Theodore Ts'o [Wed, 20 Mar 2013 13:42:11 +0000 (09:42 -0400)]
ext4: fix data=journal fast mount/umount hang
In data=journal mode, if we unmount the file system before a
transaction has a chance to complete, when the journal inode is being
evicted, we can end up calling into jbd2_log_wait_commit() for the
last transaction, after the journalling machinery has been shut down.
Arguably we should adjust ext4_should_journal_data() to return FALSE
for the journal inode, but the only place it matters is
ext4_evict_inode(), and so to save a bit of CPU time, and to make the
patch much more obviously correct by inspection(tm), we'll fix it by
explicitly not trying to waiting for a journal commit when we are
evicting the journal inode, since it's guaranteed to never succeed in
this case.
This can be easily replicated via:
mount -t ext4 -o data=journal /dev/vdb /vdb ; umount /vdb
Theodore Ts'o [Wed, 20 Mar 2013 13:39:42 +0000 (09:39 -0400)]
ext4: fix ext4_evict_inode() racing against workqueue processing code
Commit 84c17543ab56 (ext4: move work from io_end to inode) triggered a
regression when running xfstest #270 when the file system is mounted
with dioread_nolock.
The problem is that after ext4_evict_inode() calls ext4_ioend_wait(),
this guarantees that last io_end structure has been freed, but it does
not guarantee that the workqueue structure, which was moved into the
inode by commit 84c17543ab56, is actually finished. Once
ext4_flush_completed_IO() calls ext4_free_io_end() on CPU #1, this
will allow ext4_ioend_wait() to return on CPU #2, at which point the
evict_inode() codepath can race against the workqueue code on CPU #1
accessing EXT4_I(inode)->i_unwritten_work to find the next item of
work to do.
Fix this by calling cancel_work_sync() in ext4_ioend_wait(), which
will be renamed ext4_ioend_shutdown(), since it is only used by
ext4_evict_inode(). Also, move the call to ext4_ioend_shutdown()
until after truncate_inode_pages() and filemap_write_and_wait() are
called, to make sure all dirty pages have been written back and
flushed from the page cache first.
Lukas Czerner [Tue, 12 Mar 2013 16:40:04 +0000 (12:40 -0400)]
ext4: use s_extent_max_zeroout_kb value as number of kb
Currently when converting extent to initialized, we have to decide
whether to zeroout part/all of the uninitialized extent in order to
avoid extent tree growing rapidly.
The decision is made by comparing the size of the extent with the
configurable value s_extent_max_zeroout_kb which is in kibibytes units.
However when converting it to number of blocks we currently use it as it
was in bytes. This is obviously bug and it will result in ext4 _never_
zeroout extents, but rather always split and convert parts to
initialized while leaving the rest uninitialized in default setting.
Fix this by using s_extent_max_zeroout_kb as kibibytes.
Theodore Ts'o [Tue, 12 Mar 2013 03:39:59 +0000 (23:39 -0400)]
ext4: use atomic64_t for the per-flexbg free_clusters count
A user who was using a 8TB+ file system and with a very large flexbg
size (> 65536) could cause the atomic_t used in the struct flex_groups
to overflow. This was detected by PaX security patchset:
Jan Kara [Mon, 11 Mar 2013 17:24:56 +0000 (13:24 -0400)]
jbd2: fix use after free in jbd2_journal_dirty_metadata()
jbd2_journal_dirty_metadata() didn't get a reference to journal_head it
was working with. This is OK in most of the cases since the journal head
should be attached to a transaction but in rare occasions when we are
journalling data, __ext4_journalled_writepage() can race with
jbd2_journal_invalidatepage() stripping buffers from a page and thus
journal head can be freed under hands of jbd2_journal_dirty_metadata().
Fix the problem by getting own journal head reference in
jbd2_journal_dirty_metadata() (and also in jbd2_journal_set_triggers()
which can possibly have the same issue).
Reported-by: Zheng Liu <gnehzuil.liu@gmail.com> Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: stable@vger.kernel.org
Lukas Czerner [Mon, 11 Mar 2013 02:50:00 +0000 (22:50 -0400)]
ext4: reserve metadata block for every delayed write
Currently we only reserve space (data+metadata) in delayed allocation if
we're allocating from new cluster (which is always in non-bigalloc file
system) which is ok for data blocks, because we reserve the whole cluster.
However we have to reserve metadata for every delayed block we're going
to write because every block could potentially require metedata block
when we need to grow the extent tree.
Lukas Czerner [Mon, 11 Mar 2013 02:46:30 +0000 (22:46 -0400)]
ext4: update reserved space after the 'correction'
Currently in ext4_ext_map_blocks() in delayed allocation writeback
we would update the reservation and after that check whether we claimed
cluster outside of the range of the allocation and if so, we'll give the
block back to the reservation pool.
However this also means that if the number of reserved data block
dropped to zero before the correction, we would release all the metadata
reservation as well, however we might still need it because the we're
not done with the delayed allocation and there might be more blocks to
come. This will result in error messages such as:
EXT4-fs warning (device sdb): ext4_da_update_reserve_space:361: ino 12,
allocated 1 with only 0 reserved metadata blocks (releasing 1 blocks
with reserved 1 data blocks)
This will only happen on bigalloc file system and it can be easily
reproduced using fiemap-tester from xfstests like this:
Jan Kara [Mon, 11 Mar 2013 02:19:00 +0000 (22:19 -0400)]
ext4: fix WARN_ON from ext4_releasepage()
ext4_releasepage() warns when it is passed a page with PageChecked set.
However this can correctly happen when invalidate_inode_pages2_range()
invalidates pages - and we should fail the release in that case. Since
the page was dirty anyway, it won't be discarded and no harm has
happened but it's good to be safe. Also remove bogus page_has_buffers()
check - we are guaranteed page has buffers in this function.
Reported-by: Zheng Liu <gnehzuil.liu@gmail.com> Tested-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Signed-off-by: Jan Kara <jack@suse.cz>
Zheng Liu [Mon, 11 Mar 2013 01:20:23 +0000 (21:20 -0400)]
ext4: fix the wrong number of the allocated blocks in ext4_split_extent()
This commit fixes a wrong return value of the number of the allocated
blocks in ext4_split_extent. When the length of blocks we want to
allocate is greater than the length of the current extent, we return a
wrong number. Let's see what happens in the following case when we
call ext4_split_extent().
map: [48, 72]
ex: [32, 64, u]
'ex' will be split into two parts:
ex1: [32, 47, u]
ex2: [48, 64, w]
'map->m_len' is returned from this function, and the value is 24. But
the real length is 16. So it should be fixed.
Meanwhile in this commit we use right length of the allocated blocks
when get_reserved_cluster_alloc in ext4_ext_handle_uninitialized_extents
is called.
Zheng Liu [Mon, 11 Mar 2013 01:13:05 +0000 (21:13 -0400)]
ext4: update extent status tree after an extent is zeroed out
When we try to split an extent, this extent could be zeroed out and mark
as initialized. But we don't know this in ext4_map_blocks because it
only returns a length of allocated extent. Meanwhile we will mark this
extent as uninitialized because we only check m_flags.
This commit update extent status tree when we try to split an unwritten
extent. We don't need to worry about the status of this extent because
we always mark it as initialized.
Zheng Liu [Mon, 11 Mar 2013 01:08:52 +0000 (21:08 -0400)]
ext4: fix wrong m_len value after unwritten extent conversion
The ext4_ext_handle_uninitialized_extents() function was assuming the
return value of ext4_ext_map_blocks() is equal to map->m_len. This
incorrect assumption was harmless until we started use status tree as
a extent cache because we need to update status tree according to
'm_len' value.
Meanwhile this commit marks EXT4_MAP_MAPPED flag after unwritten extent
conversion. It shouldn't cause a bug because we update status tree
according to checking EXT4_MAP_UNWRITTEN flag. But it should be fixed.
After applied this commit, the following error message from self-testing
infrastructure disappears.
...
kernel: ES len assertation failed for inode: 230 retval 1 !=
map->m_len 3 in ext4_map_blocks (allocation)
...
Dmitry Monakhov [Mon, 11 Mar 2013 01:01:03 +0000 (21:01 -0400)]
ext4: add self-testing infrastructure to do a sanity check
This commit adds a self-testing infrastructure like extent tree does to
do a sanity check for extent status tree. After status tree is as a
extent cache, we'd better to make sure that it caches right result.
After applied this commit, we will get a lot of messages when we run
xfstests as below.
...
kernel: ES len assertation failed for inode: 230 retval 1 != map->m_len
3 in ext4_map_blocks (allocation)
...
kernel: ES cache assertation failed for inode: 230 es_cached ex
[974/2/4781/20] != found ex [974/1/4781/1000]
...
kernel: ES insert assertation failed for inode: 635 ex_status
[0/45/21388/w] != es_status [44/1/21432/u]
...
Dmitry Monakhov [Mon, 4 Mar 2013 05:50:47 +0000 (00:50 -0500)]
ext4: invalidate extent status tree during extent migration
mext_replace_branches() will change inode's extents layout so
we have to drop corresponding cache.
TESTCASE: 301'th xfstest was not yet accepted to official xfstest's branch
and can be found here: https://github.com/dmonakhov/xfstests/commit/7b7efeee30a41109201e2040034e71db9b66ddc0
Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>
Jan Kara [Mon, 4 Mar 2013 05:43:32 +0000 (00:43 -0500)]
ext4: remove unnecessary wait for extent conversion in ext4_fallocate()
Now that we don't merge uninitialized extents anymore,
ext4_fallocate() is free to operate on the inode while there are still
some extent conversions pending - it won't disturb them in any way.
Reviewed-by: Zheng Liu <wenqing.lz@taobao.com> Reviewed-by: Dmitry Monakhov <dmonakhov@openvz.org> Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Dmitry Monakhov [Mon, 4 Mar 2013 05:41:05 +0000 (00:41 -0500)]
ext4: add warning to ext4_convert_unwritten_extents_endio
Splitting extents inside endio is a bad thing, but unfortunately it is
still possible. In fact we are pretty close to the moment when all
related issues will be fixed. Let's warn developer if it still the
case.
Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>
Dmitry Monakhov [Mon, 4 Mar 2013 05:36:06 +0000 (00:36 -0500)]
ext4: disable merging of uninitialized extents
Derived from Jan's patch:http://permalink.gmane.org/gmane.comp.file-systems.ext4/36470
Merging of uninitialized extents creates all sorts of interesting race
possibilities when writeback / DIO races with fallocate. Thus
ext4_convert_unwritten_extents_endio() has to deal with a case where
extent to be converted needs to be split out first. That isn't nice
for two reasons:
1) It may need allocation of extent tree block so ENOSPC is possible.
2) It complicates end_io handling code
So we disable merging of uninitialized extents which allows us to simplify
the code. Extents will get merged after they are converted to initialized
ones.
Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>
Dmitry Monakhov [Mon, 4 Mar 2013 05:34:34 +0000 (00:34 -0500)]
ext4: ext4_split_extent should take care of extent zeroout
When ext4_split_extent_at() ends up doing zeroout & conversion to
initialized instead of split & conversion, ext4_split_extent() gets
confused and can wrongly mark the extent back as uninitialized
resulting in end IO code getting confused from large unwritten extents
and may result in data loss.
The example of problematic behavior is:
lblk len lblk len
ext4_split_extent() (ex=[1000,30,uninit], map=[1010,10])
ext4_split_extent_at() (split [1000,30,uninit] at 1020)
ext4_ext_insert_extent() -> ENOSPC
ext4_ext_zeroout()
-> extent [1000,30] is now initialized
ext4_split_extent_at() (split [1000,30,init] at 1010,
MARK_UNINIT1 | MARK_UNINIT2)
-> extent is split and parts marked as uninitialized
Fix the problem by rechecking extent type after the first
ext4_split_extent_at() returns. None of split_flags can not be applied
to initialized extent so this patch also add BUG_ON to prevent similar
issues in future.
Jan Kara [Sat, 2 Mar 2013 23:22:38 +0000 (18:22 -0500)]
ext4: enable quotas before orphan cleanup
When using quota feature we need to enable quotas before orphan cleanup
so that changes happening during it are properly reflected in quota
accounting.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Sat, 2 Mar 2013 22:57:08 +0000 (17:57 -0500)]
ext4: don't allow quota mount options when quota feature enabled
So far we silently ignored when quota mount options were set while quota
feature was enabled. But this can create confusion in userspace when
mount options are set but silently ignored and also creates opportunities
for bugs when we don't properly test all quota types. Actually
ext4_mark_dquot_dirty() forgets to test for quota feature so it was
dependent on journaled quota options being set. OTOH ext4_orphan_cleanup()
tries to enable journaled quota when quota options are specified which is
wrong when quota feature is enabled.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Zheng Liu [Sat, 2 Mar 2013 22:24:05 +0000 (17:24 -0500)]
ext4: fix a warning from sparse check for ext4_dir_llseek
ext4_dir_llseek is only used as a callback function, and no one calls
it directly. So make it as a static function in order to remove a
warning message from sparse check.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Theodore Ts'o [Sat, 2 Mar 2013 15:27:46 +0000 (10:27 -0500)]
ext4: use percpu counter for extent cache count
Use a percpu counter rather than atomic types for shrinker accounting.
There's no need for ultimate accuracy in the shrinker, so this
should come a little more cheaply. The percpu struct is somewhat
large, but there was a big gap before the cache-aligned
s_es_lru_lock anyway, and it fits nicely in there.
Signed-off-by: Eric Sandeen <sandeen@redhat.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Theodore Ts'o [Fri, 1 Mar 2013 04:58:56 +0000 (23:58 -0500)]
ext4: optimize ext4_es_shrink()
When the system is under memory pressure, ext4_es_srhink() will get
called very often. So optimize returning the number of items in the
file system's extent status cache by keeping a per-filesystem count,
instead of calculating it each time by scanning all of the inodes in
the extent status cache.
Also rename the slab used for the extent status cache to be
"ext4_extent_status" so it's obviousl the slab in question is created
by ext4.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: Zheng Liu <gnehzuil.liu@gmail.com>
Theodore Ts'o [Wed, 27 Feb 2013 19:54:37 +0000 (14:54 -0500)]
ext4: fix extent status tree regression for file systems > 512GB
This fixes a regression introduced by commit f7fec032aa782. The
problem was that the extents status flags caused us to mask out block
numbers smaller than 2**28 blocks. Since we didn't test with file
systems smaller than 512GB, we didn't notice this during the
development cycle.
A typical failure looks like this:
EXT4-fs error (device sdb1): htree_dirblock_to_tree:919: inode #172235804: block 152052301: comm ls: bad entry in directory: rec_len is smaller than minimal -
offset=0(0), inode=0, rec_len=0, name_len=0
... where 'debugfs -R "stat <172235804>" /dev/sdb1' reports that the
inode has block number 688923213. When viewed in hex, block number 152052301 (from the syslog) is 0x910224D, while block number 688923213
is 0x2910224D. Note the missing "0x20000000" in the block number.
Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de> Verified-by: Markus Trippelsdorf <markus@trippelsdorf.de> Reported-by: Dave Jones <davej@redhat.com> Verified-by: Dave Jones <davej@redhat.com> Cc: Zheng Liu <gnehzuil.liu@gmail.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Lukas Czerner [Fri, 22 Feb 2013 20:27:52 +0000 (15:27 -0500)]
ext4: fix free clusters calculation in bigalloc filesystem
ext4_has_free_clusters() should tell us whether there is enough free
clusters to allocate, however number of free clusters in the file system
is converted to blocks using EXT4_C2B() which is not only wrong use of
the macro (we should have used EXT4_NUM_B2C) but it's also completely
wrong concept since everything else is in cluster units.
Moreover when calculating number of root clusters we should be using
macro EXT4_NUM_B2C() instead of EXT4_B2C() otherwise the result might be
off by one. However r_blocks_count should always be a multiple of the
cluster ratio so doing a plain bit shift should be enough here. We
avoid using EXT4_B2C() because it's confusing.
As a result of the first problem number of free clusters is much bigger
than it should have been and ext4_has_free_clusters() would return 1 even
if there is really not enough free clusters available.
Fix this by removing the EXT4_C2B() conversion of free clusters and
using bit shift when calculating number of root clusters. This bug
affects number of xfstests tests covering file system ENOSPC situation
handling. With this patch most of the ENOSPC problems with bigalloc file
system disappear, especially the errors caused by delayed allocation not
having enough space when the actual allocation is finally requested.
Lukas Czerner [Mon, 18 Feb 2013 17:12:07 +0000 (12:12 -0500)]
ext4: fix xattr block allocation/release with bigalloc
Currently when new xattr block is created or released we we would call
dquot_free_block() or dquot_alloc_block() respectively, among the else
decrementing or incrementing the number of blocks assigned to the
inode by one block.
This however does not work for bigalloc file system because we always
allocate/free the whole cluster so we have to count with that in
dquot_free_block() and dquot_alloc_block() as well.
Use the clusters-to-blocks conversion EXT4_C2B() when passing number of
blocks to the dquot_alloc/free functions to fix the problem.
The problem has been revealed by xfstests #117 (and possibly others).
Zheng Liu [Mon, 18 Feb 2013 05:32:55 +0000 (00:32 -0500)]
ext4: reclaim extents from extent status tree
Although extent status is loaded on-demand, we also need to reclaim
extent from the tree when we are under a heavy memory pressure because
in some cases fragmented extent tree causes status tree costs too much
memory.
Here we maintain a lru list in super_block. When the extent status of
an inode is accessed and changed, this inode will be move to the tail
of the list. The inode will be dropped from this list when it is
cleared. In the inode, a counter is added to count the number of
cached objects in extent status tree. Here only written/unwritten/hole
extent is counted because delayed extent doesn't be reclaimed due to
fiemap, bigalloc and seek_data/hole need it. The counter will be
increased as a new extent is allocated, and it will be decreased as a
extent is freed.
In this commit we use normal shrinker framework to reclaim memory from
the status tree. ext4_es_reclaim_extents_count() traverses the lru list
to count the number of reclaimable extents. ext4_es_shrink() tries to
reclaim written/unwritten/hole extents from extent status tree. The
inode that has been shrunk is moved to the tail of lru list.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan kara <jack@suse.cz>
Zheng Liu [Mon, 18 Feb 2013 05:29:59 +0000 (00:29 -0500)]
ext4: lookup block mapping in extent status tree
After tracking all extent status, we already have a extent cache in
memory. Every time we want to lookup a block mapping, we can first
try to lookup it in extent status tree to avoid a potential disk I/O.
A new function called ext4_es_lookup_extent is defined to finish this
work. When we try to lookup a block mapping, we always call
ext4_map_blocks and/or ext4_da_map_blocks. So in these functions we
first try to lookup a block mapping in extent status tree.
A new flag EXT4_GET_BLOCKS_NO_PUT_HOLE is used in ext4_da_map_blocks
in order not to put a hole into extent status tree because this hole
will be converted to delayed extent in the tree immediately.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan kara <jack@suse.cz>
Zheng Liu [Mon, 18 Feb 2013 05:28:47 +0000 (00:28 -0500)]
ext4: track all extent status in extent status tree
By recording the phycisal block and status, extent status tree is able
to track the status of every extents. When we call _map_blocks
functions to lookup an extent or create a new written/unwritten/delayed
extent, this extent will be inserted into extent status tree.
We don't load all extents from disk in alloc_inode() because it costs
too much memory, and if a file is opened and closed frequently it will
takes too much time to load all extent information. So currently when
we create/lookup an extent, this extent will be inserted into extent
status tree. Hence, the extent status tree may not comprehensively
contain all of the extents found in the file.
Here a condition we need to take care is that an extent might contains
unwritten and delayed status simultaneously because an extent is delayed
allocated and could be allocated by fallocate. At this time we need to
keep delayed status because later we need to update delayed reservation
space using it.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan kara <jack@suse.cz>
Zheng Liu [Mon, 18 Feb 2013 05:28:04 +0000 (00:28 -0500)]
ext4: let ext4_ext_map_blocks return EXT4_MAP_UNWRITTEN flag
This commit lets ext4_ext_map_blocks return EXT4_MAP_UNWRITTEN flag
because in later commit ext4_map_blocks needs to use this flag to
determine the extent status.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>
Zheng Liu [Mon, 18 Feb 2013 05:27:26 +0000 (00:27 -0500)]
ext4: rename and improbe ext4_es_find_extent()
This commit renames ext4_es_find_extent with ext4_es_find_delayed_extent
and improve this function. First, we split input and output parameter.
Second, this function never return the first block of the next delayed
extent after 'es'.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: Jan kara <jack@suse.cz>
Zheng Liu [Mon, 18 Feb 2013 05:26:51 +0000 (00:26 -0500)]
ext4: add physical block and status member into extent status tree
This commit adds two members in extent_status structure to let it record
physical block and extent status. Here es_pblk is used to record both
of them because physical block only has 48 bits. So extent status could
be stashed into it so that we can save some memory. Now written,
unwritten, delayed and hole are defined as status.
Due to new member is added into extent status tree, all interfaces need
to be adjusted.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>
Zheng Liu [Mon, 18 Feb 2013 05:26:51 +0000 (00:26 -0500)]
ext4: refine extent status tree
This commit refines the extent status tree code.
1) A prefix 'es_' is added to to the extent status tree structure
members.
2) Refactored es_remove_extent() so that __es_remove_extent() can be
used by es_insert_extent() to remove the old extent entry(-ies) before
inserting a new one.
3) Rename extent_status_end() to ext4_es_end()
4) ext4_es_can_be_merged() is define to check whether two extents can
be merged or not.
5) Update and clarified comments.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>
Theodore Ts'o [Fri, 15 Feb 2013 04:59:26 +0000 (23:59 -0500)]
ext4: refactor code to read directory blocks into ext4_read_dirblock()
The code to read in directory blocks and verify their metadata
checksums was replicated in ten different places across
fs/ext4/namei.c, and the code was buggy in subtle ways in a number of
those replicated sites. In some cases, ext4_error() was called with a
training newline. In others, in particularly in empty_dir(), it was
possible to call ext4_dirent_csum_verify() on an index block, which
would trigger false warnings requesting the system adminsitrator to
run e2fsck.
By refactoring the code, we make the code more readable, as well as
shrinking the compiled object file by over 700 bytes and 50 lines of
code.
Theodore Ts'o [Thu, 14 Feb 2013 20:11:41 +0000 (15:11 -0500)]
ext4: use KERN_WARNING for warning messages
Some messages printed related to a WARN_ON(1) were printed using
KERN_NOTICE. Use KERN_WARNING or ext4_warning() instead so that
context related to the WARN_ON() is printed at the same printk warning
level (and log files, etc.)
Theodore Ts'o [Sat, 9 Feb 2013 21:29:20 +0000 (16:29 -0500)]
jbd2: use module parameters instead of debugfs for jbd_debug
There are multiple reasons to move away from debugfs. First of all,
we are only using it for a single parameter, and it is much more
complicated to set up (some 30 lines of code compared to 3), and one
more thing that might fail while loading the jbd2 module.
Secondly, as a module paramter it can be specified as a boot option if
jbd2 is built into the kernel, or as a parameter when the module is
loaded, and it can also be manipulated dynamically under
/sys/module/jbd2/parameters/jbd2_debug. So it is more flexible.
Ultimately we want to move away from using jbd_debug() towards
tracepoints, but for now this is still a useful simplification of the
code base.
Theodore Ts'o [Sat, 9 Feb 2013 21:28:20 +0000 (16:28 -0500)]
ext4: use module parameters instead of debugfs for mballoc_debug
There are multiple reasons to move away from debugfs. First of all,
we are only using it for a single parameter, and it is much more
complicated to set up (some 30 lines of code compared to 3), and one
more thing that might fail while loading the ext4 module.
Secondly, as a module paramter it can be specified as a boot option if
ext4 is built into the kernel, or as a parameter when the module is
loaded, and it can also be manipulated dynamically under
/sys/module/ext4/parameters/mballoc_debug. So it is more flexible.
Ultimately we want to move away from using mb_debug() towards
tracepoints, but for now this is still a useful simplification of the
code base.
Theodore Ts'o [Sat, 9 Feb 2013 21:27:09 +0000 (16:27 -0500)]
ext4: start handle at the last possible moment when creating inodes
In ext4_{create,mknod,mkdir,symlink}(), don't start the journal handle
until the inode has been succesfully allocated. In order to do this,
we need to start the handle in the ext4_new_inode(). So create a new
variant of this function, ext4_new_inode_start_handle(), so the handle
can be created at the last possible minute, before we need to modify
the inode allocation bitmap block.
Theodore Ts'o [Sat, 9 Feb 2013 20:23:03 +0000 (15:23 -0500)]
ext4: fix the number of credits needed for acl ops with inline data
Operations which modify extended attributes may need extra journal
credits if inline data is used, since there is a chance that some
extended attributes may need to get pushed to an external attribute
block.
Changes to reflect this was made in xattr.c, but they were missed in
fs/ext4/acl.c. To fix this, abstract the calculation of the number of
credits needed for xattr operations to an inline function defined in
ext4_jbd2.h, and use it in acl.c and xattr.c.
Also move the function declarations used in inline.c from xattr.h
(where they are non-obviously hidden, and caused problems since
ext4_jbd2.h needs to use the function ext4_has_inline_data), and move
them to ext4.h.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Tao Ma <boyu.mt@taobao.com> Reviewed-by: Jan Kara <jack@suse.cz>
Theodore Ts'o [Sat, 9 Feb 2013 20:06:24 +0000 (15:06 -0500)]
ext4: fix the number of credits needed for ext4_unlink() and ext4_rmdir()
The ext4_unlink() and ext4_rmdir() don't actually release the blocks
associated with the file/directory. This gets done in a separate jbd2
handle called via ext4_evict_inode(). Thus, we don't need to reserve
lots of journal credits for the truncate.
Note that using too many journal credits is non-optimal because it can
leading to the journal transmit getting closed too early, before it is
strictly necessary.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>
Theodore Ts'o [Sat, 9 Feb 2013 17:50:27 +0000 (12:50 -0500)]
ext4: fix the number of credits needed for ext4_ext_migrate()
The migration ioctl creates a temporary inode. Since this inode is
never linked to a directory, we don't need to reserve journal credits
required for modifying the directory.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>
Theodore Ts'o [Sat, 9 Feb 2013 14:24:14 +0000 (09:24 -0500)]
ext4: grab page before starting transaction handle in write_begin()
The grab_cache_page_write_begin() function can potentially sleep for a
long time, since it may need to do memory allocation which can block
if the system is under significant memory pressure, and because it may
be blocked on page writeback. If it does take a long time to grab the
page, it's better that we not hold an active jbd2 handle.
So grab a handle on the page first, and _then_ start the transaction
handle.
This commit fixes the following long transaction handle hold time:
postmark-2917 [000] .... 196.435786: jbd2_handle_stats: dev 254,32
tid 570 type 2 line_no 2541 interval 311 sync 0 requested_blocks 1
dirtied_blocks 0
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>
Theodore Ts'o [Sat, 9 Feb 2013 02:59:22 +0000 (21:59 -0500)]
ext4: pass context information to jbd2__journal_start()
So we can better understand what bits of ext4 are responsible for
long-running jbd2 handles, use jbd2__journal_start() so we can pass
context information for logging purposes.
The recommended way for finding the longer-running handles is:
This will list handles that were active for longer than 20ms. Having
longer-running handles is bad, because a commit started at the wrong
time could stall for those 20+ milliseconds, which could delay an
fsync() or an O_SYNC operation. Here is an example line from the
trace file describing a handle which lived on for 311 jiffies, or over
1.2 seconds:
postmark-2917 [000] .... 196.435786: jbd2_handle_stats: dev 254,32
tid 570 type 2 line_no 2541 interval 311 sync 0 requested_blocks 1
dirtied_blocks 0
Theodore Ts'o [Fri, 8 Feb 2013 18:00:22 +0000 (13:00 -0500)]
jbd2: add tracepoints which provide per-handle statistics
Handles which stay open a long time are problematic when it comes time
to close down a transaction so it can be committed. These tracepoints
will help us determine which ones are the problematic ones, and to
validate whether changes makes things better or worse.
Theodore Ts'o [Thu, 7 Feb 2013 03:30:23 +0000 (22:30 -0500)]
jbd2: track request delay statistics
Track the delay between when we first request that the commit begin
and when it actually begins, so we can see how much of a gap exists.
In theory, this should just be the remaining scheduling quantuum of
the thread which requested the commit (assuming it was not a
synchronous operation which triggered the commit request) plus
scheduling overhead; however, it's possible that real time processes
might get in the way of letting the kjournald thread from executing.
Theodore Ts'o [Mon, 4 Feb 2013 20:08:40 +0000 (15:08 -0500)]
ext4: optimize mballoc for large allocations
The ext4 block allocator only maintains buddy bitmaps for chunks which
are less than or equal to one quarter of a block group. That is, for
a file aystem with a 1k blocksize, and where the number of blocks in a
block group is 8192 blocks, the largest chunk size tracked by buddy
bitmaps is 2048 blocks.
For a file system with a 4k blocksize, and where the number of blocks
in a block group is 32768 blocks, the largest chunk size tracked by
buddy bitmaps is 8192 blocks.
To work around this code, mballoc.c before this commit would truncate
allocation requests to the number of blocks in a block group minus 10.
Why 10? Aside from being a completely arbitrary number, it avoids
block allocation to be a power of two larger than 25% of the block
group. If you try to explicitly fallocate 50% of the block group
size, this will demonstrate the problem; the block allocation code
will scan the all of the blocks in the file system with cr==0 (since
the request is for a natural power of two), but then completely fail
for all blocks groups, since the buddy bitmaps don't track chunk sizes
of 50% of the block group.
To fix this, in these we use ext4_mb_complex_scan_group() instead of
ext4_mb_simple_scan_group().
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger@dilger.ca>
Jan Kara [Sun, 3 Feb 2013 04:09:36 +0000 (23:09 -0500)]
ext4: make mount option parsing loop more logical
The loop looking for correct mount option entry is more logical if it is
written rewritten as an empty loop looking for correct option entry and then
code handling the option. It also saves one level of indentation for a lot of
code so we can join a couple of split lines.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Sun, 3 Feb 2013 03:52:19 +0000 (22:52 -0500)]
ext4: move several mount options to standard handling loop
Several mount option (resuid, resgid, journal_dev, journal_ioprio) are
currently handled before we enter standard option handling loop. I don't
see a reason for this so move them to normal handling loop to make things
more regular.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Akria Fujita [Sat, 2 Feb 2013 01:52:46 +0000 (20:52 -0500)]
ext4: fix smatch warning in move_extent.c's mext_replace_branches()
Commit 2147b1a6a48 resulted in a new smatch warning:
> fs/ext4/move_extent.c:693 mext_replace_branches()
> warn: variable dereferenced before check 'dext' (see line 683)
Fix this by adding a check to make sure dext is non-NULL before we
derefrence it.
Signed-off-by: Akria Fujita <a-fujita@rs.jp.nec.com>
[ modified by tytso to make sure an ext4_error is called ] Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Eric Sandeen [Wed, 30 Jan 2013 05:39:28 +0000 (00:39 -0500)]
jbd2: don't wake kjournald unnecessarily
Don't send an extra wakeup to kjournald in the case where we
already have the proper target in j_commit_request, i.e. that
transaction has already been requested for commit.
commit deeeaf13 "jbd2: fix fsync() tid wraparound bug" changed
the logic leading to a wakeup, but it caused some extra wakeups
which were found to lead to a measurable performance regression.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
[tytso@mit.edu: reworked check to make it clearer] Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Wed, 30 Jan 2013 03:48:17 +0000 (22:48 -0500)]
ext4: fix possible use-after-free with AIO
Running AIO is pinning inode in memory using file reference. Once AIO
is completed using aio_complete(), file reference is put and inode can
be freed from memory. So we have to be sure that calling aio_complete()
is the last thing we do with the inode.
CC: stable@vger.kernel.org Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com> Acked-by: Jeff Moyer <jmoyer@redhat.com> Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Guo Chao [Tue, 29 Jan 2013 02:23:24 +0000 (21:23 -0500)]
ext4: release buffer when checksum failed
Commit b0336e8d (ext4: calculate and verify checksums of directory
leaf blocks) and commit dbe89444 (ext4: Calculate and verify checksums
for htree nodes) forget to release buffer when checksum failed, at
some places.
Signed-off-by: Guo Chao <yan@linux.vnet.ibm.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Lukas Czerner [Tue, 29 Jan 2013 02:21:12 +0000 (21:21 -0500)]
ext4: remove explicit WARN_ON when ext4_map_blocks() fails
In two places we call WARN_ON() before we print out the debug message,
however we agreed that the WARN_ON() is unnecessary at those places so
remove them.
Also use ext4_warning() instead of ext4_msg() and printk().
Jan Kara [Tue, 29 Jan 2013 02:06:42 +0000 (21:06 -0500)]
ext4: fix ext4_writepage() to achieve data=ordered guarantees
So far ext4_writepage() skipped writing pages that had any delayed or
unwritten buffers attached. When blocksize < pagesize this breaks
data=ordered mode guarantees as we can have a page with one freshly
allocated buffer whose allocation is part of the committing
transaction and another buffer in the page which is delayed or
unwritten. So fix this problem by calling ext4_bio_writepage()
anyway. It will submit mapped buffers and leave others alone.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Tue, 29 Jan 2013 01:53:28 +0000 (20:53 -0500)]
ext4: Make ext4_bio_writepage() handle unprepared buffers
So far ext4_bio_writepage() unconditionally cleared dirty bit on all
buffers underlying the page. That implicitely assumes we can write all
buffers. So far that is true because callers call into
ext4_bio_writepage() make sure all buffers in the page are mapped but:
a) it's a data corruption bug waiting to happen
b) in data=ordered mode when blocksize < pagesize we do need to write
pages that may have only some of dirty buffers mapped.
So change ext4_bio_writepage() to skip buffers that cannot be written without
clearing their dirty bit.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Mon, 28 Jan 2013 18:06:48 +0000 (13:06 -0500)]
ext4: simplify mpage_add_bh_to_extent()
The argument b_size of mpage_add_bh_to_extent() was bogus since it was
always == blocksize (which we can easily derive from inode->i_blkbits).
Also second branch of condition:
if (nrblocks >= EXT4_MAX_TRANS_DATA) {
} else if ((nrblocks + (b_size >> mpd->inode->i_blkbits)) >
EXT4_MAX_TRANS_DATA) {
}
was never taken because (b_size >> mpd->inode->i_blkbits) == 1.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Mon, 28 Jan 2013 17:55:08 +0000 (12:55 -0500)]
ext4: dirty page has always buffers attached
ext4_writepage(), write_cache_pages_da(), and mpage_da_submit_io()
doesn't have to deal with the case when page doesn't have buffers. We
attach buffers to a page in ->write_begin() and ->page_mkwrite() which
covers all places where a page can become dirty.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Mon, 28 Jan 2013 14:49:15 +0000 (09:49 -0500)]
ext4: simplify list handling in ext4_do_flush_completed_IO()
The function splices i_completed_io_list to its private list
first. From that moment on we don't need any lock for working with
io_end structures because all io_end structure on the list are only
our own. So we can remove the other two lists in the function and free
io_end immediately after we are done with it.
CC: Dmitry Monakhov <dmonakhov@openvz.org> Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Mon, 28 Jan 2013 14:43:46 +0000 (09:43 -0500)]
ext4: move work from io_end to inode
It does not make much sense to have struct work in ext4_io_end_t
because we always use it for only one ext4_io_end_t per inode (the
first one in the i_completed_io list). So just move the structure to
inode itself. This also allows for a small simplification in
processing io_end structures.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Mon, 28 Jan 2013 14:38:49 +0000 (09:38 -0500)]
ext4: remove __ext4_journalled_writepage() from mpage_da_submit_io()
We don't support delayed allocation in data=journal mode. So checking for it in
mpage_da_submit_io() doesn't make really sence. If we ever decide to extend
delayed allocation support to data=journal mode, adding
__ext4_journalled_writepage() call will be the least of problems we have to
solve. Most likely we'd have to implement separate writepages call anyways
because we don't have transaction credits for writing more than a single page
so mapping of page buffers would have to be done differently.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Mon, 28 Jan 2013 14:32:54 +0000 (09:32 -0500)]
ext4: use redirty_page_for_writepage() in ext4_bio_write_page()
When we cannot write a page we should use redirty_page_for_writepage()
instead of plain set_page_dirty(). That tells writeback code we have
problems, redirties only the page (redirtying buffers is not needed),
and updates mm accounting of failed page writes.
Also move clearing of buffer dirty flag after io_submit_add_bh(). At that
moment we are sure buffer will be going to disk.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Jan Kara [Mon, 28 Jan 2013 14:30:52 +0000 (09:30 -0500)]
ext4: Always use ext4_bio_write_page() for writeout
Currently we sometimes used block_write_full_page() and sometimes
ext4_bio_write_page() for writeback (depending on mount options and call
path). Let's always use ext4_bio_write_page() to simplify things a bit.
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>
Zheng Liu [Mon, 28 Jan 2013 14:21:37 +0000 (09:21 -0500)]
ext4: add punching hole support for non-extent-mapped files
This patch add supports for indirect file support punching hole. It
is almost the same as ext4_ext_punch_hole. First, we invalidate all
pages between this hole, and then we try to deallocate all blocks of
this hole.
A recursive function is used to handle deallocation of blocks. In
this function, it iterates over the entries in inode's i_blocks or
indirect blocks, and try to free the block for each one of them.
After applying this patch, xfstest #255 will not pass w/o extent because
indirect-based file doesn't support unwritten extents.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Theodore Ts'o [Fri, 25 Jan 2013 04:24:56 +0000 (23:24 -0500)]
quota: autoload the quota_v2 module for QFMT_VFS_V1 quota format
Otherwise, ext4 file systems with the quota featured enable will get a
very confusing "No such process" error message if the quota code is
built as a module and the quota_v2 module has not been loaded.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com> Acked-by: Jan Kara <jack@suse.cz> Cc: stable@vger.kernel.org
Theodore Ts'o [Sat, 12 Jan 2013 21:19:36 +0000 (16:19 -0500)]
ext4: return ENOMEM if sb_getblk() fails
The only reason for sb_getblk() failing is if it can't allocate the
buffer_head. So ENOMEM is more appropriate than EIO. In addition,
make sure that the file system is marked as being inconsistent if
sb_getblk() fails.
Linus Torvalds [Wed, 9 Jan 2013 16:58:57 +0000 (08:58 -0800)]
Merge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm
Pull ARM fixes from Russell King.
* 'fixes' of git://git.linaro.org/people/rmk/linux-arm:
ARM: 7616/1: cache-l2x0: aurora: Use writel_relaxed instead of writel
ARM: 7615/1: cache-l2x0: aurora: Invalidate during clean operation with WT enable
ARM: 7614/1: mm: fix wrong branch from Cortex-A9 to PJ4b
ARM: 7612/1: imx: Do not select some errata that depends on !ARCH_MULTIPLATFORM
ARM: 7611/1: VIC: fix bug in VIC irqdomain code
ARM: 7610/1: versatile: bump IRQ numbers
ARM: 7609/1: disable errata work-arounds which access secure registers
ARM: 7608/1: l2x0: Only set .set_debug on PL310 r3p0 and earlier
Linus Torvalds [Wed, 9 Jan 2013 16:43:56 +0000 (08:43 -0800)]
Merge tag 'edac_fixes_for_3.8' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp
Pull EDAC fixes from Borislav Petkov:
"Two error path fixes causing a crash and a Kconfig fix for an issue
which spilled all EDAC suboptions into the 'Device Drivers' menu."
* tag 'edac_fixes_for_3.8' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp:
EDAC: Cleanup device deregistering path
EDAC: Fix EDAC Kconfig menu
EDAC: Fix kernel panic on module unloading
The check for a pmd being in the process of being split was dropped by
mistake by commit d10e63f29488 ("mm: numa: Create basic numa page
hinting infrastructure"). Put it back.
Reported-by: Dave Jones <davej@redhat.com> Debugged-by: Hillf Danton <dhillf@gmail.com> Acked-by: Andrea Arcangeli <aarcange@redhat.com> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Kirill Shutemov <kirill@shutemov.name> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Marc Dionne [Wed, 9 Jan 2013 14:16:30 +0000 (14:16 +0000)]
cred: Remove tgcred pointer from struct cred
Commit 3a50597de863 ("KEYS: Make the session and process keyrings
per-thread") removed the definition of the thread_group_cred structure,
but left a now unused pointer in struct cred.
Signed-off-by: Marc Dionne <marc.c.dionne@gmail.com> Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>