kernel-ark/drivers/md/dm-raid1.c

1450 lines
34 KiB
C
Raw Normal View History

/*
* Copyright (C) 2003 Sistina Software Limited.
* Copyright (C) 2005-2008 Red Hat, Inc. All rights reserved.
*
* This file is released under the GPL.
*/
#include "dm-bio-record.h"
#include <linux/init.h>
#include <linux/mempool.h>
#include <linux/module.h>
#include <linux/pagemap.h>
#include <linux/slab.h>
#include <linux/workqueue.h>
#include <linux/device-mapper.h>
#include <linux/dm-io.h>
#include <linux/dm-dirty-log.h>
#include <linux/dm-kcopyd.h>
#include <linux/dm-region-hash.h>
#define DM_MSG_PREFIX "raid1"
#define MAX_RECOVERY 1 /* Maximum number of regions recovered in parallel. */
#define DM_RAID1_HANDLE_ERRORS 0x01
#define errors_handled(p) ((p)->features & DM_RAID1_HANDLE_ERRORS)
static DECLARE_WAIT_QUEUE_HEAD(_kmirrord_recovery_stopped);
/*-----------------------------------------------------------------
* Mirror set structures.
*---------------------------------------------------------------*/
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
enum dm_raid1_error {
DM_RAID1_WRITE_ERROR,
DM_RAID1_FLUSH_ERROR,
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
DM_RAID1_SYNC_ERROR,
DM_RAID1_READ_ERROR
};
struct mirror {
struct mirror_set *ms;
atomic_t error_count;
unsigned long error_type;
struct dm_dev *dev;
sector_t offset;
};
struct mirror_set {
struct dm_target *ti;
struct list_head list;
uint64_t features;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
spinlock_t lock; /* protects the lists */
struct bio_list reads;
struct bio_list writes;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
struct bio_list failures;
struct bio_list holds; /* bios are waiting until suspend */
struct dm_region_hash *rh;
struct dm_kcopyd_client *kcopyd_client;
struct dm_io_client *io_client;
/* recovery */
region_t nr_regions;
int in_sync;
int log_failure;
int leg_failure;
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
atomic_t suspend;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
atomic_t default_mirror; /* Default mirror */
struct workqueue_struct *kmirrord_wq;
struct work_struct kmirrord_work;
struct timer_list timer;
unsigned long timer_pending;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
struct work_struct trigger_event;
unsigned nr_mirrors;
struct mirror mirror[0];
};
DECLARE_DM_KCOPYD_THROTTLE_WITH_MODULE_PARM(raid1_resync_throttle,
"A percentage of time allocated for raid resynchronization");
static void wakeup_mirrord(void *context)
{
struct mirror_set *ms = context;
queue_work(ms->kmirrord_wq, &ms->kmirrord_work);
}
static void delayed_wake_fn(unsigned long data)
{
struct mirror_set *ms = (struct mirror_set *) data;
clear_bit(0, &ms->timer_pending);
wakeup_mirrord(ms);
}
static void delayed_wake(struct mirror_set *ms)
{
if (test_and_set_bit(0, &ms->timer_pending))
return;
ms->timer.expires = jiffies + HZ / 5;
ms->timer.data = (unsigned long) ms;
ms->timer.function = delayed_wake_fn;
add_timer(&ms->timer);
}
static void wakeup_all_recovery_waiters(void *context)
{
wake_up_all(&_kmirrord_recovery_stopped);
}
static void queue_bio(struct mirror_set *ms, struct bio *bio, int rw)
{
unsigned long flags;
int should_wake = 0;
struct bio_list *bl;
bl = (rw == WRITE) ? &ms->writes : &ms->reads;
spin_lock_irqsave(&ms->lock, flags);
should_wake = !(bl->head);
bio_list_add(bl, bio);
spin_unlock_irqrestore(&ms->lock, flags);
if (should_wake)
wakeup_mirrord(ms);
}
static void dispatch_bios(void *context, struct bio_list *bio_list)
{
struct mirror_set *ms = context;
struct bio *bio;
while ((bio = bio_list_pop(bio_list)))
queue_bio(ms, bio, WRITE);
}
struct dm_raid1_bio_record {
struct mirror *m;
/* if details->bi_bdev == NULL, details were not saved */
struct dm_bio_details details;
region_t write_region;
};
/*
* Every mirror should look like this one.
*/
#define DEFAULT_MIRROR 0
/*
* This is yucky. We squirrel the mirror struct away inside
* bi_next for read/write buffers. This is safe since the bh
* doesn't get submitted to the lower levels of block layer.
*/
static struct mirror *bio_get_m(struct bio *bio)
{
return (struct mirror *) bio->bi_next;
}
static void bio_set_m(struct bio *bio, struct mirror *m)
{
bio->bi_next = (struct bio *) m;
}
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
static struct mirror *get_default_mirror(struct mirror_set *ms)
{
return &ms->mirror[atomic_read(&ms->default_mirror)];
}
static void set_default_mirror(struct mirror *m)
{
struct mirror_set *ms = m->ms;
struct mirror *m0 = &(ms->mirror[0]);
atomic_set(&ms->default_mirror, m - m0);
}
static struct mirror *get_valid_mirror(struct mirror_set *ms)
{
struct mirror *m;
for (m = ms->mirror; m < ms->mirror + ms->nr_mirrors; m++)
if (!atomic_read(&m->error_count))
return m;
return NULL;
}
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
/* fail_mirror
* @m: mirror device to fail
* @error_type: one of the enum's, DM_RAID1_*_ERROR
*
* If errors are being handled, record the type of
* error encountered for this device. If this type
* of error has already been recorded, we can return;
* otherwise, we must signal userspace by triggering
* an event. Additionally, if the device is the
* primary device, we must choose a new primary, but
* only if the mirror is in-sync.
*
* This function must not block.
*/
static void fail_mirror(struct mirror *m, enum dm_raid1_error error_type)
{
struct mirror_set *ms = m->ms;
struct mirror *new;
ms->leg_failure = 1;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
/*
* error_count is used for nothing more than a
* simple way to tell if a device has encountered
* errors.
*/
atomic_inc(&m->error_count);
if (test_and_set_bit(error_type, &m->error_type))
return;
if (!errors_handled(ms))
return;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
if (m != get_default_mirror(ms))
goto out;
if (!ms->in_sync) {
/*
* Better to issue requests to same failing device
* than to risk returning corrupt data.
*/
DMERR("Primary mirror (%s) failed while out-of-sync: "
"Reads may fail.", m->dev->name);
goto out;
}
new = get_valid_mirror(ms);
if (new)
set_default_mirror(new);
else
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
DMWARN("All sides of mirror have failed.");
out:
schedule_work(&ms->trigger_event);
}
static int mirror_flush(struct dm_target *ti)
{
struct mirror_set *ms = ti->private;
unsigned long error_bits;
unsigned int i;
struct dm_io_region io[ms->nr_mirrors];
struct mirror *m;
struct dm_io_request io_req = {
dm: implement REQ_FLUSH/FUA support for bio-based dm This patch converts bio-based dm to support REQ_FLUSH/FUA instead of now deprecated REQ_HARDBARRIER. * -EOPNOTSUPP handling logic dropped. * Preflush is handled as before but postflush is dropped and replaced with passing down REQ_FUA to member request_queues. This replaces one array wide cache flush w/ member specific FUA writes. * __split_and_process_bio() now calls __clone_and_map_flush() directly for flushes and guarantees all FLUSH bio's going to targets are zero ` length. * It's now guaranteed that all FLUSH bio's which are passed onto dm targets are zero length. bio_empty_barrier() tests are replaced with REQ_FLUSH tests. * Empty WRITE_BARRIERs are replaced with WRITE_FLUSHes. * Dropped unlikely() around REQ_FLUSH tests. Flushes are not unlikely enough to be marked with unlikely(). * Block layer now filters out REQ_FLUSH/FUA bio's if the request_queue doesn't support cache flushing. Advertise REQ_FLUSH | REQ_FUA capability. * Request based dm isn't converted yet. dm_init_request_based_queue() resets flush support to 0 for now. To avoid disturbing request based dm code, dm->flush_error is added for bio based dm while requested based dm continues to use dm->barrier_error. Lightly tested linear, stripe, raid1, snap and crypt targets. Please proceed with caution as I'm not familiar with the code base. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: dm-devel@redhat.com Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-09-03 09:56:19 +00:00
.bi_rw = WRITE_FLUSH,
.mem.type = DM_IO_KMEM,
.mem.ptr.addr = NULL,
.client = ms->io_client,
};
for (i = 0, m = ms->mirror; i < ms->nr_mirrors; i++, m++) {
io[i].bdev = m->dev->bdev;
io[i].sector = 0;
io[i].count = 0;
}
error_bits = -1;
dm_io(&io_req, ms->nr_mirrors, io, &error_bits);
if (unlikely(error_bits != 0)) {
for (i = 0; i < ms->nr_mirrors; i++)
if (test_bit(i, &error_bits))
fail_mirror(ms->mirror + i,
DM_RAID1_FLUSH_ERROR);
return -EIO;
}
return 0;
}
/*-----------------------------------------------------------------
* Recovery.
*
* When a mirror is first activated we may find that some regions
* are in the no-sync state. We have to recover these by
* recopying from the default mirror to all the others.
*---------------------------------------------------------------*/
static void recovery_complete(int read_err, unsigned long write_err,
void *context)
{
struct dm_region *reg = context;
struct mirror_set *ms = dm_rh_region_context(reg);
int m, bit = 0;
if (read_err) {
/* Read error means the failure of default mirror. */
DMERR_LIMIT("Unable to read primary mirror during recovery");
fail_mirror(get_default_mirror(ms), DM_RAID1_SYNC_ERROR);
}
if (write_err) {
DMERR_LIMIT("Write error during recovery (error = 0x%lx)",
write_err);
/*
* Bits correspond to devices (excluding default mirror).
* The default mirror cannot change during recovery.
*/
for (m = 0; m < ms->nr_mirrors; m++) {
if (&ms->mirror[m] == get_default_mirror(ms))
continue;
if (test_bit(bit, &write_err))
fail_mirror(ms->mirror + m,
DM_RAID1_SYNC_ERROR);
bit++;
}
}
dm_rh_recovery_end(reg, !(read_err || write_err));
}
static int recover(struct mirror_set *ms, struct dm_region *reg)
{
int r;
unsigned i;
struct dm_io_region from, to[DM_KCOPYD_MAX_REGIONS], *dest;
struct mirror *m;
unsigned long flags = 0;
region_t key = dm_rh_get_region_key(reg);
sector_t region_size = dm_rh_get_region_size(ms->rh);
/* fill in the source */
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
m = get_default_mirror(ms);
from.bdev = m->dev->bdev;
from.sector = m->offset + dm_rh_region_to_sector(ms->rh, key);
if (key == (ms->nr_regions - 1)) {
/*
* The final region may be smaller than
* region_size.
*/
from.count = ms->ti->len & (region_size - 1);
if (!from.count)
from.count = region_size;
} else
from.count = region_size;
/* fill in the destinations */
for (i = 0, dest = to; i < ms->nr_mirrors; i++) {
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
if (&ms->mirror[i] == get_default_mirror(ms))
continue;
m = ms->mirror + i;
dest->bdev = m->dev->bdev;
dest->sector = m->offset + dm_rh_region_to_sector(ms->rh, key);
dest->count = from.count;
dest++;
}
/* hand to kcopyd */
if (!errors_handled(ms))
set_bit(DM_KCOPYD_IGNORE_ERROR, &flags);
r = dm_kcopyd_copy(ms->kcopyd_client, &from, ms->nr_mirrors - 1, to,
flags, recovery_complete, reg);
return r;
}
static void do_recovery(struct mirror_set *ms)
{
struct dm_region *reg;
struct dm_dirty_log *log = dm_rh_dirty_log(ms->rh);
int r;
/*
* Start quiescing some regions.
*/
dm_rh_recovery_prepare(ms->rh);
/*
* Copy any already quiesced regions.
*/
while ((reg = dm_rh_recovery_start(ms->rh))) {
r = recover(ms, reg);
if (r)
dm_rh_recovery_end(reg, 0);
}
/*
* Update the in sync flag.
*/
if (!ms->in_sync &&
(log->type->get_sync_count(log) == ms->nr_regions)) {
/* the sync is complete */
dm_table_event(ms->ti->table);
ms->in_sync = 1;
}
}
/*-----------------------------------------------------------------
* Reads
*---------------------------------------------------------------*/
static struct mirror *choose_mirror(struct mirror_set *ms, sector_t sector)
{
struct mirror *m = get_default_mirror(ms);
do {
if (likely(!atomic_read(&m->error_count)))
return m;
if (m-- == ms->mirror)
m += ms->nr_mirrors;
} while (m != get_default_mirror(ms));
return NULL;
}
static int default_ok(struct mirror *m)
{
struct mirror *default_mirror = get_default_mirror(m->ms);
return !atomic_read(&default_mirror->error_count);
}
static int mirror_available(struct mirror_set *ms, struct bio *bio)
{
struct dm_dirty_log *log = dm_rh_dirty_log(ms->rh);
region_t region = dm_rh_bio_to_region(ms->rh, bio);
if (log->type->in_sync(log, region, 0))
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-11 22:44:27 +00:00
return choose_mirror(ms, bio->bi_iter.bi_sector) ? 1 : 0;
return 0;
}
/*
* remap a buffer to a particular mirror.
*/
static sector_t map_sector(struct mirror *m, struct bio *bio)
{
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-11 22:44:27 +00:00
if (unlikely(!bio->bi_iter.bi_size))
return 0;
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-11 22:44:27 +00:00
return m->offset + dm_target_offset(m->ms->ti, bio->bi_iter.bi_sector);
}
static void map_bio(struct mirror *m, struct bio *bio)
{
bio->bi_bdev = m->dev->bdev;
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-11 22:44:27 +00:00
bio->bi_iter.bi_sector = map_sector(m, bio);
}
static void map_region(struct dm_io_region *io, struct mirror *m,
struct bio *bio)
{
io->bdev = m->dev->bdev;
io->sector = map_sector(m, bio);
io->count = bio_sectors(bio);
}
static void hold_bio(struct mirror_set *ms, struct bio *bio)
{
/*
* Lock is required to avoid race condition during suspend
* process.
*/
spin_lock_irq(&ms->lock);
if (atomic_read(&ms->suspend)) {
spin_unlock_irq(&ms->lock);
/*
* If device is suspended, complete the bio.
*/
if (dm_noflush_suspending(ms->ti))
bio_endio(bio, DM_ENDIO_REQUEUE);
else
bio_endio(bio, -EIO);
return;
}
/*
* Hold bio until the suspend is complete.
*/
bio_list_add(&ms->holds, bio);
spin_unlock_irq(&ms->lock);
}
/*-----------------------------------------------------------------
* Reads
*---------------------------------------------------------------*/
static void read_callback(unsigned long error, void *context)
{
struct bio *bio = context;
struct mirror *m;
m = bio_get_m(bio);
bio_set_m(bio, NULL);
if (likely(!error)) {
bio_endio(bio, 0);
return;
}
fail_mirror(m, DM_RAID1_READ_ERROR);
if (likely(default_ok(m)) || mirror_available(m->ms, bio)) {
DMWARN_LIMIT("Read failure on mirror device %s. "
"Trying alternative device.",
m->dev->name);
queue_bio(m->ms, bio, bio_rw(bio));
return;
}
DMERR_LIMIT("Read failure on mirror device %s. Failing I/O.",
m->dev->name);
bio_endio(bio, -EIO);
}
/* Asynchronous read. */
static void read_async_bio(struct mirror *m, struct bio *bio)
{
struct dm_io_region io;
struct dm_io_request io_req = {
.bi_rw = READ,
.mem.type = DM_IO_BIO,
.mem.ptr.bio = bio,
.notify.fn = read_callback,
.notify.context = bio,
.client = m->ms->io_client,
};
map_region(&io, m, bio);
bio_set_m(bio, m);
BUG_ON(dm_io(&io_req, 1, &io, NULL));
}
static inline int region_in_sync(struct mirror_set *ms, region_t region,
int may_block)
{
int state = dm_rh_get_state(ms->rh, region, may_block);
return state == DM_RH_CLEAN || state == DM_RH_DIRTY;
}
static void do_reads(struct mirror_set *ms, struct bio_list *reads)
{
region_t region;
struct bio *bio;
struct mirror *m;
while ((bio = bio_list_pop(reads))) {
region = dm_rh_bio_to_region(ms->rh, bio);
m = get_default_mirror(ms);
/*
* We can only read balance if the region is in sync.
*/
if (likely(region_in_sync(ms, region, 1)))
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-11 22:44:27 +00:00
m = choose_mirror(ms, bio->bi_iter.bi_sector);
else if (m && atomic_read(&m->error_count))
m = NULL;
if (likely(m))
read_async_bio(m, bio);
else
bio_endio(bio, -EIO);
}
}
/*-----------------------------------------------------------------
* Writes.
*
* We do different things with the write io depending on the
* state of the region that it's in:
*
* SYNC: increment pending, use kcopyd to write to *all* mirrors
* RECOVERING: delay the io until recovery completes
* NOSYNC: increment pending, just write to the default mirror
*---------------------------------------------------------------*/
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
static void write_callback(unsigned long error, void *context)
{
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
unsigned i, ret = 0;
struct bio *bio = (struct bio *) context;
struct mirror_set *ms;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
int should_wake = 0;
unsigned long flags;
ms = bio_get_m(bio)->ms;
bio_set_m(bio, NULL);
/*
* NOTE: We don't decrement the pending count here,
* instead it is done by the targets endio function.
* This way we handle both writes to SYNC and NOSYNC
* regions with the same code.
*/
if (likely(!error)) {
bio_endio(bio, ret);
return;
}
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
for (i = 0; i < ms->nr_mirrors; i++)
if (test_bit(i, &error))
fail_mirror(ms->mirror + i, DM_RAID1_WRITE_ERROR);
/*
* Need to raise event. Since raising
* events can block, we need to do it in
* the main thread.
*/
spin_lock_irqsave(&ms->lock, flags);
if (!ms->failures.head)
should_wake = 1;
bio_list_add(&ms->failures, bio);
spin_unlock_irqrestore(&ms->lock, flags);
if (should_wake)
wakeup_mirrord(ms);
}
static void do_write(struct mirror_set *ms, struct bio *bio)
{
unsigned int i;
struct dm_io_region io[ms->nr_mirrors], *dest = io;
struct mirror *m;
struct dm_io_request io_req = {
dm: implement REQ_FLUSH/FUA support for bio-based dm This patch converts bio-based dm to support REQ_FLUSH/FUA instead of now deprecated REQ_HARDBARRIER. * -EOPNOTSUPP handling logic dropped. * Preflush is handled as before but postflush is dropped and replaced with passing down REQ_FUA to member request_queues. This replaces one array wide cache flush w/ member specific FUA writes. * __split_and_process_bio() now calls __clone_and_map_flush() directly for flushes and guarantees all FLUSH bio's going to targets are zero ` length. * It's now guaranteed that all FLUSH bio's which are passed onto dm targets are zero length. bio_empty_barrier() tests are replaced with REQ_FLUSH tests. * Empty WRITE_BARRIERs are replaced with WRITE_FLUSHes. * Dropped unlikely() around REQ_FLUSH tests. Flushes are not unlikely enough to be marked with unlikely(). * Block layer now filters out REQ_FLUSH/FUA bio's if the request_queue doesn't support cache flushing. Advertise REQ_FLUSH | REQ_FUA capability. * Request based dm isn't converted yet. dm_init_request_based_queue() resets flush support to 0 for now. To avoid disturbing request based dm code, dm->flush_error is added for bio based dm while requested based dm continues to use dm->barrier_error. Lightly tested linear, stripe, raid1, snap and crypt targets. Please proceed with caution as I'm not familiar with the code base. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: dm-devel@redhat.com Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
2010-09-03 09:56:19 +00:00
.bi_rw = WRITE | (bio->bi_rw & WRITE_FLUSH_FUA),
.mem.type = DM_IO_BIO,
.mem.ptr.bio = bio,
.notify.fn = write_callback,
.notify.context = bio,
.client = ms->io_client,
};
if (bio->bi_rw & REQ_DISCARD) {
io_req.bi_rw |= REQ_DISCARD;
io_req.mem.type = DM_IO_KMEM;
io_req.mem.ptr.addr = NULL;
}
for (i = 0, m = ms->mirror; i < ms->nr_mirrors; i++, m++)
map_region(dest++, m, bio);
/*
* Use default mirror because we only need it to retrieve the reference
* to the mirror set in write_callback().
*/
bio_set_m(bio, get_default_mirror(ms));
BUG_ON(dm_io(&io_req, ms->nr_mirrors, io, NULL));
}
static void do_writes(struct mirror_set *ms, struct bio_list *writes)
{
int state;
struct bio *bio;
struct bio_list sync, nosync, recover, *this_list = NULL;
struct bio_list requeue;
struct dm_dirty_log *log = dm_rh_dirty_log(ms->rh);
region_t region;
if (!writes->head)
return;
/*
* Classify each write.
*/
bio_list_init(&sync);
bio_list_init(&nosync);
bio_list_init(&recover);
bio_list_init(&requeue);
while ((bio = bio_list_pop(writes))) {
if ((bio->bi_rw & REQ_FLUSH) ||
(bio->bi_rw & REQ_DISCARD)) {
bio_list_add(&sync, bio);
continue;
}
region = dm_rh_bio_to_region(ms->rh, bio);
if (log->type->is_remote_recovering &&
log->type->is_remote_recovering(log, region)) {
bio_list_add(&requeue, bio);
continue;
}
state = dm_rh_get_state(ms->rh, region, 1);
switch (state) {
case DM_RH_CLEAN:
case DM_RH_DIRTY:
this_list = &sync;
break;
case DM_RH_NOSYNC:
this_list = &nosync;
break;
case DM_RH_RECOVERING:
this_list = &recover;
break;
}
bio_list_add(this_list, bio);
}
/*
* Add bios that are delayed due to remote recovery
* back on to the write queue
*/
if (unlikely(requeue.head)) {
spin_lock_irq(&ms->lock);
bio_list_merge(&ms->writes, &requeue);
spin_unlock_irq(&ms->lock);
delayed_wake(ms);
}
/*
* Increment the pending counts for any regions that will
* be written to (writes to recover regions are going to
* be delayed).
*/
dm_rh_inc_pending(ms->rh, &sync);
dm_rh_inc_pending(ms->rh, &nosync);
dm raid1: do not allow log_failure variable to unset after being set This patch fixes a bug which was triggering a case where the primary leg could not be changed on failure even when the mirror was in-sync. The case involves the failure of the primary device along with the transient failure of the log device. The problem is that bios can be put on the 'failures' list (due to log failure) before 'fail_mirror' is called due to the primary device failure. Normally, this is fine, but if the log device failure is transient, a subsequent iteration of the work thread, 'do_mirror', will reset 'log_failure'. The 'do_failures' function then resets the 'in_sync' variable when processing bios on the failures list. The 'in_sync' variable is what is used to determine if the primary device can be switched in the event of a failure. Since this has been reset, the primary device is incorrectly assumed to be not switchable. The case has been seen in the cluster mirror context, where one machine realizes the log device is dead before the other machines. As the responsibilities of the server migrate from one node to another (because the mirror is being reconfigured due to the failure), the new server may think for a moment that the log device is fine - thus resetting the 'log_failure' variable. In any case, it is inappropiate for us to reset the 'log_failure' variable. The above bug simply illustrates that it can actually hurt us. Cc: stable@kernel.org Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2009-09-04 19:40:32 +00:00
/*
* If the flush fails on a previous call and succeeds here,
* we must not reset the log_failure variable. We need
* userspace interaction to do that.
*/
ms->log_failure = dm_rh_flush(ms->rh) ? 1 : ms->log_failure;
/*
* Dispatch io.
*/
if (unlikely(ms->log_failure) && errors_handled(ms)) {
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
spin_lock_irq(&ms->lock);
bio_list_merge(&ms->failures, &sync);
spin_unlock_irq(&ms->lock);
wakeup_mirrord(ms);
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
} else
while ((bio = bio_list_pop(&sync)))
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
do_write(ms, bio);
while ((bio = bio_list_pop(&recover)))
dm_rh_delay(ms->rh, bio);
while ((bio = bio_list_pop(&nosync))) {
if (unlikely(ms->leg_failure) && errors_handled(ms)) {
spin_lock_irq(&ms->lock);
bio_list_add(&ms->failures, bio);
spin_unlock_irq(&ms->lock);
wakeup_mirrord(ms);
} else {
map_bio(get_default_mirror(ms), bio);
generic_make_request(bio);
}
}
}
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
static void do_failures(struct mirror_set *ms, struct bio_list *failures)
{
struct bio *bio;
if (likely(!failures->head))
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
return;
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
/*
* If the log has failed, unattempted writes are being
* put on the holds list. We can't issue those writes
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
* until a log has been marked, so we must store them.
*
* If a 'noflush' suspend is in progress, we can requeue
* the I/O's to the core. This give userspace a chance
* to reconfigure the mirror, at which point the core
* will reissue the writes. If the 'noflush' flag is
* not set, we have no choice but to return errors.
*
* Some writes on the failures list may have been
* submitted before the log failure and represent a
* failure to write to one of the devices. It is ok
* for us to treat them the same and requeue them
* as well.
*/
while ((bio = bio_list_pop(failures))) {
if (!ms->log_failure) {
ms->in_sync = 0;
dm_rh_mark_nosync(ms->rh, bio);
}
/*
* If all the legs are dead, fail the I/O.
* If we have been told to handle errors, hold the bio
* and wait for userspace to deal with the problem.
* Otherwise pretend that the I/O succeeded. (This would
* be wrong if the failed leg returned after reboot and
* got replicated back to the good legs.)
*/
if (!get_valid_mirror(ms))
bio_endio(bio, -EIO);
else if (errors_handled(ms))
hold_bio(ms, bio);
else
bio_endio(bio, 0);
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
}
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
}
static void trigger_event(struct work_struct *work)
{
struct mirror_set *ms =
container_of(work, struct mirror_set, trigger_event);
dm_table_event(ms->ti->table);
}
/*-----------------------------------------------------------------
* kmirrord
*---------------------------------------------------------------*/
static void do_mirror(struct work_struct *work)
{
struct mirror_set *ms = container_of(work, struct mirror_set,
kmirrord_work);
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
struct bio_list reads, writes, failures;
unsigned long flags;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
spin_lock_irqsave(&ms->lock, flags);
reads = ms->reads;
writes = ms->writes;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
failures = ms->failures;
bio_list_init(&ms->reads);
bio_list_init(&ms->writes);
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
bio_list_init(&ms->failures);
spin_unlock_irqrestore(&ms->lock, flags);
dm_rh_update_states(ms->rh, errors_handled(ms));
do_recovery(ms);
do_reads(ms, &reads);
do_writes(ms, &writes);
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
do_failures(ms, &failures);
}
/*-----------------------------------------------------------------
* Target functions
*---------------------------------------------------------------*/
static struct mirror_set *alloc_context(unsigned int nr_mirrors,
uint32_t region_size,
struct dm_target *ti,
struct dm_dirty_log *dl)
{
size_t len;
struct mirror_set *ms = NULL;
len = sizeof(*ms) + (sizeof(ms->mirror[0]) * nr_mirrors);
2007-07-19 08:49:03 +00:00
ms = kzalloc(len, GFP_KERNEL);
if (!ms) {
ti->error = "Cannot allocate mirror context";
return NULL;
}
spin_lock_init(&ms->lock);
bio_list_init(&ms->reads);
bio_list_init(&ms->writes);
bio_list_init(&ms->failures);
bio_list_init(&ms->holds);
ms->ti = ti;
ms->nr_mirrors = nr_mirrors;
ms->nr_regions = dm_sector_div_up(ti->len, region_size);
ms->in_sync = 0;
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
ms->log_failure = 0;
ms->leg_failure = 0;
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
atomic_set(&ms->suspend, 0);
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
atomic_set(&ms->default_mirror, DEFAULT_MIRROR);
ms->io_client = dm_io_client_create();
if (IS_ERR(ms->io_client)) {
ti->error = "Error creating dm_io client";
kfree(ms);
return NULL;
}
ms->rh = dm_region_hash_create(ms, dispatch_bios, wakeup_mirrord,
wakeup_all_recovery_waiters,
ms->ti->begin, MAX_RECOVERY,
dl, region_size, ms->nr_regions);
if (IS_ERR(ms->rh)) {
ti->error = "Error creating dirty region hash";
dm_io_client_destroy(ms->io_client);
kfree(ms);
return NULL;
}
return ms;
}
static void free_context(struct mirror_set *ms, struct dm_target *ti,
unsigned int m)
{
while (m--)
dm_put_device(ti, ms->mirror[m].dev);
dm_io_client_destroy(ms->io_client);
dm_region_hash_destroy(ms->rh);
kfree(ms);
}
static int get_mirror(struct mirror_set *ms, struct dm_target *ti,
unsigned int mirror, char **argv)
{
unsigned long long offset;
char dummy;
if (sscanf(argv[1], "%llu%c", &offset, &dummy) != 1) {
ti->error = "Invalid offset";
return -EINVAL;
}
if (dm_get_device(ti, argv[0], dm_table_get_mode(ti->table),
&ms->mirror[mirror].dev)) {
ti->error = "Device lookup failure";
return -ENXIO;
}
ms->mirror[mirror].ms = ms;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
atomic_set(&(ms->mirror[mirror].error_count), 0);
ms->mirror[mirror].error_type = 0;
ms->mirror[mirror].offset = offset;
return 0;
}
/*
* Create dirty log: log_type #log_params <log_params>
*/
static struct dm_dirty_log *create_dirty_log(struct dm_target *ti,
unsigned argc, char **argv,
unsigned *args_used)
{
unsigned param_count;
struct dm_dirty_log *dl;
char dummy;
if (argc < 2) {
ti->error = "Insufficient mirror log arguments";
return NULL;
}
if (sscanf(argv[1], "%u%c", &param_count, &dummy) != 1) {
ti->error = "Invalid mirror log argument count";
return NULL;
}
*args_used = 2 + param_count;
if (argc < *args_used) {
ti->error = "Insufficient mirror log arguments";
return NULL;
}
dl = dm_dirty_log_create(argv[0], ti, mirror_flush, param_count,
argv + 2);
if (!dl) {
ti->error = "Error creating mirror dirty log";
return NULL;
}
return dl;
}
static int parse_features(struct mirror_set *ms, unsigned argc, char **argv,
unsigned *args_used)
{
unsigned num_features;
struct dm_target *ti = ms->ti;
char dummy;
*args_used = 0;
if (!argc)
return 0;
if (sscanf(argv[0], "%u%c", &num_features, &dummy) != 1) {
ti->error = "Invalid number of features";
return -EINVAL;
}
argc--;
argv++;
(*args_used)++;
if (num_features > argc) {
ti->error = "Not enough arguments to support feature count";
return -EINVAL;
}
if (!strcmp("handle_errors", argv[0]))
ms->features |= DM_RAID1_HANDLE_ERRORS;
else {
ti->error = "Unrecognised feature requested";
return -EINVAL;
}
(*args_used)++;
return 0;
}
/*
* Construct a mirror mapping:
*
* log_type #log_params <log_params>
* #mirrors [mirror_path offset]{2,}
* [#features <features>]
*
* log_type is "core" or "disk"
* #log_params is between 1 and 3
*
* If present, features must be "handle_errors".
*/
static int mirror_ctr(struct dm_target *ti, unsigned int argc, char **argv)
{
int r;
unsigned int nr_mirrors, m, args_used;
struct mirror_set *ms;
struct dm_dirty_log *dl;
char dummy;
dl = create_dirty_log(ti, argc, argv, &args_used);
if (!dl)
return -EINVAL;
argv += args_used;
argc -= args_used;
if (!argc || sscanf(argv[0], "%u%c", &nr_mirrors, &dummy) != 1 ||
nr_mirrors < 2 || nr_mirrors > DM_KCOPYD_MAX_REGIONS + 1) {
ti->error = "Invalid number of mirrors";
dm_dirty_log_destroy(dl);
return -EINVAL;
}
argv++, argc--;
if (argc < nr_mirrors * 2) {
ti->error = "Too few mirror arguments";
dm_dirty_log_destroy(dl);
return -EINVAL;
}
ms = alloc_context(nr_mirrors, dl->type->get_region_size(dl), ti, dl);
if (!ms) {
dm_dirty_log_destroy(dl);
return -ENOMEM;
}
/* Get the mirror parameter sets */
for (m = 0; m < nr_mirrors; m++) {
r = get_mirror(ms, ti, m, argv);
if (r) {
free_context(ms, ti, m);
return r;
}
argv += 2;
argc -= 2;
}
ti->private = ms;
r = dm_set_target_max_io_len(ti, dm_rh_get_region_size(ms->rh));
if (r)
goto err_free_context;
ti->num_flush_bios = 1;
ti->num_discard_bios = 1;
ti->per_bio_data_size = sizeof(struct dm_raid1_bio_record);
ti->discard_zeroes_data_unsupported = true;
ms->kmirrord_wq = alloc_workqueue("kmirrord", WQ_MEM_RECLAIM, 0);
if (!ms->kmirrord_wq) {
DMERR("couldn't start kmirrord");
r = -ENOMEM;
goto err_free_context;
}
INIT_WORK(&ms->kmirrord_work, do_mirror);
init_timer(&ms->timer);
ms->timer_pending = 0;
dm raid1: handle write failures This patch gives mirror the ability to handle device failures during normal write operations. The 'write_callback' function is called when a write completes. If all the writes failed or succeeded, we report failure or success respectively. If some of the writes failed, we call fail_mirror; which increments the error count for the device, notes the type of error encountered (DM_RAID1_WRITE_ERROR), and selects a new primary (if necessary). Note that the primary device can never change while the mirror is not in-sync (IOW, while recovery is happening.) This means that the scenario where a failed write changes the primary and gives recovery_complete a chance to misread the primary never happens. The fact that the primary can change has necessitated the change to the default_mirror field. We need to protect against reading garbage while the primary changes. We then add the bio to a new list in the mirror set, 'failures'. For every bio in the 'failures' list, we call a new function, '__bio_mark_nosync', where we mark the region 'not-in-sync' in the log and properly set the region state as, RH_NOSYNC. Userspace must also be notified of the failure. This is done by 'raising an event' (dm_table_event()). If fail_mirror is called in process context the event can be raised right away. If in interrupt context, the event is deferred to the kmirrord thread - which raises the event if 'event_waiting' is set. Backwards compatibility is maintained by ignoring errors if the DM_FEATURES_HANDLE_ERRORS flag is not present. Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:29 +00:00
INIT_WORK(&ms->trigger_event, trigger_event);
r = parse_features(ms, argc, argv, &args_used);
if (r)
goto err_destroy_wq;
argv += args_used;
argc -= args_used;
/*
* Any read-balancing addition depends on the
* DM_RAID1_HANDLE_ERRORS flag being present.
* This is because the decision to balance depends
* on the sync state of a region. If the above
* flag is not present, we ignore errors; and
* the sync state may be inaccurate.
*/
if (argc) {
ti->error = "Too many mirror arguments";
r = -EINVAL;
goto err_destroy_wq;
}
ms->kcopyd_client = dm_kcopyd_client_create(&dm_kcopyd_throttle);
if (IS_ERR(ms->kcopyd_client)) {
r = PTR_ERR(ms->kcopyd_client);
goto err_destroy_wq;
}
wakeup_mirrord(ms);
return 0;
err_destroy_wq:
destroy_workqueue(ms->kmirrord_wq);
err_free_context:
free_context(ms, ti, ms->nr_mirrors);
return r;
}
static void mirror_dtr(struct dm_target *ti)
{
struct mirror_set *ms = (struct mirror_set *) ti->private;
del_timer_sync(&ms->timer);
flush_workqueue(ms->kmirrord_wq);
workqueue: deprecate flush[_delayed]_work_sync() flush[_delayed]_work_sync() are now spurious. Mark them deprecated and convert all users to flush[_delayed]_work(). If you're cc'd and wondering what's going on: Now all workqueues are non-reentrant and the regular flushes guarantee that the work item is not pending or running on any CPU on return, so there's no reason to use the sync flushes at all and they're going away. This patch doesn't make any functional difference. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Russell King <linux@arm.linux.org.uk> Cc: Paul Mundt <lethal@linux-sh.org> Cc: Ian Campbell <ian.campbell@citrix.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Mattia Dongili <malattia@linux.it> Cc: Kent Yoder <key@linux.vnet.ibm.com> Cc: David Airlie <airlied@linux.ie> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Karsten Keil <isdn@linux-pingi.de> Cc: Bryan Wu <bryan.wu@canonical.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mauro Carvalho Chehab <mchehab@infradead.org> Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de> Cc: David Woodhouse <dwmw2@infradead.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: linux-wireless@vger.kernel.org Cc: Anton Vorontsov <cbou@mail.ru> Cc: Sangbeom Kim <sbkim73@samsung.com> Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Eric Van Hensbergen <ericvh@gmail.com> Cc: Takashi Iwai <tiwai@suse.de> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Petr Vandrovec <petr@vandrovec.name> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Avi Kivity <avi@redhat.com>
2012-08-20 21:51:24 +00:00
flush_work(&ms->trigger_event);
dm_kcopyd_client_destroy(ms->kcopyd_client);
destroy_workqueue(ms->kmirrord_wq);
free_context(ms, ti, ms->nr_mirrors);
}
/*
* Mirror mapping function
*/
static int mirror_map(struct dm_target *ti, struct bio *bio)
{
int r, rw = bio_rw(bio);
struct mirror *m;
struct mirror_set *ms = ti->private;
struct dm_dirty_log *log = dm_rh_dirty_log(ms->rh);
struct dm_raid1_bio_record *bio_record =
dm_per_bio_data(bio, sizeof(struct dm_raid1_bio_record));
bio_record->details.bi_bdev = NULL;
if (rw == WRITE) {
/* Save region for mirror_end_io() handler */
bio_record->write_region = dm_rh_bio_to_region(ms->rh, bio);
queue_bio(ms, bio, rw);
return DM_MAPIO_SUBMITTED;
}
r = log->type->in_sync(log, dm_rh_bio_to_region(ms->rh, bio), 0);
if (r < 0 && r != -EWOULDBLOCK)
return r;
/*
* If region is not in-sync queue the bio.
*/
if (!r || (r == -EWOULDBLOCK)) {
if (rw == READA)
return -EWOULDBLOCK;
queue_bio(ms, bio, rw);
return DM_MAPIO_SUBMITTED;
}
/*
* The region is in-sync and we can perform reads directly.
* Store enough information so we can retry if it fails.
*/
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-11 22:44:27 +00:00
m = choose_mirror(ms, bio->bi_iter.bi_sector);
if (unlikely(!m))
return -EIO;
dm_bio_record(&bio_record->details, bio);
bio_record->m = m;
map_bio(m, bio);
return DM_MAPIO_REMAPPED;
}
static int mirror_end_io(struct dm_target *ti, struct bio *bio, int error)
{
int rw = bio_rw(bio);
struct mirror_set *ms = (struct mirror_set *) ti->private;
struct mirror *m = NULL;
struct dm_bio_details *bd = NULL;
struct dm_raid1_bio_record *bio_record =
dm_per_bio_data(bio, sizeof(struct dm_raid1_bio_record));
/*
* We need to dec pending if this was a write.
*/
if (rw == WRITE) {
dm raid1: fix crash with mirror recovery and discard This patch fixes a crash when a discard request is sent during mirror recovery. Firstly, some background. Generally, the following sequence happens during mirror synchronization: - function do_recovery is called - do_recovery calls dm_rh_recovery_prepare - dm_rh_recovery_prepare uses a semaphore to limit the number simultaneously recovered regions (by default the semaphore value is 1, so only one region at a time is recovered) - dm_rh_recovery_prepare calls __rh_recovery_prepare, __rh_recovery_prepare asks the log driver for the next region to recover. Then, it sets the region state to DM_RH_RECOVERING. If there are no pending I/Os on this region, the region is added to quiesced_regions list. If there are pending I/Os, the region is not added to any list. It is added to the quiesced_regions list later (by dm_rh_dec function) when all I/Os finish. - when the region is on quiesced_regions list, there are no I/Os in flight on this region. The region is popped from the list in dm_rh_recovery_start function. Then, a kcopyd job is started in the recover function. - when the kcopyd job finishes, recovery_complete is called. It calls dm_rh_recovery_end. dm_rh_recovery_end adds the region to recovered_regions or failed_recovered_regions list (depending on whether the copy operation was successful or not). The above mechanism assumes that if the region is in DM_RH_RECOVERING state, no new I/Os are started on this region. When I/O is started, dm_rh_inc_pending is called, which increases reg->pending count. When I/O is finished, dm_rh_dec is called. It decreases reg->pending count. If the count is zero and the region was in DM_RH_RECOVERING state, dm_rh_dec adds it to the quiesced_regions list. Consequently, if we call dm_rh_inc_pending/dm_rh_dec while the region is in DM_RH_RECOVERING state, it could be added to quiesced_regions list multiple times or it could be added to this list when kcopyd is copying data (it is assumed that the region is not on any list while kcopyd does its jobs). This results in memory corruption and crash. There already exist bypasses for REQ_FLUSH requests: REQ_FLUSH requests do not belong to any region, so they are always added to the sync list in do_writes. dm_rh_inc_pending does not increase count for REQ_FLUSH requests. In mirror_end_io, dm_rh_dec is never called for REQ_FLUSH requests. These bypasses avoid the crash possibility described above. These bypasses were improperly implemented for REQ_DISCARD when the mirror target gained discard support in commit 5fc2ffeabb9ee0fc0e71ff16b49f34f0ed3d05b4 (dm raid1: support discard). In do_writes, REQ_DISCARD requests is always added to the sync queue and immediately dispatched (even if the region is in DM_RH_RECOVERING). However, dm_rh_inc and dm_rh_dec is called for REQ_DISCARD resusts. So it violates the rule that no I/Os are started on DM_RH_RECOVERING regions, and causes the list corruption described above. This patch changes it so that REQ_DISCARD requests follow the same path as REQ_FLUSH. This avoids the crash. Reference: https://bugzilla.redhat.com/837607 Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Cc: stable@kernel.org Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2012-07-20 13:25:03 +00:00
if (!(bio->bi_rw & (REQ_FLUSH | REQ_DISCARD)))
dm_rh_dec(ms->rh, bio_record->write_region);
return error;
}
if (error == -EOPNOTSUPP)
goto out;
if ((error == -EWOULDBLOCK) && (bio->bi_rw & REQ_RAHEAD))
goto out;
if (unlikely(error)) {
if (!bio_record->details.bi_bdev) {
/*
* There wasn't enough memory to record necessary
* information for a retry or there was no other
* mirror in-sync.
*/
DMERR_LIMIT("Mirror read failed.");
return -EIO;
}
m = bio_record->m;
DMERR("Mirror read failed from %s. Trying alternative device.",
m->dev->name);
fail_mirror(m, DM_RAID1_READ_ERROR);
/*
* A failed read is requeued for another attempt using an intact
* mirror.
*/
if (default_ok(m) || mirror_available(ms, bio)) {
bd = &bio_record->details;
dm_bio_restore(bd, bio);
bio_record->details.bi_bdev = NULL;
atomic_inc(&bio->bi_remaining);
queue_bio(ms, bio, rw);
return DM_ENDIO_INCOMPLETE;
}
DMERR("All replicated volumes dead, failing I/O");
}
out:
bio_record->details.bi_bdev = NULL;
return error;
}
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
static void mirror_presuspend(struct dm_target *ti)
{
struct mirror_set *ms = (struct mirror_set *) ti->private;
struct dm_dirty_log *log = dm_rh_dirty_log(ms->rh);
struct bio_list holds;
struct bio *bio;
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
atomic_set(&ms->suspend, 1);
/*
* Process bios in the hold list to start recovery waiting
* for bios in the hold list. After the process, no bio has
* a chance to be added in the hold list because ms->suspend
* is set.
*/
spin_lock_irq(&ms->lock);
holds = ms->holds;
bio_list_init(&ms->holds);
spin_unlock_irq(&ms->lock);
while ((bio = bio_list_pop(&holds)))
hold_bio(ms, bio);
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
/*
* We must finish up all the work that we've
* generated (i.e. recovery work).
*/
dm_rh_stop_recovery(ms->rh);
wait_event(_kmirrord_recovery_stopped,
!dm_rh_recovery_in_flight(ms->rh));
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
if (log->type->presuspend && log->type->presuspend(log))
/* FIXME: need better error handling */
DMWARN("log presuspend failed");
/*
* Now that recovery is complete/stopped and the
* delayed bios are queued, we need to wait for
* the worker thread to complete. This way,
* we know that all of our I/O has been pushed.
*/
flush_workqueue(ms->kmirrord_wq);
}
static void mirror_postsuspend(struct dm_target *ti)
{
struct mirror_set *ms = ti->private;
struct dm_dirty_log *log = dm_rh_dirty_log(ms->rh);
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
if (log->type->postsuspend && log->type->postsuspend(log))
/* FIXME: need better error handling */
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
DMWARN("log postsuspend failed");
}
static void mirror_resume(struct dm_target *ti)
{
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
struct mirror_set *ms = ti->private;
struct dm_dirty_log *log = dm_rh_dirty_log(ms->rh);
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
atomic_set(&ms->suspend, 0);
if (log->type->resume && log->type->resume(log))
/* FIXME: need better error handling */
DMWARN("log resume failed");
dm_rh_start_recovery(ms->rh);
}
/*
* device_status_char
* @m: mirror device/leg we want the status of
*
* We return one character representing the most severe error
* we have encountered.
* A => Alive - No failures
* D => Dead - A write failure occurred leaving mirror out-of-sync
* S => Sync - A sychronization failure occurred, mirror out-of-sync
* R => Read - A read failure occurred, mirror data unaffected
*
* Returns: <char>
*/
static char device_status_char(struct mirror *m)
{
if (!atomic_read(&(m->error_count)))
return 'A';
return (test_bit(DM_RAID1_FLUSH_ERROR, &(m->error_type))) ? 'F' :
(test_bit(DM_RAID1_WRITE_ERROR, &(m->error_type))) ? 'D' :
(test_bit(DM_RAID1_SYNC_ERROR, &(m->error_type))) ? 'S' :
(test_bit(DM_RAID1_READ_ERROR, &(m->error_type))) ? 'R' : 'U';
}
dm: fix truncated status strings Avoid returning a truncated table or status string instead of setting the DM_BUFFER_FULL_FLAG when the last target of a table fills the buffer. When processing a table or status request, the function retrieve_status calls ti->type->status. If ti->type->status returns non-zero, retrieve_status assumes that the buffer overflowed and sets DM_BUFFER_FULL_FLAG. However, targets don't return non-zero values from their status method on overflow. Most targets returns always zero. If a buffer overflow happens in a target that is not the last in the table, it gets noticed during the next iteration of the loop in retrieve_status; but if a buffer overflow happens in the last target, it goes unnoticed and erroneously truncated data is returned. In the current code, the targets behave in the following way: * dm-crypt returns -ENOMEM if there is not enough space to store the key, but it returns 0 on all other overflows. * dm-thin returns errors from the status method if a disk error happened. This is incorrect because retrieve_status doesn't check the error code, it assumes that all non-zero values mean buffer overflow. * all the other targets always return 0. This patch changes the ti->type->status function to return void (because most targets don't use the return code). Overflow is detected in retrieve_status: if the status method fills up the remaining space completely, it is assumed that buffer overflow happened. Cc: stable@vger.kernel.org Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2013-03-01 22:45:44 +00:00
static void mirror_status(struct dm_target *ti, status_type_t type,
unsigned status_flags, char *result, unsigned maxlen)
{
unsigned int m, sz = 0;
struct mirror_set *ms = (struct mirror_set *) ti->private;
struct dm_dirty_log *log = dm_rh_dirty_log(ms->rh);
char buffer[ms->nr_mirrors + 1];
switch (type) {
case STATUSTYPE_INFO:
DMEMIT("%d ", ms->nr_mirrors);
for (m = 0; m < ms->nr_mirrors; m++) {
DMEMIT("%s ", ms->mirror[m].dev->name);
buffer[m] = device_status_char(&(ms->mirror[m]));
}
buffer[m] = '\0';
DMEMIT("%llu/%llu 1 %s ",
(unsigned long long)log->type->get_sync_count(log),
(unsigned long long)ms->nr_regions, buffer);
sz += log->type->status(log, type, result+sz, maxlen-sz);
break;
case STATUSTYPE_TABLE:
sz = log->type->status(log, type, result, maxlen);
DMEMIT("%d", ms->nr_mirrors);
for (m = 0; m < ms->nr_mirrors; m++)
DMEMIT(" %s %llu", ms->mirror[m].dev->name,
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
(unsigned long long)ms->mirror[m].offset);
if (ms->features & DM_RAID1_HANDLE_ERRORS)
DMEMIT(" 1 handle_errors");
}
}
static int mirror_iterate_devices(struct dm_target *ti,
iterate_devices_callout_fn fn, void *data)
{
struct mirror_set *ms = ti->private;
int ret = 0;
unsigned i;
for (i = 0; !ret && i < ms->nr_mirrors; i++)
ret = fn(ti, ms->mirror[i].dev,
ms->mirror[i].offset, ti->len, data);
return ret;
}
static struct target_type mirror_target = {
.name = "mirror",
dm: fix truncated status strings Avoid returning a truncated table or status string instead of setting the DM_BUFFER_FULL_FLAG when the last target of a table fills the buffer. When processing a table or status request, the function retrieve_status calls ti->type->status. If ti->type->status returns non-zero, retrieve_status assumes that the buffer overflowed and sets DM_BUFFER_FULL_FLAG. However, targets don't return non-zero values from their status method on overflow. Most targets returns always zero. If a buffer overflow happens in a target that is not the last in the table, it gets noticed during the next iteration of the loop in retrieve_status; but if a buffer overflow happens in the last target, it goes unnoticed and erroneously truncated data is returned. In the current code, the targets behave in the following way: * dm-crypt returns -ENOMEM if there is not enough space to store the key, but it returns 0 on all other overflows. * dm-thin returns errors from the status method if a disk error happened. This is incorrect because retrieve_status doesn't check the error code, it assumes that all non-zero values mean buffer overflow. * all the other targets always return 0. This patch changes the ti->type->status function to return void (because most targets don't use the return code). Overflow is detected in retrieve_status: if the status method fills up the remaining space completely, it is assumed that buffer overflow happened. Cc: stable@vger.kernel.org Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2013-03-01 22:45:44 +00:00
.version = {1, 13, 2},
.module = THIS_MODULE,
.ctr = mirror_ctr,
.dtr = mirror_dtr,
.map = mirror_map,
.end_io = mirror_end_io,
dm raid1: fix EIO after log failure This patch adds the ability to requeue write I/O to core device-mapper when there is a log device failure. If a write to the log produces and error, the pending writes are put on the "failures" list. Since the log is marked as failed, they will stay on the failures list until a suspend happens. Suspends come in two phases, presuspend and postsuspend. We must make sure that all the writes on the failures list are requeued in the presuspend phase (a requirement of dm core). This means that recovery must be complete (because writes may be delayed behind it) and the failures list must be requeued before we return from presuspend. The mechanisms to ensure recovery is complete (or stopped) was already in place, but needed to be moved from postsuspend to presuspend. We rely on 'flush_workqueue' to ensure that the mirror thread is complete and therefore, has requeued all writes in the failures list. Because we are using flush_workqueue, we must ensure that no additional 'queue_work' calls will produce additional I/O that we need to requeue (because once we return from presuspend, we are unable to do anything about it). 'queue_work' is called in response to the following functions: - complete_resync_work = NA, recovery is stopped - rh_dec (mirror_end_io) = NA, only calls 'queue_work' if it is ready to recover the region (recovery is stopped) or it needs to clear the region in the log* **this doesn't get called while suspending** - rh_recovery_end = NA, recovery is stopped - rh_recovery_start = NA, recovery is stopped - write_callback = 1) Writes w/o failures simply call bio_endio -> mirror_end_io -> rh_dec (see rh_dec above) 2) Writes with failures are put on the failures list and queue_work is called** ** write_callbacks don't happen during suspend ** - do_failures = NA, 'queue_work' not called if suspending - add_mirror (initialization) = NA, only done on mirror creation - queue_bio = NA, 1) delayed I/O scheduled before flush_workqueue is called. 2) No more I/Os are being issued. 3) Re-attempted READs can still be handled. (Write completions are handled through rh_dec/ write_callback - mention above - and do not use queue_bio.) Signed-off-by: Jonathan Brassow <jbrassow@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com>
2008-02-08 02:11:35 +00:00
.presuspend = mirror_presuspend,
.postsuspend = mirror_postsuspend,
.resume = mirror_resume,
.status = mirror_status,
.iterate_devices = mirror_iterate_devices,
};
static int __init dm_mirror_init(void)
{
int r;
r = dm_register_target(&mirror_target);
if (r < 0) {
DMERR("Failed to register mirror target");
goto bad_target;
}
return 0;
bad_target:
return r;
}
static void __exit dm_mirror_exit(void)
{
dm_unregister_target(&mirror_target);
}
/* Module hooks */
module_init(dm_mirror_init);
module_exit(dm_mirror_exit);
MODULE_DESCRIPTION(DM_NAME " mirror target");
MODULE_AUTHOR("Joe Thornber");
MODULE_LICENSE("GPL");