This might be unexpected but pages allocated for sbi->s_buddy_cache are
charged to current memory cgroup. So, GFP_NOFS allocation could fail if
current task has been killed by OOM or if current memory cgroup has no
free memory left. Block allocator cannot handle such failures here yet.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

the error is:
    fs/ext4/mballoc.c:475:43: error: 'struct ext4_group_info' has
no member named 'bb_bitmap'.
    so, the definition of macro DOUBLE_CHECK should before
'struct ext4_group_info', I fixed it, and I moved the macro
AGGRESSIVE_CHECK together, because I think they shoule be together.

Signed-off-by: Aihua Zhang <zhangaihua1@huawei.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

If data_err=abort option is specified for an ext3/ext4 mount,
/proc/mounts does show it as "(null)". This is caused by token2str()
returning NULL for Opt_data_err_abort (due to its pattern containing
'=').

Signed-off-by: Ales Novak <alnovak@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

ext4_reserve_inode_write() in ext4_mark_inode_dirty() could fail on
error (e.g. EIO) and iloc.bh can be NULL in this case. But the error is
ignored in the following "if" condition and ext4_expand_extra_isize()
might be called with NULL iloc.bh set, which triggers NULL pointer
dereference.

This is uncovered by commit 8b4953e1

 ("ext4: reserve code points for
the project quota feature"), which enlarges the ext4_inode size, and
run the following script on new kernel but with old mke2fs:

  #/bin/bash
  mnt=/mnt/ext4
  devname=ext4-error
  dev=/dev/mapper/$devname
  fsimg=/home/fs.img

  trap cleanup 0 1 2 3 9 15

  cleanup()
  {
          umount $mnt >/dev/null 2>&1
          dmsetup remove $devname
          losetup -d $backend_dev
          rm -f $fsimg
          exit 0
  }

  rm -f $fsimg
  fallocate -l 1g $fsimg
  backend_dev=`losetup -f --show $fsimg`
  devsize=`blockdev --getsz $backend_dev`

  good_tab="0 $devsize linear $backend_dev 0"
  error_tab="0 $devsize error $backend_dev 0"

  dmsetup create $devname --table "$good_tab"

  mkfs -t ext4 $dev
  mount -t ext4 -o errors=continue,strictatime $dev $mnt

  dmsetup load $devname --table "$error_tab" && dmsetup resume $devname
  echo 3 > /proc/sys/vm/drop_caches
  ls -l $mnt
  exit 0

[ Patch changed to simplify the function a tiny bit. -- Ted ]

Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

BUFFER_TRACE info "call ext4_handle_dirty_metadata" doesn't match the
code, so drop it.

Signed-off-by: Geliang Tang <geliangtang@163.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Signed-off-by: Adam Buchbinder <adam.buchbinder@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

On umount path, jbd2_journal_destroy() writes latest transaction ID
(->j_tail_sequence) to be used at next mount.

The bug is that ->j_tail_sequence is not holding latest transaction ID
in some cases. So, at next mount, there is chance to conflict with
remaining (not overwritten yet) transactions.

	mount (id=10)
	write transaction (id=11)
	write transaction (id=12)
	umount (id=10) <= the bug doesn't write latest ID

	mount (id=10)
	write transaction (id=11)
	crash

	mount
	[recovery process]
		transaction (id=11)
		transaction (id=12) <= valid transaction ID, but old commit
                                       must not replay

Like above, this bug become the cause of recovery failure, or FS
corruption.

So why ->j_tail_sequence doesn't point latest ID?

Because if checkpoint transactions was reclaimed by memory pressure
(i.e. bdev_try_to_free_page()), then ->j_tail_sequence is not updated.
(And another case is, __jbd2_journal_clean_checkpoint_list() is called
with empty transaction.)

So in above cases, ->j_tail_sequence is not pointing latest
transaction ID at umount path. Plus, REQ_FLUSH for checkpoint is not
done too.

So, to fix this problem with minimum changes, this patch updates
->j_tail_sequence, and issue REQ_FLUSH.  (With more complex changes,
some optimizations would be possible to avoid unnecessary REQ_FLUSH
for example though.)

BTW,

	journal->j_tail_sequence =
		++journal->j_transaction_sequence;

Increment of ->j_transaction_sequence seems to be unnecessary, but
ext3 does this.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@vger.kernel.org

Using SEEK_DATA in a huge sparse file can easily lead to sotflockups as
ext4_seek_data() iterates hole block-by-block. Fix the problem by using
returned hole size from ext4_map_blocks() and thus skip the hole in one
go.

Update also SEEK_HOLE implementation to follow the same pattern as
SEEK_DATA to make future maintenance easier.

Furthermore we add cond_resched() to both ext4_seek_data() and
ext4_seek_hole() to avoid softlockups in case evil user creates huge
fragmented file and we have to go through lots of extents.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

ext4_dax_mmap_get_block() updates bh->b_state directly instead of using
ext4_update_bh_state(). This is mostly a cosmetic issue since DAX code
always passes on-stack buffer_head but clean this up to make code more
uniform.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Currently, ext4_map_blocks() just returns 0 when it finds a hole and
allocation is not requested. However we have all the information
available to tell how large the hole actually is and there are callers
of ext4_map_blocks() which would save some block-by-block hole iteration
if they knew this information. So fill in struct ext4_map_blocks even
for holes with the information we have. We keep returning 0 for holes to
maintain backward compatibility of the function.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

ext4_ext_put_gap_in_cache() determines hole size in the extent tree,
then trims this with possible delayed allocated blocks, and inserts the
result into the extent status tree. Factor out determination of the size
of the hole in the extent tree as we will need this information in
ext4_ext_map_blocks() as well.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

We were setting referenced bit on the extent structure we return from
ext4_es_lookup_extent() which is just a private structure on stack. Thus
setting had no effect. Set the bit in the structure in the status tree
instead.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Remove counter of pending io ends as it is unused.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

When mapping blocks for direct IO, we allocate io_end structure before
mapping blocks and store pointer to it in the inode. This creates a
requirement that any AIO DIO using io_end must be protected by i_mutex.
This created problems in the past with dioread_nolock mode which was
corrupting io_end pointers. Also io_end is allocated unnecessarily in
case where we don't need to convert any extents (which is a common case
for example when overwriting file).

We fix the problem by allocating io_end only once we return unwritten
extent from block mapping function for AIO DIO (so we can save some
pointless io_end allocations) and we pass pointer to it in bh->b_private
which generic DIO code later passes to our end IO callback. That way we
remove any need for global pointer to io_end structure and thus fix the
races.

The downside of this change is that the checking for unwritten IO in
flight in ext4_extents_can_be_merged() is more racy since we now
increment i_unwritten / set EXT4_STATE_DIO_UNWRITTEN only after dropping
i_data_sem. However the check has been racy already before because
ext4_writepages() already increment i_unwritten after dropping
i_data_sem and reserved blocks save us from hitting ENOSPC in the worst
case.

Signed-off-by: Jan Kara <jack@suse.cz>

There is no need to handle starting of a transaction and deferal of DIO
completion in _ext4_get_block() function. We can move this out to get
block functions for direct IO that need it. That way we can add stricter
checks verifying things work as we expect.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Rename ext4_get_blocks_write() to ext4_get_blocks_unwritten() to better
describe what it does. Also split out get blocks functions for direct
IO. Later we move functionality from _ext4_get_blocks() there. There's no
functional change in this patch.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Currently we've used hashed aio_mutex to serialize unaligned AIO DIO.
However the code cleanups that happened after 2011 when the lock was
introduced made aio_mutex acquired at almost the same places where we
already have exclusion using i_mutex. So just use i_mutex for the
exclusion of unaligned AIO DIO.

The change moves waiting for pending unwritten extent conversion under
i_mutex. That makes special handling of O_APPEND writes unnecessary and
also avoids possible livelocking of unaligned AIO DIO with aligned one
(nothing was preventing contiguous stream of aligned AIO DIOs to let
unaligned AIO DIO wait forever).

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

On 64-bit architectures we have two 4-byte holes in struct ext4_io_end.
Order entries better to avoid this and thus make the structure occupy
64 instead of 72 bytes for 64-bit architectures.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Currently we used atomic bit operations to manipulate
__JI_COMMIT_RUNNING bit. However this is unnecessary as i_flags are
always written and read under j_list_lock. So just change the operations
to standard bit operations.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Revoke and tag descriptor blocks are just different kinds of descriptor
blocks and thus have checksum in the same place. Unify computation and
checking of checksums for these.

Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Descriptor block header is initialized in several places. Factor out the
common code into jbd2_journal_get_descriptor_buffer().

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

jbd2_journal_write_revoke_records() takes journal pointer and write_op,
although journal can be obtained from the passed transaction and
write_op is always WRITE_SYNC. Remove these superfluous arguments.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

The flags parameter is also unused.

Signed-off-by: Eric Whitney <enwlinux@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

To reduce amount of damage caused by single bad block, we limit number
of inodes sharing an xattr block to 1024. Thus there can be more xattr
blocks with the same contents when there are lots of files with the same
extended attributes. These xattr blocks naturally result in hash
collisions and can form long hash chains and we unnecessarily check each
such block only to find out we cannot use it because it is already
shared by too many inodes.

Add a reusable flag to cache entries which is cleared when a cache entry
has reached its maximum refcount.  Cache entries which are not marked
reusable are skipped by mb_cache_entry_find_{first,next}. This
significantly speeds up mbcache when there are many same xattr blocks.
For example for xattr-bench with 5 values and each process handling
20000 files, the run for 64 processes is 25x faster with this patch.
Even for 8 processes the speedup is almost 3x. We have also verified
that for situations where there is only one xattr block of each kind,
the patch doesn't have a measurable cost.

[JK: Remove handling of setting the same value since it is not needed
anymore, check for races in e_reusable setting, improve changelog,
add measurements]

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

When someone tried to set xattr to the same value (i.e., not changing
anything) we did all the work of removing original xattr, possibly
breaking references to shared xattr block, inserting new xattr, and
merging xattr blocks again. Since this is not so rare operation and it
is relatively cheap for us to detect this case, check for this and
shortcut xattr setting in that case.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Get rid of field _e_hash_list_head in cache entries and add bit field
e_referenced instead.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

This variable, introduced in commit 9c191f70

, is unnecessary: it is set
once the module has been initialized correctly, and ext4_fill_super
cannot run unless the module has been initialized correctly.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Since old mbcache code is gone, let's rename new code to mbcache since
number 2 is now meaningless. This is just a mechanical replacement.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Currently we maintain perfect LRU list by moving entry to the tail of
the list when it gets used. However these operations on cache-global
list are relatively expensive.

In this patch we switch to lazy updates of LRU list. Whenever entry gets
used, we set a referenced bit in it. When reclaiming entries, we give
referenced entries another round in the LRU. Since the list is not a
real LRU anymore, rename it to just 'list'.

In my testing this logic gives about 30% boost to workloads with mostly
unique xattr blocks (e.g. xattr-bench with 10 files and 10000 unique
xattr values).

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

So far number of entries in mbcache is limited only by the pressure from
the shrinker. Since too many entries degrade the hash table and
generally we expect that caching more entries has diminishing returns,
limit number of entries the same way as in the old mbcache to 16 * hash
table size.

Once we exceed the desired maximum number of entries, we schedule a
backround work to reclaim entries. If the background work cannot keep up
and the number of entries exceeds two times the desired maximum, we
reclaim some entries directly when allocating a new entry.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Both ext2 and ext4 are now converted to mbcache2. Remove the old mbcache
code.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

The conversion is generally straightforward. We convert filesystem from
a global cache to per-fs one. Similarly to ext4 the tricky part is that
xattr block corresponding to found mbcache entry can get freed before we
get buffer lock for that block. So we have to check whether the entry is
still valid after getting the buffer lock.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

The conversion is generally straightforward. The only tricky part is
that xattr block corresponding to found mbcache entry can get freed
before we get buffer lock for that block. So we have to check whether
the entry is still valid after getting buffer lock.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

Original mbcache was designed to have more features than what ext?
filesystems ended up using. It supported entry being in more hashes, it
had a home-grown rwlocking of each entry, and one cache could cache
entries from multiple filesystems. This genericity also resulted in more
complex locking, larger cache entries, and generally more code
complexity.

This is reimplementation of the mbcache functionality to exactly fit the
purpose ext? filesystems use it for. Cache entries are now considerably
smaller (7 instead of 13 longs), the code is considerably smaller as
well (414 vs 913 lines of code), and IMO also simpler. The new code is
also much more lightweight.

I have measured the speed using artificial xattr-bench benchmark, which
spawns P processes, each process sets xattr for F different files, and
the value of xattr is randomly chosen from a pool of V values. Averages
of runtimes for 5 runs for various combinations of parameters are below.
The first value in each cell is old mbache, the second value is the new
mbcache.

V=10
F\P	1		2		4		8		16		32		64
10	0.158,0.157	0.208,0.196	0.500,0.277	0.798,0.400	3.258,0.584	13.807,1.047	61.339,2.803
100	0.172,0.167	0.279,0.222	0.520,0.275	0.825,0.341	2.981,0.505	12.022,1.202	44.641,2.943
1000	0.185,0.174	0.297,0.239	0.445,0.283	0.767,0.340	2.329,0.480	6.342,1.198	16.440,3.888

V=100
F\P	1		2		4		8		16		32		64
10	0.162,0.153	0.200,0.186	0.362,0.257	0.671,0.496	1.433,0.943	3.801,1.345	7.938,2.501
100	0.153,0.160	0.221,0.199	0.404,0.264	0.945,0.379	1.556,0.485	3.761,1.156	7.901,2.484
1000	0.215,0.191	0.303,0.246	0.471,0.288	0.960,0.347	1.647,0.479	3.916,1.176	8.058,3.160

V=1000
F\P	1		2		4		8		16		32		64
10	0.151,0.129	0.210,0.163	0.326,0.245	0.685,0.521	1.284,0.859	3.087,2.251	6.451,4.801
100	0.154,0.153	0.211,0.191	0.276,0.282	0.687,0.506	1.202,0.877	3.259,1.954	8.738,2.887
1000	0.145,0.179	0.202,0.222	0.449,0.319	0.899,0.333	1.577,0.524	4.221,1.240	9.782,3.579

V=10000
F\P	1		2		4		8		16		32		64
10	0.161,0.154	0.198,0.190	0.296,0.256	0.662,0.480	1.192,0.818	2.989,2.200	6.362,4.746
100	0.176,0.174	0.236,0.203	0.326,0.255	0.696,0.511	1.183,0.855	4.205,3.444	19.510,17.760
1000	0.199,0.183	0.240,0.227	1.159,1.014	2.286,2.154	6.023,6.039	---,10.933	---,36.620

V=100000
F\P	1		2		4		8		16		32		64
10	0.171,0.162	0.204,0.198	0.285,0.230	0.692,0.500	1.225,0.881	2.990,2.243	6.379,4.771
100	0.151,0.171	0.220,0.210	0.295,0.255	0.720,0.518	1.226,0.844	3.423,2.831	19.234,17.544
1000	0.192,0.189	0.249,0.225	1.162,1.043	2.257,2.093	5.853,4.997	---,10.399	---,32.198

We see that the new code is faster in pretty much all the cases and
starting from 4 processes there are significant gains with the new code
resulting in upto 20-times shorter runtimes. Also for large numbers of
cached entries all values for the old code could not be measured as the
kernel started hitting softlockups and died before the test completed.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

In commit bcff2488 ("ext4: don't read blocks from disk after extents
being swapped") bh is not updated correctly in the for loop and wrong
data has been written to disk. generic/324 catches this on sub-page
block size ext4.

Fixes: bcff2488

 ("ext4: don't read blocks from disk after extentsbeing swapped")
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@vger.kernel.org

Now, ext4_free_blocks() doesn't revoke data blocks of per-file data
journalled inode and it can cause file data inconsistency problems.
Even though data blocks of per-file data journalled inode are already
forgotten by jbd2_journal_invalidatepage() in advance of invoking
ext4_free_blocks(), we still need to revoke the data blocks here.
Moreover some of the metadata blocks, which are not found by
sb_find_get_block(), are still needed to be revoked, but this is also
missing here.

Signed-off-by: Daeho Jeong <daeho.jeong@samsung.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jan Kara <jack@suse.cz>

Competing overwrite DIO in dioread_nolock mode will just overwrite
pointer to io_end in the inode. This may result in data corruption or
extent conversion happening from IO completion interrupt because we
don't properly set buffer_defer_completion() when unlocked DIO races
with locked DIO to unwritten extent.

Since unlocked DIO doesn't need io_end for anything, just avoid
allocating it and corrupting pointer from inode for locked DIO.
A cleaner fix would be to avoid these games with io_end pointer from the
inode but that requires more intrusive changes so we leave that for
later.

Cc: stable@vger.kernel.org
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

ext4 can update bh->b_state non-atomically in _ext4_get_block() and
ext4_da_get_block_prep(). Usually this is fine since bh is just a
temporary storage for mapping information on stack but in some cases it
can be fully living bh attached to a page. In such case non-atomic
update of bh->b_state can race with an atomic update which then gets
lost. Usually when we are mapping bh and thus updating bh->b_state
non-atomically, nobody else touches the bh and so things work out fine
but there is one case to especially worry about: ext4_finish_bio() uses
BH_Uptodate_Lock on the first bh in the page to synchronize handling of
PageWriteback state. So when blocksize < pagesize, we can be atomically
modifying bh->b_state of a buffer that actually isn't under IO and thus
can race e.g. with delalloc trying to map that buffer. The result is
that we can mistakenly set / clear BH_Uptodate_Lock bit resulting in the
corruption of PageWriteback state or missed unlock of BH_Uptodate_Lock.

Fix the problem by always updating bh->b_state bits atomically.

CC: stable@vger.kernel.org
Reported-by: Nikolay Borisov <kernel@kyup.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

When ext4_bread() fails, fname_crypto_str remains
allocated after return. Fix that.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
CC: Dmitry Monakhov <dmonakhov@virtuozzo.com>

The "newblock" parameter is not used in convert_initialized_extent(),
remove it.

Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>