I've bisected the deadlock when many small appends are done on jffs2 down to
this commit:
commit 6fe6900e1e
Author: Nick Piggin <npiggin@suse.de>
Date: Sun May 6 14:49:04 2007 -0700
mm: make read_cache_page synchronous
Ensure pages are uptodate after returning from read_cache_page, which allows
us to cut out most of the filesystem-internal PageUptodate calls.
I didn't have a great look down the call chains, but this appears to fixes 7
possible use-before uptodate in hfs, 2 in hfsplus, 1 in jfs, a few in
ecryptfs, 1 in jffs2, and a possible cleared data overwritten with readpage in
block2mtd. All depending on whether the filler is async and/or can return
with a !uptodate page.
It introduced a wait to read_cache_page, as well as a
read_cache_page_async function equivalent to the old read_cache_page
without any callers.
Switching jffs2_gc_fetch_page to read_cache_page_async for the old
behavior makes the deadlocks go away, but maybe reintroduces the
use-before-uptodate problem? I don't understand the mm/fs interaction
well enough to say.
[It's fine. dwmw2.]
Signed-off-by: Jason Lunz <lunz@falooley.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Ryusuke Konishi says:
The recent truncate_complete_page() clears the dirty flag from a page
before calling a_ops->invalidatepage(),
^^^^^^
static void
truncate_complete_page(struct address_space *mapping, struct page *page)
{
...
cancel_dirty_page(page, PAGE_CACHE_SIZE); <--- Inserted here at
kernel 2.6.20
if (PagePrivate(page))
do_invalidatepage(page, 0); ---> will call
a_ops->invalidatepage()
...
}
and this is disturbing nfs_wb_page_priority() from calling
nfs_writepage_locked() that is expected to handle the pending
request (=nfs_page) associated with the page.
int nfs_wb_page_priority(struct inode *inode, struct page *page, int how)
{
...
if (clear_page_dirty_for_io(page)) {
ret = nfs_writepage_locked(page, &wbc);
if (ret < 0)
goto out;
}
...
}
Since truncate_complete_page() will get rid of the page after
a_ops->invalidatepage() returns, the request (=nfs_page) associated
with the page becomes a garbage in nfs_inode->nfs_page_tree.
------------------------
Fix this by ensuring that nfs_wb_page_priority() recognises that it may
also need to clear out non-dirty pages that have an nfs_page associated
with them.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
According to the mount(2) man page, the proper error return code for the
mount(2) system call when the special device name or the mounted-on
directory name is too long is ENAMETOOLONG.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The hostname was getting truncated in the new text-based NFS mount API.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Don't filter the return code from the in-kernel rpcbind or NFS mount
clients. Return the real error code so that callers of the new NFS
text-based mount API can apply a useful retry strategy.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The new text-based NFS mount option parsing logic doesn't recognize any
valid transport protocols due to a silly mistake in the protocol token
matching logic. This prevents basic mount requests such as:
mount.nfs server:/export /mnt -o proto=tcp
from working with the new text-based NFS mount API.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Doh! We can't use cancel_delayed_work_sync because we may have been called
from an unmount that was being performed by nfs_automount_task.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
This avoids the recent NFS mount regression (returning EBUSY when
mounting the same filesystem twice with different parameters).
The best I can do given the constraints appears to be to have the kernel
first look for a superblock that matches both the fsid and the
user-specified mount options, and then spawn off a new superblock if
that search fails.
Note that this is not the same as specifying nosharecache everywhere
since nosharecache will never attempt to match an existing superblock.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Tested-by: Hua Zhong <hzhong@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
For hugepage mappings, the file offset, like the address and size, needs to
be aligned to the size of a hugepage.
In commit 68589bc353, the check for this was
moved into prepare_hugepage_range() along with the address and size checks.
But since BenH's rework of the get_unmapped_area() paths leading up to
commit 4b1d89290b, prepare_hugepage_range()
is only called for MAP_FIXED mappings, not for other mappings. This means
we're no longer ever checking for an aligned offset - I've confirmed that
mmap() will (apparently) succeed with a misaligned offset on both powerpc
and i386 at least.
This patch restores the check, removing it from prepare_hugepage_range()
and putting it back into hugetlbfs_file_mmap(). I'm putting it there,
rather than in the get_unmapped_area() path so it only needs to go in one
place, than separately in the half-dozen or so arch-specific
implementations of hugetlb_get_unmapped_area().
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Cc: Adam Litke <agl@us.ibm.com>
Cc: Andi Kleen <ak@suse.de>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This will avoid a possible fault in ecryptfs_sync_page().
In the function, eCryptfs calls sync_page() method of a lower filesystem
without checking its existence. However, there are many filesystems that
don't have this method including network filesystems such as NFS, AFS, and
so forth. They may fail when an eCryptfs page is waiting for lock.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Acked-by: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix possible NULL pointer dereference when freeing blocks in case table of
free space is used. Also fix handling of the case when we need to move
extent from one block to another one to make space for indirect extent.
BTW: Nobody seem to have ever used this code.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If UDF superblock is incorrect, we can fail to find a table of free /
allocated space and consequently Oops. Handle this situation more
gracefully by ignoring the broken UDF partition.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs:
9p: fix bad error path in conversion routines
9p: remove deprecated v9fs_fid_lookup_remove()
9p: update maintainers and documentation
9p: fix use after free
* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/driver-2.6:
sysfs: don't warn on removal of a nonexistent binary file
HOWTO: latest lxr url address changed
HOWTO: korean translation of Documentation/HOWTO
Fix Off-by-one in /sys/module/*/refcnt
sysfs: fix locking in sysfs_lookup() and sysfs_rename_dir()
This patch removes the v9fs_fid_lookup_remove which is no longer used.
Based on original patch from Adrian Bunk <bunk@stusta.de> which
used #if 0 to isolate the code.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Fix the accounting regression for CONFIG_VIRT_CPU_ACCOUNTING. It
reverts parts of commit b27f03d4bd by
converting fs/proc/array.c back to cputime_t. The new functions
task_utime and task_stime now return cputime_t instead of clock_t. If
CONFIG_VIRT_CPU_ACCOUTING is set, task->utime and task->stime are
returned directly instead of using sum_exec_runtime.
Patch is tested on s390x with and without VIRT_CPU_ACCOUTING as well as
on i386.
[ mingo@elte.hu: cleanups, comments. ]
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
de_thread:
if (atomic_read(&oldsighand->count) <= 1)
BUG_ON(atomic_read(&sig->count) != 1);
This is not safe without the rmb() in between. The results of two
correctly ordered __exit_signal()->atomic_dec_and_test()'s could be seen
out of order on our CPU.
The same is true for the "thread_group_empty()" case, __unhash_process()'s
changes could be seen before atomic_dec_and_test(&sig->count).
On some platforms (including i386) atomic_read() doesn't provide even the
compiler barrier, in that case these checks are simply racy.
Remove these BUG_ON()'s. Alternatively, we can do something like
BUG_ON( ({ smp_rmb(); atomic_read(&sig->count) != 1; }) );
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Due to inconsistent locking in the VFS between calls to lookup and
revalidate deadlock can occur in the automounter.
The inconsistency is that the directory inode mutex is held for both lookup
and revalidate calls when called via lookup_hash whereas it is held only
for lookup during a path walk. Consequently, if the mutex is held during a
call to revalidate autofs4 can't release the mutex to callback the daemon
as it can't know whether it owns the mutex.
This situation happens when a process tries to create a directory within an
automount and a second process also tries to create the same directory
between the lookup and the mkdir. Since the first process has dropped the
mutex for the daemon callback, the second process takes it during
revalidate leading to deadlock between the autofs daemon and the second
process when the daemon tries to create the mount point directory.
After spending quite a bit of time trying to resolve this on more than one
occassion, using rather complex and ulgy approaches, it turns out that just
delaying the hashing of the dentry until the create operation works fine.
Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
With this patch any thread can dequeue its own private signals via signalfd,
even if it was created by another sub-thread.
To do so, we pass "current" to dequeue_signal() if the caller is from the same
thread group. This also fixes the scheduling of posix timers broken by the
previous patch.
If the caller doesn't belong to this thread group, we can't handle __SI_TIMER
case properly anyway. Perhaps we should forbid the cross-process signalfd usage
and convert ctx->tsk to ctx->sighand.
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Cc: Roland McGrath <roland@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When ecryptfs_lookup() is called against special files, eCryptfs generates
the following errors because it tries to treat them like regular eCryptfs
files.
Error opening lower file for lower_dentry [0xffff810233a6f150], lower_mnt [0xffff810235bb4c80], and flags [0x8000]
Error opening lower_file to read header region
Error attempting to read the [user.ecryptfs] xattr from the lower file; return value = [-95]
Valid metadata not found in header region or xattr region; treating file as unencrypted
For instance, the problem can be reproduced by the steps below.
# mkdir /root/crypt /mnt/crypt
# mount -t ecryptfs /root/crypt /mnt/crypt
# mknod /mnt/crypt/c0 c 0 0
# umount /mnt/crypt
# mount -t ecryptfs /root/crypt /mnt/crypt
# ls -l /mnt/crypt
This patch fixes it by adding a check similar to directories and
symlinks.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Acked-by: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch (as960) removes the error message and stack dump logged by
sysfs_remove_bin_file() when someone tries to remove a nonexistent
file. The warning doesn't seem to be needed, since none of the other
file-, symlink-, or directory-removal routines in sysfs complain in a
comparable way.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Acked-by: Tejun Heo <htejun@gmail.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
sd children list walking in sysfs_lookup() and sd renaming in
sysfs_rename_dir() were left out during i_mutex -> sysfs_mutex
conversion. Fix them.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch uses kzalloc to zero all of struct dio rather than manually
trying to track which fields we rely on being zero. It passed aio+dio
stress testing and some bug regression testing on ext3.
This patch was introduced by Linus in the conversation that lead up to
Badari's minimal fix to manually zero .map_bh.b_state in commit:
6a648fa721
It makes the code a bit smaller. Maybe a couple fewer cachelines to
load, if we're lucky:
text data bss dec hex filename
3285925 568506 1304616 5159047 4eb887 vmlinux
3285797 568506 1304616 5158919 4eb807 vmlinux.patched
I was unable to measure a stable difference in the number of cpu cycles
spent in blockdev_direct_IO() when pushing aio+dio 256K reads at
~340MB/s.
So the resulting intent of the patch isn't a performance gain but to
avoid exposing ourselves to the risk of finding another field like
.map_bh.b_state where we rely on zeroing but don't enforce it in the
code.
Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit a491486a20 introduced a locking
problem in JFFS2 -- we up() the alloc_sem when we weren't previously
holding it. This leads to all kinds of fun behaviour later.
There was a _reason_ for the
if (1 /* alternative path needs testing */ ||
which the above-mentioned commit removed :)
Discovered and debugged by Giulio Fedel <giulio.fedel@andorsystems.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
[CIFS] Check return code on failed alloc
[CIFS] Update CIFS project web site
[CIFS] Fix hang in find_writable_file
This fixes a vulnerability in the "parent process death signal"
implementation discoverd by Wojciech Purczynski of COSEINC PTE Ltd.
and iSEC Security Research.
http://marc.info/?l=bugtraq&m=118711306802632&w=2
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This reverts commit 569a7b6c2e. The
code was correct originally. The default setting for ACLs after a
remount should be to be the same as before the remount.
Signed-off-by: Abhijith Das <adas@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Due to a mix up between the jdata attribute and inherit jdata attribute
it has not been possible to set the inherit jdata attribute on
directories. This is now fixed and the ioctl will report the inherit
jdata attribute for directories rather than the jdata attribute as it
did previously. This stems from our need to have the one bit in the
ioctl attr flags mean two different things according to whether the
underlying inode is a directory or not.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
The error path in prepare_write() was incorrect in the (very rare) event
that the transaction fails to start. The following prevents a NULL
pointer dereference,
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
The following patch fixes a bug where 0 was being used as a return code
to indicate "nothing to do" when in fact 0 was a valid block location
which might be returned by the function.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
This patch seems to fix the problem described in bugzilla bug 246114.
It was written by Steve Whitehouse with some tweaking by me.
The code was looping in the relatively new section of code designed to
search for and reuse unlinked inodes. In cases where it was finding an
appropriate inode to reuse, it was looping around and finding the same
block over and over because a "<=" check should have been a "<" when
comparing the goal block to the last unlinked block found.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
This is part 2 of the patch for bug #245832, part 1 of which is already
in the git tree.
The problem was that sdp->sd_log_num_databuf was not always being
protected by the gfs2_log_lock spinlock, but the sd_log_le_databuf
(which it is supposed to reflect) was protected. That meant there
was a timing window during which gfs2_log_flush called
databuf_lo_before_commit and the count didn't match what was
really on the linked list in that window. So when it ran out of
items on the linked list, it decremented total_dbuf from 0 to -1 and
thus never left the "while(total_dbuf)" loop.
The solution is to protect the variable sdp->sd_log_num_databuf so
that the value will always match the contents of the linked list,
and therefore the number will never go negative, and therefore, the
loop will be exited properly.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Fix a long standing bug where a blocking callback would be missed
when there's a granted lock in PR mode and waiting locks in both
PR and CW modes (and the PR lock was added to the waiting queue
before the CW lock). The logic simply compared the numerical values
of the modes to determine if a blocking callback was required, but in
the one case of PR and CW, the lower valued CW mode blocks the higher
valued PR mode. We just need to add a special check for this PR/CW
case in the tests that decide when a blocking callback is needed.
Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
The last patch to clean out 'othercon' structures only fixed half the problem.
The attached addresses the other situations too, and fixes bz#238490
Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
There's a memory leak in fs/dlm/member.c::dlm_add_member().
If "dlm_node_weight(ls->ls_name, nodeid)" returns < 0, then
we'll return without freeing the memory allocated to the (at
that point yet unused) 'memb'.
This patch frees the allocated memory in that case and thus
avoids the leak.
Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
When we build a sockaddr_storage for an IP address, clear the unused parts as
they could be used for node comparisons.
I have seen this occasionally make sctp connections fail.
Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Fix regression in recent patch "[DLM] variable allocation" which
attempts to dereference an "ls" struct when it's NULL.
Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
This patch clears the othercon pointer and frees the memory when a connnection
is closed. This could cause a small memory leak when nodes leave the cluster.
Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2:
ocfs2: set non-default s_time_gran during mount
ocfs2: Retry sendpage() if it returns EAGAIN
ocfs2: Fix rename/extend race
[2.6 patch] ocfs2_insert_extent(): remove dead code
ocfs2: Fix max offset calculations
ocfs2: check ia_size limits in setattr
ocfs2: Fix some casting errors related to file writes
ocfs2: use s_maxbytes directly in ocfs2_change_file_space()
ocfs2: Restrict inode changes in ocfs2_update_inode_atime()
ecryptfs_init() exits without doing any cleanup jobs if
ecryptfs_init_messaging() fails. In that case, eCryptfs leaves
sysfs entries, leaks memory, and causes an invalid page fault.
This patch fixes the problem.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Acked-by: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When ecryptfs_lookup() is called against special files, eCryptfs generates
the following errors because it tries to treat them like regular eCryptfs
files.
Error opening lower file for lower_dentry [0xffff810233a6f150], lower_mnt [0xffff810235bb4c80], and flags
[0x8000]
Error opening lower_file to read header region
Error attempting to read the [user.ecryptfs] xattr from the lower file; return value = [-95]
Valid metadata not found in header region or xattr region; treating file as unencrypted
For instance, the problem can be reproduced by the steps below.
# mkdir /root/crypt /mnt/crypt
# mount -t ecryptfs /root/crypt /mnt/crypt
# mknod /mnt/crypt/c0 c 0 0
# umount /mnt/crypt
# mount -t ecryptfs /root/crypt /mnt/crypt
# ls -l /mnt/crypt
This patch fixes it by adding a check similar to directories and
symlinks.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Acked-by: Michael Halcrow <mhalcrow@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Need to initialize map_bh.b_state to zero. Otherwise, in case of a faulty
user-buffer its possible to go into dio_zero_block() and submit a page by
mistake - since it checks for buffer_new().
http://marc.info/?l=linux-kernel&m=118551339032528&w=2
akpm: Linus had a (better) patch to just do a kzalloc() in there, but it got
lost. Probably this version is better for -stable anwyay.
Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com>
Acked-by: Joe Jin <joe.jin@oracle.com>
Acked-by: Zach Brown <zach.brown@oracle.com>
Cc: gurudas pai <gurudas.pai@oracle.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We need to manually set this to '1' during mount, otherwise inode_setattr()
will chop off the nanosecond portion of our timestamps.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Instead of treating EAGAIN, returned from sendpage(), as an error, this
patch retries the operation.
Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
