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>
Instead of abusing a buffer_head flag just add a variant of
sync_dirty_buffer which allows passing the exact type of write
flag required.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
This should remove the last exclusive lock from start_this_handle(),
so that we should now be able to start multiple transactions at the
same time on large SMP systems.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
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>
Saying things like "sync failed" when a device does
not support barriers makes users slightly more worried than
they need to be; rather than talking about sync failures,
let's just state the barrier-based facts.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
__GFP_NOFAIL is going away, so add our own retry loop. Also add
jbd2__journal_start() and jbd2__journal_restart() which take a gfp
mask, so that file systems can optionally (re)start transaction
handles using GFP_KERNEL. If they do this, then they need to be
prepared to handle receiving an PTR_ERR(-ENOMEM) error, and be ready
to reflect that error up to userspace.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
OCFS2 uses t_commit trigger to compute and store checksum of the just
committed blocks. When a buffer has b_frozen_data, checksum is computed
for it instead of b_data but this can result in an old checksum being
written to the filesystem in the following scenario:
1) transaction1 is opened
2) handle1 is opened
3) journal_access(handle1, bh)
- This sets jh->b_transaction to transaction1
4) modify(bh)
5) journal_dirty(handle1, bh)
6) handle1 is closed
7) start committing transaction1, opening transaction2
8) handle2 is opened
9) journal_access(handle2, bh)
- This copies off b_frozen_data to make it safe for transaction1 to commit.
jh->b_next_transaction is set to transaction2.
10) jbd2_journal_write_metadata() checksums b_frozen_data
11) the journal correctly writes b_frozen_data to the disk journal
12) handle2 is closed
- There was no dirty call for the bh on handle2, so it is never queued for
any more journal operation
13) Checkpointing finally happens, and it just spools the bh via normal buffer
writeback. This will write b_data, which was never triggered on and thus
contains a wrong (old) checksum.
This patch fixes the problem by calling the trigger at the moment data is
frozen for journal commit - i.e., either when b_frozen_data is created by
do_get_write_access or just before we write a buffer to the log if
b_frozen_data does not exist. We also rename the trigger to t_frozen as
that better describes when it is called.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
No real bugs found, just removed some dead code.
Found by gcc 4.6's new warnings.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
One of the most contended locks in the jbd2 layer is j_state_lock when
running dbench. This is especially true if using the real-time kernel
with its "sleeping spinlocks" patch that replaces spinlocks with
priority inheriting mutexes --- but it also shows up on large SMP
benchmarks.
Thanks to John Stultz for pointing this out.
Reviewed by Mingming Cao and Jan Kara.
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>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
commit_transaction has the same value as journal->j_running_transaction,
so we can simplify the assert statement.
Signed-off-by: dingdinghua <dingdinghua@nrchpc.ac.cn>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Delay discarding buffers in journal_unmap_buffer until
we know that "add to orphan" operation has definitely been
committed, otherwise the log space of committing transation
may be freed and reused before truncate get committed, updates
may get lost if crash happens.
Signed-off-by: dingdinghua <dingdinghua@nrchpc.ac.cn>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
It triggers the warning in get_page_from_freelist(), and it isn't
appropriate to use __GFP_NOFAIL here anyway.
Addresses http://bugzilla.kernel.org/show_bug.cgi?id=14843
Reported-by: Christian Casteyde <casteyde.christian@free.fr>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
jbd-debug and jbd2-debug is currently read-only (S_IRUGO), which is not
correct. Make it writable so that we can start debuging.
Signed-off-by: Yin Kangkai <kangkai.yin@intel.com>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jan Kara <jack@suse.cz>
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>
This reverts commit e4c570c4cb, as
requested by Alexey:
"I think I gave a good enough arguments to not merge it.
To iterate:
* patch makes impossible to start using ext3 on EXT3_FS=n kernels
without reboot.
* this is done only for one pointer on task_struct"
None of config options which define task_struct are tristate directly
or effectively."
Requested-by: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
journal_info in task_struct is used in journaling file system only. So
introduce CONFIG_FS_JOURNAL_INFO and make it conditional.
Signed-off-by: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
All callers really want the more logical filemap_fdatawait_range interface,
so convert them to use it and merge wait_on_page_writeback_range into
filemap_fdatawait_range.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>
Now that the SLUB seems to be fixed so that it respects the requested
alignment, use kmem_cache_alloc() to allocator if the block size of
the buffer heads to be allocated is less than the page size.
Previously, we were using 16k page on a Power system for each buffer,
even when the file system was using 1k or 4k block size.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
If there is a failed journal checksum, don't reset the journal. This
allows for userspace programs to decide how to recover from this
situation. It may be that ignoring the journal checksum failure might
be a better way of recovering the file system. Once we add per-block
checksums, we can definitely do better. Until then, a system
administrator can try backing up the file system image (or taking a
snapshot) and and trying to determine experimentally whether ignoring
the checksum failure or aborting the journal replay results in less
data loss.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: stable@kernel.org
If journal init fails, we need to free j_wbuf.
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Jan Kara <jack@suse.cz>
Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Jan Kara <jack@suse.cz>
The /proc/fs/jbd2/<dev>/history was maintained manually; by using
tracepoints, we can get all of the existing functionality of the /proc
file plus extra capabilities thanks to the ftrace infrastructure. We
save memory as a bonus.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
There are a number of kernel printk's which are printed when an ext4
filesystem is mounted and unmounted. Disable them to economize space
in the system logs. In addition, disabling the mballoc stats by
default saves a number of unneeded atomic operations for every block
allocation or deallocation.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Make all seq_operations structs const, to help mitigate against
revectoring user-triggerable function pointers.
This is derived from the grsecurity patch, although generated from scratch
because it's simpler than extracting the changes from there.
Signed-off-by: James Morris <jmorris@namei.org>
Acked-by: Serge Hallyn <serue@us.ibm.com>
Acked-by: Casey Schaufler <casey@schaufler-ca.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Previously the journal_async_commit mount option was equivalent to
using barrier=0 (and just as unsafe). This patch fixes it so that we
eliminate the barrier before the commit block (by not using ordered
mode), and explicitly issuing an empty barrier bio after writing the
commit block. Because of the journal checksum, it is safe to do this;
if the journal blocks are not all written before a power failure, the
checksum in the commit block will prevent the last transaction from
being replayed.
Using the fs_mark benchmark, using journal_async_commit shows a 50%
improvement:
FSUse% Count Size Files/sec App Overhead
8 1000 10240 30.5 28242
vs.
FSUse% Count Size Files/sec App Overhead
8 1000 10240 45.8 28620
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
lockdep annotation for a transaction start has been at the end of
jbd2_journal_start(). But a transaction is also started from
jbd2_journal_restart(). Move the lockdep annotation to start_this_handle()
which covers both cases.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
fix jiffie rounding in jbd commit timer setup code. Rounding down
could cause the timer to be fired before the corresponding transaction
has expired. That transaction can stay not committed forever if no
new transaction is created or expicit sync/umount happens.
Signed-off-by: Alex Zhuravlev (Tomas) <alex.zhuravlev@sun.com>
Signed-off-by: Andreas Dilger <adilger@sun.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The function jbd2_journal_write_metadata_buffer() calls
jbd_unlock_bh_state(bh_in) too early; this could potentially allow
another thread to call get_write_access on the buffer head, modify the
data, and dirty it, and allowing the wrong data to be written into the
journal. Fortunately, if we lose this race, the only time this will
actually cause filesystem corruption is if there is a system crash or
other unclean shutdown of the system before the next commit can take
place.
Signed-off-by: dingdinghua <dingdinghua85@gmail.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Due to on disk corruption, it can happen that journal is too short. Fail
to load it in such case so that we don't oops somewhere later.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
This patch reverts 3f31fddf, which is no longer needed because if a
race between freeing buffer and committing transaction functionality
occurs and dio gets error, currently dio falls back to buffered IO due
to the commit 6ccfa806.
Signed-off-by: Hisashi Hifumi <hifumi.hisashi@oss.ntt.co.jp>
Cc: Mingming Cao <cmm@us.ibm.com>
Acked-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The following race can happen:
CPU1 CPU2
checkpointing code checks the buffer, adds
it to an array for writeback
do_get_write_access()
...
lock_buffer()
unlock_buffer()
flush_batch() submits the buffer for IO
__jbd2_journal_file_buffer()
So a buffer under writeout is returned from
do_get_write_access(). Since the filesystem code relies on the fact
that journaled buffers cannot be written out, it does not take the
buffer lock and so it can modify buffer while it is under
writeout. That can lead to a filesystem corruption if we crash at the
right moment.
We fix the problem by clearing the buffer dirty bit under buffer_lock
even if the buffer is on BJ_None list. Actually, we clear the dirty
bit regardless the list the buffer is in and warn about the fact if
the buffer is already journalled.
Thanks for spotting the problem goes to dingdinghua <dingdinghua85@gmail.com>.
Reported-by: dingdinghua <dingdinghua85@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
The revoke records must be written using the same way as the rest of
the blocks during the commit process; that is, either marked as
synchronous writes or as asynchornous writes.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
When you are going to be submitting several sync writes, we want to
give the IO scheduler a chance to merge some of them. Instead of
using the implicitly unplugging WRITE_SYNC variant, use WRITE_SYNC_PLUG
and rely on sync_buffer() doing the unplug when someone does a
wait_on_buffer()/lock_buffer().
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Update information about locking in JBD2 revoke code. Inconsistency in
comments found by Lin Tan <tammy000@gmail.com>.
CC: Lin Tan <tammy000@gmail.com>.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
If a commit is triggered by fsync(), set a flag indicating the journal
blocks associated with the transaction should be flushed out using
WRITE_SYNC.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
If we race with commit code setting i_transaction to NULL, we could
possibly dereference it. Proper locking requires the journal pointer
(to access journal->j_list_lock), which we don't have. So we have to
change the prototype of the function so that filesystem passes us the
journal pointer. Also add a more detailed comment about why the
function jbd2_journal_begin_ordered_truncate() does what it does and
how it should be used.
Thanks to Dan Carpenter <error27@gmail.com> for pointing to the
suspitious code.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Acked-by: Joel Becker <joel.becker@oracle.com>
CC: linux-ext4@vger.kernel.org
CC: ocfs2-devel@oss.oracle.com
CC: mfasheh@suse.de
CC: Dan Carpenter <error27@gmail.com>
The function jbd2_journal_start_commit() returns 1 if either a
transaction is committing or the function has queued a transaction
commit. But it returns 0 if we raced with somebody queueing the
transaction commit as well. This resulted in ext4_sync_fs() not
functioning correctly (description from Arthur Jones):
In the case of a data=ordered umount with pending long symlinks
which are delayed due to a long list of other I/O on the backing
block device, this causes the buffer associated with the long
symlinks to not be moved to the inode dirty list in the second
phase of fsync_super. Then, before they can be dirtied again,
kjournald exits, seeing the UMOUNT flag and the dirty pages are
never written to the backing block device, causing long symlink
corruption and exposing new or previously freed block data to
userspace.
This can be reproduced with a script created by Eric Sandeen
<sandeen@redhat.com>:
#!/bin/bash
umount /mnt/test2
mount /dev/sdb4 /mnt/test2
rm -f /mnt/test2/*
dd if=/dev/zero of=/mnt/test2/bigfile bs=1M count=512
touch /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename
ln -s /mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename
/mnt/test2/link
umount /mnt/test2
mount /dev/sdb4 /mnt/test2
ls /mnt/test2/
This patch fixes jbd2_journal_start_commit() to always return 1 when
there's a transaction committing or queued for commit.
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
CC: Eric Sandeen <sandeen@redhat.com>
CC: linux-ext4@vger.kernel.org
the following warning:
fs/jbd2/journal.c: In function ‘jbd2_seq_info_show’:
fs/jbd2/journal.c:850: warning: format ‘%lu’ expects type ‘long
unsigned int’, but argument 3 has type ‘uint32_t’
is caused by wrong usage of do_div that modifies the dividend in-place
and returns the quotient. So not only would an incorrect value be
displayed, but s->journal->j_average_commit_time would also be changed
to a wrong value!
Fix it by using div_u64 instead.
Signed-off-by: Simon Holm Thøgersen <odie@cs.aau.dk>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Christoph Hellwig