_request(lreq->ping_req);
target_destroy(&lreq->t);
kfree(lreq);
}
static void linger_put(struct ceph_osd_linger_request *lreq)
{
if (lreq)
kref_put(&lreq->kref, linger_release);
}
static struct ceph_osd_linger_request *
linger_get(struct ceph_osd_linger_request *lreq)
{
kref_get(&lreq->kref);
return lreq;
}
static struct ceph_osd_linger_request *
linger_alloc(struct ceph_osd_client *osdc)
{
struct ceph_osd_linger_request *lreq;
lreq = kzalloc(sizeof(*lreq), GFP_NOIO);
if (!lreq)
return NULL;
kref_init(&lreq->kref);
mutex_init(&lreq->lock);
RB_CLEAR_NODE(&lreq->node);
RB_CLEAR_NODE(&lreq->osdc_node);
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
RB_CLEAR_NODE(&lreq->mc_node);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
INIT_LIST_HEAD(&lreq->scan_item);
libceph: support for checking on status of watch Implement ceph_osdc_watch_check() to be able to check on status of watch. Note that the time it takes for a watch/notify event to get delivered through the notify_wq is taken into account. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
INIT_LIST_HEAD(&lreq->pending_lworks);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
init_completion(&lreq->reg_commit_wait);
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
init_completion(&lreq->notify_finish_wait);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
lreq->osdc = osdc;
target_init(&lreq->t);
dout("%s lreq %p\n", __func__, lreq);
return lreq;
}
DEFINE_RB_INSDEL_FUNCS(linger, struct ceph_osd_linger_request, linger_id, node)
DEFINE_RB_FUNCS(linger_osdc, struct ceph_osd_linger_request, linger_id, osdc_node)
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
DEFINE_RB_FUNCS(linger_mc, struct ceph_osd_linger_request, linger_id, mc_node)
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
/*
* Create linger request <-> OSD session relation.
*
* @lreq has to be registered, @osd may be homeless.
*/
static void link_linger(struct ceph_osd *osd,
struct ceph_osd_linger_request *lreq)
{
verify_osd_locked(osd);
WARN_ON(!lreq->linger_id || lreq->osd);
dout("%s osd %p osd%d lreq %p linger_id %llu\n", __func__, osd,
osd->o_osd, lreq, lreq->linger_id);
if (!osd_homeless(osd))
__remove_osd_from_lru(osd);
else
atomic_inc(&osd->o_osdc->num_homeless);
get_osd(osd);
insert_linger(&osd->o_linger_requests, lreq);
lreq->osd = osd;
}
static void unlink_linger(struct ceph_osd *osd,
struct ceph_osd_linger_request *lreq)
{
verify_osd_locked(osd);
WARN_ON(lreq->osd != osd);
dout("%s osd %p osd%d lreq %p linger_id %llu\n", __func__, osd,
osd->o_osd, lreq, lreq->linger_id);
lreq->osd = NULL;
erase_linger(&osd->o_linger_requests, lreq);
put_osd(osd);
if (!osd_homeless(osd))
maybe_move_osd_to_lru(osd);
else
atomic_dec(&osd->o_osdc->num_homeless);
}
static bool __linger_registered(struct ceph_osd_linger_request *lreq)
{
verify_osdc_locked(lreq->osdc);
return !RB_EMPTY_NODE(&lreq->osdc_node);
}
static bool linger_registered(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
bool registered;
down_read(&osdc->lock);
registered = __linger_registered(lreq);
up_read(&osdc->lock);
return registered;
}
static void linger_register(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
verify_osdc_wrlocked(osdc);
WARN_ON(lreq->linger_id);
linger_get(lreq);
lreq->linger_id = ++osdc->last_linger_id;
insert_linger_osdc(&osdc->linger_requests, lreq);
}
static void linger_unregister(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
verify_osdc_wrlocked(osdc);
erase_linger_osdc(&osdc->linger_requests, lreq);
linger_put(lreq);
}
static void cancel_linger_request(struct ceph_osd_request *req)
{
struct ceph_osd_linger_request *lreq = req->r_priv;
WARN_ON(!req->r_linger);
cancel_request(req);
linger_put(lreq);
}
struct linger_work {
struct work_struct work;
struct ceph_osd_linger_request *lreq;
libceph: support for checking on status of watch Implement ceph_osdc_watch_check() to be able to check on status of watch. Note that the time it takes for a watch/notify event to get delivered through the notify_wq is taken into account. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
struct list_head pending_item;
unsigned long queued_stamp;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
union {
struct {
u64 notify_id;
u64 notifier_id;
void *payload; /* points into @msg front */
size_t payload_len;
struct ceph_msg *msg; /* for ceph_msg_put() */
} notify;
struct {
int err;
} error;
};
};
static struct linger_work *lwork_alloc(struct ceph_osd_linger_request *lreq,
work_func_t workfn)
{
struct linger_work *lwork;
lwork = kzalloc(sizeof(*lwork), GFP_NOIO);
if (!lwork)
return NULL;
INIT_WORK(&lwork->work, workfn);
libceph: support for checking on status of watch Implement ceph_osdc_watch_check() to be able to check on status of watch. Note that the time it takes for a watch/notify event to get delivered through the notify_wq is taken into account. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
INIT_LIST_HEAD(&lwork->pending_item);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
lwork->lreq = linger_get(lreq);
return lwork;
}
static void lwork_free(struct linger_work *lwork)
{
struct ceph_osd_linger_request *lreq = lwork->lreq;
libceph: support for checking on status of watch Implement ceph_osdc_watch_check() to be able to check on status of watch. Note that the time it takes for a watch/notify event to get delivered through the notify_wq is taken into account. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
mutex_lock(&lreq->lock);
list_del(&lwork->pending_item);
mutex_unlock(&lreq->lock);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
linger_put(lreq);
kfree(lwork);
}
static void lwork_queue(struct linger_work *lwork)
{
struct ceph_osd_linger_request *lreq = lwork->lreq;
struct ceph_osd_client *osdc = lreq->osdc;
verify_lreq_locked(lreq);
libceph: support for checking on status of watch Implement ceph_osdc_watch_check() to be able to check on status of watch. Note that the time it takes for a watch/notify event to get delivered through the notify_wq is taken into account. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
WARN_ON(!list_empty(&lwork->pending_item));
lwork->queued_stamp = jiffies;
list_add_tail(&lwork->pending_item, &lreq->pending_lworks);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
queue_work(osdc->notify_wq, &lwork->work);
}
static void do_watch_notify(struct work_struct *w)
{
struct linger_work *lwork = container_of(w, struct linger_work, work);
struct ceph_osd_linger_request *lreq = lwork->lreq;
if (!linger_registered(lreq)) {
dout("%s lreq %p not registered\n", __func__, lreq);
goto out;
}
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
WARN_ON(!lreq->is_watch);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
dout("%s lreq %p notify_id %llu notifier_id %llu payload_len %zu\n",
__func__, lreq, lwork->notify.notify_id, lwork->notify.notifier_id,
lwork->notify.payload_len);
lreq->wcb(lreq->data, lwork->notify.notify_id, lreq->linger_id,
lwork->notify.notifier_id, lwork->notify.payload,
lwork->notify.payload_len);
out:
ceph_msg_put(lwork->notify.msg);
lwork_free(lwork);
}
static void do_watch_error(struct work_struct *w)
{
struct linger_work *lwork = container_of(w, struct linger_work, work);
struct ceph_osd_linger_request *lreq = lwork->lreq;
if (!linger_registered(lreq)) {
dout("%s lreq %p not registered\n", __func__, lreq);
goto out;
}
dout("%s lreq %p err %d\n", __func__, lreq, lwork->error.err);
lreq->errcb(lreq->data, lreq->linger_id, lwork->error.err);
out:
lwork_free(lwork);
}
static void queue_watch_error(struct ceph_osd_linger_request *lreq)
{
struct linger_work *lwork;
lwork = lwork_alloc(lreq, do_watch_error);
if (!lwork) {
pr_err("failed to allocate error-lwork\n");
return;
}
lwork->error.err = lreq->last_error;
lwork_queue(lwork);
}
static void linger_reg_commit_complete(struct ceph_osd_linger_request *lreq,
int result)
{
if (!completion_done(&lreq->reg_commit_wait)) {
lreq->reg_commit_error = (result <= 0 ? result : 0);
complete_all(&lreq->reg_commit_wait);
}
}
static void linger_commit_cb(struct ceph_osd_request *req)
{
struct ceph_osd_linger_request *lreq = req->r_priv;
mutex_lock(&lreq->lock);
dout("%s lreq %p linger_id %llu result %d\n", __func__, lreq,
lreq->linger_id, req->r_result);
linger_reg_commit_complete(lreq, req->r_result);
lreq->committed = true;
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
if (!lreq->is_watch) {
struct ceph_osd_data *osd_data =
osd_req_op_data(req, 0, notify, response_data);
void *p = page_address(osd_data->pages[0]);
WARN_ON(req->r_ops[0].op != CEPH_OSD_OP_NOTIFY ||
osd_data->type != CEPH_OSD_DATA_TYPE_PAGES);
/* make note of the notify_id */
if (req->r_ops[0].outdata_len >= sizeof(u64)) {
lreq->notify_id = ceph_decode_64(&p);
dout("lreq %p notify_id %llu\n", lreq,
lreq->notify_id);
} else {
dout("lreq %p no notify_id\n", lreq);
}
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
mutex_unlock(&lreq->lock);
linger_put(lreq);
}
static int normalize_watch_error(int err)
{
/*
* Translate ENOENT -> ENOTCONN so that a delete->disconnection
* notification and a failure to reconnect because we raced with
* the delete appear the same to the user.
*/
if (err == -ENOENT)
err = -ENOTCONN;
return err;
}
static void linger_reconnect_cb(struct ceph_osd_request *req)
{
struct ceph_osd_linger_request *lreq = req->r_priv;
mutex_lock(&lreq->lock);
dout("%s lreq %p linger_id %llu result %d last_error %d\n", __func__,
lreq, lreq->linger_id, req->r_result, lreq->last_error);
if (req->r_result < 0) {
if (!lreq->last_error) {
lreq->last_error = normalize_watch_error(req->r_result);
queue_watch_error(lreq);
}
}
mutex_unlock(&lreq->lock);
linger_put(lreq);
}
static void send_linger(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_request *req = lreq->reg_req;
struct ceph_osd_req_op *op = &req->r_ops[0];
verify_osdc_wrlocked(req->r_osdc);
dout("%s lreq %p linger_id %llu\n", __func__, lreq, lreq->linger_id);
if (req->r_osd)
cancel_linger_request(req);
request_reinit(req);
ceph_oid_copy(&req->r_base_oid, &lreq->t.base_oid);
ceph_oloc_copy(&req->r_base_oloc, &lreq->t.base_oloc);
req->r_flags = lreq->t.flags;
req->r_mtime = lreq->mtime;
mutex_lock(&lreq->lock);
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
if (lreq->is_watch && lreq->committed) {
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
WARN_ON(op->op != CEPH_OSD_OP_WATCH ||
op->watch.cookie != lreq->linger_id);
op->watch.op = CEPH_OSD_WATCH_OP_RECONNECT;
op->watch.gen = ++lreq->register_gen;
dout("lreq %p reconnect register_gen %u\n", lreq,
op->watch.gen);
req->r_callback = linger_reconnect_cb;
} else {
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
if (!lreq->is_watch)
lreq->notify_id = 0;
else
WARN_ON(op->watch.op != CEPH_OSD_WATCH_OP_WATCH);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
dout("lreq %p register\n", lreq);
req->r_callback = linger_commit_cb;
}
mutex_unlock(&lreq->lock);
req->r_priv = linger_get(lreq);
req->r_linger = true;
submit_request(req, true);
}
static void linger_ping_cb(struct ceph_osd_request *req)
{
struct ceph_osd_linger_request *lreq = req->r_priv;
mutex_lock(&lreq->lock);
dout("%s lreq %p linger_id %llu result %d ping_sent %lu last_error %d\n",
__func__, lreq, lreq->linger_id, req->r_result, lreq->ping_sent,
lreq->last_error);
if (lreq->register_gen == req->r_ops[0].watch.gen) {
libceph: support for checking on status of watch Implement ceph_osdc_watch_check() to be able to check on status of watch. Note that the time it takes for a watch/notify event to get delivered through the notify_wq is taken into account. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
if (!req->r_result) {
lreq->watch_valid_thru = lreq->ping_sent;
} else if (!lreq->last_error) {
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
lreq->last_error = normalize_watch_error(req->r_result);
queue_watch_error(lreq);
}
} else {
dout("lreq %p register_gen %u ignoring old pong %u\n", lreq,
lreq->register_gen, req->r_ops[0].watch.gen);
}
mutex_unlock(&lreq->lock);
linger_put(lreq);
}
static void send_linger_ping(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
struct ceph_osd_request *req = lreq->ping_req;
struct ceph_osd_req_op *op = &req->r_ops[0];
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD)) {
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
dout("%s PAUSERD\n", __func__);
return;
}
lreq->ping_sent = jiffies;
dout("%s lreq %p linger_id %llu ping_sent %lu register_gen %u\n",
__func__, lreq, lreq->linger_id, lreq->ping_sent,
lreq->register_gen);
if (req->r_osd)
cancel_linger_request(req);
request_reinit(req);
target_copy(&req->r_t, &lreq->t);
WARN_ON(op->op != CEPH_OSD_OP_WATCH ||
op->watch.cookie != lreq->linger_id ||
op->watch.op != CEPH_OSD_WATCH_OP_PING);
op->watch.gen = lreq->register_gen;
req->r_callback = linger_ping_cb;
req->r_priv = linger_get(lreq);
req->r_linger = true;
ceph_osdc_get_request(req);
account_request(req);
req->r_tid = atomic64_inc_return(&osdc->last_tid);
link_request(lreq->osd, req);
send_request(req);
}
static void linger_submit(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
struct ceph_osd *osd;
libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
calc_target(osdc, &lreq->t, NULL, false);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
osd = lookup_create_osd(osdc, lreq->t.osd, true);
link_linger(osd, lreq);
send_linger(lreq);
}
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
static void cancel_linger_map_check(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
struct ceph_osd_linger_request *lookup_lreq;
verify_osdc_wrlocked(osdc);
lookup_lreq = lookup_linger_mc(&osdc->linger_map_checks,
lreq->linger_id);
if (!lookup_lreq)
return;
WARN_ON(lookup_lreq != lreq);
erase_linger_mc(&osdc->linger_map_checks, lreq);
linger_put(lreq);
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
/*
* @lreq has to be both registered and linked.
*/
static void __linger_cancel(struct ceph_osd_linger_request *lreq)
{
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
if (lreq->is_watch && lreq->ping_req->r_osd)
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
cancel_linger_request(lreq->ping_req);
if (lreq->reg_req->r_osd)
cancel_linger_request(lreq->reg_req);
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
cancel_linger_map_check(lreq);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
unlink_linger(lreq->osd, lreq);
linger_unregister(lreq);
}
static void linger_cancel(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
down_write(&osdc->lock);
if (__linger_registered(lreq))
__linger_cancel(lreq);
up_write(&osdc->lock);
}
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
static void send_linger_map_check(struct ceph_osd_linger_request *lreq);
static void check_linger_pool_dne(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
struct ceph_osdmap *map = osdc->osdmap;
verify_osdc_wrlocked(osdc);
WARN_ON(!map->epoch);
if (lreq->register_gen) {
lreq->map_dne_bound = map->epoch;
dout("%s lreq %p linger_id %llu pool disappeared\n", __func__,
lreq, lreq->linger_id);
} else {
dout("%s lreq %p linger_id %llu map_dne_bound %u have %u\n",
__func__, lreq, lreq->linger_id, lreq->map_dne_bound,
map->epoch);
}
if (lreq->map_dne_bound) {
if (map->epoch >= lreq->map_dne_bound) {
/* we had a new enough map */
pr_info("linger_id %llu pool does not exist\n",
lreq->linger_id);
linger_reg_commit_complete(lreq, -ENOENT);
__linger_cancel(lreq);
}
} else {
send_linger_map_check(lreq);
}
}
static void linger_map_check_cb(struct ceph_mon_generic_request *greq)
{
struct ceph_osd_client *osdc = &greq->monc->client->osdc;
struct ceph_osd_linger_request *lreq;
u64 linger_id = greq->private_data;
WARN_ON(greq->result || !greq->u.newest);
down_write(&osdc->lock);
lreq = lookup_linger_mc(&osdc->linger_map_checks, linger_id);
if (!lreq) {
dout("%s linger_id %llu dne\n", __func__, linger_id);
goto out_unlock;
}
dout("%s lreq %p linger_id %llu map_dne_bound %u newest %llu\n",
__func__, lreq, lreq->linger_id, lreq->map_dne_bound,
greq->u.newest);
if (!lreq->map_dne_bound)
lreq->map_dne_bound = greq->u.newest;
erase_linger_mc(&osdc->linger_map_checks, lreq);
check_linger_pool_dne(lreq);
linger_put(lreq);
out_unlock:
up_write(&osdc->lock);
}
static void send_linger_map_check(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
struct ceph_osd_linger_request *lookup_lreq;
int ret;
verify_osdc_wrlocked(osdc);
lookup_lreq = lookup_linger_mc(&osdc->linger_map_checks,
lreq->linger_id);
if (lookup_lreq) {
WARN_ON(lookup_lreq != lreq);
return;
}
linger_get(lreq);
insert_linger_mc(&osdc->linger_map_checks, lreq);
ret = ceph_monc_get_version_async(&osdc->client->monc, "osdmap",
linger_map_check_cb, lreq->linger_id);
WARN_ON(ret);
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
static int linger_reg_commit_wait(struct ceph_osd_linger_request *lreq)
{
int ret;
dout("%s lreq %p linger_id %llu\n", __func__, lreq, lreq->linger_id);
ret = wait_for_completion_interruptible(&lreq->reg_commit_wait);
return ret ?: lreq->reg_commit_error;
}
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
static int linger_notify_finish_wait(struct ceph_osd_linger_request *lreq)
{
int ret;
dout("%s lreq %p linger_id %llu\n", __func__, lreq, lreq->linger_id);
ret = wait_for_completion_interruptible(&lreq->notify_finish_wait);
return ret ?: lreq->notify_finish_error;
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
libceph: schedule tick from ceph_osdc_init() Both homeless OSD sessions and watch/notify v2, introduced in later commits, require periodic ticks which don't depend on ->num_requests. Schedule the initial tick from ceph_osdc_init() and reschedule from handle_timeout() unconditionally. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
* Timeout callback, called every N seconds. When 1 or more OSD
* requests has been active for more than N seconds, we send a keepalive
* (tag + timestamp) to its OSD to ensure any communications channel
* reset is detected.
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
*/
static void handle_timeout(struct work_struct *work)
{
struct ceph_osd_client *osdc =
container_of(work, struct ceph_osd_client, timeout_work.work);
libceph: store timeouts in jiffies, verify user input There are currently three libceph-level timeouts that the user can specify on mount: mount_timeout, osd_idle_ttl and osdkeepalive. All of these are in seconds and no checking is done on user input: negative values are accepted, we multiply them all by HZ which may or may not overflow, arbitrarily large jiffies then get added together, etc. There is also a bug in the way mount_timeout=0 is handled. It's supposed to mean "infinite timeout", but that's not how wait.h APIs treat it and so __ceph_open_session() for example will busy loop without much chance of being interrupted if none of ceph-mons are there. Fix all this by verifying user input, storing timeouts capped by msecs_to_jiffies() in jiffies and using the new ceph_timeout_jiffies() helper for all user-specified waits to handle infinite timeouts correctly. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-05-15 09:02:17 +00:00
struct ceph_options *opts = osdc->client->options;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
unsigned long cutoff = jiffies - opts->osd_keepalive_timeout;
libceph: osd_request_timeout option osd_request_timeout specifies how many seconds to wait for a response from OSDs before returning -ETIMEDOUT from an OSD request. 0 (default) means no limit. osd_request_timeout is osdkeepalive-precise -- in-flight requests are swept through every osdkeepalive seconds. With ack vs commit behaviour gone, abort_request() is really simple. This is based on a patch from Artur Molchanov <artur.molchanov@synesis.ru>. Tested-by: Artur Molchanov <artur.molchanov@synesis.ru> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-12 16:11:07 +00:00
unsigned long expiry_cutoff = jiffies - opts->osd_request_timeout;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
LIST_HEAD(slow_osds);
struct rb_node *n, *p;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("%s osdc %p\n", __func__, osdc);
down_write(&osdc->lock);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
/*
* ping osds that are a bit slow. this ensures that if there
* is a break in the TCP connection we will notice, and reopen
* a connection with that osd (from the fault callback).
*/
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
for (n = rb_first(&osdc->osds); n; n = rb_next(n)) {
struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node);
bool found = false;
libceph: osd_request_timeout option osd_request_timeout specifies how many seconds to wait for a response from OSDs before returning -ETIMEDOUT from an OSD request. 0 (default) means no limit. osd_request_timeout is osdkeepalive-precise -- in-flight requests are swept through every osdkeepalive seconds. With ack vs commit behaviour gone, abort_request() is really simple. This is based on a patch from Artur Molchanov <artur.molchanov@synesis.ru>. Tested-by: Artur Molchanov <artur.molchanov@synesis.ru> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-12 16:11:07 +00:00
for (p = rb_first(&osd->o_requests); p; ) {
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd_request *req =
rb_entry(p, struct ceph_osd_request, r_node);
libceph: osd_request_timeout option osd_request_timeout specifies how many seconds to wait for a response from OSDs before returning -ETIMEDOUT from an OSD request. 0 (default) means no limit. osd_request_timeout is osdkeepalive-precise -- in-flight requests are swept through every osdkeepalive seconds. With ack vs commit behaviour gone, abort_request() is really simple. This is based on a patch from Artur Molchanov <artur.molchanov@synesis.ru>. Tested-by: Artur Molchanov <artur.molchanov@synesis.ru> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-12 16:11:07 +00:00
p = rb_next(p); /* abort_request() */
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
if (time_before(req->r_stamp, cutoff)) {
dout(" req %p tid %llu on osd%d is laggy\n",
req, req->r_tid, osd->o_osd);
found = true;
}
libceph: osd_request_timeout option osd_request_timeout specifies how many seconds to wait for a response from OSDs before returning -ETIMEDOUT from an OSD request. 0 (default) means no limit. osd_request_timeout is osdkeepalive-precise -- in-flight requests are swept through every osdkeepalive seconds. With ack vs commit behaviour gone, abort_request() is really simple. This is based on a patch from Artur Molchanov <artur.molchanov@synesis.ru>. Tested-by: Artur Molchanov <artur.molchanov@synesis.ru> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-12 16:11:07 +00:00
if (opts->osd_request_timeout &&
time_before(req->r_start_stamp, expiry_cutoff)) {
pr_err_ratelimited("tid %llu on osd%d timeout\n",
req->r_tid, osd->o_osd);
abort_request(req, -ETIMEDOUT);
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
for (p = rb_first(&osd->o_linger_requests); p; p = rb_next(p)) {
struct ceph_osd_linger_request *lreq =
rb_entry(p, struct ceph_osd_linger_request, node);
dout(" lreq %p linger_id %llu is served by osd%d\n",
lreq, lreq->linger_id, osd->o_osd);
found = true;
mutex_lock(&lreq->lock);
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
if (lreq->is_watch && lreq->committed && !lreq->last_error)
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
send_linger_ping(lreq);
mutex_unlock(&lreq->lock);
}
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
if (found)
list_move_tail(&osd->o_keepalive_item, &slow_osds);
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
libceph: osd_request_timeout option osd_request_timeout specifies how many seconds to wait for a response from OSDs before returning -ETIMEDOUT from an OSD request. 0 (default) means no limit. osd_request_timeout is osdkeepalive-precise -- in-flight requests are swept through every osdkeepalive seconds. With ack vs commit behaviour gone, abort_request() is really simple. This is based on a patch from Artur Molchanov <artur.molchanov@synesis.ru>. Tested-by: Artur Molchanov <artur.molchanov@synesis.ru> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-12 16:11:07 +00:00
if (opts->osd_request_timeout) {
for (p = rb_first(&osdc->homeless_osd.o_requests); p; ) {
struct ceph_osd_request *req =
rb_entry(p, struct ceph_osd_request, r_node);
p = rb_next(p); /* abort_request() */
if (time_before(req->r_start_stamp, expiry_cutoff)) {
pr_err_ratelimited("tid %llu on osd%d timeout\n",
req->r_tid, osdc->homeless_osd.o_osd);
abort_request(req, -ETIMEDOUT);
}
}
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
if (atomic_read(&osdc->num_homeless) || !list_empty(&slow_osds))
maybe_request_map(osdc);
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
while (!list_empty(&slow_osds)) {
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd *osd = list_first_entry(&slow_osds,
struct ceph_osd,
o_keepalive_item);
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
list_del_init(&osd->o_keepalive_item);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
ceph_con_keepalive(&osd->o_con);
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
up_write(&osdc->lock);
libceph: schedule tick from ceph_osdc_init() Both homeless OSD sessions and watch/notify v2, introduced in later commits, require periodic ticks which don't depend on ->num_requests. Schedule the initial tick from ceph_osdc_init() and reschedule from handle_timeout() unconditionally. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
schedule_delayed_work(&osdc->timeout_work,
osdc->client->options->osd_keepalive_timeout);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
static void handle_osds_timeout(struct work_struct *work)
{
struct ceph_osd_client *osdc =
container_of(work, struct ceph_osd_client,
osds_timeout_work.work);
libceph: store timeouts in jiffies, verify user input There are currently three libceph-level timeouts that the user can specify on mount: mount_timeout, osd_idle_ttl and osdkeepalive. All of these are in seconds and no checking is done on user input: negative values are accepted, we multiply them all by HZ which may or may not overflow, arbitrarily large jiffies then get added together, etc. There is also a bug in the way mount_timeout=0 is handled. It's supposed to mean "infinite timeout", but that's not how wait.h APIs treat it and so __ceph_open_session() for example will busy loop without much chance of being interrupted if none of ceph-mons are there. Fix all this by verifying user input, storing timeouts capped by msecs_to_jiffies() in jiffies and using the new ceph_timeout_jiffies() helper for all user-specified waits to handle infinite timeouts correctly. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-05-15 09:02:17 +00:00
unsigned long delay = osdc->client->options->osd_idle_ttl / 4;
libceph: open-code remove_{all,old}_osds() They are called only once, from ceph_osdc_stop() and handle_osds_timeout() respectively. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
struct ceph_osd *osd, *nosd;
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
libceph: open-code remove_{all,old}_osds() They are called only once, from ceph_osdc_stop() and handle_osds_timeout() respectively. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
dout("%s osdc %p\n", __func__, osdc);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
down_write(&osdc->lock);
libceph: open-code remove_{all,old}_osds() They are called only once, from ceph_osdc_stop() and handle_osds_timeout() respectively. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
list_for_each_entry_safe(osd, nosd, &osdc->osd_lru, o_osd_lru) {
if (time_before(jiffies, osd->lru_ttl))
break;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
WARN_ON(!RB_EMPTY_ROOT(&osd->o_requests));
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
WARN_ON(!RB_EMPTY_ROOT(&osd->o_linger_requests));
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
close_osd(osd);
libceph: open-code remove_{all,old}_osds() They are called only once, from ceph_osdc_stop() and handle_osds_timeout() respectively. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
}
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
up_write(&osdc->lock);
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
schedule_delayed_work(&osdc->osds_timeout_work,
round_jiffies_relative(delay));
}
libceph: follow redirect replies from osds Follow redirect replies from osds, for details see ceph.git commit fbbe3ad1220799b7bb00ea30fce581c5eadaf034. v1 (current) version of redirect reply consists of oloc and oid, which expands to pool, key, nspace, hash and oid. However, server-side code that would populate anything other than pool doesn't exist yet, and hence this commit adds support for pool redirects only. To make sure that future server-side updates don't break us, we decode all fields and, if any of key, nspace, hash or oid have a non-default value, error out with "corrupt osd_op_reply ..." message. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-27 15:40:20 +00:00
static int ceph_oloc_decode(void **p, void *end,
struct ceph_object_locator *oloc)
{
u8 struct_v, struct_cv;
u32 len;
void *struct_end;
int ret = 0;
ceph_decode_need(p, end, 1 + 1 + 4, e_inval);
struct_v = ceph_decode_8(p);
struct_cv = ceph_decode_8(p);
if (struct_v < 3) {
pr_warn("got v %d < 3 cv %d of ceph_object_locator\n",
struct_v, struct_cv);
goto e_inval;
}
if (struct_cv > 6) {
pr_warn("got v %d cv %d > 6 of ceph_object_locator\n",
struct_v, struct_cv);
goto e_inval;
}
len = ceph_decode_32(p);
ceph_decode_need(p, end, len, e_inval);
struct_end = *p + len;
oloc->pool = ceph_decode_64(p);
*p += 4; /* skip preferred */
len = ceph_decode_32(p);
if (len > 0) {
pr_warn("ceph_object_locator::key is set\n");
goto e_inval;
}
if (struct_v >= 5) {
libceph: make sure redirect does not change namespace Signed-off-by: Yan, Zheng <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-06-13 11:05:13 +00:00
bool changed = false;
libceph: follow redirect replies from osds Follow redirect replies from osds, for details see ceph.git commit fbbe3ad1220799b7bb00ea30fce581c5eadaf034. v1 (current) version of redirect reply consists of oloc and oid, which expands to pool, key, nspace, hash and oid. However, server-side code that would populate anything other than pool doesn't exist yet, and hence this commit adds support for pool redirects only. To make sure that future server-side updates don't break us, we decode all fields and, if any of key, nspace, hash or oid have a non-default value, error out with "corrupt osd_op_reply ..." message. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-27 15:40:20 +00:00
len = ceph_decode_32(p);
if (len > 0) {
libceph: rados pool namespace support Add pool namesapce pointer to struct ceph_file_layout and struct ceph_object_locator. Pool namespace is used by when mapping object to PG, it's also used when composing OSD request. The namespace pointer in struct ceph_file_layout is RCU protected. So libceph can read namespace without taking lock. Signed-off-by: Yan, Zheng <zyan@redhat.com> [idryomov@gmail.com: ceph_oloc_destroy(), misc minor changes] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-14 03:24:31 +00:00
ceph_decode_need(p, end, len, e_inval);
libceph: make sure redirect does not change namespace Signed-off-by: Yan, Zheng <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-06-13 11:05:13 +00:00
if (!oloc->pool_ns ||
ceph_compare_string(oloc->pool_ns, *p, len))
changed = true;
libceph: rados pool namespace support Add pool namesapce pointer to struct ceph_file_layout and struct ceph_object_locator. Pool namespace is used by when mapping object to PG, it's also used when composing OSD request. The namespace pointer in struct ceph_file_layout is RCU protected. So libceph can read namespace without taking lock. Signed-off-by: Yan, Zheng <zyan@redhat.com> [idryomov@gmail.com: ceph_oloc_destroy(), misc minor changes] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-14 03:24:31 +00:00
*p += len;
libceph: make sure redirect does not change namespace Signed-off-by: Yan, Zheng <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-06-13 11:05:13 +00:00
} else {
if (oloc->pool_ns)
changed = true;
}
if (changed) {
/* redirect changes namespace */
pr_warn("ceph_object_locator::nspace is changed\n");
goto e_inval;
libceph: follow redirect replies from osds Follow redirect replies from osds, for details see ceph.git commit fbbe3ad1220799b7bb00ea30fce581c5eadaf034. v1 (current) version of redirect reply consists of oloc and oid, which expands to pool, key, nspace, hash and oid. However, server-side code that would populate anything other than pool doesn't exist yet, and hence this commit adds support for pool redirects only. To make sure that future server-side updates don't break us, we decode all fields and, if any of key, nspace, hash or oid have a non-default value, error out with "corrupt osd_op_reply ..." message. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-27 15:40:20 +00:00
}
}
if (struct_v >= 6) {
s64 hash = ceph_decode_64(p);
if (hash != -1) {
pr_warn("ceph_object_locator::hash is set\n");
goto e_inval;
}
}
/* skip the rest */
*p = struct_end;
out:
return ret;
e_inval:
ret = -EINVAL;
goto out;
}
static int ceph_redirect_decode(void **p, void *end,
struct ceph_request_redirect *redir)
{
u8 struct_v, struct_cv;
u32 len;
void *struct_end;
int ret;
ceph_decode_need(p, end, 1 + 1 + 4, e_inval);
struct_v = ceph_decode_8(p);
struct_cv = ceph_decode_8(p);
if (struct_cv > 1) {
pr_warn("got v %d cv %d > 1 of ceph_request_redirect\n",
struct_v, struct_cv);
goto e_inval;
}
len = ceph_decode_32(p);
ceph_decode_need(p, end, len, e_inval);
struct_end = *p + len;
ret = ceph_oloc_decode(p, end, &redir->oloc);
if (ret)
goto out;
len = ceph_decode_32(p);
if (len > 0) {
pr_warn("ceph_request_redirect::object_name is set\n");
goto e_inval;
}
len = ceph_decode_32(p);
*p += len; /* skip osd_instructions */
/* skip the rest */
*p = struct_end;
out:
return ret;
e_inval:
ret = -EINVAL;
goto out;
}
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
struct MOSDOpReply {
struct ceph_pg pgid;
u64 flags;
int result;
u32 epoch;
int num_ops;
u32 outdata_len[CEPH_OSD_MAX_OPS];
s32 rval[CEPH_OSD_MAX_OPS];
int retry_attempt;
struct ceph_eversion replay_version;
u64 user_version;
struct ceph_request_redirect redirect;
};
ceph: fix sync vs canceled write If we cancel a write, trigger the safe completions to prevent a sync from blocking indefinitely in ceph_osdc_sync(). Signed-off-by: Sage Weil <sage@newdream.net>
2011-06-03 16:37:09 +00:00
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
static int decode_MOSDOpReply(const struct ceph_msg *msg, struct MOSDOpReply *m)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
void *p = msg->front.iov_base;
void *const end = p + msg->front.iov_len;
u16 version = le16_to_cpu(msg->hdr.version);
struct ceph_eversion bad_replay_version;
libceph: MOSDOpReply v7 encoding Empty request_redirect_t (struct ceph_request_redirect in the kernel client) is now encoded with a bool. NEW_OSDOPREPLY_ENCODING feature bit overlaps with already supported CRUSH_TUNABLES5. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2016-02-03 14:25:48 +00:00
u8 decode_redir;
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
u32 len;
int ret;
int i;
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
ceph_decode_32_safe(&p, end, len, e_inval);
ceph_decode_need(&p, end, len, e_inval);
p += len; /* skip oid */
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
ret = ceph_decode_pgid(&p, end, &m->pgid);
if (ret)
return ret;
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
ceph_decode_64_safe(&p, end, m->flags, e_inval);
ceph_decode_32_safe(&p, end, m->result, e_inval);
ceph_decode_need(&p, end, sizeof(bad_replay_version), e_inval);
memcpy(&bad_replay_version, p, sizeof(bad_replay_version));
p += sizeof(bad_replay_version);
ceph_decode_32_safe(&p, end, m->epoch, e_inval);
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
ceph_decode_32_safe(&p, end, m->num_ops, e_inval);
if (m->num_ops > ARRAY_SIZE(m->outdata_len))
goto e_inval;
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
ceph_decode_need(&p, end, m->num_ops * sizeof(struct ceph_osd_op),
e_inval);
for (i = 0; i < m->num_ops; i++) {
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
struct ceph_osd_op *op = p;
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
m->outdata_len[i] = le32_to_cpu(op->payload_len);
libceph: update osd request/reply encoding Use the new version of the encoding for osd requests and replies. In the process, update the way we are tracking request ops and reply lengths and results in the struct ceph_osd_request. Update the rbd and fs/ceph users appropriately. The main changes are: - we keep pointers into the request memory for fields we need to update each time the request is sent out over the wire - we keep information about the result in an array in the request struct where the users can easily get at it. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-02-26 00:11:12 +00:00
p += sizeof(*op);
}
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
ceph_decode_32_safe(&p, end, m->retry_attempt, e_inval);
for (i = 0; i < m->num_ops; i++)
ceph_decode_32_safe(&p, end, m->rval[i], e_inval);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
if (version >= 5) {
ceph_decode_need(&p, end, sizeof(m->replay_version), e_inval);
memcpy(&m->replay_version, p, sizeof(m->replay_version));
p += sizeof(m->replay_version);
ceph_decode_64_safe(&p, end, m->user_version, e_inval);
} else {
m->replay_version = bad_replay_version; /* struct */
m->user_version = le64_to_cpu(m->replay_version.version);
}
libceph: fix safe completion handle_reply() calls complete_request() only if the first OSD reply has ONDISK flag. Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com> Reviewed-by: Sage Weil <sage@inktank.com>
2013-05-31 07:54:44 +00:00
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
if (version >= 6) {
if (version >= 7)
ceph_decode_8_safe(&p, end, decode_redir, e_inval);
libceph: MOSDOpReply v7 encoding Empty request_redirect_t (struct ceph_request_redirect in the kernel client) is now encoded with a bool. NEW_OSDOPREPLY_ENCODING feature bit overlaps with already supported CRUSH_TUNABLES5. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2016-02-03 14:25:48 +00:00
else
decode_redir = 1;
} else {
decode_redir = 0;
}
if (decode_redir) {
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
ret = ceph_redirect_decode(&p, end, &m->redirect);
if (ret)
return ret;
libceph: follow redirect replies from osds Follow redirect replies from osds, for details see ceph.git commit fbbe3ad1220799b7bb00ea30fce581c5eadaf034. v1 (current) version of redirect reply consists of oloc and oid, which expands to pool, key, nspace, hash and oid. However, server-side code that would populate anything other than pool doesn't exist yet, and hence this commit adds support for pool redirects only. To make sure that future server-side updates don't break us, we decode all fields and, if any of key, nspace, hash or oid have a non-default value, error out with "corrupt osd_op_reply ..." message. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-27 15:40:20 +00:00
} else {
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
ceph_oloc_init(&m->redirect.oloc);
libceph: follow redirect replies from osds Follow redirect replies from osds, for details see ceph.git commit fbbe3ad1220799b7bb00ea30fce581c5eadaf034. v1 (current) version of redirect reply consists of oloc and oid, which expands to pool, key, nspace, hash and oid. However, server-side code that would populate anything other than pool doesn't exist yet, and hence this commit adds support for pool redirects only. To make sure that future server-side updates don't break us, we decode all fields and, if any of key, nspace, hash or oid have a non-default value, error out with "corrupt osd_op_reply ..." message. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-27 15:40:20 +00:00
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
return 0;
e_inval:
return -EINVAL;
}
/*
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
* Handle MOSDOpReply. Set ->r_result and call the callback if it is
* specified.
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
*/
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void handle_reply(struct ceph_osd *osd, struct ceph_msg *msg)
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd_client *osdc = osd->o_osdc;
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
struct ceph_osd_request *req;
struct MOSDOpReply m;
u64 tid = le64_to_cpu(msg->hdr.tid);
u32 data_len = 0;
int ret;
int i;
dout("%s msg %p tid %llu\n", __func__, msg, tid);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
down_read(&osdc->lock);
if (!osd_registered(osd)) {
dout("%s osd%d unknown\n", __func__, osd->o_osd);
goto out_unlock_osdc;
}
WARN_ON(osd->o_osd != le64_to_cpu(msg->hdr.src.num));
mutex_lock(&osd->lock);
req = lookup_request(&osd->o_requests, tid);
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
if (!req) {
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("%s osd%d tid %llu unknown\n", __func__, osd->o_osd, tid);
goto out_unlock_session;
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
}
libceph: make sure redirect does not change namespace Signed-off-by: Yan, Zheng <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-06-13 11:05:13 +00:00
m.redirect.oloc.pool_ns = req->r_t.target_oloc.pool_ns;
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
ret = decode_MOSDOpReply(msg, &m);
libceph: make sure redirect does not change namespace Signed-off-by: Yan, Zheng <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-06-13 11:05:13 +00:00
m.redirect.oloc.pool_ns = NULL;
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
if (ret) {
pr_err("failed to decode MOSDOpReply for tid %llu: %d\n",
req->r_tid, ret);
ceph_msg_dump(msg);
goto fail_request;
}
dout("%s req %p tid %llu flags 0x%llx pgid %llu.%x epoch %u attempt %d v %u'%llu uv %llu\n",
__func__, req, req->r_tid, m.flags, m.pgid.pool, m.pgid.seed,
m.epoch, m.retry_attempt, le32_to_cpu(m.replay_version.epoch),
le64_to_cpu(m.replay_version.version), m.user_version);
if (m.retry_attempt >= 0) {
if (m.retry_attempt != req->r_attempts - 1) {
dout("req %p tid %llu retry_attempt %d != %d, ignoring\n",
req, req->r_tid, m.retry_attempt,
req->r_attempts - 1);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
goto out_unlock_session;
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
}
} else {
WARN_ON(1); /* MOSDOpReply v4 is assumed */
}
if (!ceph_oloc_empty(&m.redirect.oloc)) {
dout("req %p tid %llu redirect pool %lld\n", req, req->r_tid,
m.redirect.oloc.pool);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
unlink_request(osd, req);
mutex_unlock(&osd->lock);
libceph: follow redirect replies from osds Follow redirect replies from osds, for details see ceph.git commit fbbe3ad1220799b7bb00ea30fce581c5eadaf034. v1 (current) version of redirect reply consists of oloc and oid, which expands to pool, key, nspace, hash and oid. However, server-side code that would populate anything other than pool doesn't exist yet, and hence this commit adds support for pool redirects only. To make sure that future server-side updates don't break us, we decode all fields and, if any of key, nspace, hash or oid have a non-default value, error out with "corrupt osd_op_reply ..." message. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-27 15:40:20 +00:00
libceph: rados pool namespace support Add pool namesapce pointer to struct ceph_file_layout and struct ceph_object_locator. Pool namespace is used by when mapping object to PG, it's also used when composing OSD request. The namespace pointer in struct ceph_file_layout is RCU protected. So libceph can read namespace without taking lock. Signed-off-by: Yan, Zheng <zyan@redhat.com> [idryomov@gmail.com: ceph_oloc_destroy(), misc minor changes] Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-14 03:24:31 +00:00
/*
* Not ceph_oloc_copy() - changing pool_ns is not
* supported.
*/
req->r_t.target_oloc.pool = m.redirect.oloc.pool;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
req->r_flags |= CEPH_OSD_FLAG_REDIRECTED;
req->r_tid = 0;
__submit_request(req, false);
goto out_unlock_osdc;
libceph: follow redirect replies from osds Follow redirect replies from osds, for details see ceph.git commit fbbe3ad1220799b7bb00ea30fce581c5eadaf034. v1 (current) version of redirect reply consists of oloc and oid, which expands to pool, key, nspace, hash and oid. However, server-side code that would populate anything other than pool doesn't exist yet, and hence this commit adds support for pool redirects only. To make sure that future server-side updates don't break us, we decode all fields and, if any of key, nspace, hash or oid have a non-default value, error out with "corrupt osd_op_reply ..." message. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-27 15:40:20 +00:00
}
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
if (m.num_ops != req->r_num_ops) {
pr_err("num_ops %d != %d for tid %llu\n", m.num_ops,
req->r_num_ops, req->r_tid);
goto fail_request;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
for (i = 0; i < req->r_num_ops; i++) {
dout(" req %p tid %llu op %d rval %d len %u\n", req,
req->r_tid, i, m.rval[i], m.outdata_len[i]);
req->r_ops[i].rval = m.rval[i];
req->r_ops[i].outdata_len = m.outdata_len[i];
data_len += m.outdata_len[i];
}
if (data_len != le32_to_cpu(msg->hdr.data_len)) {
pr_err("sum of lens %u != %u for tid %llu\n", data_len,
le32_to_cpu(msg->hdr.data_len), req->r_tid);
goto fail_request;
}
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
dout("%s req %p tid %llu result %d data_len %u\n", __func__,
req, req->r_tid, m.result, data_len);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
/*
* Since we only ever request ONDISK, we should only ever get
* one (type of) reply back.
*/
WARN_ON(!(m.flags & CEPH_OSD_FLAG_ONDISK));
req->r_result = m.result ?: data_len;
finish_request(req);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
mutex_unlock(&osd->lock);
up_read(&osdc->lock);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
__complete_request(req);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
return;
libceph: redo callbacks and factor out MOSDOpReply decoding If you specify ACK | ONDISK and set ->r_unsafe_callback, both ->r_callback and ->r_unsafe_callback(true) are called on ack. This is very confusing. Redo this so that only one of them is called: ->r_unsafe_callback(true), on ack ->r_unsafe_callback(false), on commit or ->r_callback, on ack|commit Decode everything in decode_MOSDOpReply() to reduce clutter. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
fail_request:
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
complete_request(req, -EIO);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
out_unlock_session:
mutex_unlock(&osd->lock);
out_unlock_osdc:
up_read(&osdc->lock);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
static void set_pool_was_full(struct ceph_osd_client *osdc)
{
struct rb_node *n;
for (n = rb_first(&osdc->osdmap->pg_pools); n; n = rb_next(n)) {
struct ceph_pg_pool_info *pi =
rb_entry(n, struct ceph_pg_pool_info, node);
pi->was_full = __pool_full(pi);
}
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static bool pool_cleared_full(struct ceph_osd_client *osdc, s64 pool_id)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_pg_pool_info *pi;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
pi = ceph_pg_pool_by_id(osdc->osdmap, pool_id);
if (!pi)
return false;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
return pi->was_full && !__pool_full(pi);
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
static enum calc_target_result
recalc_linger_target(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_client *osdc = lreq->osdc;
enum calc_target_result ct_res;
libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
ct_res = calc_target(osdc, &lreq->t, NULL, true);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
if (ct_res == CALC_TARGET_NEED_RESEND) {
struct ceph_osd *osd;
osd = lookup_create_osd(osdc, lreq->t.osd, true);
if (osd != lreq->osd) {
unlink_linger(lreq->osd, lreq);
link_linger(osd, lreq);
}
}
return ct_res;
}
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
/*
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
* Requeue requests whose mapping to an OSD has changed.
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
*/
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void scan_requests(struct ceph_osd *osd,
bool force_resend,
bool cleared_full,
bool check_pool_cleared_full,
struct rb_root *need_resend,
struct list_head *need_resend_linger)
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd_client *osdc = osd->o_osdc;
struct rb_node *n;
bool force_resend_writes;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
for (n = rb_first(&osd->o_linger_requests); n; ) {
struct ceph_osd_linger_request *lreq =
rb_entry(n, struct ceph_osd_linger_request, node);
enum calc_target_result ct_res;
n = rb_next(n); /* recalc_linger_target() */
dout("%s lreq %p linger_id %llu\n", __func__, lreq,
lreq->linger_id);
ct_res = recalc_linger_target(lreq);
switch (ct_res) {
case CALC_TARGET_NO_ACTION:
force_resend_writes = cleared_full ||
(check_pool_cleared_full &&
pool_cleared_full(osdc, lreq->t.base_oloc.pool));
if (!force_resend && !force_resend_writes)
break;
/* fall through */
case CALC_TARGET_NEED_RESEND:
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
cancel_linger_map_check(lreq);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
/*
* scan_requests() for the previous epoch(s)
* may have already added it to the list, since
* it's not unlinked here.
*/
if (list_empty(&lreq->scan_item))
list_add_tail(&lreq->scan_item, need_resend_linger);
break;
case CALC_TARGET_POOL_DNE:
libceph: delete from need_resend_linger before check_linger_pool_dne() When processing a map update consisting of multiple incrementals, we may end up running check_linger_pool_dne() on a lingering request that was previously added to need_resend_linger list. If it is concluded that the target pool doesn't exist, the request is killed off while still on need_resend_linger list, which leads to a crash on a NULL lreq->osd in kick_requests(): libceph: linger_id 18446462598732840961 pool does not exist BUG: unable to handle kernel NULL pointer dereference at 0000000000000010 IP: ceph_osdc_handle_map+0x4ae/0x870 Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
list_del_init(&lreq->scan_item);
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
check_linger_pool_dne(lreq);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
break;
}
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
for (n = rb_first(&osd->o_requests); n; ) {
struct ceph_osd_request *req =
rb_entry(n, struct ceph_osd_request, r_node);
enum calc_target_result ct_res;
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
n = rb_next(n); /* unlink_request(), check_pool_dne() */
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("%s req %p tid %llu\n", __func__, req, req->r_tid);
libceph: resend on PG splits if OSD has RESEND_ON_SPLIT Note that ceph_osd_request_target fields are updated regardless of RESEND_ON_SPLIT. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
ct_res = calc_target(osdc, &req->r_t, &req->r_osd->o_con,
false);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
switch (ct_res) {
case CALC_TARGET_NO_ACTION:
force_resend_writes = cleared_full ||
(check_pool_cleared_full &&
pool_cleared_full(osdc, req->r_t.base_oloc.pool));
if (!force_resend &&
(!(req->r_flags & CEPH_OSD_FLAG_WRITE) ||
!force_resend_writes))
break;
/* fall through */
case CALC_TARGET_NEED_RESEND:
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
cancel_map_check(req);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
unlink_request(osd, req);
insert_request(need_resend, req);
break;
case CALC_TARGET_POOL_DNE:
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
check_pool_dne(req);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
break;
libceph: request a new osdmap if lingering request maps to no osd This commit does two things. First, if there are any homeless lingering requests, we now request a new osdmap even if the osdmap that is being processed brought no changes, i.e. if a given lingering request turned homeless in one of the previous epochs and remained homeless in the current epoch. Not doing so leaves us with a stale osdmap and as a result we may miss our window for reestablishing the watch and lose notifies. MON=1 OSD=1: # cat linger-needmap.sh #!/bin/bash rbd create --size 1 test DEV=$(rbd map test) ceph osd out 0 rbd map dne/dne # obtain a new osdmap as a side effect (!) sleep 1 ceph osd in 0 rbd resize --size 2 test # rbd info test | grep size -> 2M # blockdev --getsize $DEV -> 1M N.B.: Not obtaining a new osdmap in between "osd out" and "osd in" above is enough to make it miss that resize notify, but that is a bug^Wlimitation of ceph watch/notify v1. Second, homeless lingering requests are now kicked just like those lingering requests whose mapping has changed. This is mainly to recognize that a homeless lingering request makes no sense and to preserve the invariant that a registered lingering request is not sitting on any of r_req_lru_item lists. This spares us a WARN_ON, which commit ba9d114ec557 ("libceph: clear r_req_lru_item in __unregister_linger_request()") tried to fix the _wrong_ way. Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2015-05-11 14:53:10 +00:00
}
libceph: fix osd request queuing on osdmap updates If we send a request to osd A, and the request's pg remaps to osd B and then back to A in quick succession, we need to resend the request to A. The old code was only calling kick_requests after processing all incremental maps in a message, so it was very possible to not resend a request that needed to be resent. This would make the osd eventually time out (at least with the current default of osd timeouts enabled). The correct approach is to scan requests on every map incremental. This patch refactors the kick code in a few ways: - all requests are either on req_lru (in flight), req_unsent (ready to send), or req_notarget (currently map to no up osd) - mapping always done by map_request (previous map_osds) - if the mapping changes, we requeue. requests are resent only after all map incrementals are processed. - some osd reset code is moved out of kick_requests into a separate function - the "kick this osd" functionality is moved to kick_osd_requests, as it is unrelated to scanning for request->pg->osd mapping changes Signed-off-by: Sage Weil <sage@newdream.net>
2011-01-18 04:34:08 +00:00
}
ceph: reset osd after relevant messages timed out This simplifies the process of timing out messages. We keep lru of current messages that are in flight. If a timeout has passed, we reset the osd connection, so that messages will be retransmitted. This is a failsafe in case we hit some sort of problem sending out message to the OSD. Normally, we'll get notification via an updated osdmap if there are problems. If a request is older than the keepalive timeout, send a keepalive to ensure we detect any breaks in the TCP connection. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-26 23:32:31 +00:00
}
libceph: fix osd request queuing on osdmap updates If we send a request to osd A, and the request's pg remaps to osd B and then back to A in quick succession, we need to resend the request to A. The old code was only calling kick_requests after processing all incremental maps in a message, so it was very possible to not resend a request that needed to be resent. This would make the osd eventually time out (at least with the current default of osd timeouts enabled). The correct approach is to scan requests on every map incremental. This patch refactors the kick code in a few ways: - all requests are either on req_lru (in flight), req_unsent (ready to send), or req_notarget (currently map to no up osd) - mapping always done by map_request (previous map_osds) - if the mapping changes, we requeue. requests are resent only after all map incrementals are processed. - some osd reset code is moved out of kick_requests into a separate function - the "kick this osd" functionality is moved to kick_osd_requests, as it is unrelated to scanning for request->pg->osd mapping changes Signed-off-by: Sage Weil <sage@newdream.net>
2011-01-18 04:34:08 +00:00
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
static int handle_one_map(struct ceph_osd_client *osdc,
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
void *p, void *end, bool incremental,
struct rb_root *need_resend,
struct list_head *need_resend_linger)
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
{
struct ceph_osdmap *newmap;
struct rb_node *n;
bool skipped_map = false;
bool was_full;
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
was_full = ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL);
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
set_pool_was_full(osdc);
if (incremental)
newmap = osdmap_apply_incremental(&p, end, osdc->osdmap);
else
newmap = ceph_osdmap_decode(&p, end);
if (IS_ERR(newmap))
return PTR_ERR(newmap);
if (newmap != osdc->osdmap) {
/*
* Preserve ->was_full before destroying the old map.
* For pools that weren't in the old map, ->was_full
* should be false.
*/
for (n = rb_first(&newmap->pg_pools); n; n = rb_next(n)) {
struct ceph_pg_pool_info *pi =
rb_entry(n, struct ceph_pg_pool_info, node);
struct ceph_pg_pool_info *old_pi;
old_pi = ceph_pg_pool_by_id(osdc->osdmap, pi->id);
if (old_pi)
pi->was_full = old_pi->was_full;
else
WARN_ON(pi->was_full);
}
if (osdc->osdmap->epoch &&
osdc->osdmap->epoch + 1 < newmap->epoch) {
WARN_ON(incremental);
skipped_map = true;
}
ceph_osdmap_destroy(osdc->osdmap);
osdc->osdmap = newmap;
}
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
was_full &= !ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
scan_requests(&osdc->homeless_osd, skipped_map, was_full, true,
need_resend, need_resend_linger);
for (n = rb_first(&osdc->osds); n; ) {
struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node);
n = rb_next(n); /* close_osd() */
scan_requests(osd, skipped_map, was_full, true, need_resend,
need_resend_linger);
if (!ceph_osd_is_up(osdc->osdmap, osd->o_osd) ||
memcmp(&osd->o_con.peer_addr,
ceph_osd_addr(osdc->osdmap, osd->o_osd),
sizeof(struct ceph_entity_addr)))
close_osd(osd);
}
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
return 0;
}
libceph: fix osd request queuing on osdmap updates If we send a request to osd A, and the request's pg remaps to osd B and then back to A in quick succession, we need to resend the request to A. The old code was only calling kick_requests after processing all incremental maps in a message, so it was very possible to not resend a request that needed to be resent. This would make the osd eventually time out (at least with the current default of osd timeouts enabled). The correct approach is to scan requests on every map incremental. This patch refactors the kick code in a few ways: - all requests are either on req_lru (in flight), req_unsent (ready to send), or req_notarget (currently map to no up osd) - mapping always done by map_request (previous map_osds) - if the mapping changes, we requeue. requests are resent only after all map incrementals are processed. - some osd reset code is moved out of kick_requests into a separate function - the "kick this osd" functionality is moved to kick_osd_requests, as it is unrelated to scanning for request->pg->osd mapping changes Signed-off-by: Sage Weil <sage@newdream.net>
2011-01-18 04:34:08 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static void kick_requests(struct ceph_osd_client *osdc,
struct rb_root *need_resend,
struct list_head *need_resend_linger)
{
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
struct ceph_osd_linger_request *lreq, *nlreq;
libceph: make sure need_resend targets reflect latest map Otherwise we may miss events like PG splits, pool deletions, etc when we get multiple incremental maps at once. Because check_pool_dne() can now be fed an unlinked request, finish_request() needed to be taught to handle unlinked requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
enum calc_target_result ct_res;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct rb_node *n;
libceph: make sure need_resend targets reflect latest map Otherwise we may miss events like PG splits, pool deletions, etc when we get multiple incremental maps at once. Because check_pool_dne() can now be fed an unlinked request, finish_request() needed to be taught to handle unlinked requests. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:55 +00:00
/* make sure need_resend targets reflect latest map */
for (n = rb_first(need_resend); n; ) {
struct ceph_osd_request *req =
rb_entry(n, struct ceph_osd_request, r_node);
n = rb_next(n);
if (req->r_t.epoch < osdc->osdmap->epoch) {
ct_res = calc_target(osdc, &req->r_t, NULL, false);
if (ct_res == CALC_TARGET_POOL_DNE) {
erase_request(need_resend, req);
check_pool_dne(req);
}
}
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
for (n = rb_first(need_resend); n; ) {
struct ceph_osd_request *req =
rb_entry(n, struct ceph_osd_request, r_node);
struct ceph_osd *osd;
n = rb_next(n);
erase_request(need_resend, req); /* before link_request() */
osd = lookup_create_osd(osdc, req->r_t.osd, true);
link_request(osd, req);
if (!req->r_linger) {
if (!osd_homeless(osd) && !req->r_t.paused)
send_request(req);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
} else {
cancel_linger_request(req);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
}
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
list_for_each_entry_safe(lreq, nlreq, need_resend_linger, scan_item) {
if (!osd_homeless(lreq->osd))
send_linger(lreq);
list_del_init(&lreq->scan_item);
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* Process updated osd map.
*
* The message contains any number of incremental and full maps, normally
* indicating some sort of topology change in the cluster. Kick requests
* off to different OSDs as needed.
*/
void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg)
{
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
void *p = msg->front.iov_base;
void *const end = p + msg->front.iov_len;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
u32 nr_maps, maplen;
u32 epoch;
struct ceph_fsid fsid;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct rb_root need_resend = RB_ROOT;
LIST_HEAD(need_resend_linger);
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
bool handled_incremental = false;
bool was_pauserd, was_pausewr;
bool pauserd, pausewr;
int err;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
dout("%s have %u\n", __func__, osdc->osdmap->epoch);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
down_write(&osdc->lock);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/* verify fsid */
ceph_decode_need(&p, end, sizeof(fsid), bad);
ceph_decode_copy(&p, &fsid, sizeof(fsid));
ceph: fix debugfs entry, simplify fsid checks We may first learn our fsid from any of the mon, osd, or mds maps (whichever the monitor sends first). Consolidate checks in a single helper. Initialize the client debugfs entry then, since we need the fsid (and global_id) for the directory name. Also remove dead mount code. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:50:41 +00:00
if (ceph_check_fsid(osdc->client, &fsid) < 0)
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
goto bad;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
was_pauserd = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD);
was_pausewr = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR) ||
ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
have_pool_full(osdc);
libceph: resend all writes after the osdmap loses the full flag With the current full handling, there is a race between osds and clients getting the first map marked full. If the osd wins, it will return -ENOSPC to any writes, but the client may already have writes in flight. This results in the client getting the error and propagating it up the stack. For rbd, the block layer turns this into EIO, which can cause corruption in filesystems above it. To avoid this race, osds are being changed to drop writes that came from clients with an osdmap older than the last osdmap marked full. In order for this to work, clients must resend all writes after they encounter a full -> not full transition in the osdmap. osds will wait for an updated map instead of processing a request from a client with a newer map, so resent writes will not be dropped by the osd unless there is another not full -> full transition. This approach requires both osds and clients to be fixed to avoid the race. Old clients talking to osds with this fix may hang instead of returning EIO and potentially corrupting an fs. New clients talking to old osds have the same behavior as before if they encounter this race. Fixes: http://tracker.ceph.com/issues/6938 Reviewed-by: Sage Weil <sage@inktank.com> Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
2013-12-10 17:35:13 +00:00
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/* incremental maps */
ceph_decode_32_safe(&p, end, nr_maps, bad);
dout(" %d inc maps\n", nr_maps);
while (nr_maps > 0) {
ceph_decode_need(&p, end, 2*sizeof(u32), bad);
ceph: convert encode/decode macros to inlines This avoids the fugly pass by reference and makes the code a bit easier to read. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-14 16:59:09 +00:00
epoch = ceph_decode_32(&p);
maplen = ceph_decode_32(&p);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
ceph_decode_need(&p, end, maplen, bad);
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
if (osdc->osdmap->epoch &&
osdc->osdmap->epoch + 1 == epoch) {
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
dout("applying incremental map %u len %d\n",
epoch, maplen);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
err = handle_one_map(osdc, p, p + maplen, true,
&need_resend, &need_resend_linger);
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
if (err)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
goto bad;
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
handled_incremental = true;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
} else {
dout("ignoring incremental map %u len %d\n",
epoch, maplen);
}
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
p += maplen;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
nr_maps--;
}
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
if (handled_incremental)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
goto done;
/* full maps */
ceph_decode_32_safe(&p, end, nr_maps, bad);
dout(" %d full maps\n", nr_maps);
while (nr_maps) {
ceph_decode_need(&p, end, 2*sizeof(u32), bad);
ceph: convert encode/decode macros to inlines This avoids the fugly pass by reference and makes the code a bit easier to read. Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-14 16:59:09 +00:00
epoch = ceph_decode_32(&p);
maplen = ceph_decode_32(&p);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
ceph_decode_need(&p, end, maplen, bad);
if (nr_maps > 1) {
dout("skipping non-latest full map %u len %d\n",
epoch, maplen);
libceph: allocate dummy osdmap in ceph_osdc_init() This leads to a simpler osdmap handling code, particularly when dealing with pi->was_full, which is introduced in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
} else if (osdc->osdmap->epoch >= epoch) {
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
dout("skipping full map %u len %d, "
"older than our %u\n", epoch, maplen,
osdc->osdmap->epoch);
} else {
dout("taking full map %u len %d\n", epoch, maplen);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
err = handle_one_map(osdc, p, p + maplen, false,
&need_resend, &need_resend_linger);
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
if (err)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
goto bad;
}
p += maplen;
nr_maps--;
}
done:
libceph: subscribe to osdmap when cluster is full When the cluster is marked full, subscribe to subsequent map updates to ensure we find out promptly when it is no longer full. This will prevent us from spewing ENOSPC for (much) longer than necessary. Signed-off-by: Sage Weil <sage@newdream.net>
2011-05-12 16:29:18 +00:00
/*
* subscribe to subsequent osdmap updates if full to ensure
* we find out when we are no longer full and stop returning
* ENOSPC.
*/
libceph: change ceph_osdmap_flag() to take osdc For the benefit of every single caller, take osdc instead of map. Also, now that osdc->osdmap can't ever be NULL, drop the check. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
pauserd = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD);
pausewr = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR) ||
ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) ||
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
have_pool_full(osdc);
libceph: add an epoch_barrier field to struct ceph_osd_client Cephfs can get cap update requests that contain a new epoch barrier in them. When that happens we want to pause all OSD traffic until the right map epoch arrives. Add an epoch_barrier field to ceph_osd_client that is protected by the osdc->lock rwsem. When the barrier is set, and the current OSD map epoch is below that, pause the request target when submitting the request or when revisiting it. Add a way for upper layers (cephfs) to update the epoch_barrier as well. If we get a new map, compare the new epoch against the barrier before kicking requests and request another map if the map epoch is still lower than the one we want. If we get a map with a full pool, or at quota condition, then set the barrier to the current epoch value. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-18 13:21:16 +00:00
if (was_pauserd || was_pausewr || pauserd || pausewr ||
osdc->osdmap->epoch < osdc->epoch_barrier)
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
maybe_request_map(osdc);
libceph: subscribe to osdmap when cluster is full When the cluster is marked full, subscribe to subsequent map updates to ensure we find out promptly when it is no longer full. This will prevent us from spewing ENOSPC for (much) longer than necessary. Signed-off-by: Sage Weil <sage@newdream.net>
2011-05-12 16:29:18 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
kick_requests(osdc, &need_resend, &need_resend_linger);
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
libceph: abort already submitted but abortable requests when map or pool goes full When a Ceph volume hits capacity, a flag is set in the OSD map to indicate that, and a new map is sprayed around the cluster. With cephfs we want it to shut down any abortable requests that are in progress with an -ENOSPC error as they'd just hang otherwise. Add a new ceph_osdc_abort_on_full helper function to handle this. It will first check whether there is an out-of-space condition in the cluster and then walk the tree and abort any request that has r_abort_on_full set with a -ENOSPC error. Call this new function directly whenever we get a new OSD map. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-04-04 12:39:39 +00:00
ceph_osdc_abort_on_full(osdc);
libceph: handle_one_map() Separate osdmap handling from decoding and iterating over a bag of maps in a fresh MOSDMap message. This sets up the scene for the updated OSD client. Of particular importance here is the addition of pi->was_full, which can be used to answer "did this pool go full -> not-full in this map?". This is the key bit for supporting pool quotas. We won't be able to downgrade map_sem for much longer, so drop downgrade_write(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
ceph_monc_got_map(&osdc->client->monc, CEPH_SUB_OSDMAP,
osdc->osdmap->epoch);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
up_write(&osdc->lock);
ceph: use complete_all and wake_up_all This fixes an issue triggered by running concurrent syncs. One of the syncs would go through while the other would just hang indefinitely. In any case, we never actually want to wake a single waiter, so the *_all functions should be used. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-07-27 20:11:08 +00:00
wake_up_all(&osdc->client->auth_wq);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
return;
bad:
pr_err("osdc handle_map corrupt msg\n");
ceph: hex dump corrupt server data to KERN_DEBUG Also, print fsid using standard format, NOT hex dump. Signed-off-by: Sage Weil <sage@newdream.net>
2009-12-14 23:13:47 +00:00
ceph_msg_dump(msg);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
up_write(&osdc->lock);
}
/*
* Resubmit requests pending on the given osd.
*/
static void kick_osd_requests(struct ceph_osd *osd)
{
struct rb_node *n;
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
clear_backoffs(osd);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
for (n = rb_first(&osd->o_requests); n; ) {
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd_request *req =
rb_entry(n, struct ceph_osd_request, r_node);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
n = rb_next(n); /* cancel_linger_request() */
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
if (!req->r_linger) {
if (!req->r_t.paused)
send_request(req);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
} else {
cancel_linger_request(req);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
}
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
for (n = rb_first(&osd->o_linger_requests); n; n = rb_next(n)) {
struct ceph_osd_linger_request *lreq =
rb_entry(n, struct ceph_osd_linger_request, node);
send_linger(lreq);
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
}
/*
* If the osd connection drops, we need to resubmit all requests.
*/
static void osd_fault(struct ceph_connection *con)
{
struct ceph_osd *osd = con->private;
struct ceph_osd_client *osdc = osd->o_osdc;
dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd);
down_write(&osdc->lock);
if (!osd_registered(osd)) {
dout("%s osd%d unknown\n", __func__, osd->o_osd);
goto out_unlock;
}
if (!reopen_osd(osd))
kick_osd_requests(osd);
maybe_request_map(osdc);
out_unlock:
up_write(&osdc->lock);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
struct MOSDBackoff {
struct ceph_spg spgid;
u32 map_epoch;
u8 op;
u64 id;
struct ceph_hobject_id *begin;
struct ceph_hobject_id *end;
};
static int decode_MOSDBackoff(const struct ceph_msg *msg, struct MOSDBackoff *m)
{
void *p = msg->front.iov_base;
void *const end = p + msg->front.iov_len;
u8 struct_v;
u32 struct_len;
int ret;
ret = ceph_start_decoding(&p, end, 1, "spg_t", &struct_v, &struct_len);
if (ret)
return ret;
ret = ceph_decode_pgid(&p, end, &m->spgid.pgid);
if (ret)
return ret;
ceph_decode_8_safe(&p, end, m->spgid.shard, e_inval);
ceph_decode_32_safe(&p, end, m->map_epoch, e_inval);
ceph_decode_8_safe(&p, end, m->op, e_inval);
ceph_decode_64_safe(&p, end, m->id, e_inval);
m->begin = kzalloc(sizeof(*m->begin), GFP_NOIO);
if (!m->begin)
return -ENOMEM;
ret = decode_hoid(&p, end, m->begin);
if (ret) {
free_hoid(m->begin);
return ret;
}
m->end = kzalloc(sizeof(*m->end), GFP_NOIO);
if (!m->end) {
free_hoid(m->begin);
return -ENOMEM;
}
ret = decode_hoid(&p, end, m->end);
if (ret) {
free_hoid(m->begin);
free_hoid(m->end);
return ret;
}
return 0;
e_inval:
return -EINVAL;
}
static struct ceph_msg *create_backoff_message(
const struct ceph_osd_backoff *backoff,
u32 map_epoch)
{
struct ceph_msg *msg;
void *p, *end;
int msg_size;
msg_size = CEPH_ENCODING_START_BLK_LEN +
CEPH_PGID_ENCODING_LEN + 1; /* spgid */
msg_size += 4 + 1 + 8; /* map_epoch, op, id */
msg_size += CEPH_ENCODING_START_BLK_LEN +
hoid_encoding_size(backoff->begin);
msg_size += CEPH_ENCODING_START_BLK_LEN +
hoid_encoding_size(backoff->end);
msg = ceph_msg_new(CEPH_MSG_OSD_BACKOFF, msg_size, GFP_NOIO, true);
if (!msg)
return NULL;
p = msg->front.iov_base;
end = p + msg->front_alloc_len;
encode_spgid(&p, &backoff->spgid);
ceph_encode_32(&p, map_epoch);
ceph_encode_8(&p, CEPH_OSD_BACKOFF_OP_ACK_BLOCK);
ceph_encode_64(&p, backoff->id);
encode_hoid(&p, end, backoff->begin);
encode_hoid(&p, end, backoff->end);
BUG_ON(p != end);
msg->front.iov_len = p - msg->front.iov_base;
msg->hdr.version = cpu_to_le16(1); /* MOSDBackoff v1 */
msg->hdr.front_len = cpu_to_le32(msg->front.iov_len);
return msg;
}
static void handle_backoff_block(struct ceph_osd *osd, struct MOSDBackoff *m)
{
struct ceph_spg_mapping *spg;
struct ceph_osd_backoff *backoff;
struct ceph_msg *msg;
dout("%s osd%d spgid %llu.%xs%d id %llu\n", __func__, osd->o_osd,
m->spgid.pgid.pool, m->spgid.pgid.seed, m->spgid.shard, m->id);
spg = lookup_spg_mapping(&osd->o_backoff_mappings, &m->spgid);
if (!spg) {
spg = alloc_spg_mapping();
if (!spg) {
pr_err("%s failed to allocate spg\n", __func__);
return;
}
spg->spgid = m->spgid; /* struct */
insert_spg_mapping(&osd->o_backoff_mappings, spg);
}
backoff = alloc_backoff();
if (!backoff) {
pr_err("%s failed to allocate backoff\n", __func__);
return;
}
backoff->spgid = m->spgid; /* struct */
backoff->id = m->id;
backoff->begin = m->begin;
m->begin = NULL; /* backoff now owns this */
backoff->end = m->end;
m->end = NULL; /* ditto */
insert_backoff(&spg->backoffs, backoff);
insert_backoff_by_id(&osd->o_backoffs_by_id, backoff);
/*
* Ack with original backoff's epoch so that the OSD can
* discard this if there was a PG split.
*/
msg = create_backoff_message(backoff, m->map_epoch);
if (!msg) {
pr_err("%s failed to allocate msg\n", __func__);
return;
}
ceph_con_send(&osd->o_con, msg);
}
static bool target_contained_by(const struct ceph_osd_request_target *t,
const struct ceph_hobject_id *begin,
const struct ceph_hobject_id *end)
{
struct ceph_hobject_id hoid;
int cmp;
hoid_fill_from_target(&hoid, t);
cmp = hoid_compare(&hoid, begin);
return !cmp || (cmp > 0 && hoid_compare(&hoid, end) < 0);
}
static void handle_backoff_unblock(struct ceph_osd *osd,
const struct MOSDBackoff *m)
{
struct ceph_spg_mapping *spg;
struct ceph_osd_backoff *backoff;
struct rb_node *n;
dout("%s osd%d spgid %llu.%xs%d id %llu\n", __func__, osd->o_osd,
m->spgid.pgid.pool, m->spgid.pgid.seed, m->spgid.shard, m->id);
backoff = lookup_backoff_by_id(&osd->o_backoffs_by_id, m->id);
if (!backoff) {
pr_err("%s osd%d spgid %llu.%xs%d id %llu backoff dne\n",
__func__, osd->o_osd, m->spgid.pgid.pool,
m->spgid.pgid.seed, m->spgid.shard, m->id);
return;
}
if (hoid_compare(backoff->begin, m->begin) &&
hoid_compare(backoff->end, m->end)) {
pr_err("%s osd%d spgid %llu.%xs%d id %llu bad range?\n",
__func__, osd->o_osd, m->spgid.pgid.pool,
m->spgid.pgid.seed, m->spgid.shard, m->id);
/* unblock it anyway... */
}
spg = lookup_spg_mapping(&osd->o_backoff_mappings, &backoff->spgid);
BUG_ON(!spg);
erase_backoff(&spg->backoffs, backoff);
erase_backoff_by_id(&osd->o_backoffs_by_id, backoff);
free_backoff(backoff);
if (RB_EMPTY_ROOT(&spg->backoffs)) {
erase_spg_mapping(&osd->o_backoff_mappings, spg);
free_spg_mapping(spg);
}
for (n = rb_first(&osd->o_requests); n; n = rb_next(n)) {
struct ceph_osd_request *req =
rb_entry(n, struct ceph_osd_request, r_node);
if (!ceph_spg_compare(&req->r_t.spgid, &m->spgid)) {
/*
* Match against @m, not @backoff -- the PG may
* have split on the OSD.
*/
if (target_contained_by(&req->r_t, m->begin, m->end)) {
/*
* If no other installed backoff applies,
* resend.
*/
send_request(req);
}
}
}
}
static void handle_backoff(struct ceph_osd *osd, struct ceph_msg *msg)
{
struct ceph_osd_client *osdc = osd->o_osdc;
struct MOSDBackoff m;
int ret;
down_read(&osdc->lock);
if (!osd_registered(osd)) {
dout("%s osd%d unknown\n", __func__, osd->o_osd);
up_read(&osdc->lock);
return;
}
WARN_ON(osd->o_osd != le64_to_cpu(msg->hdr.src.num));
mutex_lock(&osd->lock);
ret = decode_MOSDBackoff(msg, &m);
if (ret) {
pr_err("failed to decode MOSDBackoff: %d\n", ret);
ceph_msg_dump(msg);
goto out_unlock;
}
switch (m.op) {
case CEPH_OSD_BACKOFF_OP_BLOCK:
handle_backoff_block(osd, &m);
break;
case CEPH_OSD_BACKOFF_OP_UNBLOCK:
handle_backoff_unblock(osd, &m);
break;
default:
pr_err("%s osd%d unknown op %d\n", __func__, osd->o_osd, m.op);
}
free_hoid(m.begin);
free_hoid(m.end);
out_unlock:
mutex_unlock(&osd->lock);
up_read(&osdc->lock);
}
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
/*
* Process osd watch notifications
*/
libceph: kill ceph_osdc_create_event() "one_shot" parameter There is only one caller of ceph_osdc_create_event(), and it provides 0 as its "one_shot" argument. Get rid of that argument and just use 0 in its place. Replace the code in handle_watch_notify() that executes if one_shot is nonzero in the event with a BUG_ON() call. While modifying "osd_client.c", give handle_watch_notify() static scope. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-15 17:42:30 +00:00
static void handle_watch_notify(struct ceph_osd_client *osdc,
struct ceph_msg *msg)
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
{
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
void *p = msg->front.iov_base;
void *const end = p + msg->front.iov_len;
struct ceph_osd_linger_request *lreq;
struct linger_work *lwork;
u8 proto_ver, opcode;
u64 cookie, notify_id;
u64 notifier_id = 0;
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
s32 return_code = 0;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
void *payload = NULL;
u32 payload_len = 0;
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
ceph_decode_8_safe(&p, end, proto_ver, bad);
ceph_decode_8_safe(&p, end, opcode, bad);
ceph_decode_64_safe(&p, end, cookie, bad);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
p += 8; /* skip ver */
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
ceph_decode_64_safe(&p, end, notify_id, bad);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
if (proto_ver >= 1) {
ceph_decode_32_safe(&p, end, payload_len, bad);
ceph_decode_need(&p, end, payload_len, bad);
payload = p;
p += payload_len;
}
if (le16_to_cpu(msg->hdr.version) >= 2)
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
ceph_decode_32_safe(&p, end, return_code, bad);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
if (le16_to_cpu(msg->hdr.version) >= 3)
ceph_decode_64_safe(&p, end, notifier_id, bad);
down_read(&osdc->lock);
lreq = lookup_linger_osdc(&osdc->linger_requests, cookie);
if (!lreq) {
dout("%s opcode %d cookie %llu dne\n", __func__, opcode,
cookie);
goto out_unlock_osdc;
}
mutex_lock(&lreq->lock);
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
dout("%s opcode %d cookie %llu lreq %p is_watch %d\n", __func__,
opcode, cookie, lreq, lreq->is_watch);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
if (opcode == CEPH_WATCH_EVENT_DISCONNECT) {
if (!lreq->last_error) {
lreq->last_error = -ENOTCONN;
queue_watch_error(lreq);
}
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
} else if (!lreq->is_watch) {
/* CEPH_WATCH_EVENT_NOTIFY_COMPLETE */
if (lreq->notify_id && lreq->notify_id != notify_id) {
dout("lreq %p notify_id %llu != %llu, ignoring\n", lreq,
lreq->notify_id, notify_id);
} else if (!completion_done(&lreq->notify_finish_wait)) {
struct ceph_msg_data *data =
list_first_entry_or_null(&msg->data,
struct ceph_msg_data,
links);
if (data) {
if (lreq->preply_pages) {
WARN_ON(data->type !=
CEPH_MSG_DATA_PAGES);
*lreq->preply_pages = data->pages;
*lreq->preply_len = data->length;
} else {
ceph_release_page_vector(data->pages,
calc_pages_for(0, data->length));
}
}
lreq->notify_finish_error = return_code;
complete_all(&lreq->notify_finish_wait);
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
} else {
/* CEPH_WATCH_EVENT_NOTIFY */
lwork = lwork_alloc(lreq, do_watch_notify);
if (!lwork) {
pr_err("failed to allocate notify-lwork\n");
goto out_unlock_lreq;
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
}
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
lwork->notify.notify_id = notify_id;
lwork->notify.notifier_id = notifier_id;
lwork->notify.payload = payload;
lwork->notify.payload_len = payload_len;
lwork->notify.msg = ceph_msg_get(msg);
lwork_queue(lwork);
libceph: don't try checking queue_work() return value queue_work() doesn't "fail to queue", it returns false if work was already on a queue, which can't happen here since we allocate event_work right before we queue it. So don't bother at all. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-09-11 08:18:53 +00:00
}
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
out_unlock_lreq:
mutex_unlock(&lreq->lock);
out_unlock_osdc:
up_read(&osdc->lock);
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
return;
bad:
pr_err("osdc handle_watch_notify corrupt msg\n");
}
libceph: set response data fields earlier When an incoming message is destined for the osd client, the messenger calls the osd client's alloc_msg method. That function looks up which request has the tid matching the incoming message, and returns the request message that was preallocated to receive the response. The response message is therefore known before the request is even started. Between the start of the request and the receipt of the response, the request and its data fields will not change, so there's no reason we need to hold off setting them. In fact it's preferable to set them just once because it's more obvious that they're unchanging. So set up the fields describing where incoming data is to land in a response message at the beginning of ceph_osdc_start_request(). Define a helper function that sets these fields, and use it to set the fields for both outgoing data in the request message and incoming data in the response. This resolves: http://tracker.ceph.com/issues/4284 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-05 00:29:06 +00:00
/*
* Register request, send initial attempt.
*/
int ceph_osdc_start_request(struct ceph_osd_client *osdc,
struct ceph_osd_request *req,
bool nofail)
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
down_read(&osdc->lock);
submit_request(req, false);
up_read(&osdc->lock);
libceph: factor out logic from ceph_osdc_start_request() Factor out logic from ceph_osdc_start_request() into a new helper, __ceph_osdc_start_request(). ceph_osdc_start_request() now amounts to taking locks and calling __ceph_osdc_start_request(). Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-31 17:33:39 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
return 0;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:14:15 +00:00
EXPORT_SYMBOL(ceph_osdc_start_request);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: introduce ceph_osdc_cancel_request() Introduce ceph_osdc_cancel_request() intended for canceling requests from the higher layers (rbd and cephfs). Because higher layers are in charge and are supposed to know what and when they are canceling, the request is not completed, only unref'ed and removed from the libceph data structures. __cancel_request() is no longer called before __unregister_request(), because __unregister_request() unconditionally revokes r_request and there is no point in trying to do it twice. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-19 07:38:13 +00:00
/*
libceph: always signal completion when done r_safe_completion is currently, and has always been, signaled only if on-disk ack was requested. It's there for fsync and syncfs, which wait for in-flight writes to flush - all data write requests set ONDISK. However, the pool perm check code introduced in 4.2 sends a write request with only ACK set. An unfortunately timed syncfs can then hang forever: r_safe_completion won't be signaled because only an unsafe reply was requested. We could patch ceph_osdc_sync() to skip !ONDISK write requests, but that is somewhat incomplete and yet another special case. Instead, rename this completion to r_done_completion and always signal it when the OSD client is done with the request, whether unsafe, safe, or error. This is a bit cleaner and helps with the cancellation code. Reported-by: Yan, Zheng <zyan@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-12-02 13:01:55 +00:00
* Unregister a registered request. The request is not completed:
* ->r_result isn't set and __complete_request() isn't called.
libceph: introduce ceph_osdc_cancel_request() Introduce ceph_osdc_cancel_request() intended for canceling requests from the higher layers (rbd and cephfs). Because higher layers are in charge and are supposed to know what and when they are canceling, the request is not completed, only unref'ed and removed from the libceph data structures. __cancel_request() is no longer called before __unregister_request(), because __unregister_request() unconditionally revokes r_request and there is no point in trying to do it twice. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-19 07:38:13 +00:00
*/
void ceph_osdc_cancel_request(struct ceph_osd_request *req)
{
struct ceph_osd_client *osdc = req->r_osdc;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
down_write(&osdc->lock);
if (req->r_osd)
cancel_request(req);
up_write(&osdc->lock);
libceph: introduce ceph_osdc_cancel_request() Introduce ceph_osdc_cancel_request() intended for canceling requests from the higher layers (rbd and cephfs). Because higher layers are in charge and are supposed to know what and when they are canceling, the request is not completed, only unref'ed and removed from the libceph data structures. __cancel_request() is no longer called before __unregister_request(), because __unregister_request() unconditionally revokes r_request and there is no point in trying to do it twice. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-19 07:38:13 +00:00
}
EXPORT_SYMBOL(ceph_osdc_cancel_request);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
libceph: wait_request_timeout() The unwatch timeout is currently implemented in rbd. With watch/unwatch code moving into libceph, we are going to need a ceph_osdc_wait_request() variant with a timeout. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
* @timeout: in jiffies, 0 means "wait forever"
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
*/
libceph: wait_request_timeout() The unwatch timeout is currently implemented in rbd. With watch/unwatch code moving into libceph, we are going to need a ceph_osdc_wait_request() variant with a timeout. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
static int wait_request_timeout(struct ceph_osd_request *req,
unsigned long timeout)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
libceph: wait_request_timeout() The unwatch timeout is currently implemented in rbd. With watch/unwatch code moving into libceph, we are going to need a ceph_osdc_wait_request() variant with a timeout. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
long left;
libceph: introduce ceph_osdc_cancel_request() Introduce ceph_osdc_cancel_request() intended for canceling requests from the higher layers (rbd and cephfs). Because higher layers are in charge and are supposed to know what and when they are canceling, the request is not completed, only unref'ed and removed from the libceph data structures. __cancel_request() is no longer called before __unregister_request(), because __unregister_request() unconditionally revokes r_request and there is no point in trying to do it twice. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-19 07:38:13 +00:00
libceph: wait_request_timeout() The unwatch timeout is currently implemented in rbd. With watch/unwatch code moving into libceph, we are going to need a ceph_osdc_wait_request() variant with a timeout. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
dout("%s req %p tid %llu\n", __func__, req, req->r_tid);
libceph: make ceph_osdc_wait_request() uninterruptible Ceph_osdc_wait_request() is used when cephfs issues sync IO. In most cases, the sync IO should be uninterruptible. The fix is use killale wait function in ceph_osdc_wait_request(). Signed-off-by: Yan, Zheng <zyan@redhat.com>
2016-05-13 03:04:33 +00:00
left = wait_for_completion_killable_timeout(&req->r_completion,
libceph: wait_request_timeout() The unwatch timeout is currently implemented in rbd. With watch/unwatch code moving into libceph, we are going to need a ceph_osdc_wait_request() variant with a timeout. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
ceph_timeout_jiffies(timeout));
if (left <= 0) {
left = left ?: -ETIMEDOUT;
libceph: introduce ceph_osdc_cancel_request() Introduce ceph_osdc_cancel_request() intended for canceling requests from the higher layers (rbd and cephfs). Because higher layers are in charge and are supposed to know what and when they are canceling, the request is not completed, only unref'ed and removed from the libceph data structures. __cancel_request() is no longer called before __unregister_request(), because __unregister_request() unconditionally revokes r_request and there is no point in trying to do it twice. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2014-06-19 07:38:13 +00:00
ceph_osdc_cancel_request(req);
libceph: wait_request_timeout() The unwatch timeout is currently implemented in rbd. With watch/unwatch code moving into libceph, we are going to need a ceph_osdc_wait_request() variant with a timeout. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
} else {
left = req->r_result; /* completed */
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: wait_request_timeout() The unwatch timeout is currently implemented in rbd. With watch/unwatch code moving into libceph, we are going to need a ceph_osdc_wait_request() variant with a timeout. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
return left;
}
/*
* wait for a request to complete
*/
int ceph_osdc_wait_request(struct ceph_osd_client *osdc,
struct ceph_osd_request *req)
{
return wait_request_timeout(req, 0);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:14:15 +00:00
EXPORT_SYMBOL(ceph_osdc_wait_request);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* sync - wait for all in-flight requests to flush. avoid starvation.
*/
void ceph_osdc_sync(struct ceph_osd_client *osdc)
{
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct rb_node *n, *p;
u64 last_tid = atomic64_read(&osdc->last_tid);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
again:
down_read(&osdc->lock);
for (n = rb_first(&osdc->osds); n; n = rb_next(n)) {
struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node);
mutex_lock(&osd->lock);
for (p = rb_first(&osd->o_requests); p; p = rb_next(p)) {
struct ceph_osd_request *req =
rb_entry(p, struct ceph_osd_request, r_node);
if (req->r_tid > last_tid)
break;
if (!(req->r_flags & CEPH_OSD_FLAG_WRITE))
continue;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
ceph_osdc_get_request(req);
mutex_unlock(&osd->lock);
up_read(&osdc->lock);
dout("%s waiting on req %p tid %llu last_tid %llu\n",
__func__, req, req->r_tid, last_tid);
libceph: get rid of ack vs commit - CEPH_OSD_FLAG_ACK shouldn't be set anymore, so assert on it - remove support for handling ack replies (OSDs will send ack replies only if clients request them) - drop the "do lingering callbacks under osd->lock" logic from handle_reply() -- lreq->lock is sufficient in all three cases Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:46:08 +00:00
wait_for_completion(&req->r_completion);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
ceph_osdc_put_request(req);
goto again;
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
mutex_unlock(&osd->lock);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
up_read(&osdc->lock);
dout("%s done last_tid %llu\n", __func__, last_tid);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:14:15 +00:00
EXPORT_SYMBOL(ceph_osdc_sync);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
static struct ceph_osd_request *
alloc_linger_request(struct ceph_osd_linger_request *lreq)
{
struct ceph_osd_request *req;
req = ceph_osdc_alloc_request(lreq->osdc, NULL, 1, false, GFP_NOIO);
if (!req)
return NULL;
ceph_oid_copy(&req->r_base_oid, &lreq->t.base_oid);
ceph_oloc_copy(&req->r_base_oloc, &lreq->t.base_oloc);
if (ceph_osdc_alloc_messages(req, GFP_NOIO)) {
ceph_osdc_put_request(req);
return NULL;
}
return req;
}
/*
* Returns a handle, caller owns a ref.
*/
struct ceph_osd_linger_request *
ceph_osdc_watch(struct ceph_osd_client *osdc,
struct ceph_object_id *oid,
struct ceph_object_locator *oloc,
rados_watchcb2_t wcb,
rados_watcherrcb_t errcb,
void *data)
{
struct ceph_osd_linger_request *lreq;
int ret;
lreq = linger_alloc(osdc);
if (!lreq)
return ERR_PTR(-ENOMEM);
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
lreq->is_watch = true;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
lreq->wcb = wcb;
lreq->errcb = errcb;
lreq->data = data;
libceph: support for checking on status of watch Implement ceph_osdc_watch_check() to be able to check on status of watch. Note that the time it takes for a watch/notify event to get delivered through the notify_wq is taken into account. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
lreq->watch_valid_thru = jiffies;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
ceph_oid_copy(&lreq->t.base_oid, oid);
ceph_oloc_copy(&lreq->t.base_oloc, oloc);
libceph, rbd, ceph: WRITE | ONDISK -> WRITE CEPH_OSD_FLAG_ONDISK is set in account_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:48:41 +00:00
lreq->t.flags = CEPH_OSD_FLAG_WRITE;
fs: ceph: CURRENT_TIME with ktime_get_real_ts() CURRENT_TIME is not y2038 safe. The macro will be deleted and all the references to it will be replaced by ktime_get_* apis. struct timespec is also not y2038 safe. Retain timespec for timestamp representation here as ceph uses it internally everywhere. These references will be changed to use struct timespec64 in a separate patch. The current_fs_time() api is being changed to use vfs struct inode* as an argument instead of struct super_block*. Set the new mds client request r_stamp field using ktime_get_real_ts() instead of using current_fs_time(). Also, since r_stamp is used as mtime on the server, use timespec_trunc() to truncate the timestamp, using the right granularity from the superblock. This api will be transitioned to be y2038 safe along with vfs. Link: http://lkml.kernel.org/r/1491613030-11599-5-git-send-email-deepa.kernel@gmail.com Signed-off-by: Deepa Dinamani <deepa.kernel@gmail.com> Reviewed-by: Arnd Bergmann <arnd@arndb.de> M: Ilya Dryomov <idryomov@gmail.com> M: "Yan, Zheng" <zyan@redhat.com> M: Sage Weil <sage@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-08 22:59:19 +00:00
ktime_get_real_ts(&lreq->mtime);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
lreq->reg_req = alloc_linger_request(lreq);
if (!lreq->reg_req) {
ret = -ENOMEM;
goto err_put_lreq;
}
lreq->ping_req = alloc_linger_request(lreq);
if (!lreq->ping_req) {
ret = -ENOMEM;
goto err_put_lreq;
}
down_write(&osdc->lock);
linger_register(lreq); /* before osd_req_op_* */
osd_req_op_watch_init(lreq->reg_req, 0, lreq->linger_id,
CEPH_OSD_WATCH_OP_WATCH);
osd_req_op_watch_init(lreq->ping_req, 0, lreq->linger_id,
CEPH_OSD_WATCH_OP_PING);
linger_submit(lreq);
up_write(&osdc->lock);
ret = linger_reg_commit_wait(lreq);
if (ret) {
linger_cancel(lreq);
goto err_put_lreq;
}
return lreq;
err_put_lreq:
linger_put(lreq);
return ERR_PTR(ret);
}
EXPORT_SYMBOL(ceph_osdc_watch);
/*
* Releases a ref.
*
* Times out after mount_timeout to preserve rbd unmap behaviour
* introduced in 2894e1d76974 ("rbd: timeout watch teardown on unmap
* with mount_timeout").
*/
int ceph_osdc_unwatch(struct ceph_osd_client *osdc,
struct ceph_osd_linger_request *lreq)
{
struct ceph_options *opts = osdc->client->options;
struct ceph_osd_request *req;
int ret;
req = ceph_osdc_alloc_request(osdc, NULL, 1, false, GFP_NOIO);
if (!req)
return -ENOMEM;
ceph_oid_copy(&req->r_base_oid, &lreq->t.base_oid);
ceph_oloc_copy(&req->r_base_oloc, &lreq->t.base_oloc);
libceph, rbd, ceph: WRITE | ONDISK -> WRITE CEPH_OSD_FLAG_ONDISK is set in account_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:48:41 +00:00
req->r_flags = CEPH_OSD_FLAG_WRITE;
fs: ceph: CURRENT_TIME with ktime_get_real_ts() CURRENT_TIME is not y2038 safe. The macro will be deleted and all the references to it will be replaced by ktime_get_* apis. struct timespec is also not y2038 safe. Retain timespec for timestamp representation here as ceph uses it internally everywhere. These references will be changed to use struct timespec64 in a separate patch. The current_fs_time() api is being changed to use vfs struct inode* as an argument instead of struct super_block*. Set the new mds client request r_stamp field using ktime_get_real_ts() instead of using current_fs_time(). Also, since r_stamp is used as mtime on the server, use timespec_trunc() to truncate the timestamp, using the right granularity from the superblock. This api will be transitioned to be y2038 safe along with vfs. Link: http://lkml.kernel.org/r/1491613030-11599-5-git-send-email-deepa.kernel@gmail.com Signed-off-by: Deepa Dinamani <deepa.kernel@gmail.com> Reviewed-by: Arnd Bergmann <arnd@arndb.de> M: Ilya Dryomov <idryomov@gmail.com> M: "Yan, Zheng" <zyan@redhat.com> M: Sage Weil <sage@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-08 22:59:19 +00:00
ktime_get_real_ts(&req->r_mtime);
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
osd_req_op_watch_init(req, 0, lreq->linger_id,
CEPH_OSD_WATCH_OP_UNWATCH);
ret = ceph_osdc_alloc_messages(req, GFP_NOIO);
if (ret)
goto out_put_req;
ceph_osdc_start_request(osdc, req, false);
linger_cancel(lreq);
linger_put(lreq);
ret = wait_request_timeout(req, opts->mount_timeout);
out_put_req:
ceph_osdc_put_request(req);
return ret;
}
EXPORT_SYMBOL(ceph_osdc_unwatch);
static int osd_req_op_notify_ack_init(struct ceph_osd_request *req, int which,
u64 notify_id, u64 cookie, void *payload,
size_t payload_len)
{
struct ceph_osd_req_op *op;
struct ceph_pagelist *pl;
int ret;
op = _osd_req_op_init(req, which, CEPH_OSD_OP_NOTIFY_ACK, 0);
pl = kmalloc(sizeof(*pl), GFP_NOIO);
if (!pl)
return -ENOMEM;
ceph_pagelist_init(pl);
ret = ceph_pagelist_encode_64(pl, notify_id);
ret |= ceph_pagelist_encode_64(pl, cookie);
if (payload) {
ret |= ceph_pagelist_encode_32(pl, payload_len);
ret |= ceph_pagelist_append(pl, payload, payload_len);
} else {
ret |= ceph_pagelist_encode_32(pl, 0);
}
if (ret) {
ceph_pagelist_release(pl);
return -ENOMEM;
}
ceph_osd_data_pagelist_init(&op->notify_ack.request_data, pl);
op->indata_len = pl->length;
return 0;
}
int ceph_osdc_notify_ack(struct ceph_osd_client *osdc,
struct ceph_object_id *oid,
struct ceph_object_locator *oloc,
u64 notify_id,
u64 cookie,
void *payload,
size_t payload_len)
{
struct ceph_osd_request *req;
int ret;
req = ceph_osdc_alloc_request(osdc, NULL, 1, false, GFP_NOIO);
if (!req)
return -ENOMEM;
ceph_oid_copy(&req->r_base_oid, oid);
ceph_oloc_copy(&req->r_base_oloc, oloc);
req->r_flags = CEPH_OSD_FLAG_READ;
ret = ceph_osdc_alloc_messages(req, GFP_NOIO);
if (ret)
goto out_put_req;
ret = osd_req_op_notify_ack_init(req, 0, notify_id, cookie, payload,
payload_len);
if (ret)
goto out_put_req;
ceph_osdc_start_request(osdc, req, false);
ret = ceph_osdc_wait_request(osdc, req);
out_put_req:
ceph_osdc_put_request(req);
return ret;
}
EXPORT_SYMBOL(ceph_osdc_notify_ack);
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
static int osd_req_op_notify_init(struct ceph_osd_request *req, int which,
u64 cookie, u32 prot_ver, u32 timeout,
void *payload, size_t payload_len)
{
struct ceph_osd_req_op *op;
struct ceph_pagelist *pl;
int ret;
op = _osd_req_op_init(req, which, CEPH_OSD_OP_NOTIFY, 0);
op->notify.cookie = cookie;
pl = kmalloc(sizeof(*pl), GFP_NOIO);
if (!pl)
return -ENOMEM;
ceph_pagelist_init(pl);
ret = ceph_pagelist_encode_32(pl, 1); /* prot_ver */
ret |= ceph_pagelist_encode_32(pl, timeout);
ret |= ceph_pagelist_encode_32(pl, payload_len);
ret |= ceph_pagelist_append(pl, payload, payload_len);
if (ret) {
ceph_pagelist_release(pl);
return -ENOMEM;
}
ceph_osd_data_pagelist_init(&op->notify.request_data, pl);
op->indata_len = pl->length;
return 0;
}
/*
* @timeout: in seconds
*
* @preply_{pages,len} are initialized both on success and error.
* The caller is responsible for:
*
* ceph_release_page_vector(reply_pages, calc_pages_for(0, reply_len))
*/
int ceph_osdc_notify(struct ceph_osd_client *osdc,
struct ceph_object_id *oid,
struct ceph_object_locator *oloc,
void *payload,
size_t payload_len,
u32 timeout,
struct page ***preply_pages,
size_t *preply_len)
{
struct ceph_osd_linger_request *lreq;
struct page **pages;
int ret;
WARN_ON(!timeout);
if (preply_pages) {
*preply_pages = NULL;
*preply_len = 0;
}
lreq = linger_alloc(osdc);
if (!lreq)
return -ENOMEM;
lreq->preply_pages = preply_pages;
lreq->preply_len = preply_len;
ceph_oid_copy(&lreq->t.base_oid, oid);
ceph_oloc_copy(&lreq->t.base_oloc, oloc);
lreq->t.flags = CEPH_OSD_FLAG_READ;
lreq->reg_req = alloc_linger_request(lreq);
if (!lreq->reg_req) {
ret = -ENOMEM;
goto out_put_lreq;
}
/* for notify_id */
pages = ceph_alloc_page_vector(1, GFP_NOIO);
if (IS_ERR(pages)) {
ret = PTR_ERR(pages);
goto out_put_lreq;
}
down_write(&osdc->lock);
linger_register(lreq); /* before osd_req_op_* */
ret = osd_req_op_notify_init(lreq->reg_req, 0, lreq->linger_id, 1,
timeout, payload, payload_len);
if (ret) {
linger_unregister(lreq);
up_write(&osdc->lock);
ceph_release_page_vector(pages, 1);
goto out_put_lreq;
}
ceph_osd_data_pages_init(osd_req_op_data(lreq->reg_req, 0, notify,
response_data),
pages, PAGE_SIZE, 0, false, true);
linger_submit(lreq);
up_write(&osdc->lock);
ret = linger_reg_commit_wait(lreq);
if (!ret)
ret = linger_notify_finish_wait(lreq);
else
dout("lreq %p failed to initiate notify %d\n", lreq, ret);
linger_cancel(lreq);
out_put_lreq:
linger_put(lreq);
return ret;
}
EXPORT_SYMBOL(ceph_osdc_notify);
libceph: support for checking on status of watch Implement ceph_osdc_watch_check() to be able to check on status of watch. Note that the time it takes for a watch/notify event to get delivered through the notify_wq is taken into account. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
/*
* Return the number of milliseconds since the watch was last
* confirmed, or an error. If there is an error, the watch is no
* longer valid, and should be destroyed with ceph_osdc_unwatch().
*/
int ceph_osdc_watch_check(struct ceph_osd_client *osdc,
struct ceph_osd_linger_request *lreq)
{
unsigned long stamp, age;
int ret;
down_read(&osdc->lock);
mutex_lock(&lreq->lock);
stamp = lreq->watch_valid_thru;
if (!list_empty(&lreq->pending_lworks)) {
struct linger_work *lwork =
list_first_entry(&lreq->pending_lworks,
struct linger_work,
pending_item);
if (time_before(lwork->queued_stamp, stamp))
stamp = lwork->queued_stamp;
}
age = jiffies - stamp;
dout("%s lreq %p linger_id %llu age %lu last_error %d\n", __func__,
lreq, lreq->linger_id, age, lreq->last_error);
/* we are truncating to msecs, so return a safe upper bound */
ret = lreq->last_error ?: 1 + jiffies_to_msecs(age);
mutex_unlock(&lreq->lock);
up_read(&osdc->lock);
return ret;
}
libceph: support for CEPH_OSD_OP_LIST_WATCHERS Add support for this Ceph OSD op, needed to support the RBD exclusive lock feature. Signed-off-by: Douglas Fuller <dfuller@redhat.com> [idryomov@gmail.com: refactor, misc fixes throughout] Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-07-17 20:18:07 +00:00
static int decode_watcher(void **p, void *end, struct ceph_watch_item *item)
{
u8 struct_v;
u32 struct_len;
int ret;
ret = ceph_start_decoding(p, end, 2, "watch_item_t",
&struct_v, &struct_len);
if (ret)
return ret;
ceph_decode_copy(p, &item->name, sizeof(item->name));
item->cookie = ceph_decode_64(p);
*p += 4; /* skip timeout_seconds */
if (struct_v >= 2) {
ceph_decode_copy(p, &item->addr, sizeof(item->addr));
ceph_decode_addr(&item->addr);
}
dout("%s %s%llu cookie %llu addr %s\n", __func__,
ENTITY_NAME(item->name), item->cookie,
ceph_pr_addr(&item->addr.in_addr));
return 0;
}
static int decode_watchers(void **p, void *end,
struct ceph_watch_item **watchers,
u32 *num_watchers)
{
u8 struct_v;
u32 struct_len;
int i;
int ret;
ret = ceph_start_decoding(p, end, 1, "obj_list_watch_response_t",
&struct_v, &struct_len);
if (ret)
return ret;
*num_watchers = ceph_decode_32(p);
*watchers = kcalloc(*num_watchers, sizeof(**watchers), GFP_NOIO);
if (!*watchers)
return -ENOMEM;
for (i = 0; i < *num_watchers; i++) {
ret = decode_watcher(p, end, *watchers + i);
if (ret) {
kfree(*watchers);
return ret;
}
}
return 0;
}
/*
* On success, the caller is responsible for:
*
* kfree(watchers);
*/
int ceph_osdc_list_watchers(struct ceph_osd_client *osdc,
struct ceph_object_id *oid,
struct ceph_object_locator *oloc,
struct ceph_watch_item **watchers,
u32 *num_watchers)
{
struct ceph_osd_request *req;
struct page **pages;
int ret;
req = ceph_osdc_alloc_request(osdc, NULL, 1, false, GFP_NOIO);
if (!req)
return -ENOMEM;
ceph_oid_copy(&req->r_base_oid, oid);
ceph_oloc_copy(&req->r_base_oloc, oloc);
req->r_flags = CEPH_OSD_FLAG_READ;
ret = ceph_osdc_alloc_messages(req, GFP_NOIO);
if (ret)
goto out_put_req;
pages = ceph_alloc_page_vector(1, GFP_NOIO);
if (IS_ERR(pages)) {
ret = PTR_ERR(pages);
goto out_put_req;
}
osd_req_op_init(req, 0, CEPH_OSD_OP_LIST_WATCHERS, 0);
ceph_osd_data_pages_init(osd_req_op_data(req, 0, list_watchers,
response_data),
pages, PAGE_SIZE, 0, false, true);
ceph_osdc_start_request(osdc, req, false);
ret = ceph_osdc_wait_request(osdc, req);
if (ret >= 0) {
void *p = page_address(pages[0]);
void *const end = p + req->r_ops[0].outdata_len;
ret = decode_watchers(&p, end, watchers, num_watchers);
}
out_put_req:
ceph_osdc_put_request(req);
return ret;
}
EXPORT_SYMBOL(ceph_osdc_list_watchers);
libceph: add function to ensure notifies are complete Without a way to flush the osd client's notify workqueue, a watch event that is unregistered could continue receiving callbacks indefinitely. Unregistering the event simply means no new notifies are added to the queue, but there may still be events in the queue that will call the watch callback for the event. If the queue is flushed after the event is unregistered, the caller can be sure no more watch callbacks will occur for the canceled watch. Signed-off-by: Josh Durgin <josh.durgin@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2013-08-29 04:43:09 +00:00
/*
* Call all pending notify callbacks - for use after a watch is
* unregistered, to make sure no more callbacks for it will be invoked
*/
ceph: remove bogus extern Sparse complained about this bogus extern on definition of a function. Signed-off-by: Stephen Hemminger <stephen@networkplumber.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2014-06-11 03:30:13 +00:00
void ceph_osdc_flush_notifies(struct ceph_osd_client *osdc)
libceph: add function to ensure notifies are complete Without a way to flush the osd client's notify workqueue, a watch event that is unregistered could continue receiving callbacks indefinitely. Unregistering the event simply means no new notifies are added to the queue, but there may still be events in the queue that will call the watch callback for the event. If the queue is flushed after the event is unregistered, the caller can be sure no more watch callbacks will occur for the canceled watch. Signed-off-by: Josh Durgin <josh.durgin@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2013-08-29 04:43:09 +00:00
{
rbd: retry watch re-registration periodically Revamp watch code to support retrying watch re-registration: - add rbd_dev->watch_state for more robust errcb handling - store watch cookie separately to avoid dereferencing watch_handle which is set to NULL on unwatch - move re-register code into a delayed work and retry re-registration every second, unless the client is blacklisted Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Tested-by: Mike Christie <mchristi@redhat.com>
2016-08-12 14:11:41 +00:00
dout("%s osdc %p\n", __func__, osdc);
libceph: add function to ensure notifies are complete Without a way to flush the osd client's notify workqueue, a watch event that is unregistered could continue receiving callbacks indefinitely. Unregistering the event simply means no new notifies are added to the queue, but there may still be events in the queue that will call the watch callback for the event. If the queue is flushed after the event is unregistered, the caller can be sure no more watch callbacks will occur for the canceled watch. Signed-off-by: Josh Durgin <josh.durgin@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2013-08-29 04:43:09 +00:00
flush_workqueue(osdc->notify_wq);
}
EXPORT_SYMBOL(ceph_osdc_flush_notifies);
libceph: replace ceph_monc_request_next_osdmap() ... with a wrapper around maybe_request_map() - no need for two osdmap-specific functions. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:28 +00:00
void ceph_osdc_maybe_request_map(struct ceph_osd_client *osdc)
{
down_read(&osdc->lock);
maybe_request_map(osdc);
up_read(&osdc->lock);
}
EXPORT_SYMBOL(ceph_osdc_maybe_request_map);
libceph: add function to ensure notifies are complete Without a way to flush the osd client's notify workqueue, a watch event that is unregistered could continue receiving callbacks indefinitely. Unregistering the event simply means no new notifies are added to the queue, but there may still be events in the queue that will call the watch callback for the event. If the queue is flushed after the event is unregistered, the caller can be sure no more watch callbacks will occur for the canceled watch. Signed-off-by: Josh Durgin <josh.durgin@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@linaro.org>
2013-08-29 04:43:09 +00:00
libceph: add ceph_osdc_call() single-page helper Add a convenience function to osd_client to send Ceph OSD 'class' ops. The interface assumes that the request and reply data each consist of single pages. Signed-off-by: Douglas Fuller <dfuller@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-06-17 18:49:45 +00:00
/*
* Execute an OSD class method on an object.
*
* @flags: CEPH_OSD_FLAG_*
libceph: pass reply buffer length through ceph_osdc_call() To spare checking for "this reply fits into a page, but does it fit into my buffer?" in some callers, osd_req_op_cls_response_data_pages() needs to know how big it is. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jason Dillaman <dillaman@redhat.com>
2017-01-25 17:16:21 +00:00
* @resp_len: in/out param for reply length
libceph: add ceph_osdc_call() single-page helper Add a convenience function to osd_client to send Ceph OSD 'class' ops. The interface assumes that the request and reply data each consist of single pages. Signed-off-by: Douglas Fuller <dfuller@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-06-17 18:49:45 +00:00
*/
int ceph_osdc_call(struct ceph_osd_client *osdc,
struct ceph_object_id *oid,
struct ceph_object_locator *oloc,
const char *class, const char *method,
unsigned int flags,
struct page *req_page, size_t req_len,
struct page *resp_page, size_t *resp_len)
{
struct ceph_osd_request *req;
int ret;
libceph: pass reply buffer length through ceph_osdc_call() To spare checking for "this reply fits into a page, but does it fit into my buffer?" in some callers, osd_req_op_cls_response_data_pages() needs to know how big it is. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jason Dillaman <dillaman@redhat.com>
2017-01-25 17:16:21 +00:00
if (req_len > PAGE_SIZE || (resp_page && *resp_len > PAGE_SIZE))
return -E2BIG;
libceph: add ceph_osdc_call() single-page helper Add a convenience function to osd_client to send Ceph OSD 'class' ops. The interface assumes that the request and reply data each consist of single pages. Signed-off-by: Douglas Fuller <dfuller@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-06-17 18:49:45 +00:00
req = ceph_osdc_alloc_request(osdc, NULL, 1, false, GFP_NOIO);
if (!req)
return -ENOMEM;
ceph_oid_copy(&req->r_base_oid, oid);
ceph_oloc_copy(&req->r_base_oloc, oloc);
req->r_flags = flags;
ret = ceph_osdc_alloc_messages(req, GFP_NOIO);
if (ret)
goto out_put_req;
libceph, rbd: add error handling for osd_req_op_cls_init() Add proper error handling for osd_req_op_cls_init() to replace BUG_ON statement when failing from memory allocation. Signed-off-by: Chengguang Xu <cgxu519@gmx.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-04-12 04:04:55 +00:00
ret = osd_req_op_cls_init(req, 0, CEPH_OSD_OP_CALL, class, method);
if (ret)
goto out_put_req;
libceph: add ceph_osdc_call() single-page helper Add a convenience function to osd_client to send Ceph OSD 'class' ops. The interface assumes that the request and reply data each consist of single pages. Signed-off-by: Douglas Fuller <dfuller@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-06-17 18:49:45 +00:00
if (req_page)
osd_req_op_cls_request_data_pages(req, 0, &req_page, req_len,
0, false, false);
if (resp_page)
osd_req_op_cls_response_data_pages(req, 0, &resp_page,
libceph: pass reply buffer length through ceph_osdc_call() To spare checking for "this reply fits into a page, but does it fit into my buffer?" in some callers, osd_req_op_cls_response_data_pages() needs to know how big it is. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jason Dillaman <dillaman@redhat.com>
2017-01-25 17:16:21 +00:00
*resp_len, 0, false, false);
libceph: add ceph_osdc_call() single-page helper Add a convenience function to osd_client to send Ceph OSD 'class' ops. The interface assumes that the request and reply data each consist of single pages. Signed-off-by: Douglas Fuller <dfuller@redhat.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-06-17 18:49:45 +00:00
ceph_osdc_start_request(osdc, req, false);
ret = ceph_osdc_wait_request(osdc, req);
if (ret >= 0) {
ret = req->r_ops[0].rval;
if (resp_page)
*resp_len = req->r_ops[0].outdata_len;
}
out_put_req:
ceph_osdc_put_request(req);
return ret;
}
EXPORT_SYMBOL(ceph_osdc_call);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* init, shutdown
*/
int ceph_osdc_init(struct ceph_osd_client *osdc, struct ceph_client *client)
{
int err;
dout("init\n");
osdc->client = client;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
init_rwsem(&osdc->lock);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
osdc->osds = RB_ROOT;
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
INIT_LIST_HEAD(&osdc->osd_lru);
libceph: protect osdc->osd_lru list with a spinlock OSD client is getting moved from the big per-client lock to a set of per-session locks. The big rwlock would only be held for read most of the time, so a global osdc->osd_lru needs additional protection. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
spin_lock_init(&osdc->osd_lru_lock);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
osd_init(&osdc->homeless_osd);
osdc->homeless_osd.o_osdc = osdc;
osdc->homeless_osd.o_osd = CEPH_HOMELESS_OSD;
libceph: initialize last_linger_id with a large integer osdc->last_linger_id is a counter for lreq->linger_id, which is used for watch cookies. Starting with a large integer should ease the task of telling apart kernel and userspace clients. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-11-08 14:15:24 +00:00
osdc->last_linger_id = CEPH_LINGER_ID_START;
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
osdc->linger_requests = RB_ROOT;
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
osdc->map_checks = RB_ROOT;
osdc->linger_map_checks = RB_ROOT;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout);
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout);
ceph: handle errors during osd client init Unwind initializing if we get ENOMEM during client initialization. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-18 22:52:18 +00:00
err = -ENOMEM;
libceph: allocate dummy osdmap in ceph_osdc_init() This leads to a simpler osdmap handling code, particularly when dealing with pi->was_full, which is introduced in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
osdc->osdmap = ceph_osdmap_alloc();
if (!osdc->osdmap)
goto out;
libceph: osdc->req_mempool should be backed by a slab pool ceph_osd_request_cache was introduced a long time ago. Also, osd_req is about to get a flexible array member, which ceph_osd_request_cache is going to be aware of. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:25:31 +00:00
osdc->req_mempool = mempool_create_slab_pool(10,
ceph_osd_request_cache);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
if (!osdc->req_mempool)
libceph: allocate dummy osdmap in ceph_osdc_init() This leads to a simpler osdmap handling code, particularly when dealing with pi->was_full, which is introduced in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
goto out_map;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: initialize msgpool message types Initialize the type field for messages in a msgpool. The caller was doing this for osd ops, but not for the reply messages. Reported-by: Alex Elder <elder@inktank.com> Signed-off-by: Sage Weil <sage@inktank.com>
2012-07-09 21:22:34 +00:00
err = ceph_msgpool_init(&osdc->msgpool_op, CEPH_MSG_OSD_OP,
libceph: change how osd_op_reply message size is calculated For a message pool message, preallocate a page, just like we do for osd_op. For a normal message, take ceph_object_id into account and don't bother subtracting CEPH_OSD_SLAB_OPS ceph_osd_ops. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 16:32:56 +00:00
PAGE_SIZE, 10, true, "osd_op");
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
if (err < 0)
ceph: handle errors during osd client init Unwind initializing if we get ENOMEM during client initialization. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-18 22:52:18 +00:00
goto out_mempool;
libceph: initialize msgpool message types Initialize the type field for messages in a msgpool. The caller was doing this for osd ops, but not for the reply messages. Reported-by: Alex Elder <elder@inktank.com> Signed-off-by: Sage Weil <sage@inktank.com>
2012-07-09 21:22:34 +00:00
err = ceph_msgpool_init(&osdc->msgpool_op_reply, CEPH_MSG_OSD_OPREPLY,
libceph: change how osd_op_reply message size is calculated For a message pool message, preallocate a page, just like we do for osd_op. For a normal message, take ceph_object_id into account and don't bother subtracting CEPH_OSD_SLAB_OPS ceph_osd_ops. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-27 16:32:56 +00:00
PAGE_SIZE, 10, true, "osd_op_reply");
ceph: use single osd op reply msg Use a single ceph_msg for the osd reply, even when we are getting multiple replies. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-01 21:02:00 +00:00
if (err < 0)
goto out_msgpool;
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
libceph: create_singlethread_workqueue() doesn't return ERR_PTRs create_singlethread_workqueue() returns NULL on error, and it doesn't return ERR_PTRs. I tweaked the error handling a little to be consistent with earlier in the function. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Sage Weil <sage@inktank.com>
2013-08-15 05:58:59 +00:00
err = -ENOMEM;
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
osdc->notify_wq = create_singlethread_workqueue("ceph-watch-notify");
libceph: create_singlethread_workqueue() doesn't return ERR_PTRs create_singlethread_workqueue() returns NULL on error, and it doesn't return ERR_PTRs. I tweaked the error handling a little to be consistent with earlier in the function. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Sage Weil <sage@inktank.com>
2013-08-15 05:58:59 +00:00
if (!osdc->notify_wq)
libceph: fix error handling in ceph_osdc_init() msgpool_op_reply message pool isn't destroyed if workqueue construction fails. Fix it. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-31 15:49:22 +00:00
goto out_msgpool_reply;
libceph: defer __complete_request() to a workqueue In the common case, req->r_callback is called by handle_reply() on the ceph-msgr worker thread without any locks. If handle_reply() fails, it is called with both osd->lock and osdc->lock. In the map check case, it is called with just osdc->lock but held for write. Finally, if the request is aborted because of -ENOSPC or by ceph_osdc_abort_requests(), it is called directly on the submitter's thread, again with both locks. req->r_callback on the submitter's thread is relatively new (introduced in 4.12) and ripe for deadlocks -- e.g. writeback worker thread waiting on itself: inode_wait_for_writeback+0x26/0x40 evict+0xb5/0x1a0 iput+0x1d2/0x220 ceph_put_wrbuffer_cap_refs+0xe0/0x2c0 [ceph] writepages_finish+0x2d3/0x410 [ceph] __complete_request+0x26/0x60 [libceph] complete_request+0x2e/0x70 [libceph] __submit_request+0x256/0x330 [libceph] submit_request+0x2b/0x30 [libceph] ceph_osdc_start_request+0x25/0x40 [libceph] ceph_writepages_start+0xdfe/0x1320 [ceph] do_writepages+0x1f/0x70 __writeback_single_inode+0x45/0x330 writeback_sb_inodes+0x26a/0x600 __writeback_inodes_wb+0x92/0xc0 wb_writeback+0x274/0x330 wb_workfn+0x2d5/0x3b0 Defer __complete_request() to a workqueue in all failure cases so it's never on the same thread as ceph_osdc_start_request() and always called with no locks held. Link: http://tracker.ceph.com/issues/23978 Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-21 14:00:29 +00:00
osdc->completion_wq = create_singlethread_workqueue("ceph-completion");
if (!osdc->completion_wq)
goto out_notify_wq;
libceph: schedule tick from ceph_osdc_init() Both homeless OSD sessions and watch/notify v2, introduced in later commits, require periodic ticks which don't depend on ->num_requests. Schedule the initial tick from ceph_osdc_init() and reschedule from handle_timeout() unconditionally. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
schedule_delayed_work(&osdc->timeout_work,
osdc->client->options->osd_keepalive_timeout);
libceph: move schedule_delayed_work() in ceph_osdc_init() ceph_osdc_stop() isn't called if ceph_osdc_init() fails, so we end up with handle_osds_timeout() running on invalid memory if any one of the allocations fails. Call schedule_delayed_work() after everything is setup, just before returning. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:24 +00:00
schedule_delayed_work(&osdc->osds_timeout_work,
round_jiffies_relative(osdc->client->options->osd_idle_ttl));
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
return 0;
ceph: handle errors during osd client init Unwind initializing if we get ENOMEM during client initialization. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-18 22:52:18 +00:00
libceph: defer __complete_request() to a workqueue In the common case, req->r_callback is called by handle_reply() on the ceph-msgr worker thread without any locks. If handle_reply() fails, it is called with both osd->lock and osdc->lock. In the map check case, it is called with just osdc->lock but held for write. Finally, if the request is aborted because of -ENOSPC or by ceph_osdc_abort_requests(), it is called directly on the submitter's thread, again with both locks. req->r_callback on the submitter's thread is relatively new (introduced in 4.12) and ripe for deadlocks -- e.g. writeback worker thread waiting on itself: inode_wait_for_writeback+0x26/0x40 evict+0xb5/0x1a0 iput+0x1d2/0x220 ceph_put_wrbuffer_cap_refs+0xe0/0x2c0 [ceph] writepages_finish+0x2d3/0x410 [ceph] __complete_request+0x26/0x60 [libceph] complete_request+0x2e/0x70 [libceph] __submit_request+0x256/0x330 [libceph] submit_request+0x2b/0x30 [libceph] ceph_osdc_start_request+0x25/0x40 [libceph] ceph_writepages_start+0xdfe/0x1320 [ceph] do_writepages+0x1f/0x70 __writeback_single_inode+0x45/0x330 writeback_sb_inodes+0x26a/0x600 __writeback_inodes_wb+0x92/0xc0 wb_writeback+0x274/0x330 wb_workfn+0x2d5/0x3b0 Defer __complete_request() to a workqueue in all failure cases so it's never on the same thread as ceph_osdc_start_request() and always called with no locks held. Link: http://tracker.ceph.com/issues/23978 Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-21 14:00:29 +00:00
out_notify_wq:
destroy_workqueue(osdc->notify_wq);
libceph: fix error handling in ceph_osdc_init() msgpool_op_reply message pool isn't destroyed if workqueue construction fails. Fix it. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-31 15:49:22 +00:00
out_msgpool_reply:
ceph_msgpool_destroy(&osdc->msgpool_op_reply);
ceph: use single osd op reply msg Use a single ceph_msg for the osd reply, even when we are getting multiple replies. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-01 21:02:00 +00:00
out_msgpool:
ceph_msgpool_destroy(&osdc->msgpool_op);
ceph: handle errors during osd client init Unwind initializing if we get ENOMEM during client initialization. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-18 22:52:18 +00:00
out_mempool:
mempool_destroy(osdc->req_mempool);
libceph: allocate dummy osdmap in ceph_osdc_init() This leads to a simpler osdmap handling code, particularly when dealing with pi->was_full, which is introduced in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
out_map:
ceph_osdmap_destroy(osdc->osdmap);
ceph: handle errors during osd client init Unwind initializing if we get ENOMEM during client initialization. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-18 22:52:18 +00:00
out:
return err;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
void ceph_osdc_stop(struct ceph_osd_client *osdc)
{
libceph: defer __complete_request() to a workqueue In the common case, req->r_callback is called by handle_reply() on the ceph-msgr worker thread without any locks. If handle_reply() fails, it is called with both osd->lock and osdc->lock. In the map check case, it is called with just osdc->lock but held for write. Finally, if the request is aborted because of -ENOSPC or by ceph_osdc_abort_requests(), it is called directly on the submitter's thread, again with both locks. req->r_callback on the submitter's thread is relatively new (introduced in 4.12) and ripe for deadlocks -- e.g. writeback worker thread waiting on itself: inode_wait_for_writeback+0x26/0x40 evict+0xb5/0x1a0 iput+0x1d2/0x220 ceph_put_wrbuffer_cap_refs+0xe0/0x2c0 [ceph] writepages_finish+0x2d3/0x410 [ceph] __complete_request+0x26/0x60 [libceph] complete_request+0x2e/0x70 [libceph] __submit_request+0x256/0x330 [libceph] submit_request+0x2b/0x30 [libceph] ceph_osdc_start_request+0x25/0x40 [libceph] ceph_writepages_start+0xdfe/0x1320 [ceph] do_writepages+0x1f/0x70 __writeback_single_inode+0x45/0x330 writeback_sb_inodes+0x26a/0x600 __writeback_inodes_wb+0x92/0xc0 wb_writeback+0x274/0x330 wb_workfn+0x2d5/0x3b0 Defer __complete_request() to a workqueue in all failure cases so it's never on the same thread as ceph_osdc_start_request() and always called with no locks held. Link: http://tracker.ceph.com/issues/23978 Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Acked-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: "Yan, Zheng" <zyan@redhat.com>
2018-05-21 14:00:29 +00:00
destroy_workqueue(osdc->completion_wq);
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
destroy_workqueue(osdc->notify_wq);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
cancel_delayed_work_sync(&osdc->timeout_work);
ceph: put unused osd connections on lru Instead of removing osd connection immediately when the requests list is empty, put the osd connection on an lru. Only if that osd has not been used for more than a specified time, will it be removed. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 19:00:26 +00:00
cancel_delayed_work_sync(&osdc->osds_timeout_work);
libceph: open-code remove_{all,old}_osds() They are called only once, from ceph_osdc_stop() and handle_osds_timeout() respectively. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
down_write(&osdc->lock);
libceph: open-code remove_{all,old}_osds() They are called only once, from ceph_osdc_stop() and handle_osds_timeout() respectively. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
while (!RB_EMPTY_ROOT(&osdc->osds)) {
struct ceph_osd *osd = rb_entry(rb_first(&osdc->osds),
struct ceph_osd, o_node);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
close_osd(osd);
libceph: open-code remove_{all,old}_osds() They are called only once, from ceph_osdc_stop() and handle_osds_timeout() respectively. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
up_write(&osdc->lock);
libceph: convert ceph_osd.o_ref from atomic_t to refcount_t refcount_t type and corresponding API should be used instead of atomic_t when the variable is used as a reference counter. This allows to avoid accidental refcounter overflows that might lead to use-after-free situations. Signed-off-by: Elena Reshetova <elena.reshetova@intel.com> Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com> Signed-off-by: Kees Cook <keescook@chromium.org> Signed-off-by: David Windsor <dwindsor@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-03-17 12:10:28 +00:00
WARN_ON(refcount_read(&osdc->homeless_osd.o_ref) != 1);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
osd_cleanup(&osdc->homeless_osd);
WARN_ON(!list_empty(&osdc->osd_lru));
libceph, rbd: ceph_osd_linger_request, watch/notify v2 This adds support and switches rbd to a new, more reliable version of watch/notify protocol. As with the OSD client update, this is mostly about getting the right structures linked into the right places so that reconnects are properly sent when needed. watch/notify v2 also requires sending regular pings to the OSDs - send_linger_ping(). A major change from the old watch/notify implementation is the introduction of ceph_osd_linger_request - linger requests no longer piggy back on ceph_osd_request. ceph_osd_event has been merged into ceph_osd_linger_request. All the details are now hidden within libceph, the interface consists of a simple pair of watch/unwatch functions and ceph_osdc_notify_ack(). ceph_osdc_watch() does return ceph_osd_linger_request, but only to keep the lifetime management simple. ceph_osdc_notify_ack() accepts an optional data payload, which is relayed back to the notifier. Portions of this patch are loosely based on work by Douglas Fuller <dfuller@redhat.com> and Mike Christie <michaelc@cs.wisc.edu>. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 23:15:02 +00:00
WARN_ON(!RB_EMPTY_ROOT(&osdc->linger_requests));
libceph: pool deletion detection This adds the "map check" infrastructure for sending osdmap version checks on CALC_TARGET_POOL_DNE and completing in-flight requests with -ENOENT if the target pool doesn't exist or has just been deleted. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
WARN_ON(!RB_EMPTY_ROOT(&osdc->map_checks));
WARN_ON(!RB_EMPTY_ROOT(&osdc->linger_map_checks));
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
WARN_ON(atomic_read(&osdc->num_requests));
WARN_ON(atomic_read(&osdc->num_homeless));
libceph: open-code remove_{all,old}_osds() They are called only once, from ceph_osdc_stop() and handle_osds_timeout() respectively. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:22 +00:00
libceph: allocate dummy osdmap in ceph_osdc_init() This leads to a simpler osdmap handling code, particularly when dealing with pi->was_full, which is introduced in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:25 +00:00
ceph_osdmap_destroy(osdc->osdmap);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
mempool_destroy(osdc->req_mempool);
ceph_msgpool_destroy(&osdc->msgpool_op);
ceph: use single osd op reply msg Use a single ceph_msg for the osd reply, even when we are getting multiple replies. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-01 21:02:00 +00:00
ceph_msgpool_destroy(&osdc->msgpool_op_reply);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
/*
* Read some contiguous pages. If we cross a stripe boundary, shorten
* *plen. Return number of bytes read, or error.
*/
int ceph_osdc_readpages(struct ceph_osd_client *osdc,
struct ceph_vino vino, struct ceph_file_layout *layout,
u64 off, u64 *plen,
u32 truncate_seq, u64 truncate_size,
ceph: make page alignment explicit in osd interface We used to infer alignment of IOs within a page based on the file offset, which assumed they matched. This broke with direct IO that was not aligned to pages (e.g., 512-byte aligned IO). We were also trusting the alignment specified in the OSD reply, which could have been adjusted by the server. Explicitly specify the page alignment when setting up OSD IO requests. Signed-off-by: Sage Weil <sage@newdream.net>
2010-11-09 20:43:12 +00:00
struct page **pages, int num_pages, int page_align)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
struct ceph_osd_request *req;
int rc = 0;
dout("readpages on ino %llx.%llx on %llu~%llu\n", vino.ino,
vino.snap, off, *plen);
libceph: specify position of extent operation allow specifying position of extent operation in multi-operations osd request. This is required for cephfs to convert inline data to normal data (compare xattr, then write object). Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 06:40:37 +00:00
req = ceph_osdc_new_request(osdc, layout, vino, off, plen, 0, 1,
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ,
libceph: don't build request in ceph_osdc_new_request() This patch moves the call to ceph_osdc_build_request() out of ceph_osdc_new_request() and into its caller. This is in order to defer formatting osd operation information into the request message until just before request is started. The only unusual (ab)user of ceph_osdc_build_request() is ceph_writepages_start(), where the final length of write request may change (downward) based on the current inode size or the oldest snapshot context with dirty data for the inode. The remaining callers don't change anything in the request after has been built. This means the ops array is now supplied by the caller. It also means there is no need to pass the mtime to ceph_osdc_new_request() (it gets provided to ceph_osdc_build_request()). And rather than passing a do_sync flag, have the number of ops in the ops array supplied imply adding a second STARTSYNC operation after the READ or WRITE requested. This and some of the patches that follow are related to having the messenger (only) be responsible for filling the content of the message header, as described here: http://tracker.ceph.com/issues/4589 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 19:09:05 +00:00
NULL, truncate_seq, truncate_size,
libceph: don't assign page info in ceph_osdc_new_request() Currently ceph_osdc_new_request() assigns an osd request's r_num_pages and r_alignment fields. The only thing it does after that is call ceph_osdc_build_request(), and that doesn't need those fields to be assigned. Move the assignment of those fields out of ceph_osdc_new_request() and into its caller. As a result, the page_align parameter is no longer used, so get rid of it. Note that in ceph_sync_write(), the value for req->r_num_pages had already been calculated earlier (as num_pages, and fortunately it was computed the same way). So don't bother recomputing it, but because it's not needed earlier, move that calculation after the call to ceph_osdc_new_request(). Hold off making the assignment to r_alignment, doing it instead r_pages and r_num_pages are getting set. Similarly, in start_read(), nr_pages already holds the number of pages in the array (and is calculated the same way), so there's no need to recompute it. Move the assignment of the page alignment down with the others there as well. This and the next few patches are preparation work for: http://tracker.ceph.com/issues/4127 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-02 00:00:15 +00:00
false);
ceph: propagate layout error on osd request creation If we are creating an osd request and get an invalid layout, return an EINVAL to the caller. We switch up the return to have an error code instead of NULL implying -ENOMEM. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2012-09-25 04:01:02 +00:00
if (IS_ERR(req))
return PTR_ERR(req);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/* it may be a short read due to an object boundary */
libceph: kill off osd data write_request parameters In the incremental move toward supporting distinct data items in an osd request some of the functions had "write_request" parameters to indicate, basically, whether the data belonged to in_data or the out_data. Now that we maintain the data fields in the op structure there is no need to indicate the direction, so get rid of the "write_request" parameters. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
osd_req_op_extent_osd_data_pages(req, 0,
libceph: combine initializing and setting osd data This ends up being a rather large patch but what it's doing is somewhat straightforward. Basically, this is replacing two calls with one. The first of the two calls is initializing a struct ceph_osd_data with data (either a page array, a page list, or a bio list); the second is setting an osd request op so it associates that data with one of the op's parameters. In place of those two will be a single function that initializes the op directly. That means we sort of fan out a set of the needed functions: - extent ops with pages data - extent ops with pagelist data - extent ops with bio list data and - class ops with page data for receiving a response We also have define another one, but it's only used internally: - class ops with pagelist data for request parameters Note that we *still* haven't gotten rid of the osd request's r_data_in and r_data_out fields. All the osd ops refer to them for their data. For now, these data fields are pointers assigned to the appropriate r_data_* field when these new functions are called. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-05 06:27:12 +00:00
pages, *plen, page_align, false, false);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: record byte count not page count Record the byte count for an osd request rather than the page count. The number of pages can always be derived from the byte count (and alignment/offset) but the reverse is not true. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-07 21:38:25 +00:00
dout("readpages final extent is %llu~%llu (%llu bytes align %d)\n",
libceph: define osd data initialization helpers Define and use functions that encapsulate the initializion of a ceph_osd_data structure. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
off, *plen, *plen, page_align);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
rc = ceph_osdc_start_request(osdc, req, false);
if (!rc)
rc = ceph_osdc_wait_request(osdc, req);
ceph_osdc_put_request(req);
dout("readpages result %d\n", rc);
return rc;
}
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:14:15 +00:00
EXPORT_SYMBOL(ceph_osdc_readpages);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* do a synchronous write on N pages
*/
int ceph_osdc_writepages(struct ceph_osd_client *osdc, struct ceph_vino vino,
struct ceph_file_layout *layout,
struct ceph_snap_context *snapc,
u64 off, u64 len,
u32 truncate_seq, u64 truncate_size,
struct timespec *mtime,
ceph: kill ceph_osdc_writepages() "flags" parameter There is only one caller of ceph_osdc_writepages(), and it always passes 0 as its "flags" argument. Get rid of that argument and replace its use in ceph_osdc_writepages() with 0. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-15 17:42:29 +00:00
struct page **pages, int num_pages)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
struct ceph_osd_request *req;
int rc = 0;
ceph: make page alignment explicit in osd interface We used to infer alignment of IOs within a page based on the file offset, which assumed they matched. This broke with direct IO that was not aligned to pages (e.g., 512-byte aligned IO). We were also trusting the alignment specified in the OSD reply, which could have been adjusted by the server. Explicitly specify the page alignment when setting up OSD IO requests. Signed-off-by: Sage Weil <sage@newdream.net>
2010-11-09 20:43:12 +00:00
int page_align = off & ~PAGE_MASK;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: specify position of extent operation allow specifying position of extent operation in multi-operations osd request. This is required for cephfs to convert inline data to normal data (compare xattr, then write object). Signed-off-by: Yan, Zheng <zyan@redhat.com> Reviewed-by: Ilya Dryomov <idryomov@redhat.com>
2014-11-13 06:40:37 +00:00
req = ceph_osdc_new_request(osdc, layout, vino, off, &len, 0, 1,
libceph, rbd, ceph: WRITE | ONDISK -> WRITE CEPH_OSD_FLAG_ONDISK is set in account_request(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Sage Weil <sage@redhat.com>
2017-02-11 17:48:41 +00:00
CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE,
libceph: don't build request in ceph_osdc_new_request() This patch moves the call to ceph_osdc_build_request() out of ceph_osdc_new_request() and into its caller. This is in order to defer formatting osd operation information into the request message until just before request is started. The only unusual (ab)user of ceph_osdc_build_request() is ceph_writepages_start(), where the final length of write request may change (downward) based on the current inode size or the oldest snapshot context with dirty data for the inode. The remaining callers don't change anything in the request after has been built. This means the ops array is now supplied by the caller. It also means there is no need to pass the mtime to ceph_osdc_new_request() (it gets provided to ceph_osdc_build_request()). And rather than passing a do_sync flag, have the number of ops in the ops array supplied imply adding a second STARTSYNC operation after the READ or WRITE requested. This and some of the patches that follow are related to having the messenger (only) be responsible for filling the content of the message header, as described here: http://tracker.ceph.com/issues/4589 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-14 19:09:05 +00:00
snapc, truncate_seq, truncate_size,
libceph: don't assign page info in ceph_osdc_new_request() Currently ceph_osdc_new_request() assigns an osd request's r_num_pages and r_alignment fields. The only thing it does after that is call ceph_osdc_build_request(), and that doesn't need those fields to be assigned. Move the assignment of those fields out of ceph_osdc_new_request() and into its caller. As a result, the page_align parameter is no longer used, so get rid of it. Note that in ceph_sync_write(), the value for req->r_num_pages had already been calculated earlier (as num_pages, and fortunately it was computed the same way). So don't bother recomputing it, but because it's not needed earlier, move that calculation after the call to ceph_osdc_new_request(). Hold off making the assignment to r_alignment, doing it instead r_pages and r_num_pages are getting set. Similarly, in start_read(), nr_pages already holds the number of pages in the array (and is calculated the same way), so there's no need to recompute it. Move the assignment of the page alignment down with the others there as well. This and the next few patches are preparation work for: http://tracker.ceph.com/issues/4127 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-03-02 00:00:15 +00:00
true);
ceph: propagate layout error on osd request creation If we are creating an osd request and get an invalid layout, return an EINVAL to the caller. We switch up the return to have an error code instead of NULL implying -ENOMEM. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2012-09-25 04:01:02 +00:00
if (IS_ERR(req))
return PTR_ERR(req);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/* it may be a short write due to an object boundary */
libceph: kill off osd data write_request parameters In the incremental move toward supporting distinct data items in an osd request some of the functions had "write_request" parameters to indicate, basically, whether the data belonged to in_data or the out_data. Now that we maintain the data fields in the op structure there is no need to indicate the direction, so get rid of the "write_request" parameters. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-15 19:50:36 +00:00
osd_req_op_extent_osd_data_pages(req, 0, pages, len, page_align,
libceph: define osd data initialization helpers Define and use functions that encapsulate the initializion of a ceph_osd_data structure. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-03 06:28:57 +00:00
false, false);
dout("writepages %llu~%llu (%llu bytes)\n", off, len, len);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: switch to calc_target(), part 2 The crux of this is getting rid of ceph_osdc_build_request(), so that MOSDOp can be encoded not before but after calc_target() calculates the actual target. Encoding now happens within ceph_osdc_start_request(). Also nuked is the accompanying bunch of pointers into the encoded buffer that was used to update fields on each send - instead, the entire front is re-encoded. If we want to support target->name_len != base->name_len in the future, there is no other way, because oid is surrounded by other fields in the encoded buffer. Encoding OSD ops and adding data items to the request message were mixed together in osd_req_encode_op(). While we want to re-encode OSD ops, we don't want to add duplicate data items to the message when resending, so all call to ceph_osdc_msg_data_add() are factored out into a new setup_request_data(). Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-05-25 22:29:52 +00:00
req->r_mtime = *mtime;
ceph: kill ceph_osdc_writepages() "nofail" parameter There is only one caller of ceph_osdc_writepages(), and it always passes the value true as its "nofail" argument. Get rid of that argument and replace its use in ceph_osdc_writepages() with the constant value true. This and a number of cleanup patches that follow resolve: http://tracker.ceph.com/issues/4126 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-15 17:42:29 +00:00
rc = ceph_osdc_start_request(osdc, req, true);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
if (!rc)
rc = ceph_osdc_wait_request(osdc, req);
ceph_osdc_put_request(req);
if (rc == 0)
rc = len;
dout("writepages result %d\n", rc);
return rc;
}
ceph: factor out libceph from Ceph file system This factors out protocol and low-level storage parts of ceph into a separate libceph module living in net/ceph and include/linux/ceph. This is mostly a matter of moving files around. However, a few key pieces of the interface change as well: - ceph_client becomes ceph_fs_client and ceph_client, where the latter captures the mon and osd clients, and the fs_client gets the mds client and file system specific pieces. - Mount option parsing and debugfs setup is correspondingly broken into two pieces. - The mon client gets a generic handler callback for otherwise unknown messages (mds map, in this case). - The basic supported/required feature bits can be expanded (and are by ceph_fs_client). No functional change, aside from some subtle error handling cases that got cleaned up in the refactoring process. Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-06 22:14:15 +00:00
EXPORT_SYMBOL(ceph_osdc_writepages);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph, ceph: add __init attribution to init funcitons Add __init attribution to the functions which are called only once during initiating/registering operations and deleting unnecessary symbol exports. Signed-off-by: Chengguang Xu <cgxu519@gmx.com> Reviewed-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2018-03-10 12:32:05 +00:00
int __init ceph_osdc_setup(void)
libceph: use slab cache for osd client requests Create a slab cache to manage allocation of ceph_osdc_request structures. This resolves: http://tracker.ceph.com/issues/3926 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-05-01 17:43:04 +00:00
{
libceph: enable large, variable-sized OSD requests Turn r_ops into a flexible array member to enable large, consisting of up to 16 ops, OSD requests. The use case is scattered writeback in cephfs and, as far as the kernel client is concerned, 16 is just a made up number. r_ops had size 3 for copyup+hint+write, but copyup is really a special case - it can only happen once. ceph_osd_request_cache is therefore stuffed with num_ops=2 requests, anything bigger than that is allocated with kmalloc(). req_mempool is backed by ceph_osd_request_cache, which means either num_ops=1 or num_ops=2 for use_mempool=true - all existing users (ceph_writepages_start(), ceph_osdc_writepages()) are fine with that. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:50:15 +00:00
size_t size = sizeof(struct ceph_osd_request) +
CEPH_OSD_SLAB_OPS * sizeof(struct ceph_osd_req_op);
libceph: use slab cache for osd client requests Create a slab cache to manage allocation of ceph_osdc_request structures. This resolves: http://tracker.ceph.com/issues/3926 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-05-01 17:43:04 +00:00
BUG_ON(ceph_osd_request_cache);
libceph: enable large, variable-sized OSD requests Turn r_ops into a flexible array member to enable large, consisting of up to 16 ops, OSD requests. The use case is scattered writeback in cephfs and, as far as the kernel client is concerned, 16 is just a made up number. r_ops had size 3 for copyup+hint+write, but copyup is really a special case - it can only happen once. ceph_osd_request_cache is therefore stuffed with num_ops=2 requests, anything bigger than that is allocated with kmalloc(). req_mempool is backed by ceph_osd_request_cache, which means either num_ops=1 or num_ops=2 for use_mempool=true - all existing users (ceph_writepages_start(), ceph_osdc_writepages()) are fine with that. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-02-09 16:50:15 +00:00
ceph_osd_request_cache = kmem_cache_create("ceph_osd_request", size,
0, 0, NULL);
libceph: use slab cache for osd client requests Create a slab cache to manage allocation of ceph_osdc_request structures. This resolves: http://tracker.ceph.com/issues/3926 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-05-01 17:43:04 +00:00
return ceph_osd_request_cache ? 0 : -ENOMEM;
}
void ceph_osdc_cleanup(void)
{
BUG_ON(!ceph_osd_request_cache);
kmem_cache_destroy(ceph_osd_request_cache);
ceph_osd_request_cache = NULL;
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
/*
* handle incoming message
*/
static void dispatch(struct ceph_connection *con, struct ceph_msg *msg)
{
struct ceph_osd *osd = con->private;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_osd_client *osdc = osd->o_osdc;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
int type = le16_to_cpu(msg->hdr.type);
switch (type) {
case CEPH_MSG_OSD_MAP:
ceph_osdc_handle_map(osdc, msg);
break;
case CEPH_MSG_OSD_OPREPLY:
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
handle_reply(osd, msg);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
break;
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
case CEPH_MSG_OSD_BACKOFF:
handle_backoff(osd, msg);
break;
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
case CEPH_MSG_WATCH_NOTIFY:
handle_watch_notify(osdc, msg);
break;
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
default:
pr_err("received unknown message type %d %s\n", type,
ceph_msg_type_name(type));
}
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
ceph_msg_put(msg);
}
ceph: fix up unexpected message handling Fix skipping of unexpected message types from osd, mon. Clean up pr_info and debug output. Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-20 05:43:23 +00:00
/*
libceph: check data_len in ->alloc_msg() Only ->alloc_msg() should check data_len of the incoming message against the preallocated ceph_msg, doing it in the messenger is not right. The contract is that either ->alloc_msg() returns a ceph_msg which will fit all of the portions of the incoming message, or it returns NULL and possibly sets skip, signaling whether NULL is due to an -ENOMEM. ->alloc_msg() should be the only place where we make the skip/no-skip decision. I stumbled upon this while looking at con/osd ref counting. Right now, if we get a non-extent message with a larger data portion than we are prepared for, ->alloc_msg() returns a ceph_msg, and then, when we skip it in the messenger, we don't put the con/osd ref acquired in ceph_con_in_msg_alloc() (which is normally put in process_message()), so this also fixes a memory leak. An existing BUG_ON in ceph_msg_data_cursor_init() ensures we don't corrupt random memory should a buggy ->alloc_msg() return an unfit ceph_msg. While at it, I changed the "unknown tid" dout() to a pr_warn() to make sure all skips are seen and unified format strings. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-09-02 08:37:09 +00:00
* Lookup and return message for incoming reply. Don't try to do
* anything about a larger than preallocated data portion of the
* message at the moment - for now, just skip the message.
ceph: fix up unexpected message handling Fix skipping of unexpected message types from osd, mon. Clean up pr_info and debug output. Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-20 05:43:23 +00:00
*/
static struct ceph_msg *get_reply(struct ceph_connection *con,
ceph: allocate middle of message before stating to read Both front and middle parts of the message are now being allocated at the ceph_alloc_msg(). Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
2010-01-08 21:58:34 +00:00
struct ceph_msg_header *hdr,
int *skip)
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
{
struct ceph_osd *osd = con->private;
struct ceph_osd_client *osdc = osd->o_osdc;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
struct ceph_msg *m = NULL;
ceph: alloc message data pages and check if tid exists Now doing it in the same callback that is also responsible for allocating the 'front' part of the message. If we get a message that we haven't got a corresponding tid for, mark it for skipping. Moving the mutex unlock/lock from the osd alloc_msg callback to the calling function in the messenger. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
2010-01-11 22:47:13 +00:00
struct ceph_osd_request *req;
libceph: rename front to front_len in get_reply() Rename front local variable to front_len in get_reply() to make its purpose more clear. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-09 18:08:21 +00:00
int front_len = le32_to_cpu(hdr->front_len);
ceph: fix up unexpected message handling Fix skipping of unexpected message types from osd, mon. Clean up pr_info and debug output. Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-20 05:43:23 +00:00
int data_len = le32_to_cpu(hdr->data_len);
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
u64 tid = le64_to_cpu(hdr->tid);
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
down_read(&osdc->lock);
if (!osd_registered(osd)) {
dout("%s osd%d unknown, skipping\n", __func__, osd->o_osd);
*skip = 1;
goto out_unlock_osdc;
}
WARN_ON(osd->o_osd != le64_to_cpu(hdr->src.num));
mutex_lock(&osd->lock);
req = lookup_request(&osd->o_requests, tid);
ceph: alloc message data pages and check if tid exists Now doing it in the same callback that is also responsible for allocating the 'front' part of the message. If we get a message that we haven't got a corresponding tid for, mark it for skipping. Moving the mutex unlock/lock from the osd alloc_msg callback to the calling function in the messenger. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
2010-01-11 22:47:13 +00:00
if (!req) {
libceph: don't spam dmesg with stray reply warnings Commit d15f9d694b77 ("libceph: check data_len in ->alloc_msg()") mistakenly bumped the log level on the "tid %llu unknown, skipping" message. Turn it back into a dout() - stray replies are perfectly normal when OSDs flap, crash, get killed for testing purposes, etc. Cc: stable@vger.kernel.org # 4.3+ Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2016-02-19 10:38:57 +00:00
dout("%s osd%d tid %llu unknown, skipping\n", __func__,
osd->o_osd, tid);
libceph: check data_len in ->alloc_msg() Only ->alloc_msg() should check data_len of the incoming message against the preallocated ceph_msg, doing it in the messenger is not right. The contract is that either ->alloc_msg() returns a ceph_msg which will fit all of the portions of the incoming message, or it returns NULL and possibly sets skip, signaling whether NULL is due to an -ENOMEM. ->alloc_msg() should be the only place where we make the skip/no-skip decision. I stumbled upon this while looking at con/osd ref counting. Right now, if we get a non-extent message with a larger data portion than we are prepared for, ->alloc_msg() returns a ceph_msg, and then, when we skip it in the messenger, we don't put the con/osd ref acquired in ceph_con_in_msg_alloc() (which is normally put in process_message()), so this also fixes a memory leak. An existing BUG_ON in ceph_msg_data_cursor_init() ensures we don't corrupt random memory should a buggy ->alloc_msg() return an unfit ceph_msg. While at it, I changed the "unknown tid" dout() to a pr_warn() to make sure all skips are seen and unified format strings. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-09-02 08:37:09 +00:00
*skip = 1;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
goto out_unlock_session;
ceph: alloc message data pages and check if tid exists Now doing it in the same callback that is also responsible for allocating the 'front' part of the message. If we get a message that we haven't got a corresponding tid for, mark it for skipping. Moving the mutex unlock/lock from the osd alloc_msg callback to the calling function in the messenger. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
2010-01-11 22:47:13 +00:00
}
ceph: use single osd op reply msg Use a single ceph_msg for the osd reply, even when we are getting multiple replies. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-01 21:02:00 +00:00
libceph: drop ceph_osd_request->r_con_filling_msg A field in an osd request keeps track of whether a connection is currently filling the request's reply message. This patch gets rid of that field. An osd request includes two messages--a request and a reply--and they're both associated with the connection that existed to its the target osd at the time the request was created. An osd request can be dropped early, even when it's in flight. And at that time both messages are released. It's possible the reply message has been supplied to its connection to receive an incoming response message at the time the osd request gets dropped. So ceph_osdc_release_request() revokes that message from the connection before releasing it so things get cleaned up properly. Previously this may have caused a problem, because the connection that a message was associated with might have gone away before the revoke request. And to avoid any problems using that connection, the osd client held a reference to it when it supplies its response message. However since this commit: 38941f80 libceph: have messages point to their connection all messages hold a reference to the connection they are associated with whenever the connection is actively operating on the message (i.e. while the message is queued to send or sending, and when it data is being received into it). And if a message has no connection associated with it, ceph_msg_revoke_incoming() won't do anything when asked to revoke it. As a result, there is no need to keep an additional reference to the connection associated with a message when we hand the message to the messenger when it calls our alloc_msg() method to receive something. If the connection *were* operating on it, it would have its own reference, and if not, there's no work to be done when we need to revoke it. So get rid of the osd request's r_con_filling_msg field. This resolves: http://tracker.ceph.com/issues/4647 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-04-01 21:12:14 +00:00
ceph_msg_revoke_incoming(req->r_reply);
ceph: alloc message data pages and check if tid exists Now doing it in the same callback that is also responsible for allocating the 'front' part of the message. If we get a message that we haven't got a corresponding tid for, mark it for skipping. Moving the mutex unlock/lock from the osd alloc_msg callback to the calling function in the messenger. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
2010-01-11 22:47:13 +00:00
libceph: fix preallocation check in get_reply() The check that makes sure that we have enough memory allocated to read in the entire header of the message in question is currently busted. It compares front_len of the incoming message with iov_len field of ceph_msg::front structure, which is used primarily to indicate the amount of data already read in, and not the size of the allocated buffer. Under certain conditions (e.g. a short read from a socket followed by that socket's shutdown and owning ceph_connection reset) this results in a warning similar to [85688.975866] libceph: get_reply front 198 > preallocated 122 (4#0) and, through another bug, leads to forever hung tasks and forced reboots. Fix this by comparing front_len with front_alloc_len field of struct ceph_msg, which stores the actual size of the buffer. Fixes: http://tracker.ceph.com/issues/5425 Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-09 18:08:21 +00:00
if (front_len > req->r_reply->front_alloc_len) {
libceph: check data_len in ->alloc_msg() Only ->alloc_msg() should check data_len of the incoming message against the preallocated ceph_msg, doing it in the messenger is not right. The contract is that either ->alloc_msg() returns a ceph_msg which will fit all of the portions of the incoming message, or it returns NULL and possibly sets skip, signaling whether NULL is due to an -ENOMEM. ->alloc_msg() should be the only place where we make the skip/no-skip decision. I stumbled upon this while looking at con/osd ref counting. Right now, if we get a non-extent message with a larger data portion than we are prepared for, ->alloc_msg() returns a ceph_msg, and then, when we skip it in the messenger, we don't put the con/osd ref acquired in ceph_con_in_msg_alloc() (which is normally put in process_message()), so this also fixes a memory leak. An existing BUG_ON in ceph_msg_data_cursor_init() ensures we don't corrupt random memory should a buggy ->alloc_msg() return an unfit ceph_msg. While at it, I changed the "unknown tid" dout() to a pr_warn() to make sure all skips are seen and unified format strings. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-09-02 08:37:09 +00:00
pr_warn("%s osd%d tid %llu front %d > preallocated %d\n",
__func__, osd->o_osd, req->r_tid, front_len,
req->r_reply->front_alloc_len);
libceph: rename front to front_len in get_reply() Rename front local variable to front_len in get_reply() to make its purpose more clear. Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2014-01-09 18:08:21 +00:00
m = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, front_len, GFP_NOFS,
false);
ceph: make ceph_msg_new return NULL on failure; clean up, fix callers Returning ERR_PTR(-ENOMEM) is useless extra work. Return NULL on failure instead, and fix up the callers (about half of which were wrong anyway). Signed-off-by: Sage Weil <sage@newdream.net>
2010-04-01 23:06:19 +00:00
if (!m)
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
goto out_unlock_session;
ceph: use single osd op reply msg Use a single ceph_msg for the osd reply, even when we are getting multiple replies. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-01 21:02:00 +00:00
ceph_msg_put(req->r_reply);
req->r_reply = m;
}
libceph: separate read and write data An osd request defines information about where data to be read should be placed as well as where data to write comes from. Currently these are represented by common fields. Keep information about data for writing separate from data to be read by splitting these into data_in and data_out fields. This is the key patch in this whole series, in that it actually identifies which osd requests generate outgoing data and which generate incoming data. It's less obvious (currently) that an osd CALL op generates both outgoing and incoming data; that's the focus of some upcoming work. This resolves: http://tracker.ceph.com/issues/4127 Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
2013-02-14 18:16:43 +00:00
libceph: check data_len in ->alloc_msg() Only ->alloc_msg() should check data_len of the incoming message against the preallocated ceph_msg, doing it in the messenger is not right. The contract is that either ->alloc_msg() returns a ceph_msg which will fit all of the portions of the incoming message, or it returns NULL and possibly sets skip, signaling whether NULL is due to an -ENOMEM. ->alloc_msg() should be the only place where we make the skip/no-skip decision. I stumbled upon this while looking at con/osd ref counting. Right now, if we get a non-extent message with a larger data portion than we are prepared for, ->alloc_msg() returns a ceph_msg, and then, when we skip it in the messenger, we don't put the con/osd ref acquired in ceph_con_in_msg_alloc() (which is normally put in process_message()), so this also fixes a memory leak. An existing BUG_ON in ceph_msg_data_cursor_init() ensures we don't corrupt random memory should a buggy ->alloc_msg() return an unfit ceph_msg. While at it, I changed the "unknown tid" dout() to a pr_warn() to make sure all skips are seen and unified format strings. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-09-02 08:37:09 +00:00
if (data_len > req->r_reply->data_length) {
pr_warn("%s osd%d tid %llu data %d > preallocated %zu, skipping\n",
__func__, osd->o_osd, req->r_tid, data_len,
req->r_reply->data_length);
m = NULL;
*skip = 1;
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
goto out_unlock_session;
ceph: alloc message data pages and check if tid exists Now doing it in the same callback that is also responsible for allocating the 'front' part of the message. If we get a message that we haven't got a corresponding tid for, mark it for skipping. Moving the mutex unlock/lock from the osd alloc_msg callback to the calling function in the messenger. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
2010-01-11 22:47:13 +00:00
}
libceph: check data_len in ->alloc_msg() Only ->alloc_msg() should check data_len of the incoming message against the preallocated ceph_msg, doing it in the messenger is not right. The contract is that either ->alloc_msg() returns a ceph_msg which will fit all of the portions of the incoming message, or it returns NULL and possibly sets skip, signaling whether NULL is due to an -ENOMEM. ->alloc_msg() should be the only place where we make the skip/no-skip decision. I stumbled upon this while looking at con/osd ref counting. Right now, if we get a non-extent message with a larger data portion than we are prepared for, ->alloc_msg() returns a ceph_msg, and then, when we skip it in the messenger, we don't put the con/osd ref acquired in ceph_con_in_msg_alloc() (which is normally put in process_message()), so this also fixes a memory leak. An existing BUG_ON in ceph_msg_data_cursor_init() ensures we don't corrupt random memory should a buggy ->alloc_msg() return an unfit ceph_msg. While at it, I changed the "unknown tid" dout() to a pr_warn() to make sure all skips are seen and unified format strings. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Alex Elder <elder@linaro.org>
2015-09-02 08:37:09 +00:00
m = ceph_msg_get(req->r_reply);
ceph: use single osd op reply msg Use a single ceph_msg for the osd reply, even when we are getting multiple replies. Signed-off-by: Sage Weil <sage@newdream.net>
2010-03-01 21:02:00 +00:00
dout("get_reply tid %lld %p\n", tid, m);
ceph: alloc message data pages and check if tid exists Now doing it in the same callback that is also responsible for allocating the 'front' part of the message. If we get a message that we haven't got a corresponding tid for, mark it for skipping. Moving the mutex unlock/lock from the osd alloc_msg callback to the calling function in the messenger. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
2010-01-11 22:47:13 +00:00
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
out_unlock_session:
mutex_unlock(&osd->lock);
out_unlock_osdc:
up_read(&osdc->lock);
ceph: allocate middle of message before stating to read Both front and middle parts of the message are now being allocated at the ceph_alloc_msg(). Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
2010-01-08 21:58:34 +00:00
return m;
ceph: fix up unexpected message handling Fix skipping of unexpected message types from osd, mon. Clean up pr_info and debug output. Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-20 05:43:23 +00:00
}
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
/*
* TODO: switch to a msg-owned pagelist
*/
static struct ceph_msg *alloc_msg_with_page_vector(struct ceph_msg_header *hdr)
{
struct ceph_msg *m;
int type = le16_to_cpu(hdr->type);
u32 front_len = le32_to_cpu(hdr->front_len);
u32 data_len = le32_to_cpu(hdr->data_len);
m = ceph_msg_new(type, front_len, GFP_NOIO, false);
if (!m)
return NULL;
if (data_len) {
struct page **pages;
struct ceph_osd_data osd_data;
pages = ceph_alloc_page_vector(calc_pages_for(0, data_len),
GFP_NOIO);
libceph: fix return value check in alloc_msg_with_page_vector() In case of error, the function ceph_alloc_page_vector() returns ERR_PTR() and never returns NULL. The NULL test in the return value check should be replaced with IS_ERR(). Fixes: 1907920324f1 ('libceph: support for sending notifies') Signed-off-by: Wei Yongjun <weiyj.lk@gmail.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-07-30 00:37:57 +00:00
if (IS_ERR(pages)) {
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
ceph_msg_put(m);
return NULL;
}
ceph_osd_data_pages_init(&osd_data, pages, data_len, 0, false,
false);
ceph_osdc_msg_data_add(m, &osd_data);
}
return m;
}
ceph: fix up unexpected message handling Fix skipping of unexpected message types from osd, mon. Clean up pr_info and debug output. Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-20 05:43:23 +00:00
static struct ceph_msg *alloc_msg(struct ceph_connection *con,
struct ceph_msg_header *hdr,
int *skip)
{
struct ceph_osd *osd = con->private;
int type = le16_to_cpu(hdr->type);
libceph: tweak ceph_alloc_msg() The function ceph_alloc_msg() is only used to allocate a message that will be assigned to a connection's in_msg pointer. Rename the function so this implied usage is more clear. In addition, make that assignment inside the function (again, since that's precisely what it's intended to be used for). This allows us to return what is now provided via the passed-in address of a "skip" variable. The return type is now Boolean to be explicit that there are only two possible outcomes. Make sure the result of an ->alloc_msg method call always sets the value of *skip properly. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-06-04 19:43:32 +00:00
*skip = 0;
ceph: fix up unexpected message handling Fix skipping of unexpected message types from osd, mon. Clean up pr_info and debug output. Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-20 05:43:23 +00:00
switch (type) {
case CEPH_MSG_OSD_MAP:
libceph: respect RADOS_BACKOFF backoffs Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-19 10:18:05 +00:00
case CEPH_MSG_OSD_BACKOFF:
libceph: add lingering request and watch/notify event framework Lingering requests are requests that are sent to the OSD normally but tracked also after we get a successful request. This keeps the OSD connection open and resends the original request if the object moves to another OSD. The OSD can then send notification messages back to us if another client initiates a notify. This framework will be used by RBD so that the client gets notification when a snapshot is created by another node or tool. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2011-03-21 22:07:16 +00:00
case CEPH_MSG_WATCH_NOTIFY:
libceph: support for sending notifies Implement ceph_osdc_notify() for sending notifies. Due to the fact that the current messenger can't do read-in into pagelists (it can only do write-out from them), I had to go with a page vector for a NOTIFY_COMPLETE payload, for now. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:27 +00:00
return alloc_msg_with_page_vector(hdr);
ceph: fix up unexpected message handling Fix skipping of unexpected message types from osd, mon. Clean up pr_info and debug output. Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-20 05:43:23 +00:00
case CEPH_MSG_OSD_OPREPLY:
return get_reply(con, hdr, skip);
default:
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
pr_warn("%s osd%d unknown msg type %d, skipping\n", __func__,
osd->o_osd, type);
ceph: fix up unexpected message handling Fix skipping of unexpected message types from osd, mon. Clean up pr_info and debug output. Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-20 05:43:23 +00:00
*skip = 1;
return NULL;
}
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
}
/*
* Wrappers to refcount containing ceph_osd struct
*/
static struct ceph_connection *get_osd_con(struct ceph_connection *con)
{
struct ceph_osd *osd = con->private;
if (get_osd(osd))
return con;
return NULL;
}
static void put_osd_con(struct ceph_connection *con)
{
struct ceph_osd *osd = con->private;
put_osd(osd);
}
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
/*
* authentication
*/
ceph: have get_authorizer methods return pointers Have the get_authorizer auth_client method return a ceph_auth pointer rather than an integer, pointer-encoding any returned error value. This is to pave the way for making use of the returned value in an upcoming patch. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-16 20:16:39 +00:00
/*
* Note: returned pointer is the address of a structure that's
* managed separately. Caller must *not* attempt to free it.
*/
static struct ceph_auth_handshake *get_authorizer(struct ceph_connection *con,
ceph: use info returned by get_authorizer Rather than passing a bunch of arguments to be filled in with the content of the ceph_auth_handshake buffer now returned by the get_authorizer method, just use the returned information in the caller, and drop the unnecessary arguments. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-16 20:16:39 +00:00
int *proto, int force_new)
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
{
struct ceph_osd *o = con->private;
struct ceph_osd_client *osdc = o->o_osdc;
struct ceph_auth_client *ac = osdc->client->monc.auth;
ceph: messenger: reduce args to create_authorizer Make use of the new ceph_auth_handshake structure in order to reduce the number of arguments passed to the create_authorizor method in ceph_auth_client_ops. Use a local variable of that type as a shorthand in the get_authorizer method definitions. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-16 20:16:39 +00:00
struct ceph_auth_handshake *auth = &o->o_auth;
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
ceph: messenger: reduce args to create_authorizer Make use of the new ceph_auth_handshake structure in order to reduce the number of arguments passed to the create_authorizor method in ceph_auth_client_ops. Use a local variable of that type as a shorthand in the get_authorizer method definitions. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-16 20:16:39 +00:00
if (force_new && auth->authorizer) {
libceph: make authorizer destruction independent of ceph_auth_client Starting the kernel client with cephx disabled and then enabling cephx and restarting userspace daemons can result in a crash: [262671.478162] BUG: unable to handle kernel paging request at ffffebe000000000 [262671.531460] IP: [<ffffffff811cd04a>] kfree+0x5a/0x130 [262671.584334] PGD 0 [262671.635847] Oops: 0000 [#1] SMP [262672.055841] CPU: 22 PID: 2961272 Comm: kworker/22:2 Not tainted 4.2.0-34-generic #39~14.04.1-Ubuntu [262672.162338] Hardware name: Dell Inc. PowerEdge R720/068CDY, BIOS 2.4.3 07/09/2014 [262672.268937] Workqueue: ceph-msgr con_work [libceph] [262672.322290] task: ffff88081c2d0dc0 ti: ffff880149ae8000 task.ti: ffff880149ae8000 [262672.428330] RIP: 0010:[<ffffffff811cd04a>] [<ffffffff811cd04a>] kfree+0x5a/0x130 [262672.535880] RSP: 0018:ffff880149aeba58 EFLAGS: 00010286 [262672.589486] RAX: 000001e000000000 RBX: 0000000000000012 RCX: ffff8807e7461018 [262672.695980] RDX: 000077ff80000000 RSI: ffff88081af2be04 RDI: 0000000000000012 [262672.803668] RBP: ffff880149aeba78 R08: 0000000000000000 R09: 0000000000000000 [262672.912299] R10: ffffebe000000000 R11: ffff880819a60e78 R12: ffff8800aec8df40 [262673.021769] R13: ffffffffc035f70f R14: ffff8807e5b138e0 R15: ffff880da9785840 [262673.131722] FS: 0000000000000000(0000) GS:ffff88081fac0000(0000) knlGS:0000000000000000 [262673.245377] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [262673.303281] CR2: ffffebe000000000 CR3: 0000000001c0d000 CR4: 00000000001406e0 [262673.417556] Stack: [262673.472943] ffff880149aeba88 ffff88081af2be04 ffff8800aec8df40 ffff88081af2be04 [262673.583767] ffff880149aeba98 ffffffffc035f70f ffff880149aebac8 ffff8800aec8df00 [262673.694546] ffff880149aebac8 ffffffffc035c89e ffff8807e5b138e0 ffff8805b047f800 [262673.805230] Call Trace: [262673.859116] [<ffffffffc035f70f>] ceph_x_destroy_authorizer+0x1f/0x50 [libceph] [262673.968705] [<ffffffffc035c89e>] ceph_auth_destroy_authorizer+0x3e/0x60 [libceph] [262674.078852] [<ffffffffc0352805>] put_osd+0x45/0x80 [libceph] [262674.134249] [<ffffffffc035290e>] remove_osd+0xae/0x140 [libceph] [262674.189124] [<ffffffffc0352aa3>] __reset_osd+0x103/0x150 [libceph] [262674.243749] [<ffffffffc0354703>] kick_requests+0x223/0x460 [libceph] [262674.297485] [<ffffffffc03559e2>] ceph_osdc_handle_map+0x282/0x5e0 [libceph] [262674.350813] [<ffffffffc035022e>] dispatch+0x4e/0x720 [libceph] [262674.403312] [<ffffffffc034bd91>] try_read+0x3d1/0x1090 [libceph] [262674.454712] [<ffffffff810ab7c2>] ? dequeue_entity+0x152/0x690 [262674.505096] [<ffffffffc034cb1b>] con_work+0xcb/0x1300 [libceph] [262674.555104] [<ffffffff8108fb3e>] process_one_work+0x14e/0x3d0 [262674.604072] [<ffffffff810901ea>] worker_thread+0x11a/0x470 [262674.652187] [<ffffffff810900d0>] ? rescuer_thread+0x310/0x310 [262674.699022] [<ffffffff810957a2>] kthread+0xd2/0xf0 [262674.744494] [<ffffffff810956d0>] ? kthread_create_on_node+0x1c0/0x1c0 [262674.789543] [<ffffffff817bd81f>] ret_from_fork+0x3f/0x70 [262674.834094] [<ffffffff810956d0>] ? kthread_create_on_node+0x1c0/0x1c0 What happens is the following: (1) new MON session is established (2) old "none" ac is destroyed (3) new "cephx" ac is constructed ... (4) old OSD session (w/ "none" authorizer) is put ceph_auth_destroy_authorizer(ac, osd->o_auth.authorizer) osd->o_auth.authorizer in the "none" case is just a bare pointer into ac, which contains a single static copy for all services. By the time we get to (4), "none" ac, freed in (2), is long gone. On top of that, a new vtable installed in (3) points us at ceph_x_destroy_authorizer(), so we end up trying to destroy a "none" authorizer with a "cephx" destructor operating on invalid memory! To fix this, decouple authorizer destruction from ac and do away with a single static "none" authorizer by making a copy for each OSD or MDS session. Authorizers themselves are independent of ac and so there is no reason for destroy_authorizer() to be an ac op. Make it an op on the authorizer itself by turning ceph_authorizer into a real struct. Fixes: http://tracker.ceph.com/issues/15447 Reported-by: Alan Zhang <alan.zhang@linux.com> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2016-04-11 17:34:49 +00:00
ceph_auth_destroy_authorizer(auth->authorizer);
ceph: messenger: reduce args to create_authorizer Make use of the new ceph_auth_handshake structure in order to reduce the number of arguments passed to the create_authorizor method in ceph_auth_client_ops. Use a local variable of that type as a shorthand in the get_authorizer method definitions. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-16 20:16:39 +00:00
auth->authorizer = NULL;
}
libceph: wrap auth ops in wrapper functions Use wrapper functions that check whether the auth op exists so that callers do not need a bunch of conditional checks. Simplifies the external interface. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-03-25 17:26:14 +00:00
if (!auth->authorizer) {
int ret = ceph_auth_create_authorizer(ac, CEPH_ENTITY_TYPE_OSD,
auth);
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
if (ret)
ceph: have get_authorizer methods return pointers Have the get_authorizer auth_client method return a ceph_auth pointer rather than an integer, pointer-encoding any returned error value. This is to pave the way for making use of the returned value in an upcoming patch. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-16 20:16:39 +00:00
return ERR_PTR(ret);
libceph: wrap auth ops in wrapper functions Use wrapper functions that check whether the auth op exists so that callers do not need a bunch of conditional checks. Simplifies the external interface. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-03-25 17:26:14 +00:00
} else {
int ret = ceph_auth_update_authorizer(ac, CEPH_ENTITY_TYPE_OSD,
libceph: add update_authorizer auth method Currently the messenger calls out to a get_authorizer con op, which will create a new authorizer if it doesn't yet have one. In the meantime, when we rotate our service keys, the authorizer doesn't get updated. Eventually it will be rejected by the server on a new connection attempt and get invalidated, and we will then rebuild a new authorizer, but this is not ideal. Instead, if we do have an authorizer, call a new update_authorizer op that will verify that the current authorizer is using the latest secret. If it is not, we will build a new one that does. This avoids the transient failure. This fixes one of the sorry sequence of events for bug http://tracker.ceph.com/issues/4282 Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-03-25 17:26:01 +00:00
auth);
if (ret)
return ERR_PTR(ret);
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
}
*proto = ac->protocol;
ceph: messenger: reduce args to create_authorizer Make use of the new ceph_auth_handshake structure in order to reduce the number of arguments passed to the create_authorizor method in ceph_auth_client_ops. Use a local variable of that type as a shorthand in the get_authorizer method definitions. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-16 20:16:39 +00:00
ceph: have get_authorizer methods return pointers Have the get_authorizer auth_client method return a ceph_auth pointer rather than an integer, pointer-encoding any returned error value. This is to pave the way for making use of the returned value in an upcoming patch. Signed-off-by: Alex Elder <elder@inktank.com> Reviewed-by: Sage Weil <sage@inktank.com>
2012-05-16 20:16:39 +00:00
return auth;
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
}
libceph: drop len argument of *verify_authorizer_reply() The length of the reply is protocol-dependent - for cephx it's ceph_x_authorize_reply. Nothing sensible can be passed from the messenger layer anyway. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2016-12-02 15:35:09 +00:00
static int verify_authorizer_reply(struct ceph_connection *con)
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
{
struct ceph_osd *o = con->private;
struct ceph_osd_client *osdc = o->o_osdc;
struct ceph_auth_client *ac = osdc->client->monc.auth;
libceph: drop len argument of *verify_authorizer_reply() The length of the reply is protocol-dependent - for cephx it's ceph_x_authorize_reply. Nothing sensible can be passed from the messenger layer anyway. Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Sage Weil <sage@redhat.com>
2016-12-02 15:35:09 +00:00
return ceph_auth_verify_authorizer_reply(ac, o->o_auth.authorizer);
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
}
ceph: allow renewal of auth credentials Add infrastructure to allow the mon_client to periodically renew its auth credentials. Also add a messenger callback that will force such a renewal if a peer rejects our authenticator. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 00:21:06 +00:00
static int invalidate_authorizer(struct ceph_connection *con)
{
struct ceph_osd *o = con->private;
struct ceph_osd_client *osdc = o->o_osdc;
struct ceph_auth_client *ac = osdc->client->monc.auth;
libceph: wrap auth ops in wrapper functions Use wrapper functions that check whether the auth op exists so that callers do not need a bunch of conditional checks. Simplifies the external interface. Signed-off-by: Sage Weil <sage@inktank.com> Reviewed-by: Alex Elder <elder@inktank.com>
2013-03-25 17:26:14 +00:00
ceph_auth_invalidate_authorizer(ac, CEPH_ENTITY_TYPE_OSD);
ceph: allow renewal of auth credentials Add infrastructure to allow the mon_client to periodically renew its auth credentials. Also add a messenger callback that will force such a renewal if a peer rejects our authenticator. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 00:21:06 +00:00
return ceph_monc_validate_auth(&osdc->client->monc);
}
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
static void osd_reencode_message(struct ceph_msg *msg)
{
libceph: don't call encode_request_finish() on MOSDBackoff messages encode_request_finish() is for MOSDOp messages. Calling it on MOSDBackoff ack-block messages corrupts them. Fixes: a02a946dfe96 ("libceph: respect RADOS_BACKOFF backoffs") Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-07-14 14:08:54 +00:00
int type = le16_to_cpu(msg->hdr.type);
if (type == CEPH_MSG_OSD_OP)
encode_request_finish(msg);
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
}
libceph: msg signing callouts don't need con argument We can use msg->con instead - at the point we sign an outgoing message or check the signature on the incoming one, msg->con is always set. We wouldn't know how to sign a message without an associated session (i.e. msg->con == NULL) and being able to sign a message using an explicitly provided authorizer is of no use. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2015-10-26 21:23:56 +00:00
static int osd_sign_message(struct ceph_msg *msg)
libceph: message signature support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2014-11-04 08:33:37 +00:00
{
libceph: msg signing callouts don't need con argument We can use msg->con instead - at the point we sign an outgoing message or check the signature on the incoming one, msg->con is always set. We wouldn't know how to sign a message without an associated session (i.e. msg->con == NULL) and being able to sign a message using an explicitly provided authorizer is of no use. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2015-10-26 21:23:56 +00:00
struct ceph_osd *o = msg->con->private;
libceph: message signature support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2014-11-04 08:33:37 +00:00
struct ceph_auth_handshake *auth = &o->o_auth;
libceph: msg signing callouts don't need con argument We can use msg->con instead - at the point we sign an outgoing message or check the signature on the incoming one, msg->con is always set. We wouldn't know how to sign a message without an associated session (i.e. msg->con == NULL) and being able to sign a message using an explicitly provided authorizer is of no use. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2015-10-26 21:23:56 +00:00
libceph: message signature support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2014-11-04 08:33:37 +00:00
return ceph_auth_sign_message(auth, msg);
}
libceph: msg signing callouts don't need con argument We can use msg->con instead - at the point we sign an outgoing message or check the signature on the incoming one, msg->con is always set. We wouldn't know how to sign a message without an associated session (i.e. msg->con == NULL) and being able to sign a message using an explicitly provided authorizer is of no use. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2015-10-26 21:23:56 +00:00
static int osd_check_message_signature(struct ceph_msg *msg)
libceph: message signature support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2014-11-04 08:33:37 +00:00
{
libceph: msg signing callouts don't need con argument We can use msg->con instead - at the point we sign an outgoing message or check the signature on the incoming one, msg->con is always set. We wouldn't know how to sign a message without an associated session (i.e. msg->con == NULL) and being able to sign a message using an explicitly provided authorizer is of no use. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2015-10-26 21:23:56 +00:00
struct ceph_osd *o = msg->con->private;
libceph: message signature support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2014-11-04 08:33:37 +00:00
struct ceph_auth_handshake *auth = &o->o_auth;
libceph: msg signing callouts don't need con argument We can use msg->con instead - at the point we sign an outgoing message or check the signature on the incoming one, msg->con is always set. We wouldn't know how to sign a message without an associated session (i.e. msg->con == NULL) and being able to sign a message using an explicitly provided authorizer is of no use. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2015-10-26 21:23:56 +00:00
libceph: message signature support Signed-off-by: Yan, Zheng <zyan@redhat.com>
2014-11-04 08:33:37 +00:00
return ceph_auth_check_message_signature(auth, msg);
}
ceph: Storage class should be before const qualifier The C99 specification states in section 6.11.5: The placement of a storage-class specifier other than at the beginning of the declaration specifiers in a declaration is an obsolescent feature. Signed-off-by: Tobias Klauser <tklauser@distanz.ch> Signed-off-by: Sage Weil <sage@newdream.net>
2010-05-20 08:40:19 +00:00
static const struct ceph_connection_operations osd_con_ops = {
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
.get = get_osd_con,
.put = put_osd_con,
.dispatch = dispatch,
ceph: negotiate authentication protocol; implement AUTH_NONE protocol When we open a monitor session, we send an initial AUTH message listing the auth protocols we support, our entity name, and (possibly) a previously assigned global_id. The monitor chooses a protocol and responds with an initial message. Initially implement AUTH_NONE, a dummy protocol that provides no security, but works within the new framework. It generates 'authorizers' that are used when connecting to (mds, osd) services that simply state our entity name and global_id. This is a wire protocol change. Signed-off-by: Sage Weil <sage@newdream.net>
2009-11-19 00:19:57 +00:00
.get_authorizer = get_authorizer,
.verify_authorizer_reply = verify_authorizer_reply,
ceph: allow renewal of auth credentials Add infrastructure to allow the mon_client to periodically renew its auth credentials. Also add a messenger callback that will force such a renewal if a peer rejects our authenticator. Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net> Signed-off-by: Sage Weil <sage@newdream.net>
2010-02-03 00:21:06 +00:00
.invalidate_authorizer = invalidate_authorizer,
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
.alloc_msg = alloc_msg,
libceph: MOSDOp v8 encoding (actual spgid + full hash) Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2017-06-15 14:30:54 +00:00
.reencode_message = osd_reencode_message,
libceph: msg signing callouts don't need con argument We can use msg->con instead - at the point we sign an outgoing message or check the signature on the incoming one, msg->con is always set. We wouldn't know how to sign a message without an associated session (i.e. msg->con == NULL) and being able to sign a message using an explicitly provided authorizer is of no use. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2015-10-26 21:23:56 +00:00
.sign_message = osd_sign_message,
.check_message_signature = osd_check_message_signature,
libceph: a major OSD client update This is a major sync up, up to ~Jewel. The highlights are: - per-session request trees (vs a global per-client tree) - per-session locking (vs a global per-client rwlock) - homeless OSD session - no ad-hoc global per-client lists - support for pool quotas - foundation for watch/notify v2 support - foundation for map check (pool deletion detection) support The switchover is incomplete: lingering requests can be setup and teared down but aren't ever reestablished. This functionality is restored with the introduction of the new lingering infrastructure (ceph_osd_linger_request, linger_work, etc) in a later commit. Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
2016-04-28 14:07:26 +00:00
.fault = osd_fault,
ceph: OSD client The OSD client is responsible for reading and writing data from/to the object storage pool. This includes determining where objects are stored in the cluster, and ensuring that requests are retried or redirected in the event of a node failure or data migration. If an OSD does not respond before a timeout expires, keepalive messages are sent across the lossless, ordered communications channel to ensure that any break in the TCP is discovered. If the session does reset, a reconnection is attempted and affected requests are resent (by the message transport layer). Signed-off-by: Sage Weil <sage@newdream.net>
2009-10-06 18:31:10 +00:00
};
Reference in New Issue Copy Permalink