After submit_one_bio(), `bio' can go away. However submit_extent_page()
leave `bio' referable if submit_one_bio() failed (e.g. -ENOMEM on OOM).
It will cause invalid paging request when submit_extent_page() is called
next time.

I reproduced ENOMEM case with the following script (need
CONFIG_FAIL_PAGE_ALLOC, and CONFIG_FAULT_INJECTION_DEBUG_FS).

  #!/bin/bash

  dmesgout=dmesg.txt
  start=100000
  end=300000
  step=1000

  # btrfs options
  device=/dev/vdb1
  directory=/mnt/btrfs

  # fault-injection options
  percent=100
  times=3

  mkdir -p $directory || exit 1
  mount -o compress $device $directory || exit 1

  rm -f $directory/file || exit 1
  dd if=/dev/zero of=$directory/file bs=1M count=512 || exit 1

  for interval in `seq $start $step $end`; do
          dmesg -C
          echo 1 > /proc/sys/vm/drop_caches
          sync
          export FAILCMD_TYPE=fail_page_alloc
          ./failcmd.sh -p $percent -t $times -i $interval \
                  --ignore-gfp-highmem=N --ignore-gfp-wait=N --min-order=0 \
                  -- \
                  cat $directory/file > /dev/null
          dmesg > ${dmesgout}
          if grep -q BUG: ${dmesgout}; then
                  cat ${dmesgout}
                  exit 1
          fi
  done

  umount $directory
  exit 0
Signed-off-by: NNaohiro Aota <naota@elisp.net>
Tested-by: NSatoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

1: ref_count is simple than current RBIO_HOLD_BBIO_MAP_BIT flag
   to keep btrfs_bio's memory in raid56 recovery implement.
2: free function for bbio will make code clean and flexible, plus
   forced data type checking in compile.

Changelog v1->v2:
 Rename following by David Sterba's suggestion:
 put_btrfs_bio() -> btrfs_put_bio()
 get_btrfs_bio() -> btrfs_get_bio()
 bbio->ref_count -> bbio->refs
Signed-off-by: NZhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

Currently there's a 4B hole in the structure between refs and state and there
are only 16 bits used so we can make it unsigned. This will get a better
packing and may save some stack space for local variables.

The size of extent_state gets reduced by 8B and there are usually a lot
of slab objects.

struct extent_state {
	u64                        start;                /*     0     8 */
	u64                        end;                  /*     8     8 */
	struct rb_node             rb_node;              /*    16    24 */
	wait_queue_head_t          wq;                   /*    40    24 */
	/* --- cacheline 1 boundary (64 bytes) --- */
	atomic_t                   refs;                 /*    64     4 */

	/* XXX 4 bytes hole, try to pack */

	long unsigned int          state;                /*    72     8 */
	u64                        private;              /*    80     8 */

	/* size: 88, cachelines: 2, members: 7 */
	/* sum members: 84, holes: 1, sum holes: 4 */
	/* last cacheline: 24 bytes */
};
Signed-off-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NChris Mason <clm@fb.com>

Because we're using globally known nodesize. Do the same for the sanity
test function variant.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>

Make the extent buffer allocation interface consistent.  Cloned eb will
set a valid fs_info.  For dummy eb, we can drop the length parameter and
set it from fs_info.

The built-in sanity checks may pass a NULL fs_info that's queried for
nodesize, but we know it's 4096.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>

Same mask from all callers.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>

We try to allocate an extent state structure before acquiring the extent
state tree's spinlock as we might need a new one later and therefore avoid
doing later an atomic allocation while holding the tree's spinlock. However
we returned -ENOMEM if that initial non-atomic allocation failed, which is
a bit excessive since we might end up not needing the pre-allocated extent
state at all - for the case where the tree doesn't have any extent states
that cover the input range and cover too any other range. Therefore don't
return -ENOMEM if that pre-allocation fails.
Signed-off-by: NFilipe Manana <fdmanana@suse.com>
Signed-off-by: NChris Mason <clm@fb.com>

We try to allocate an extent state before acquiring the tree's spinlock
just in case we end up needing to split an existing extent state into two.
If that allocation failed, we would return -ENOMEM.
However, our only single caller (transaction/log commit code), passes in
an extent state that was cached from a call to find_first_extent_bit() and
that has a very high chance to match exactly the input range (always true
for a transaction commit and very often, but not always, true for a log
commit) - in this case we end up not needing at all that initial extent
state used for an eventual split. Therefore just don't return -ENOMEM if
we can't allocate the temporary extent state, since we might not need it
at all, and if we end up needing one, we'll do it later anyway.
Signed-off-by: NFilipe Manana <fdmanana@suse.com>
Signed-off-by: NChris Mason <clm@fb.com>

Right now the only caller of find_first_extent_bit() that is interested
in caching extent states (transaction or log commit), never gets an extent
state cached. This is because find_first_extent_bit() only caches states
that have at least one of the flags EXTENT_IOBITS or EXTENT_BOUNDARY, and
the transaction/log commit caller always passes a tree that doesn't have
ever extent states with any of those flags (they can only have one of the
following flags: EXTENT_DIRTY, EXTENT_NEW or EXTENT_NEED_WAIT).

This change together with the following one in the patch series (titled
"Btrfs: avoid returning -ENOMEM in convert_extent_bit() too early") will
help reduce significantly the chances of calls to convert_extent_bit()
fail with -ENOMEM when called from the transaction/log commit code.
Signed-off-by: NFilipe Manana <fdmanana@suse.com>
Signed-off-by: NChris Mason <clm@fb.com>

If we fail in submit_compressed_extents() before calling btrfs_submit_compressed_write(),
we start and end the writeback for the pages (clear their dirty flag, unlock them, etc)
but we don't tag the pages, nor the inode's mapping, with an error. This makes it
impossible for a caller of filemap_fdatawait_range() (fsync, or transaction commit
for e.g.) know that there was an error.

Note that the return value of submit_compressed_extents() is useless, as that function
is executed by a workqueue task and not directly by the fill_delalloc callback. This
means the writepage/s callbacks of the inode's address space operations don't get that
return value.
Signed-off-by: NFilipe Manana <fdmanana@suse.com>
Signed-off-by: NChris Mason <clm@fb.com>

While we have a transaction ongoing, the VM might decide at any time
to call btree_inode->i_mapping->a_ops->writepages(), which will start
writeback of dirty pages belonging to btree nodes/leafs. This call
might return an error or the writeback might finish with an error
before we attempt to commit the running transaction. If this happens,
we might have no way of knowing that such error happened when we are
committing the transaction - because the pages might no longer be
marked dirty nor tagged for writeback (if a subsequent modification
to the extent buffer didn't happen before the transaction commit) which
makes filemap_fdata[write|wait]_range unable to find such pages (even
if they're marked with SetPageError).
So if this happens we must abort the transaction, otherwise we commit
a super block with btree roots that point to btree nodes/leafs whose
content on disk is invalid - either garbage or the content of some
node/leaf from a past generation that got cowed or deleted and is no
longer valid (for this later case we end up getting error messages like
"parent transid verify failed on 10826481664 wanted 25748 found 29562"
when reading btree nodes/leafs from disk).

Note that setting and checking AS_EIO/AS_ENOSPC in the btree inode's
i_mapping would not be enough because we need to distinguish between
log tree extents (not fatal) vs non-log tree extents (fatal) and
because the next call to filemap_fdatawait_range() will catch and clear
such errors in the mapping - and that call might be from a log sync and
not from a transaction commit, which means we would not know about the
error at transaction commit time. Also, checking for the eb flag
EXTENT_BUFFER_IOERR at transaction commit time isn't done and would
not be completely reliable, as the eb might be removed from memory and
read back when trying to get it, which clears that flag right before
reading the eb's pages from disk, making us not know about the previous
write error.

Using the new 3 flags for the btree inode also makes us achieve the
goal of AS_EIO/AS_ENOSPC when writepages() returns success, started
writeback for all dirty pages and before filemap_fdatawait_range() is
called, the writeback for all dirty pages had already finished with
errors - because we were not using AS_EIO/AS_ENOSPC,
filemap_fdatawait_range() would return success, as it could not know
that writeback errors happened (the pages were no longer tagged for
writeback).
Signed-off-by: NFilipe Manana <fdmanana@suse.com>
Signed-off-by: NChris Mason <clm@fb.com>

This is actually inspired by Filipe's patch.  When write_one_eb() fails on
submit_extent_page(), it'll give up writing this eb and mark it with
EXTENT_BUFFER_IOERR.  So if it's not the last page that encounter the failure,
there are some left pages which remain DIRTY, and if a later COW on this eb
happens, ie. eb is COWed and freed, it'd run into BUG_ON in
btrfs_release_extent_buffer_page() for the DIRTY page, ie. BUG_ON(PageDirty(page));

This adds the missing clear_page_dirty_for_io() for the rest pages of eb.
Signed-off-by: NLiu Bo <bo.li.liu@oracle.com>
Reviewed-by: NFilipe Manana <fdmanana@suse.com>
Signed-off-by: NChris Mason <clm@fb.com>

If submit_extent_page() fails in write_one_eb(), we end up with the current
page not marked dirty anymore, unlocked and marked for writeback. But we never
end up calling end_page_writeback() against the page, which will make calls to
filemap_fdatawait_range (e.g. at transaction commit time) hang forever waiting
for the writeback bit to be cleared from the page.
Signed-off-by: NFilipe Manana <fdmanana@suse.com>
Reviewed-by: NLiu Bo <bo.li.liu@oracle.com>
Signed-off-by: NChris Mason <clm@fb.com>

It used to be more complex but now it's just a simple array access.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>

All callers use the same value, simplify the function.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>

After the data is written successfully, we should cleanup the read failure record
in that range because
- If we set data COW for the file, the range that the failure record pointed to is
  mapped to a new place, so it is invalid.
- If we set no data COW for the file, and if there is no error during writting,
  the corrupted data is corrected, so the failure record can be removed. And if
  some errors happen on the mirrors, we also needn't worry about it because the
  failure record will be recreated if we read the same place again.

Sometimes, we may fail to correct the data, so the failure records will be left
in the tree, we need free them when we free the inode or the memory leak happens.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

This patch implement data repair function when direct read fails.

The detail of the implementation is:
- When we find the data is not right, we try to read the data from the other
  mirror.
- When the io on the mirror ends, we will insert the endio work into the
  dedicated btrfs workqueue, not common read endio workqueue, because the
  original endio work is still blocked in the btrfs endio workqueue, if we
  insert the endio work of the io on the mirror into that workqueue, deadlock
  would happen.
- After we get right data, we write it back to the corrupted mirror.
- And if the data on the new mirror is still corrupted, we will try next
  mirror until we read right data or all the mirrors are traversed.
- After the above work, we set the uptodate flag according to the result.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

We could not use clean_io_failure in the direct IO path because it got the
filesystem information from the page structure, but the page in the direct
IO bio didn't have the filesystem information in its structure. So we need
modify it and pass all the information it need by parameters.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

The original code of repair_io_failure was just used for buffered read,
because it got some filesystem data from page structure, it is safe for
the page in the page cache. But when we do a direct read, the pages in bio
are not in the page cache, that is there is no filesystem data in the page
structure. In order to implement direct read data repair, we need modify
repair_io_failure and pass all filesystem data it need by function
parameters.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

The data repair function of direct read will be implemented later, and some code
in bio_readpage_error will be reused, so split bio_readpage_error into
several functions which will be used in direct read repair later.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

We forgot to free failure record and bio after submitting re-read bio failed,
fix it.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

Direct IO splits the original bio to several sub-bios because of the limit of
raid stripe, and the filesystem will wait for all sub-bios and then run final
end io process.

But it was very hard to implement the data repair when dio read failure happens,
because at the final end io function, we didn't know which mirror the data was
read from. So in order to implement the data repair, we have to move the file data
check in the final end io function to the sub-bio end io function, in which we can
get the mirror number of the device we access. This patch did this work as the
first step of the direct io data repair implementation.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

The current code would load checksum data for several times when we split
a whole direct read io because of the limit of the raid stripe, it would
make us search the csum tree for several times. In fact, it just wasted time,
and made the contention of the csum tree root be more serious. This patch
improves this problem by loading the data at once.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <clm@fb.com>

We have been iterating all references for each extent we have in a file when we
do fiemap to see if it is shared.  This is fine when you have a few clones or a
few snapshots, but when you have 5k snapshots suddenly fiemap just sits there
and stares at you.  So add btrfs_check_shared which will use the backref walking
code but will short circuit as soon as it finds a root or inode that doesn't
match the one we currently have.  This makes fiemap on my testbox go from
looking at me blankly for a day to spitting out actual output in a reasonable
amount of time.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fb.com>
Signed-off-by: NChris Mason <clm@fb.com>

We've defined a 'offset' out of bio_for_each_segment_all.

This is just a clean rename, no function changes.
Signed-off-by: NLiu Bo <bo.li.liu@oracle.com>
Signed-off-by: NChris Mason <clm@fb.com>

The tree field of struct extent_state was only used to figure out if
an extent state was connected to an inode's io tree or not. For this
we can just use the rb_node field itself.

On a x86_64 system with this change the sizeof(struct extent_state) is
reduced from 96 bytes down to 88 bytes, meaning that with a page size
of 4096 bytes we can now store 46 extent states per page instead of 42.
Signed-off-by: NFilipe Manana <fdmanana@suse.com>
Signed-off-by: NChris Mason <clm@fb.com>

btrfs_set_key_type and btrfs_key_type are used inconsistently along with
open coded variants. Other members of btrfs_key are accessed directly
without any helpers anyway.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NChris Mason <clm@fb.com>

The crash is

------------[ cut here ]------------
kernel BUG at fs/btrfs/extent_io.c:2124!
[...]
Workqueue: btrfs-endio normal_work_helper [btrfs]
RIP: 0010:[<ffffffffa02d6055>]  [<ffffffffa02d6055>] end_bio_extent_readpage+0xb45/0xcd0 [btrfs]

This is in fact a regression.

It is because we forgot to increase @offset properly in reading corrupted block,
so that the @offset remains, and this leads to checksum errors while reading
left blocks queued up in the same bio, and then ends up with hiting the above
BUG_ON.
Reported-by: NChris Murphy <lists@colorremedies.com>
Signed-off-by: NLiu Bo <bo.li.liu@oracle.com>
Signed-off-by: NChris Mason <clm@fb.com>

When page aligned start and len passed to extent_fiemap(), the result is
good, but when start and len is not aligned, e.g. start = 1 and len =
4095 is passed to extent_fiemap(), it returns no extent.

The problem is that start and len is all rounded down which causes the
problem. This patch will round down start and round up (start + len) to
return right extent.
Reported-by: NChandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: NQu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NChris Mason <clm@fb.com>

The current "wait_on_bit" interface requires an 'action'
function to be provided which does the actual waiting.
There are over 20 such functions, many of them identical.
Most cases can be satisfied by one of just two functions, one
which uses io_schedule() and one which just uses schedule().

So:
 Rename wait_on_bit and        wait_on_bit_lock to
        wait_on_bit_action and wait_on_bit_lock_action
 to make it explicit that they need an action function.

 Introduce new wait_on_bit{,_lock} and wait_on_bit{,_lock}_io
 which are *not* given an action function but implicitly use
 a standard one.
 The decision to error-out if a signal is pending is now made
 based on the 'mode' argument rather than being encoded in the action
 function.

 All instances of the old wait_on_bit and wait_on_bit_lock which
 can use the new version have been changed accordingly and their
 action functions have been discarded.
 wait_on_bit{_lock} does not return any specific error code in the
 event of a signal so the caller must check for non-zero and
 interpolate their own error code as appropriate.

The wait_on_bit() call in __fscache_wait_on_invalidate() was
ambiguous as it specified TASK_UNINTERRUPTIBLE but used
fscache_wait_bit_interruptible as an action function.
David Howells confirms this should be uniformly
"uninterruptible"

The main remaining user of wait_on_bit{,_lock}_action is NFS
which needs to use a freezer-aware schedule() call.

A comment in fs/gfs2/glock.c notes that having multiple 'action'
functions is useful as they display differently in the 'wchan'
field of 'ps'. (and /proc/$PID/wchan).
As the new bit_wait{,_io} functions are tagged "__sched", they
will not show up at all, but something higher in the stack.  So
the distinction will still be visible, only with different
function names (gds2_glock_wait versus gfs2_glock_dq_wait in the
gfs2/glock.c case).

Since first version of this patch (against 3.15) two new action
functions appeared, on in NFS and one in CIFS.  CIFS also now
uses an action function that makes the same freezer aware
schedule call as NFS.
Signed-off-by: NNeilBrown <neilb@suse.de>
Acked-by: David Howells <dhowells@redhat.com> (fscache, keys)
Acked-by: Steven Whitehouse <swhiteho@redhat.com> (gfs2)
Acked-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Steve French <sfrench@samba.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/20140707051603.28027.72349.stgit@notabene.brownSigned-off-by: NIngo Molnar <mingo@kernel.org>

If this condition in end_extent_writepage() is false:

	if (tree->ops && tree->ops->writepage_end_io_hook)

we will then test an uninitialized "ret" at:

	ret = ret < 0 ? ret : -EIO;

The test for ret is for the case where ->writepage_end_io_hook
failed, and we'd choose that ret as the error; but if
there is no ->writepage_end_io_hook, nothing sets ret.

Initializing ret to 0 should be sufficient; if
writepage_end_io_hook wasn't set, (!uptodate) means
non-zero err was passed in, so we choose -EIO in that case.
Signed-of-by: NEric Sandeen <sandeen@redhat.com>
Signed-off-by: NChris Mason <clm@fb.com>

This new function reads the content of an extent directly to user memory.
Signed-off-by: NGerhard Heift <Gerhard@Heift.Name>
Signed-off-by: NChris Mason <clm@fb.com>
Acked-by: NDavid Sterba <dsterba@suse.cz>

__extent_writepage has two unrelated parts.  First it does the delayed
allocation dance and second it does the mapping and IO for the page
we're actually writing.

This splits it up into those two parts so the stack from one doesn't
impact the stack from the other.
Signed-off-by: NChris Mason <clm@fb.com>

This adds noinline_for_stack to two helpers used by
btree_write_cache_pages.  It shaves us down from 424 bytes on the
stack to 280.
Signed-off-by: NChris Mason <clm@fb.com>

We need to NULL the cached_state after freeing it, otherwise
we might free it again if find_delalloc_range doesn't find anything.
Signed-off-by: NChris Mason <clm@fb.com>
cc: stable@vger.kernel.org

This exercises the various parts of the new qgroup accounting code.  We do some
basic stuff and do some things with the shared refs to make sure all that code
works.  I had to add a bunch of infrastructure because I needed to be able to
insert items into a fake tree without having to do all the hard work myself,
hopefully this will be usefull in the future.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fb.com>
Signed-off-by: NChris Mason <clm@fb.com>

According to commit 865ffef3
(fs: fix fsync() error reporting),
it's not stable to just check error pages because pages can be
truncated or invalidated, we should also mark mapping with error
flag so that a later fsync can catch the error.
Signed-off-by: NLiu Bo <bo.li.liu@oracle.com>
Signed-off-by: NChris Mason <clm@fb.com>

When running low on available disk space and having several processes
doing buffered file IO, I got the following trace in dmesg:

[ 4202.720152] INFO: task kworker/u8:1:5450 blocked for more than 120 seconds.
[ 4202.720401]       Not tainted 3.13.0-fdm-btrfs-next-26+ #1
[ 4202.720596] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[ 4202.720874] kworker/u8:1    D 0000000000000001     0  5450      2 0x00000000
[ 4202.720904] Workqueue: btrfs-flush_delalloc normal_work_helper [btrfs]
[ 4202.720908]  ffff8801f62ddc38 0000000000000082 ffff880203ac2490 00000000001d3f40
[ 4202.720913]  ffff8801f62ddfd8 00000000001d3f40 ffff8800c4f0c920 ffff880203ac2490
[ 4202.720918]  00000000001d4a40 ffff88020fe85a40 ffff88020fe85ab8 0000000000000001
[ 4202.720922] Call Trace:
[ 4202.720931]  [<ffffffff816a3cb9>] schedule+0x29/0x70
[ 4202.720950]  [<ffffffffa01ec48d>] btrfs_start_ordered_extent+0x6d/0x110 [btrfs]
[ 4202.720956]  [<ffffffff8108e620>] ? bit_waitqueue+0xc0/0xc0
[ 4202.720972]  [<ffffffffa01ec559>] btrfs_run_ordered_extent_work+0x29/0x40 [btrfs]
[ 4202.720988]  [<ffffffffa0201987>] normal_work_helper+0x137/0x2c0 [btrfs]
[ 4202.720994]  [<ffffffff810680e5>] process_one_work+0x1f5/0x530
(...)
[ 4202.721027] 2 locks held by kworker/u8:1/5450:
[ 4202.721028]  #0:  (%s-%s){++++..}, at: [<ffffffff81068083>] process_one_work+0x193/0x530
[ 4202.721037]  #1:  ((&work->normal_work)){+.+...}, at: [<ffffffff81068083>] process_one_work+0x193/0x530
[ 4202.721054] INFO: task btrfs:7891 blocked for more than 120 seconds.
[ 4202.721258]       Not tainted 3.13.0-fdm-btrfs-next-26+ #1
[ 4202.721444] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[ 4202.721699] btrfs           D 0000000000000001     0  7891   7890 0x00000001
[ 4202.721704]  ffff88018c2119e8 0000000000000086 ffff8800a33d2490 00000000001d3f40
[ 4202.721710]  ffff88018c211fd8 00000000001d3f40 ffff8802144b0000 ffff8800a33d2490
[ 4202.721714]  ffff8800d8576640 ffff88020fe85bc0 ffff88020fe85bc8 7fffffffffffffff
[ 4202.721718] Call Trace:
[ 4202.721723]  [<ffffffff816a3cb9>] schedule+0x29/0x70
[ 4202.721727]  [<ffffffff816a2ebc>] schedule_timeout+0x1dc/0x270
[ 4202.721732]  [<ffffffff8109bd79>] ? mark_held_locks+0xb9/0x140
[ 4202.721736]  [<ffffffff816a90c0>] ? _raw_spin_unlock_irq+0x30/0x40
[ 4202.721740]  [<ffffffff8109bf0d>] ? trace_hardirqs_on_caller+0x10d/0x1d0
[ 4202.721744]  [<ffffffff816a488f>] wait_for_completion+0xdf/0x120
[ 4202.721749]  [<ffffffff8107fa90>] ? try_to_wake_up+0x310/0x310
[ 4202.721765]  [<ffffffffa01ebee4>] btrfs_wait_ordered_extents+0x1f4/0x280 [btrfs]
[ 4202.721781]  [<ffffffffa020526e>] btrfs_mksubvol.isra.62+0x30e/0x5a0 [btrfs]
[ 4202.721786]  [<ffffffff8108e620>] ? bit_waitqueue+0xc0/0xc0
[ 4202.721799]  [<ffffffffa02056a9>] btrfs_ioctl_snap_create_transid+0x1a9/0x1b0 [btrfs]
[ 4202.721813]  [<ffffffffa020583a>] btrfs_ioctl_snap_create_v2+0x10a/0x170 [btrfs]
(...)

It turns out that extent_io.c:__extent_writepage(), which ends up being called
through filemap_fdatawrite_range() in btrfs_start_ordered_extent(), was getting
-ENOSPC when calling the fill_delalloc callback. In this situation, it returned
without the writepage_end_io_hook callback (inode.c:btrfs_writepage_end_io_hook)
ever being called for the respective page, which prevents the ordered extent's
bytes_left count from ever reaching 0, and therefore a finish_ordered_fn work
is never queued into the endio_write_workers queue. This makes the task that
called btrfs_start_ordered_extent() hang forever on the wait queue of the ordered
extent.

This is fairly easy to reproduce using a small filesystem and fsstress on
a quad core vm:

    mkfs.btrfs -f -b `expr 2100 \* 1024 \* 1024` /dev/sdd
    mount /dev/sdd /mnt

    fsstress -p 6 -d /mnt -n 100000 -x \
        "btrfs subvolume snapshot -r /mnt /mnt/mysnap" \
	    -f allocsp=0 \
	    -f bulkstat=0 \
	    -f bulkstat1=0 \
	    -f chown=0 \
	    -f creat=1 \
	    -f dread=0 \
	    -f dwrite=0 \
	    -f fallocate=1 \
	    -f fdatasync=0 \
	    -f fiemap=0 \
	    -f freesp=0 \
	    -f fsync=0 \
	    -f getattr=0 \
	    -f getdents=0 \
	    -f link=0 \
	    -f mkdir=0 \
	    -f mknod=0 \
	    -f punch=1 \
	    -f read=0 \
	    -f readlink=0 \
	    -f rename=0 \
	    -f resvsp=0 \
	    -f rmdir=0 \
	    -f setxattr=0 \
	    -f stat=0 \
	    -f symlink=0 \
	    -f sync=0 \
	    -f truncate=1 \
	    -f unlink=0 \
	    -f unresvsp=0 \
	    -f write=4

So just ensure that if an error happens while writing the extent page
we call the writepage_end_io_hook callback. Also make it return the
error code and ensure the caller (extent_write_cache_pages) processes
all pages in the page vector even if an error happens only for some
of them, so that ordered extents end up released.
Signed-off-by: NFilipe David Borba Manana <fdmanana@gmail.com>
Signed-off-by: NChris Mason <clm@fb.com>

aops->write_begin may allocate a new page and make it visible only to have
mark_page_accessed called almost immediately after.  Once the page is
visible the atomic operations are necessary which is noticable overhead
when writing to an in-memory filesystem like tmpfs but should also be
noticable with fast storage.  The objective of the patch is to initialse
the accessed information with non-atomic operations before the page is
visible.

The bulk of filesystems directly or indirectly use
grab_cache_page_write_begin or find_or_create_page for the initial
allocation of a page cache page.  This patch adds an init_page_accessed()
helper which behaves like the first call to mark_page_accessed() but may
called before the page is visible and can be done non-atomically.

The primary APIs of concern in this care are the following and are used
by most filesystems.

	find_get_page
	find_lock_page
	find_or_create_page
	grab_cache_page_nowait
	grab_cache_page_write_begin

All of them are very similar in detail to the patch creates a core helper
pagecache_get_page() which takes a flags parameter that affects its
behavior such as whether the page should be marked accessed or not.  Then
old API is preserved but is basically a thin wrapper around this core
function.

Each of the filesystems are then updated to avoid calling
mark_page_accessed when it is known that the VM interfaces have already
done the job.  There is a slight snag in that the timing of the
mark_page_accessed() has now changed so in rare cases it's possible a page
gets to the end of the LRU as PageReferenced where as previously it might
have been repromoted.  This is expected to be rare but it's worth the
filesystem people thinking about it in case they see a problem with the
timing change.  It is also the case that some filesystems may be marking
pages accessed that previously did not but it makes sense that filesystems
have consistent behaviour in this regard.

The test case used to evaulate this is a simple dd of a large file done
multiple times with the file deleted on each iterations.  The size of the
file is 1/10th physical memory to avoid dirty page balancing.  In the
async case it will be possible that the workload completes without even
hitting the disk and will have variable results but highlight the impact
of mark_page_accessed for async IO.  The sync results are expected to be
more stable.  The exception is tmpfs where the normal case is for the "IO"
to not hit the disk.

The test machine was single socket and UMA to avoid any scheduling or NUMA
artifacts.  Throughput and wall times are presented for sync IO, only wall
times are shown for async as the granularity reported by dd and the
variability is unsuitable for comparison.  As async results were variable
do to writback timings, I'm only reporting the maximum figures.  The sync
results were stable enough to make the mean and stddev uninteresting.

The performance results are reported based on a run with no profiling.
Profile data is based on a separate run with oprofile running.

async dd
                                    3.15.0-rc3            3.15.0-rc3
                                       vanilla           accessed-v2
ext3    Max      elapsed     13.9900 (  0.00%)     11.5900 ( 17.16%)
tmpfs	Max      elapsed      0.5100 (  0.00%)      0.4900 (  3.92%)
btrfs   Max      elapsed     12.8100 (  0.00%)     12.7800 (  0.23%)
ext4	Max      elapsed     18.6000 (  0.00%)     13.3400 ( 28.28%)
xfs	Max      elapsed     12.5600 (  0.00%)      2.0900 ( 83.36%)

The XFS figure is a bit strange as it managed to avoid a worst case by
sheer luck but the average figures looked reasonable.

        samples percentage
ext3       86107    0.9783  vmlinux-3.15.0-rc4-vanilla        mark_page_accessed
ext3       23833    0.2710  vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed
ext3        5036    0.0573  vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed
ext4       64566    0.8961  vmlinux-3.15.0-rc4-vanilla        mark_page_accessed
ext4        5322    0.0713  vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed
ext4        2869    0.0384  vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed
xfs        62126    1.7675  vmlinux-3.15.0-rc4-vanilla        mark_page_accessed
xfs         1904    0.0554  vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed
xfs          103    0.0030  vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed
btrfs      10655    0.1338  vmlinux-3.15.0-rc4-vanilla        mark_page_accessed
btrfs       2020    0.0273  vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed
btrfs        587    0.0079  vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed
tmpfs      59562    3.2628  vmlinux-3.15.0-rc4-vanilla        mark_page_accessed
tmpfs       1210    0.0696  vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed
tmpfs         94    0.0054  vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed

[akpm@linux-foundation.org: don't run init_page_accessed() against an uninitialised pointer]
Signed-off-by: NMel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jan Kara <jack@suse.cz>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Hugh Dickins <hughd@google.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Tested-by: NPrabhakar Lad <prabhakar.csengg@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>