The function jbd2_journal_unregister_shrinker() was getting called
twice when the file system was getting unmounted. On Power and ARM
platforms this was causing kernel crash when unmounting the file
system, when a percpu_counter was destroyed twice.
Fix this by removing jbd2_journal_[un]register_shrinker() functions,
and inlining the shrinker setup and teardown into
journal_init_common() and jbd2_journal_destroy(). This means that
ext4 and ocfs2 now no longer need to know about registering and
unregistering jbd2's shrinker.
Also, while we're at it, rename the percpu counter from
j_jh_shrink_count to j_checkpoint_jh_count, since this makes it
clearer what this counter is intended to track.
Link: https://lore.kernel.org/r/20210705145025.3363130-1-tytso@mit.edu
Fixes: 4ba3fcdde7 ("jbd2,ext4: add a shrinker to release checkpointed buffers")
Reported-by: Jon Hunter <jonathanh@nvidia.com>
Reported-by: Sachin Sant <sachinp@linux.vnet.ibm.com>
Tested-by: Sachin Sant <sachinp@linux.vnet.ibm.com>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Now that __try_to_free_cp_buf() remove checkpointed buffer or transaction
when the buffer is not 'busy', which is only called by
journal_clean_one_cp_list(). This patch simplify this function by remove
__try_to_free_cp_buf() and invoke __cp_buffer_busy() directly.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20210610112440.3438139-7-yi.zhang@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Current metadata buffer release logic in bdev_try_to_free_page() have
a lot of use-after-free issues when umount filesystem concurrently, and
it is difficult to fix directly because ext4 is the only user of
s_op->bdev_try_to_free_page callback and we may have to add more special
refcount or lock that is only used by ext4 into the common vfs layer,
which is unacceptable.
One better solution is remove the bdev_try_to_free_page callback, but
the real problem is we cannot easily release journal_head on the
checkpointed buffer, so try_to_free_buffers() cannot release buffers and
page under memory pressure, which is more likely to trigger
out-of-memory. So we cannot remove the callback directly before we find
another way to release journal_head.
This patch introduce a shrinker to free journal_head on the checkpointed
transaction. After the journal_head got freed, try_to_free_buffers()
could free buffer properly.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Suggested-by: Jan Kara <jack@suse.cz>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20210610112440.3438139-6-yi.zhang@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Now that __jbd2_journal_remove_checkpoint() can detect buffer io error
and mark journal checkpoint error, then we abort the journal later
before updating log tail to ensure the filesystem works consistently.
So we could remove other redundant buffer io error checkes.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20210610112440.3438139-5-yi.zhang@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Although we merged c044f3d836 ("jbd2: abort journal if free a async
write error metadata buffer"), there is a race between
jbd2_journal_try_to_free_buffers() and jbd2_journal_destroy(), so the
jbd2_log_do_checkpoint() may still fail to detect the buffer write
io error flag which may lead to filesystem inconsistency.
jbd2_journal_try_to_free_buffers() ext4_put_super()
jbd2_journal_destroy()
__jbd2_journal_remove_checkpoint()
detect buffer write error jbd2_log_do_checkpoint()
jbd2_cleanup_journal_tail()
<--- lead to inconsistency
jbd2_journal_abort()
Fix this issue by introducing a new atomic flag which only have one
JBD2_CHECKPOINT_IO_ERROR bit now, and set it in
__jbd2_journal_remove_checkpoint() when freeing a checkpoint buffer
which has write_io_error flag. Then jbd2_journal_destroy() will detect
this mark and abort the journal to prevent updating log tail.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20210610112440.3438139-3-yi.zhang@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The 'out' lable just return the 'ret' value and seems not required, so
remove this label and switch to return appropriate value immediately.
This patch also do some minor cleanup, no logical change.
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20210610112440.3438139-2-yi.zhang@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
There is no point in allocating memory for a synchronous flush.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Acked-by: Damien Le Moal <damien.lemoal@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Add missing __acquires() and __releases() annotations. Also, in an
"this should never happen" WARN_ON check, if it *does* actually
happen, we need to release j_state_lock since this function is always
supposed to release that lock. Otherwise, things will quickly grind
to a halt after the WARN_ON trips.
Fixes: 96f1e09745 ("jbd2: avoid long hold times of j_state_lock...")
Cc: stable@kernel.org
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The argument isn't used by any caller, and drivers don't fill out
bi_sector for flush requests either.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
JBD2_REC_ERR flag used to indicate the errno has been updated when jbd2
aborted, and then __ext4_abort() and ext4_handle_error() can invoke
panic if ERRORS_PANIC is specified. But if the journal has been aborted
with zero errno, jbd2_journal_abort() didn't set this flag so we can
no longer panic. Fix this by always record the proper errno in the
journal superblock.
Fixes: 4327ba52af ("ext4, jbd2: ensure entering into panic after recording an error in superblock")
Signed-off-by: zhangyi (F) <yi.zhang@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20191204124614.45424-3-yi.zhang@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The function is now just a trivial wrapper returning
journal->j_max_transaction_buffers. Drop it.
Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20191105164437.32602-19-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
At the beginning, nblocks has been assigned. There is no need
to repeat the assignment in the while loop, and remove it.
Signed-off-by: Liu Song <liu.song11@zte.com.cn>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>
This issue was found when I tried to put checkpoint work in a separate thread,
the deadlock below happened:
Thread1 | Thread2
__jbd2_log_wait_for_space |
jbd2_log_do_checkpoint (hold j_checkpoint_mutex)|
if (jh->b_transaction != NULL) |
... |
jbd2_log_start_commit(journal, tid); |jbd2_update_log_tail
| will lock j_checkpoint_mutex,
| but will be blocked here.
|
jbd2_log_wait_commit(journal, tid); |
wait_event(journal->j_wait_done_commit, |
!tid_gt(tid, journal->j_commit_sequence)); |
... |wake_up(j_wait_done_commit)
} |
then deadlock occurs, Thread1 will never be waken up.
To fix this issue, drop j_checkpoint_mutex in jbd2_log_do_checkpoint()
when we are going to wait for transaction commit.
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Xiaoguang Wang <xiaoguang.wang@linux.alibaba.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The code cleaning transaction's lists of checkpoint buffers has a bug
where it increases bh refcount only after releasing
journal->j_list_lock. Thus the following race is possible:
CPU0 CPU1
jbd2_log_do_checkpoint()
jbd2_journal_try_to_free_buffers()
__journal_try_to_free_buffer(bh)
...
while (transaction->t_checkpoint_io_list)
...
if (buffer_locked(bh)) {
<-- IO completes now, buffer gets unlocked -->
spin_unlock(&journal->j_list_lock);
spin_lock(&journal->j_list_lock);
__jbd2_journal_remove_checkpoint(jh);
spin_unlock(&journal->j_list_lock);
try_to_free_buffers(page);
get_bh(bh) <-- accesses freed bh
Fix the problem by grabbing bh reference before unlocking
journal->j_list_lock.
Fixes: dc6e8d669c ("jbd2: don't call get_bh() before calling __jbd2_journal_remove_checkpoint()")
Fixes: be1158cc61 ("jbd2: fold __process_buffer() into jbd2_log_do_checkpoint()")
Reported-by: syzbot+7f4a27091759e2fe7453@syzkaller.appspotmail.com
CC: stable@vger.kernel.org
Reviewed-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
A number of ext4 source files were skipped due because their copyright
permission statements didn't match the expected text used by the
automated conversion utilities. I've added SPDX tags for the rest.
While looking at some of these files, I've noticed that we have quite
a bit of variation on the licenses that were used --- in particular
some of the Red Hat licenses on the jbd2 files use a GPL2+ license,
and we have some files that have a LGPL-2.1 license (which was quite
surprising).
I've not attempted to do any license changes. Even if it is perfectly
legal to relicense to GPL 2.0-only for consistency's sake, that should
be done with ext4 developer community discussion.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Remove the WRITE_* and READ_SYNC wrappers, and just use the flags
directly. Where applicable this also drops usage of the
bio_set_op_attrs wrapper.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@fb.com>
Unlike comments and expectation of callers journal_clean_one_cp_list()
returned 1 not only if it freed the transaction but also if it freed
some buffers in the transaction. That could make
__jbd2_journal_clean_checkpoint_list() skip processing
t_checkpoint_io_list and continue with processing the next transaction.
This is mostly a cosmetic issue since the only result is we can
sometimes free less memory than we could. But it's still worth fixing.
Fix journal_clean_one_cp_list() to return 1 only if the transaction was
really freed.
Fixes: 50849db32a
Signed-off-by: Jan Kara <jack@suse.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@vger.kernel.org
Commit 6f6a6fda29 "jbd2: fix ocfs2 corrupt when updating journal
superblock fails" changed jbd2_cleanup_journal_tail() to return EIO
when the journal is aborted. That makes logic in
jbd2_log_do_checkpoint() bail out which is fine, except that
jbd2_journal_destroy() expects jbd2_log_do_checkpoint() to always make
a progress in cleaning the journal. Without it jbd2_journal_destroy()
just loops in an infinite loop.
Fix jbd2_journal_destroy() to cleanup journal checkpoint lists of
jbd2_log_do_checkpoint() fails with error.
Reported-by: Eryu Guan <guaneryu@gmail.com>
Tested-by: Eryu Guan <guaneryu@gmail.com>
Fixes: 6f6a6fda29
Signed-off-by: Jan Kara <jack@suse.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
If updating journal superblock fails after journal data has been
flushed, the error is omitted and this will mislead the caller as a
normal case. In ocfs2, the checkpoint will be treated successfully
and the other node can get the lock to update. Since the sb_start is
still pointing to the old log block, it will rewrite the journal data
during journal recovery by the other node. Thus the new updates will
be overwritten and ocfs2 corrupts. So in above case we have to return
the error, and ocfs2_commit_cache will take care of the error and
prevent the other node to do update first. And only after recovering
journal it can do the new updates.
The issue discussion mail can be found at:
https://oss.oracle.com/pipermail/ocfs2-devel/2015-June/010856.htmlhttp://comments.gmane.org/gmane.comp.file-systems.ext4/48841
[ Fixed bug in patch which allowed a non-negative error return from
jbd2_cleanup_journal_tail() to leak out of jbd2_fjournal_flush(); this
was causing xfstests ext4/306 to fail. -- Ted ]
Reported-by: Yiwen Jiang <jiangyiwen@huawei.com>
Signed-off-by: Joseph Qi <joseph.qi@huawei.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Tested-by: Yiwen Jiang <jiangyiwen@huawei.com>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: stable@vger.kernel.org
__jbd2_journal_clean_checkpoint_list() returns number of buffers it
freed but noone was using the value so just stop doing that. This
also allows for simplifying the calling convention for
journal_clean_once_cp_list().
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Yuanhan has reported that when he is running fsync(2) heavy workload
creating new files over ramdisk, significant amount of time is spent in
__jbd2_journal_clean_checkpoint_list() trying to clean old transactions
(but they cannot be cleaned up because flusher hasn't yet checkpointed
those buffers). The workload can be generated by:
fs_mark -d /fs/ram0/1 -D 2 -N 2560 -n 1000000 -L 1 -S 1 -s 4096
Reduce the amount of scanning by stopping to scan the transaction list
once we find a transaction that cannot be checkpointed. Note that this
way of cleaning is still enough to keep freeing space in the journal
after fully checkpointed transactions.
Reported-and-tested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
If EIO happens after we have dropped j_state_lock, we won't notice
that the journal has been aborted. So it is reasonable to move this
check after we have grabbed the j_checkpoint_mutex and re-grabbed the
j_state_lock. This patch helps to prevent false positive complain
after EIO.
#DMESG:
__jbd2_log_wait_for_space: needed 8448 blocks and only had 8386 space available
__jbd2_log_wait_for_space: no way to get more journal space in ram1-8
------------[ cut here ]------------
WARNING: CPU: 15 PID: 6739 at fs/jbd2/checkpoint.c:168 __jbd2_log_wait_for_space+0x188/0x200()
Modules linked in: brd iTCO_wdt lpc_ich mfd_core igb ptp dm_mirror dm_region_hash dm_log dm_mod
CPU: 15 PID: 6739 Comm: fsstress Tainted: G W 3.17.0-rc2-00429-g684de57 #139
Hardware name: Intel Corporation W2600CR/W2600CR, BIOS SE5C600.86B.99.99.x028.061320111235 06/13/2011
00000000000000a8 ffff88077aaab878 ffffffff815c1a8c 00000000000000a8
0000000000000000 ffff88077aaab8b8 ffffffff8106ce8c ffff88077aaab898
ffff8807c57e6000 ffff8807c57e6028 0000000000002100 ffff8807c57e62f0
Call Trace:
[<ffffffff815c1a8c>] dump_stack+0x51/0x6d
[<ffffffff8106ce8c>] warn_slowpath_common+0x8c/0xc0
[<ffffffff8106ceda>] warn_slowpath_null+0x1a/0x20
[<ffffffff812419f8>] __jbd2_log_wait_for_space+0x188/0x200
[<ffffffff8123be9a>] start_this_handle+0x4da/0x7b0
[<ffffffff810990e5>] ? local_clock+0x25/0x30
[<ffffffff810aba87>] ? lockdep_init_map+0xe7/0x180
[<ffffffff8123c5bc>] jbd2__journal_start+0xdc/0x1d0
[<ffffffff811f2414>] ? __ext4_new_inode+0x7f4/0x1330
[<ffffffff81222a38>] __ext4_journal_start_sb+0xf8/0x110
[<ffffffff811f2414>] __ext4_new_inode+0x7f4/0x1330
[<ffffffff810ac359>] ? lock_release_holdtime+0x29/0x190
[<ffffffff812025bb>] ext4_create+0x8b/0x150
[<ffffffff8117fe3b>] vfs_create+0x7b/0xb0
[<ffffffff8118097b>] do_last+0x7db/0xcf0
[<ffffffff8117e31d>] ? inode_permission+0x4d/0x50
[<ffffffff811845d2>] path_openat+0x242/0x590
[<ffffffff81191a76>] ? __alloc_fd+0x36/0x140
[<ffffffff81184a6a>] do_filp_open+0x4a/0xb0
[<ffffffff81191b61>] ? __alloc_fd+0x121/0x140
[<ffffffff81172f20>] do_sys_open+0x170/0x220
[<ffffffff8117300e>] SyS_open+0x1e/0x20
[<ffffffff811715d6>] SyS_creat+0x16/0x20
[<ffffffff815c7e12>] system_call_fastpath+0x16/0x1b
---[ end trace cd71c831f82059db ]---
Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
When we discover written out buffer in transaction checkpoint list we
don't have to recheck validity of a transaction. Either this is the
last buffer in a transaction - and then we are done - or this isn't
and then we can just take another buffer from the checkpoint list
without dropping j_list_lock.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
The __jbd2_journal_remove_checkpoint() doesn't require an elevated
b_count; indeed, until the jh structure gets released by the call to
jbd2_journal_put_journal_head(), the bh's b_count is elevated by
virtue of the existence of the jh structure.
Suggested-by: Jan Kara <jack@suse.cz>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
__wait_cp_io() is only called by jbd2_log_do_checkpoint(). Fold it in
to make it a bit easier to understand.
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
__process_buffer() is only called by jbd2_log_do_checkpoint(), and it
had a very complex locking protocol where it would be called with the
j_list_lock, and sometimes exit with the lock held (if the return code
was 0), or release the lock.
This was confusing both to humans and to smatch (which erronously
complained that the lock was taken twice).
Folding __process_buffer() to the caller allows us to simplify the
control flow, making the resulting function easier to read and reason
about, and dropping the compiled size of fs/jbd2/checkpoint.c by 150
bytes (over 4% of the text size).
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>
While trying to debug an an issue under extreme I/O loading
on preempt-rt kernels, the following backtrace was observed
via SysRQ output:
rm D ffff8802203afbc0 4600 4878 4748 0x00000000
ffff8802217bfb78 0000000000000082 ffff88021fc2bb80 ffff88021fc2bb80
ffff88021fc2bb80 ffff8802217bffd8 ffff8802217bffd8 ffff8802217bffd8
ffff88021f1d4c80 ffff88021fc2bb80 ffff8802217bfb88 ffff88022437b000
Call Trace:
[<ffffffff8172dc34>] schedule+0x24/0x70
[<ffffffff81225b5d>] jbd2_log_wait_commit+0xbd/0x140
[<ffffffff81060390>] ? __init_waitqueue_head+0x50/0x50
[<ffffffff81223635>] jbd2_log_do_checkpoint+0xf5/0x520
[<ffffffff81223b09>] __jbd2_log_wait_for_space+0xa9/0x1f0
[<ffffffff8121dc40>] start_this_handle.isra.10+0x2e0/0x530
[<ffffffff81060390>] ? __init_waitqueue_head+0x50/0x50
[<ffffffff8121e0a3>] jbd2__journal_start+0xc3/0x110
[<ffffffff811de7ce>] ? ext4_rmdir+0x6e/0x230
[<ffffffff8121e0fe>] jbd2_journal_start+0xe/0x10
[<ffffffff811f308b>] ext4_journal_start_sb+0x5b/0x160
[<ffffffff811de7ce>] ext4_rmdir+0x6e/0x230
[<ffffffff811435c5>] vfs_rmdir+0xd5/0x140
[<ffffffff8114370f>] do_rmdir+0xdf/0x120
[<ffffffff8105c6b4>] ? task_work_run+0x44/0x80
[<ffffffff81002889>] ? do_notify_resume+0x89/0x100
[<ffffffff817361ae>] ? int_signal+0x12/0x17
[<ffffffff81145d85>] sys_unlinkat+0x25/0x40
[<ffffffff81735f22>] system_call_fastpath+0x16/0x1b
What is interesting here, is that we call log_wait_commit, from
within wait_for_space, but we are still holding the checkpoint_mutex
as it surrounds mostly the whole of wait_for_space. And then, as we
are waiting, journal_commit_transaction can run, and if the JBD2_FLUSHED
bit is set, then we will also try to take the same checkpoint_mutex.
It seems that we need to drop the checkpoint_mutex while sitting in
jbd2_log_wait_commit, if we want to guarantee that progress can be made
by jbd2_journal_commit_transaction(). There does not seem to be
anything preempt-rt specific about this, other then perhaps increasing
the odds of it happening.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
j_wait_logspace and j_wait_checkpoint are unused. Remove them.
Reviewed-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
__jbd2_log_space_left() and jbd_space_needed() were kind of odd.
jbd_space_needed() accounted also credits needed for currently
committing transaction while it didn't account for credits needed for
control blocks. __jbd2_log_space_left() then accounted for control
blocks as a fraction of free space. Since results of these two
functions are always only compared against each other, this works
correct but is somewhat strange. Move the estimates so that
jbd_space_needed() returns number of blocks needed for a transaction
including control blocks and __jbd2_log_space_left() returns free
space in the journal (with the committing transaction already
subtracted). Rename functions to jbd2_log_space_left() and
jbd2_space_needed() while we are changing them.
Reviewed-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Similarly as for metadata buffers, also log descriptor buffers don't
really need the journal head. So strip it and remove BJ_LogCtl list.
Reviewed-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
When writing metadata to the journal, we create temporary buffer heads
for that task. We also attach journal heads to these buffer heads but
the only purpose of the journal heads is to keep buffers linked in
transaction's BJ_IO list. We remove the need for journal heads by
reusing buffer_head's b_assoc_buffers list for that purpose. Also
since BJ_IO list is just a temporary list for transaction commit, we
use a private list in jbd2_journal_commit_transaction() for that thus
removing BJ_IO list from transaction completely.
Reviewed-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
All accesses to checkpointing entries in journal_head are protected
by j_list_lock. Thus __jbd2_journal_remove_checkpoint() doesn't really
need bh_state lock.
Also the only part of journal head that the rest of checkpointing code
needs to check is jh->b_transaction which is safe to read under
j_list_lock.
So we can safely remove bh_state lock from all of checkpointing code which
makes it considerably prettier.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
BH_JWrite bit should be set when buffer is written to the journal. So
checkpointing shouldn't set this bit when writing out buffer. This didn't
cause any observable bug since BH_JWrite bit is used only for debugging
purposes but it's good to have this consistent.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
When we reach jbd2_cleanup_journal_tail(), there is no guarantee that
checkpointed buffers are on a stable storage - especially if buffers were
written out by jbd2_log_do_checkpoint(), they are likely to be only in disk's
caches. Thus when we update journal superblock effectively removing old
transaction from journal, this write of superblock can get to stable storage
before those checkpointed buffers which can result in filesystem corruption
after a crash. Thus we must unconditionally issue a cache flush before we
update journal superblock in these cases.
A similar problem can also occur if journal superblock is written only in
disk's caches, other transaction starts reusing space of the transaction
cleaned from the log and power failure happens. Subsequent journal replay would
still try to replay the old transaction but some of it's blocks may be already
overwritten by the new transaction. For this reason we must use WRITE_FUA when
updating log tail and we must first write new log tail to disk and update
in-memory information only after that.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
There are three case of updating journal superblock. In the first case, we want
to mark journal as empty (setting s_sequence to 0), in the second case we want
to update log tail, in the third case we want to update s_errno. Split these
cases into separate functions. It makes the code slightly more straightforward
and later patches will make the distinction even more important.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
There is normally only a handful of these active at any one time, but
putting them in a separate slab cache makes debugging memory
corruption problems easier. Manish Katiyar also wanted this make it
easier to test memory failure scenarios in the jbd2 layer.
Cc: Manish Katiyar <mkatiyar@gmail.com>
Signed-off-by: Yongqiang Yang <xiaoqiangnk@gmail.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
This patch adds trace_jbd2_drop_transaction and
trace_jbd2_update_superblock_end because there are similar tracepoints
in jbd and they are needed in jbd2 as well.
Reviewed-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Seiji Aguchi <seiji.aguchi@hds.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
In journal checkpoint, we write the buffer and wait for its finish.
But in cfq, the async queue has a very low priority, and in our test,
if there are too many sync queues and every queue is filled up with
requests, the write request will be delayed for quite a long time and
all the tasks which are waiting for journal space will end with errors like:
INFO: task attr_set:3816 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
attr_set D ffff880028393480 0 3816 1 0x00000000
ffff8802073fbae8 0000000000000086 ffff8802140847c8 ffff8800283934e8
ffff8802073fb9d8 ffffffff8103e456 ffff8802140847b8 ffff8801ed728080
ffff8801db4bc080 ffff8801ed728450 ffff880028393480 0000000000000002
Call Trace:
[<ffffffff8103e456>] ? __dequeue_entity+0x33/0x38
[<ffffffff8103caad>] ? need_resched+0x23/0x2d
[<ffffffff814006a6>] ? thread_return+0xa2/0xbc
[<ffffffffa01f6224>] ? jbd2_journal_dirty_metadata+0x116/0x126 [jbd2]
[<ffffffffa01f6224>] ? jbd2_journal_dirty_metadata+0x116/0x126 [jbd2]
[<ffffffff81400d31>] __mutex_lock_common+0x14e/0x1a9
[<ffffffffa021dbfb>] ? brelse+0x13/0x15 [ext4]
[<ffffffff81400ddb>] __mutex_lock_slowpath+0x19/0x1b
[<ffffffff81400b2d>] mutex_lock+0x1b/0x32
[<ffffffffa01f927b>] __jbd2_journal_insert_checkpoint+0xe3/0x20c [jbd2]
[<ffffffffa01f547b>] start_this_handle+0x438/0x527 [jbd2]
[<ffffffff8106f491>] ? autoremove_wake_function+0x0/0x3e
[<ffffffffa01f560b>] jbd2_journal_start+0xa1/0xcc [jbd2]
[<ffffffffa02353be>] ext4_journal_start_sb+0x57/0x81 [ext4]
[<ffffffffa024a314>] ext4_xattr_set+0x6c/0xe3 [ext4]
[<ffffffffa024aaff>] ext4_xattr_user_set+0x42/0x4b [ext4]
[<ffffffff81145adb>] generic_setxattr+0x6b/0x76
[<ffffffff81146ac0>] __vfs_setxattr_noperm+0x47/0xc0
[<ffffffff81146bb8>] vfs_setxattr+0x7f/0x9a
[<ffffffff81146c88>] setxattr+0xb5/0xe8
[<ffffffff81137467>] ? do_filp_open+0x571/0xa6e
[<ffffffff81146d26>] sys_fsetxattr+0x6b/0x91
[<ffffffff81002d32>] system_call_fastpath+0x16/0x1b
So this patch tries to use WRITE_SYNC in __flush_batch so that the request will
be moved into sync queue and handled by cfq timely. We also use the new plug,
sot that all the WRITE_SYNC requests can be given as a whole when we unplug it.
Signed-off-by: Tao Ma <boyu.mt@taobao.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: Jan Kara <jack@suse.cz>
Reported-by: Robin Dong <sanbai@taobao.com>
jbd2_journal_remove_journal_head() can oops when trying to access
journal_head returned by bh2jh(). This is caused for example by the
following race:
TASK1 TASK2
jbd2_journal_commit_transaction()
...
processing t_forget list
__jbd2_journal_refile_buffer(jh);
if (!jh->b_transaction) {
jbd_unlock_bh_state(bh);
jbd2_journal_try_to_free_buffers()
jbd2_journal_grab_journal_head(bh)
jbd_lock_bh_state(bh)
__journal_try_to_free_buffer()
jbd2_journal_put_journal_head(jh)
jbd2_journal_remove_journal_head(bh);
jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and
buffer is not part of any transaction and thus frees journal_head
before TASK1 gets to doing so. Note that even buffer_head can be
released by try_to_free_buffers() after
jbd2_journal_put_journal_head() which adds even larger opportunity for
oops (but I didn't see this happen in reality).
Fix the problem by making transactions hold their own journal_head
reference (in b_jcount). That way we don't have to remove journal_head
explicitely via jbd2_journal_remove_journal_head() and instead just
remove journal_head when b_jcount drops to zero. The result of this is
that [__]jbd2_journal_refile_buffer(),
[__]jbd2_journal_unfile_buffer(), and
__jdb2_journal_remove_checkpoint() can free journal_head which needs
modification of a few callers. Also we have to be careful because once
journal_head is removed, buffer_head might be freed as well. So we
have to get our own buffer_head reference where it matters.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
An attempt to modify the file system during the call to
jbd2_destroy_journal() can lead to a system lockup. So add some
checking to make it much more obvious when this happens to and to
determine where the offending code is located.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
All the blkdev_issue_* helpers can only sanely be used for synchronous
caller. To issue cache flushes or barriers asynchronously the caller needs
to set up a bio by itself with a completion callback to move the asynchronous
state machine ahead. So drop the BLKDEV_IFL_WAIT flag that is always
specified when calling blkdev_issue_* and also remove the now unused flags
argument to blkdev_issue_flush and blkdev_issue_zeroout. For
blkdev_issue_discard we need to keep it for the secure discard flag, which
gains a more descriptive name and loses the bitops vs flag confusion.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
These flags aren't real I/O types, but tell ll_rw_block to always
lock the buffer instead of giving up on a failed trylock.
Instead add a new write_dirty_buffer helper that implements this semantic
and use it from the existing SWRITE* callers. Note that the ll_rw_block
code had a bug where it didn't promote WRITE_SYNC_PLUG properly, which
this patch fixes.
In the ufs code clean up the helper that used to call ll_rw_block
to mirror sync_dirty_buffer, which is the function it implements for
compound buffers.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Lockstat reports have shown that j_state_lock is a major source of
lock contention, especially on systems with more than 4 CPU cores. So
change it to be a read/write spinlock.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
By using an atomic_t for t_updates and t_outstanding credits, this
should allow us to not need to take transaction t_handle_lock in
jbd2_journal_stop().
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The patch just convert all blkdev_issue_xxx function to common
set of flags. Wait/allocation semantics preserved.
Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This is a bit complicated because we are trying to optimize when we
send barriers to the fs data disk. We could just throw in an extra
barrier to the data disk whenever we send a barrier to the journal
disk, but that's not always strictly necessary.
We only need to send a barrier during a commit when there are data
blocks which are must be written out due to an inode written in
ordered mode, or if fsync() depends on the commit to force data blocks
to disk. Finally, before we drop transactions from the beginning of
the journal during a checkpoint operation, we need to guarantee that
any blocks that were flushed out to the data disk are firmly on the
rust platter before we drop the transaction from the journal.
Thanks to Oleg Drokin for pointing out this flaw in ext3/ext4.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
