Skip to contentSkip to conversation

fstests: generic/366: add a new test case to verify if certain fio load will hang the filesystem

PATCH In review Qu Wenruo 4 participants 5 messages Oct 28, 2024 – Oct 30, 2024 Btrfs file system

lore
Share & tools
lore

Qu Wenruo

author
#0 lore raw

PATCH v2 1/1 fstests: generic/366: add a new test case to verify if certain fio load will hang the filesystem

[BUG]
During the development to make btrfs pass generic/563 (which needs to make btrfs to support partial folios), generic/095 causes hangs during tests.

The call trace for the hanging process looks like this:

  __switch_to+0xf8/0x168
  __schedule+0x328/0x8a8
  schedule+0x54/0x140
  io_schedule+0x44/0x68
  folio_wait_bit_common+0x198/0x3f8
  __folio_lock+0x24/0x40
  extent_write_cache_pages+0x2e0/0x4c0 [btrfs]
  btrfs_writepages+0x94/0x158 [btrfs]
  do_writepages+0x74/0x190
  filemap_fdatawrite_wbc+0x88/0xc8
  __filemap_fdatawrite_range+0x6c/0xa8
  filemap_fdatawrite_range+0x1c/0x30
  btrfs_start_ordered_extent+0x264/0x2e0 [btrfs]
  btrfs_lock_and_flush_ordered_range+0x8c/0x160 [btrfs]
  __get_extent_map+0xa0/0x220 [btrfs]
  btrfs_do_readpage+0x1bc/0x5d8 [btrfs]
  btrfs_read_folio+0x50/0xa0 [btrfs]
  filemap_read_folio+0x54/0x110
  filemap_update_page+0x2e0/0x3b8
  filemap_get_pages+0x228/0x4d8
  filemap_read+0x11c/0x3b8
  btrfs_file_read_iter+0x74/0x90 [btrfs]
  new_sync_read+0xd0/0x1d0
  vfs_read+0x1a0/0x1f0

[CAUSE]
The root cause is a btrfs specific behavior that during a folio read, we can trigger writeback of the same folio, which will try to lock the same folio already locked by the read process.

The fix is already sent to the mailing list:
https://lore.kernel.org/linux-btrfs/62bf73ada7be2888d45a787c2b6fd252103a5d25.1729725088.git.wqu@suse.com/

This problem can only happen if all the following conditions are met:

- The sector size of btrfs is smaller than page size To have partial uptodate folios.

- Btrfs won't read the full folio if buffered write is block aligned This is done by the not yet merged patch:
https://lore.kernel.org/linux-btrfs/ac2639ec4e9ac176d33e95ef7ecf008fa6be5461.1727833878.git.wqu@suse.com/

[TEST CASE]
During the debugging of that generic/095 hang, I extracted a minimal reproducer which is much smaller and faster, although it still requires several runs to trigger a hang.

The test case will run the fio workload 32 times by default, which is more than enough to trigger the hang.

---
Changelog:
v2:
- Fix the duplicated _fixed_by_kernel_commit line
- Fix a typo in the commit message
- Update the comments to show the workload and how it hangs btrfs

---
 tests/generic/366     | 106 ++++++++++++++++++++++++++++++++++++++++++
 tests/generic/366.out |   2 +
 2 files changed, 108 insertions(+)
 create mode 100755 tests/generic/366
 create mode 100644 tests/generic/366.out
tests/generic/366 new file +106 -0
new file mode 100755
index 00000000..9d31c1c8
@@ -0,0 +1,106 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (C) 2024 SUSE Linux Products GmbH. All Rights Reserved.
+#
+# FS QA Test 366
+#
+# Test if mixed direct read, direct write and buffered write on the same file will
+# hang the filesystem.
+#
+# This is exposed by an incoming btrfs feature, which allows a folio to be
+# partial uptodate if the buffered write range is block aligned but not yet
+# full folio aligned.
+#
+# Such behavior makes btrfs to hang reliably under generic/095.
+# This is the extracted minimal reproducer for 4k block size and 64K page size.
+#
+. ./common/preamble
+_begin_fstest auto quick
+
+. ./common/filter
+
+_require_scratch
+_require_odirect
+_require_aio
+
+_fixed_by_kernel_commit xxxxxxxxxxxx \
+	"btrfs: avoid deadlock when reading a partial uptodate folio"
+
+iterations=$((32 * LOAD_FACTOR))
+
+fio_config=$tmp.fio
+fio_out=$tmp.fio.out
+blksz=`$here/src/min_dio_alignment $SCRATCH_MNT $SCRATCH_DEV`
+cat >$fio_config <<EOF
+[global]
+bs=8k
+iodepth=1
+randrepeat=1
+size=256k
+directory=$SCRATCH_MNT
+numjobs=1
+[job1]
+ioengine=sync
+bs=512
+direct=1
+rw=randread
+filename=file1
+[job2]
+ioengine=libaio
+rw=randwrite
+direct=1
+filename=file1
+[job3]
+ioengine=posixaio
+rw=randwrite
+filename=file1
+EOF
+_require_fio $fio_config
+
+for (( i = 0; i < $iterations; i++)); do
+	_scratch_mkfs >>$seqres.full 2>&1
+	_scratch_mount
+	# There's a known EIO failure to report collisions between directio and buffered
+	# writes to userspace, refer to upstream linux 5a9d929d6e13. So ignore EIO error
+	# at here.
+	#
+	# And for btrfs if sector size < page size, if we have a partial
+	# uptodate folio caused by a buffered write, e.g:
+	#
+	#    0          16K         32K          48K         64K
+	#    |                                   |///////////|
+	#					     \- sector Uptodate|Dirty
+	#
+	# Then writeback happens and finished, but btrfs' ordered extent not
+	# yet finished.
+	# In that case, the folio can be released from the page cache (since
+	# the folio is not under IO/lock).
+	#
+	# Then new buffered writes into the folio happened, re-dirty the folio:
+	#   0          16K         32K          48K         64K
+	#   |//////////|                        |///////////|
+	#      \- sector Uptodate|Dirty              \- No sector flags
+	#                                               extent map PINNED
+	#                                               OE still here
+	#
+	# Now read is triggered on that folio.
+	# Btrfs will need to wait for any existing ordered extents in the folio range,
+	# that wait will also trigger writeback if the folio is dirty.
+	# That writeback will happen for range [48K, 64K), but since the whole folio
+	# is locked for read, writeback will also try to lock the same folio, causing
+	# a deadlock.
+	$FIO_PROG $fio_config --ignore_error=,EIO --output=$fio_out
+	# umount before checking dmesg in case umount triggers any WARNING or Oops
+	_scratch_unmount
+
+	_check_dmesg _filter_aiodio_dmesg
+
+	echo "=== fio $i/$iterations ===" >> $seqres.full
+	cat $fio_out >> $seqres.full
+done
+
+echo "Silence is golden"
+
+# success, all done
+status=0
+exit
tests/generic/366.out new file +2 -0
new file mode 100644
index 00000000..1fe90e06
@@ -0,0 +1,2 @@
+QA output created by 366
+Silence is golden
Signature · 1 line
2.46.0
#1 lore raw
Quoted 85 lines · Qu Wenruo

[BUG]
During the development to make btrfs pass generic/563 (which needs to make btrfs to support partial folios), generic/095 causes hangs during tests.

The call trace for the hanging process looks like this:

  __switch_to+0xf8/0x168
  __schedule+0x328/0x8a8
  schedule+0x54/0x140
  io_schedule+0x44/0x68
  folio_wait_bit_common+0x198/0x3f8
  __folio_lock+0x24/0x40
  extent_write_cache_pages+0x2e0/0x4c0 [btrfs]
  btrfs_writepages+0x94/0x158 [btrfs]
  do_writepages+0x74/0x190
  filemap_fdatawrite_wbc+0x88/0xc8
  __filemap_fdatawrite_range+0x6c/0xa8
  filemap_fdatawrite_range+0x1c/0x30
  btrfs_start_ordered_extent+0x264/0x2e0 [btrfs]
  btrfs_lock_and_flush_ordered_range+0x8c/0x160 [btrfs]
  __get_extent_map+0xa0/0x220 [btrfs]
  btrfs_do_readpage+0x1bc/0x5d8 [btrfs]
  btrfs_read_folio+0x50/0xa0 [btrfs]
  filemap_read_folio+0x54/0x110
  filemap_update_page+0x2e0/0x3b8
  filemap_get_pages+0x228/0x4d8
  filemap_read+0x11c/0x3b8
  btrfs_file_read_iter+0x74/0x90 [btrfs]
  new_sync_read+0xd0/0x1d0
  vfs_read+0x1a0/0x1f0

[CAUSE]
The root cause is a btrfs specific behavior that during a folio read, we can trigger writeback of the same folio, which will try to lock the same folio already locked by the read process.

The fix is already sent to the mailing list:
https://lore.kernel.org/linux-btrfs/62bf73ada7be2888d45a787c2b6fd252103a5d25.1729725088.git.wqu@suse.com/

This problem can only happen if all the following conditions are met:

- The sector size of btrfs is smaller than page size To have partial uptodate folios.

- Btrfs won't read the full folio if buffered write is block aligned This is done by the not yet merged patch:
https://lore.kernel.org/linux-btrfs/ac2639ec4e9ac176d33e95ef7ecf008fa6be5461.1727833878.git.wqu@suse.com/

[TEST CASE]
During the debugging of that generic/095 hang, I extracted a minimal reproducer which is much smaller and faster, although it still requires several runs to trigger a hang.

The test case will run the fio workload 32 times by default, which is more than enough to trigger the hang.

---
Changelog:
v2:
- Fix the duplicated _fixed_by_kernel_commit line
- Fix a typo in the commit message
- Update the comments to show the workload and how it hangs btrfs

---
 tests/generic/366     | 106 ++++++++++++++++++++++++++++++++++++++++++
 tests/generic/366.out |   2 +
 2 files changed, 108 insertions(+)
 create mode 100755 tests/generic/366
 create mode 100644 tests/generic/366.out
tests/generic/366 new file +8 -0
new file mode 100755
index 00000000..9d31c1c8
@@ -0,0 +1,106 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (C) 2024 SUSE Linux Products GmbH. All Rights Reserved.
+#
+# FS QA Test 366
+#
+# Test if mixed direct read, direct write and buffered write on the same file will
+# hang the filesystem.

tests/generic/366 line 18

Quoted 10 lines
tests/generic/366 new file +10 -0
+#
+# This is exposed by an incoming btrfs feature, which allows a folio to be
+# partial uptodate if the buffered write range is block aligned but not yet
+# full folio aligned.
+#
+# Such behavior makes btrfs to hang reliably under generic/095.
+# This is the extracted minimal reproducer for 4k block size and 64K page size.
+#
+. ./common/preamble
+_begin_fstest auto quick
                            ^^
                            rw

I think we can keep the "rw" group as g/095 does.

Quoted 64 lines
tests/generic/366 new file +64 -0
+
+. ./common/filter
+
+_require_scratch
+_require_odirect
+_require_aio
+
+_fixed_by_kernel_commit xxxxxxxxxxxx \
+	"btrfs: avoid deadlock when reading a partial uptodate folio"
+
+iterations=$((32 * LOAD_FACTOR))
+
+fio_config=$tmp.fio
+fio_out=$tmp.fio.out
+blksz=`$here/src/min_dio_alignment $SCRATCH_MNT $SCRATCH_DEV`
+cat >$fio_config <<EOF
+[global]
+bs=8k
+iodepth=1
+randrepeat=1
+size=256k
+directory=$SCRATCH_MNT
+numjobs=1
+[job1]
+ioengine=sync
+bs=512
+direct=1
+rw=randread
+filename=file1
+[job2]
+ioengine=libaio
+rw=randwrite
+direct=1
+filename=file1
+[job3]
+ioengine=posixaio
+rw=randwrite
+filename=file1
+EOF
+_require_fio $fio_config
+
+for (( i = 0; i < $iterations; i++)); do
+	_scratch_mkfs >>$seqres.full 2>&1
+	_scratch_mount
+	# There's a known EIO failure to report collisions between directio and buffered
+	# writes to userspace, refer to upstream linux 5a9d929d6e13. So ignore EIO error
+	# at here.
+	#
+	# And for btrfs if sector size < page size, if we have a partial
+	# uptodate folio caused by a buffered write, e.g:
+	#
+	#    0          16K         32K          48K         64K
+	#    |                                   |///////////|
+	#					     \- sector Uptodate|Dirty
+	#
+	# Then writeback happens and finished, but btrfs' ordered extent not
+	# yet finished.
+	# In that case, the folio can be released from the page cache (since
+	# the folio is not under IO/lock).
+	#
+	# Then new buffered writes into the folio happened, re-dirty the folio:
+	#   0          16K         32K          48K         64K
+	#   |//////////|                        |///////////|
+	#      \- sector Uptodate|Dirty              \- No sector flags

tests/generic/366 line 92

Quoted 10 lines
tests/generic/366 new file +10 -0
+	#                                               extent map PINNED
+	#                                               OE still here
+	#
+	# Now read is triggered on that folio.
+	# Btrfs will need to wait for any existing ordered extents in the folio range,
+	# that wait will also trigger writeback if the folio is dirty.
+	# That writeback will happen for range [48K, 64K), but since the whole folio
+	# is locked for read, writeback will also try to lock the same folio, causing
+	# a deadlock.
+	$FIO_PROG $fio_config --ignore_error=,EIO --output=$fio_out

Looks like this test doesn't mix DIO and buffered IO, so this EIO ignoring might not be needed.

tests/generic/366 line 96

Quoted 4 lines
tests/generic/366 new file +4 -0
+	# umount before checking dmesg in case umount triggers any WARNING or Oops
+	_scratch_unmount
+
+	_check_dmesg _filter_aiodio_dmesg

This test removed mmap test part, so this dmesg filter might not be needed either ? If so, don't need to import above "./common/filter" either.

Others looks good to me.

Thanks,
Zorro

Quoted 22 lines
tests/generic/366 new file +10 -0
+
+	echo "=== fio $i/$iterations ===" >> $seqres.full
+	cat $fio_out >> $seqres.full
+done
+
+echo "Silence is golden"
+
+# success, all done
+status=0
+exit
tests/generic/366.out new file +2 -0
new file mode 100644
index 00000000..1fe90e06
@@ -0,0 +1,2 @@
+QA output created by 366
+Silence is golden
Signature · 1 line
2.46.0
#2 lore raw
Quoted 1 line · Zorro Lang

On Tue, Oct 29, 2024 at 10:05:25AM +1030, Qu Wenruo wrote:

[...]

Quoted 7 lines

tests/generic/366 line 18

Quoted 3 lines
tests/generic/366 new file +3 -0
+#
+. ./common/preamble
+_begin_fstest auto quick
                             ^^
                             rw

I think we can keep the "rw" group as g/095 does.

If that's the only missing part, mind to add that rw group when merging?

The remaining two points may not need modification, explained inlined.

Quoted 10 lines

tests/generic/366 line 92

Quoted 7 lines
tests/generic/366 new file +7 -0
+	# Now read is triggered on that folio.
+	# Btrfs will need to wait for any existing ordered extents in the folio range,
+	# that wait will also trigger writeback if the folio is dirty.
+	# That writeback will happen for range [48K, 64K), but since the whole folio
+	# is locked for read, writeback will also try to lock the same folio, causing
+	# a deadlock.
+	$FIO_PROG $fio_config --ignore_error=,EIO --output=$fio_out

Looks like this test doesn't mix DIO and buffered IO, so this EIO ignoring might not be needed.

I'm not sure about the EIO part, but at least job2 and job3 are all doing writes, one is buffered and the other is direct. So it's still mixing both.

Quoted 8 lines

tests/generic/366 line 96

Quoted 5 lines
tests/generic/366 new file +4 -0
+	# umount before checking dmesg in case umount triggers any WARNING or Oops
+	_scratch_unmount
+
+	_check_dmesg _filter_aiodio_dmesg

This test removed mmap test part, so this dmesg filter might not be needed either ? If so, don't need to import above "./common/filter" either.

Since it's still mixing buffered and direct IO, I can hit the aiodio error message just like this:

[91619.077752] BTRFS warning (device dm-2): read-write for sector size
4096 with page size 65536 is experimental
[91619.086204] BTRFS info (device dm-2): checking UUID tree
[91619.473510] Page cache invalidation failure on direct I/O.  Possible

data corruption due to collision with buffered I/O!

So I'm afraid we need the filter too.

Thanks,
Qu

Quoted 30 lines

Others looks good to me.

Thanks,
Zorro

Quoted 22 lines
tests/generic/366 new file +10 -0
+
+	echo "=== fio $i/$iterations ===" >> $seqres.full
+	cat $fio_out >> $seqres.full
+done
+
+echo "Silence is golden"
+
+# success, all done
+status=0
+exit
tests/generic/366.out new file +2 -1
new file mode 100644
index 00000000..1fe90e06
@@ -0,0 +1,2 @@
+QA output created by 366
+Silence is golden
--

2.46.0

Reviewed-by
#3 lore raw
Quoted 44 lines · Qu Wenruo
Quoted 1 line · Zorro Lang

On Tue, Oct 29, 2024 at 10:05:25AM +1030, Qu Wenruo wrote:

[...]

Quoted 7 lines

tests/generic/366 line 18

Quoted 3 lines
tests/generic/366 new file +3 -0
+#
+. ./common/preamble
+_begin_fstest auto quick
                             ^^
                             rw

I think we can keep the "rw" group as g/095 does.

If that's the only missing part, mind to add that rw group when merging?

The remaining two points may not need modification, explained inlined.

Quoted 10 lines

tests/generic/366 line 92

Quoted 7 lines
tests/generic/366 new file +7 -0
+	# Now read is triggered on that folio.
+	# Btrfs will need to wait for any existing ordered extents in the folio range,
+	# that wait will also trigger writeback if the folio is dirty.
+	# That writeback will happen for range [48K, 64K), but since the whole folio
+	# is locked for read, writeback will also try to lock the same folio, causing
+	# a deadlock.
+	$FIO_PROG $fio_config --ignore_error=,EIO --output=$fio_out

Looks like this test doesn't mix DIO and buffered IO, so this EIO ignoring might not be needed.

I'm not sure about the EIO part, but at least job2 and job3 are all doing writes, one is buffered and the other is direct. So it's still mixing both.

Quoted 8 lines

tests/generic/366 line 96

Quoted 5 lines
tests/generic/366 new file +4 -0
+	# umount before checking dmesg in case umount triggers any WARNING or Oops
+	_scratch_unmount
+
+	_check_dmesg _filter_aiodio_dmesg

This test removed mmap test part, so this dmesg filter might not be needed either ? If so, don't need to import above "./common/filter" either.

Since it's still mixing buffered and direct IO, I can hit the aiodio error message just like this:

Quoted 7 lines
[91619.077752] BTRFS warning (device dm-2): read-write for sector size
4096 with page size 65536 is experimental
[91619.086204] BTRFS info (device dm-2): checking UUID tree
[91619.473510] Page cache invalidation failure on direct I/O.  Possible

data corruption due to collision with buffered I/O!

So I'm afraid we need the filter too.

Oh, thanks for your confirming. I can merge this patch with that tiny change. If there's not more review points from btrfs list, I'll merge it. It looks good to me now.

Quoted 35 lines

Thanks,
Qu

Quoted 30 lines

Others looks good to me.

Thanks,
Zorro

Quoted 22 lines
tests/generic/366 new file +10 -0
+
+	echo "=== fio $i/$iterations ===" >> $seqres.full
+	cat $fio_out >> $seqres.full
+done
+
+echo "Silence is golden"
+
+# success, all done
+status=0
+exit
tests/generic/366.out new file +2 -1
new file mode 100644
index 00000000..1fe90e06
@@ -0,0 +1,2 @@
+QA output created by 366
+Silence is golden
--

2.46.0

Reviewed-by on v2 1/1
#4 lore raw
In reply to Qu Wenruo
Quoted 170 lines · Qu Wenruo

[BUG]
During the development to make btrfs pass generic/563 (which needs to make btrfs to support partial folios), generic/095 causes hangs during tests.

The call trace for the hanging process looks like this:

   __switch_to+0xf8/0x168
   __schedule+0x328/0x8a8
   schedule+0x54/0x140
   io_schedule+0x44/0x68
   folio_wait_bit_common+0x198/0x3f8
   __folio_lock+0x24/0x40
   extent_write_cache_pages+0x2e0/0x4c0 [btrfs]
   btrfs_writepages+0x94/0x158 [btrfs]
   do_writepages+0x74/0x190
   filemap_fdatawrite_wbc+0x88/0xc8
   __filemap_fdatawrite_range+0x6c/0xa8
   filemap_fdatawrite_range+0x1c/0x30
   btrfs_start_ordered_extent+0x264/0x2e0 [btrfs]
   btrfs_lock_and_flush_ordered_range+0x8c/0x160 [btrfs]
   __get_extent_map+0xa0/0x220 [btrfs]
   btrfs_do_readpage+0x1bc/0x5d8 [btrfs]
   btrfs_read_folio+0x50/0xa0 [btrfs]
   filemap_read_folio+0x54/0x110
   filemap_update_page+0x2e0/0x3b8
   filemap_get_pages+0x228/0x4d8
   filemap_read+0x11c/0x3b8
   btrfs_file_read_iter+0x74/0x90 [btrfs]
   new_sync_read+0xd0/0x1d0
   vfs_read+0x1a0/0x1f0

[CAUSE]
The root cause is a btrfs specific behavior that during a folio read, we can trigger writeback of the same folio, which will try to lock the same folio already locked by the read process.

The fix is already sent to the mailing list:
https://lore.kernel.org/linux-btrfs/62bf73ada7be2888d45a787c2b6fd252103a5d25.1729725088.git.wqu@suse.com/

This problem can only happen if all the following conditions are met:

- The sector size of btrfs is smaller than page size To have partial uptodate folios.

- Btrfs won't read the full folio if buffered write is block aligned This is done by the not yet merged patch:

   https://lore.kernel.org/linux-btrfs/ac2639ec4e9ac176d33e95ef7ecf008fa6be5461.1727833878.git.wqu@suse.com/

[TEST CASE]
During the debugging of that generic/095 hang, I extracted a minimal reproducer which is much smaller and faster, although it still requires several runs to trigger a hang.

The test case will run the fio workload 32 times by default, which is more than enough to trigger the hang.

---
Changelog:
v2:
- Fix the duplicated _fixed_by_kernel_commit line
- Fix a typo in the commit message
- Update the comments to show the workload and how it hangs btrfs

---
  tests/generic/366     | 106 ++++++++++++++++++++++++++++++++++++++++++
  tests/generic/366.out |   2 +
  2 files changed, 108 insertions(+)

create mode 100755 tests/generic/366
create mode 100644 tests/generic/366.out

tests/generic/366 new file +93 -0
new file mode 100755
index 00000000..9d31c1c8
@@ -0,0 +1,106 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (C) 2024 SUSE Linux Products GmbH. All Rights Reserved.
+#
+# FS QA Test 366
+#
+# Test if mixed direct read, direct write and buffered write on the same file will
+# hang the filesystem.
+#
+# This is exposed by an incoming btrfs feature, which allows a folio to be
+# partial uptodate if the buffered write range is block aligned but not yet
+# full folio aligned.
+#
+# Such behavior makes btrfs to hang reliably under generic/095.
+# This is the extracted minimal reproducer for 4k block size and 64K page size.
+#
+. ./common/preamble
+_begin_fstest auto quick
+
+. ./common/filter
+
+_require_scratch
+_require_odirect
+_require_aio
+
+_fixed_by_kernel_commit xxxxxxxxxxxx \
+	"btrfs: avoid deadlock when reading a partial uptodate folio"
+
+iterations=$((32 * LOAD_FACTOR))
+
+fio_config=$tmp.fio
+fio_out=$tmp.fio.out
+blksz=`$here/src/min_dio_alignment $SCRATCH_MNT $SCRATCH_DEV`
+cat >$fio_config <<EOF
+[global]
+bs=8k
+iodepth=1
+randrepeat=1
+size=256k
+directory=$SCRATCH_MNT
+numjobs=1
+[job1]
+ioengine=sync
+bs=512
+direct=1
+rw=randread
+filename=file1
+[job2]
+ioengine=libaio
+rw=randwrite
+direct=1
+filename=file1
+[job3]
+ioengine=posixaio
+rw=randwrite
+filename=file1
+EOF
+_require_fio $fio_config
+
+for (( i = 0; i < $iterations; i++)); do
+	_scratch_mkfs >>$seqres.full 2>&1
+	_scratch_mount
+	# There's a known EIO failure to report collisions between directio and buffered
+	# writes to userspace, refer to upstream linux 5a9d929d6e13. So ignore EIO error
+	# at here.
+	#
+	# And for btrfs if sector size < page size, if we have a partial
+	# uptodate folio caused by a buffered write, e.g:
+	#
+	#    0          16K         32K          48K         64K
+	#    |                                   |///////////|
+	#					     \- sector Uptodate|Dirty
+	#
+	# Then writeback happens and finished, but btrfs' ordered extent not
+	# yet finished.
+	# In that case, the folio can be released from the page cache (since
+	# the folio is not under IO/lock).
+	#
+	# Then new buffered writes into the folio happened, re-dirty the folio:
+	#   0          16K         32K          48K         64K
+	#   |//////////|                        |///////////|
+	#      \- sector Uptodate|Dirty              \- No sector flags
+	#                                               extent map PINNED
+	#                                               OE still here
+	#
+	# Now read is triggered on that folio.
+	# Btrfs will need to wait for any existing ordered extents in the folio range,
+	# that wait will also trigger writeback if the folio is dirty.
+	# That writeback will happen for range [48K, 64K), but since the whole folio
+	# is locked for read, writeback will also try to lock the same folio, causing
+	# a deadlock.
+	$FIO_PROG $fio_config --ignore_error=,EIO --output=$fio_out
+	# umount before checking dmesg in case umount triggers any WARNING or Oops

tests/generic/366 line 103

Quoted 10 lines
tests/generic/366 new file +10 -0
+	_scratch_unmount
+
+	_check_dmesg _filter_aiodio_dmesg
+
+	echo "=== fio $i/$iterations ===" >> $seqres.full
+	cat $fio_out >> $seqres.full
+done
+
+echo "Silence is golden"
+

Thanks!

Quoted 11 lines
tests/generic/366 new file +3 -0
+# success, all done
+status=0
+exit
tests/generic/366.out new file +2 -0
new file mode 100644
index 00000000..1fe90e06
@@ -0,0 +1,2 @@
+QA output created by 366
+Silence is golden