Skip to contentSkip to conversation

fs/dax: Fix "don't skip locked entries when scanning entries"

PATCH Applied Alistair Popple 4 participants 5 messages May 23, 2025 – May 30, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fs/dax: Fix "don't skip locked entries when scanning entries"

Commit 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries") introduced a new function, wait_entry_unlocked_exclusive(), which waits for the current entry to become unlocked without advancing the XArray iterator state.

Waiting for the entry to become unlocked requires dropping the XArray lock. This requires calling xas_pause() prior to dropping the lock which leaves the xas in a suitable state for the next iteration. However this has the side-effect of advancing the xas state to the next index. Normally this isn't an issue because xas_for_each() contains code to detect this state and thus avoid advancing the index a second time on the next loop iteration.

However both callers of and wait_entry_unlocked_exclusive() itself subsequently use the xas state to reload the entry. As xas_pause() updated the state to the next index this will cause the current entry which is being waited on to be skipped. This caused the following warning to fire intermittently when running xftest generic/068 on an XFS filesystem with FS DAX enabled:

[   35.067397] ------------[ cut here ]------------
[   35.068229] WARNING: CPU: 21 PID: 1640 at mm/truncate.c:89 truncate_folio_batch_exceptionals+0xd8/0x1e0
[   35.069717] Modules linked in: nd_pmem dax_pmem nd_btt nd_e820 libnvdimm
[   35.071006] CPU: 21 UID: 0 PID: 1640 Comm: fstest Not tainted 6.15.0-rc7+ #77 PREEMPT(voluntary)
[   35.072613] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/204
[   35.074845] RIP: 0010:truncate_folio_batch_exceptionals+0xd8/0x1e0
[   35.075962] Code: a1 00 00 00 f6 47 0d 20 0f 84 97 00 00 00 4c 63 e8 41 39 c4 7f 0b eb 61 49 83 c5 01 45 39 ec 7e 58 42 f68
[   35.079522] RSP: 0018:ffffb04e426c7850 EFLAGS: 00010202
[   35.080359] RAX: 0000000000000000 RBX: ffff9d21e3481908 RCX: ffffb04e426c77f4
[   35.081477] RDX: ffffb04e426c79e8 RSI: ffffb04e426c79e0 RDI: ffff9d21e34816e8
[   35.082590] RBP: ffffb04e426c79e0 R08: 0000000000000001 R09: 0000000000000003
[   35.083733] R10: 0000000000000000 R11: 822b53c0f7a49868 R12: 000000000000001f
[   35.084850] R13: 0000000000000000 R14: ffffb04e426c78e8 R15: fffffffffffffffe
[   35.085953] FS:  00007f9134c87740(0000) GS:ffff9d22abba0000(0000) knlGS:0000000000000000
[   35.087346] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   35.088244] CR2: 00007f9134c86000 CR3: 000000040afff000 CR4: 00000000000006f0
[   35.089354] Call Trace:
[   35.089749]  <TASK>
[   35.090168]  truncate_inode_pages_range+0xfc/0x4d0
[   35.091078]  truncate_pagecache+0x47/0x60
[   35.091735]  xfs_setattr_size+0xc7/0x3e0
[   35.092648]  xfs_vn_setattr+0x1ea/0x270
[   35.093437]  notify_change+0x1f4/0x510
[   35.094219]  ? do_truncate+0x97/0xe0
[   35.094879]  do_truncate+0x97/0xe0
[   35.095640]  path_openat+0xabd/0xca0
[   35.096278]  do_filp_open+0xd7/0x190
[   35.096860]  do_sys_openat2+0x8a/0xe0
[   35.097459]  __x64_sys_openat+0x6d/0xa0
[   35.098076]  do_syscall_64+0xbb/0x1d0
[   35.098647]  entry_SYSCALL_64_after_hwframe+0x77/0x7f
[   35.099444] RIP: 0033:0x7f9134d81fc1
[   35.100033] Code: 75 57 89 f0 25 00 00 41 00 3d 00 00 41 00 74 49 80 3d 2a 26 0e 00 00 74 6d 89 da 48 89 ee bf 9c ff ff ff5
[   35.102993] RSP: 002b:00007ffcd41e0d10 EFLAGS: 00000202 ORIG_RAX: 0000000000000101
[   35.104263] RAX: ffffffffffffffda RBX: 0000000000000242 RCX: 00007f9134d81fc1
[   35.105452] RDX: 0000000000000242 RSI: 00007ffcd41e1200 RDI: 00000000ffffff9c
[   35.106663] RBP: 00007ffcd41e1200 R08: 0000000000000000 R09: 0000000000000064
[   35.107923] R10: 00000000000001a4 R11: 0000000000000202 R12: 0000000000000066
[   35.109112] R13: 0000000000100000 R14: 0000000000100000 R15: 0000000000000400
[   35.110357]  </TASK>
[   35.110769] irq event stamp: 8415587
[   35.111486] hardirqs last  enabled at (8415599): [<ffffffff8d74b562>] __up_console_sem+0x52/0x60
[   35.113067] hardirqs last disabled at (8415610): [<ffffffff8d74b547>] __up_console_sem+0x37/0x60
[   35.114575] softirqs last  enabled at (8415300): [<ffffffff8d6ac625>] handle_softirqs+0x315/0x3f0
[   35.115933] softirqs last disabled at (8415291): [<ffffffff8d6ac811>] __irq_exit_rcu+0xa1/0xc0
[   35.117316] ---[ end trace 0000000000000000 ]---

Fix this by using xas_reset() instead, which is equivalent in implementation to xas_pause() but does not advance the XArray state.

Fixes 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries")
Cc Dan Williams <[email protected]>
Cc Alison Schofield <[email protected]>
Cc Matthew Wilcow (Oracle) <[email protected]>
Cc Balbir Singh <[email protected]>
Cc "Darrick J. Wong" <[email protected]>
Cc Dave Chinner <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Jan Kara <[email protected]>
Cc John Hubbard <[email protected]>
Cc Ted Ts'o <[email protected]>
Cc Alexander Viro <[email protected]>
Cc Christian Brauner <[email protected]>

---

Hi Andrew,

Apologies for finding this so late in the cycle. This is a very intermittent issue for me, and it seems it was only exposed by a recent upgrade to my test machine/setup. The user visible impact is the same as for the original commit this fixes. That is possible file data corruption if a device has a FS DAX page pinned for DMA.

So in other words it means my original fix was not 100% effective. The issue that commit fixed has existed for a long time without being reported, so not sure if this is worth trying to get into v6.15 or not.

Either way I figured it would be best to send this ASAP, which means I am still waiting for a complete xfstest run to complete (although the failing test does now pass cleanly).

---
 fs/dax.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/dax.c +1 -1
index 676303419e9e..f8d8b1afd232 100644
@@ -257,7 +257,7 @@ static void *wait_entry_unlocked_exclusive(struct xa_state *xas, void *entry)
 		wq = dax_entry_waitqueue(xas, entry, &ewait.key);
 		prepare_to_wait_exclusive(wq, &ewait.wait,
 					TASK_UNINTERRUPTIBLE);
-		xas_pause(xas);
+		xas_reset(xas);
 		xas_unlock_irq(xas);
 		schedule();
 		finish_wait(wq, &ewait.wait);
Signature · 1 line
2.47.2
reviewer Reviewed-by on v1 1/1
#1 lore raw
Quoted 70 lines · Alistair Popple

Commit 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries") introduced a new function, wait_entry_unlocked_exclusive(), which waits for the current entry to become unlocked without advancing the XArray iterator state.

Waiting for the entry to become unlocked requires dropping the XArray lock. This requires calling xas_pause() prior to dropping the lock which leaves the xas in a suitable state for the next iteration. However this has the side-effect of advancing the xas state to the next index. Normally this isn't an issue because xas_for_each() contains code to detect this state and thus avoid advancing the index a second time on the next loop iteration.

However both callers of and wait_entry_unlocked_exclusive() itself subsequently use the xas state to reload the entry. As xas_pause() updated the state to the next index this will cause the current entry which is being waited on to be skipped. This caused the following warning to fire intermittently when running xftest generic/068 on an XFS filesystem with FS DAX enabled:

[   35.067397] ------------[ cut here ]------------
[   35.068229] WARNING: CPU: 21 PID: 1640 at mm/truncate.c:89 truncate_folio_batch_exceptionals+0xd8/0x1e0
[   35.069717] Modules linked in: nd_pmem dax_pmem nd_btt nd_e820 libnvdimm
[   35.071006] CPU: 21 UID: 0 PID: 1640 Comm: fstest Not tainted 6.15.0-rc7+ #77 PREEMPT(voluntary)
[   35.072613] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/204
[   35.074845] RIP: 0010:truncate_folio_batch_exceptionals+0xd8/0x1e0
[   35.075962] Code: a1 00 00 00 f6 47 0d 20 0f 84 97 00 00 00 4c 63 e8 41 39 c4 7f 0b eb 61 49 83 c5 01 45 39 ec 7e 58 42 f68
[   35.079522] RSP: 0018:ffffb04e426c7850 EFLAGS: 00010202
[   35.080359] RAX: 0000000000000000 RBX: ffff9d21e3481908 RCX: ffffb04e426c77f4
[   35.081477] RDX: ffffb04e426c79e8 RSI: ffffb04e426c79e0 RDI: ffff9d21e34816e8
[   35.082590] RBP: ffffb04e426c79e0 R08: 0000000000000001 R09: 0000000000000003
[   35.083733] R10: 0000000000000000 R11: 822b53c0f7a49868 R12: 000000000000001f
[   35.084850] R13: 0000000000000000 R14: ffffb04e426c78e8 R15: fffffffffffffffe
[   35.085953] FS:  00007f9134c87740(0000) GS:ffff9d22abba0000(0000) knlGS:0000000000000000
[   35.087346] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   35.088244] CR2: 00007f9134c86000 CR3: 000000040afff000 CR4: 00000000000006f0
[   35.089354] Call Trace:
[   35.089749]  <TASK>
[   35.090168]  truncate_inode_pages_range+0xfc/0x4d0
[   35.091078]  truncate_pagecache+0x47/0x60
[   35.091735]  xfs_setattr_size+0xc7/0x3e0
[   35.092648]  xfs_vn_setattr+0x1ea/0x270
[   35.093437]  notify_change+0x1f4/0x510
[   35.094219]  ? do_truncate+0x97/0xe0
[   35.094879]  do_truncate+0x97/0xe0
[   35.095640]  path_openat+0xabd/0xca0
[   35.096278]  do_filp_open+0xd7/0x190
[   35.096860]  do_sys_openat2+0x8a/0xe0
[   35.097459]  __x64_sys_openat+0x6d/0xa0
[   35.098076]  do_syscall_64+0xbb/0x1d0
[   35.098647]  entry_SYSCALL_64_after_hwframe+0x77/0x7f
[   35.099444] RIP: 0033:0x7f9134d81fc1
[   35.100033] Code: 75 57 89 f0 25 00 00 41 00 3d 00 00 41 00 74 49 80 3d 2a 26 0e 00 00 74 6d 89 da 48 89 ee bf 9c ff ff ff5
[   35.102993] RSP: 002b:00007ffcd41e0d10 EFLAGS: 00000202 ORIG_RAX: 0000000000000101
[   35.104263] RAX: ffffffffffffffda RBX: 0000000000000242 RCX: 00007f9134d81fc1
[   35.105452] RDX: 0000000000000242 RSI: 00007ffcd41e1200 RDI: 00000000ffffff9c
[   35.106663] RBP: 00007ffcd41e1200 R08: 0000000000000000 R09: 0000000000000064
[   35.107923] R10: 00000000000001a4 R11: 0000000000000202 R12: 0000000000000066
[   35.109112] R13: 0000000000100000 R14: 0000000000100000 R15: 0000000000000400
[   35.110357]  </TASK>
[   35.110769] irq event stamp: 8415587
[   35.111486] hardirqs last  enabled at (8415599): [<ffffffff8d74b562>] __up_console_sem+0x52/0x60
[   35.113067] hardirqs last disabled at (8415610): [<ffffffff8d74b547>] __up_console_sem+0x37/0x60
[   35.114575] softirqs last  enabled at (8415300): [<ffffffff8d6ac625>] handle_softirqs+0x315/0x3f0
[   35.115933] softirqs last disabled at (8415291): [<ffffffff8d6ac811>] __irq_exit_rcu+0xa1/0xc0
[   35.117316] ---[ end trace 0000000000000000 ]---

Fix this by using xas_reset() instead, which is equivalent in implementation to xas_pause() but does not advance the XArray state.

Quoted 14 lines
Fixes 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries")
Cc Dan Williams <[email protected]>
Cc Alison Schofield <[email protected]>
Cc Matthew Wilcow (Oracle) <[email protected]>
Cc Balbir Singh <[email protected]>
Cc "Darrick J. Wong" <[email protected]>
Cc Dave Chinner <[email protected]>
Cc David Hildenbrand <[email protected]>
Cc Jan Kara <[email protected]>
Cc John Hubbard <[email protected]>
Cc Ted Ts'o <[email protected]>
Cc Alexander Viro <[email protected]>
Cc Christian Brauner <[email protected]>

Looks good to me. Feel free to add:

Honza

Quoted 37 lines

---

Hi Andrew,

Apologies for finding this so late in the cycle. This is a very intermittent issue for me, and it seems it was only exposed by a recent upgrade to my test machine/setup. The user visible impact is the same as for the original commit this fixes. That is possible file data corruption if a device has a FS DAX page pinned for DMA.

So in other words it means my original fix was not 100% effective. The issue that commit fixed has existed for a long time without being reported, so not sure if this is worth trying to get into v6.15 or not.

Either way I figured it would be best to send this ASAP, which means I am still waiting for a complete xfstest run to complete (although the failing test does now pass cleanly).

---
 fs/dax.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/dax.c +1 -1
index 676303419e9e..f8d8b1afd232 100644
@@ -257,7 +257,7 @@ static void *wait_entry_unlocked_exclusive(struct xa_state *xas, void *entry)
 		wq = dax_entry_waitqueue(xas, entry, &ewait.key);
 		prepare_to_wait_exclusive(wq, &ewait.wait,
 					TASK_UNINTERRUPTIBLE);
-		xas_pause(xas);
+		xas_reset(xas);
 		xas_unlock_irq(xas);
 		schedule();
 		finish_wait(wq, &ewait.wait);
Signature · 1 line
2.47.2
Signature · 2 lines
Jan Kara <[email protected]>
SUSE Labs, CR
Reviewed-by on v1 1/1
#2 lore raw
In reply to Alistair Popple
Quoted 22 lines · Alistair Popple

Commit 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries") introduced a new function, wait_entry_unlocked_exclusive(), which waits for the current entry to become unlocked without advancing the XArray iterator state.

Waiting for the entry to become unlocked requires dropping the XArray lock. This requires calling xas_pause() prior to dropping the lock which leaves the xas in a suitable state for the next iteration. However this has the side-effect of advancing the xas state to the next index. Normally this isn't an issue because xas_for_each() contains code to detect this state and thus avoid advancing the index a second time on the next loop iteration.

However both callers of and wait_entry_unlocked_exclusive() itself subsequently use the xas state to reload the entry. As xas_pause() updated the state to the next index this will cause the current entry which is being waited on to be skipped. This caused the following warning to fire intermittently when running xftest generic/068 on an XFS filesystem with FS DAX enabled:

[   35.067397] ------------[ cut here ]------------
[   35.068229] WARNING: CPU: 21 PID: 1640 at mm/truncate.c:89 truncate_folio_batch_exceptionals+0xd8/0x1e0

[..]

Quoted 6 lines

Fix this by using xas_reset() instead, which is equivalent in implementation to xas_pause() but does not advance the XArray state.

Fixes 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries")

[..]

Quoted 25 lines

Hi Andrew,

Apologies for finding this so late in the cycle. This is a very intermittent issue for me, and it seems it was only exposed by a recent upgrade to my test machine/setup. The user visible impact is the same as for the original commit this fixes. That is possible file data corruption if a device has a FS DAX page pinned for DMA.

So in other words it means my original fix was not 100% effective. The issue that commit fixed has existed for a long time without being reported, so not sure if this is worth trying to get into v6.15 or not.

Either way I figured it would be best to send this ASAP, which means I am still waiting for a complete xfstest run to complete (although the failing test does now pass cleanly).

---
 fs/dax.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/dax.c +0 -0
index 676303419e9e..f8d8b1afd232 100644

fs/dax.c static void *wait_entry_unlocked_exclusive(struct xa_state *xas, void *entry) · line 263

Quoted 8 lines
fs/dax.c +1 -1
@@ -257,7 +257,7 @@ static void *wait_entry_unlocked_exclusive(struct xa_state *xas, void *entry)
 		wq = dax_entry_waitqueue(xas, entry, &ewait.key);
 		prepare_to_wait_exclusive(wq, &ewait.wait,
 					TASK_UNINTERRUPTIBLE);
-		xas_pause(xas);
+		xas_reset(xas);
 		xas_unlock_irq(xas);
 		schedule();
 		finish_wait(wq, &ewait.wait);

This looks super-subtle, but so did the original fix commit 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries"). The resolution is the same to make sure the xarray state does not mistakenly advance when the lock is dropped.

You can add:

#3 lore raw
Quoted 61 lines · Dan Williams
Quoted 22 lines · Alistair Popple

Commit 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries") introduced a new function, wait_entry_unlocked_exclusive(), which waits for the current entry to become unlocked without advancing the XArray iterator state.

Waiting for the entry to become unlocked requires dropping the XArray lock. This requires calling xas_pause() prior to dropping the lock which leaves the xas in a suitable state for the next iteration. However this has the side-effect of advancing the xas state to the next index. Normally this isn't an issue because xas_for_each() contains code to detect this state and thus avoid advancing the index a second time on the next loop iteration.

However both callers of and wait_entry_unlocked_exclusive() itself subsequently use the xas state to reload the entry. As xas_pause() updated the state to the next index this will cause the current entry which is being waited on to be skipped. This caused the following warning to fire intermittently when running xftest generic/068 on an XFS filesystem with FS DAX enabled:

[   35.067397] ------------[ cut here ]------------
[   35.068229] WARNING: CPU: 21 PID: 1640 at mm/truncate.c:89 truncate_folio_batch_exceptionals+0xd8/0x1e0

[..]

Quoted 6 lines

Fix this by using xas_reset() instead, which is equivalent in implementation to xas_pause() but does not advance the XArray state.

Fixes 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries")

[..]

Quoted 30 lines

Hi Andrew,

Apologies for finding this so late in the cycle. This is a very intermittent issue for me, and it seems it was only exposed by a recent upgrade to my test machine/setup. The user visible impact is the same as for the original commit this fixes. That is possible file data corruption if a device has a FS DAX page pinned for DMA.

So in other words it means my original fix was not 100% effective. The issue that commit fixed has existed for a long time without being reported, so not sure if this is worth trying to get into v6.15 or not.

Either way I figured it would be best to send this ASAP, which means I am still waiting for a complete xfstest run to complete (although the failing test does now pass cleanly).

---
 fs/dax.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/dax.c +1 -1
index 676303419e9e..f8d8b1afd232 100644
@@ -257,7 +257,7 @@ static void *wait_entry_unlocked_exclusive(struct xa_state *xas, void *entry)
 		wq = dax_entry_waitqueue(xas, entry, &ewait.key);
 		prepare_to_wait_exclusive(wq, &ewait.wait,
 					TASK_UNINTERRUPTIBLE);
-		xas_pause(xas);
+		xas_reset(xas);
Quoted 8 lines

fs/dax.c static void *wait_entry_unlocked_exclusive(struct xa_state *xas, void *entry) · line 263

Quoted 3 lines
fs/dax.c +0 -0
 		xas_unlock_irq(xas);
 		schedule();
 		finish_wait(wq, &ewait.wait);

This looks super-subtle, but so did the original fix commit 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries"). The resolution is the same to make sure the xarray state does not mistakenly advance when the lock is dropped.

Yes, the XArray iterator code is super subtle. The fact it took at least two engineers three attempts which were also reviewed to get this (hopefully!) right suggests it's certainly not obvious.

That said I don't have any good suggestions for making it better other than renaming functions which will create a lot of churn.

Quoted 3 lines

You can add:

Thanks!

maintainer Applied
#4 lore raw
In reply to Alistair Popple
Quoted 13 lines · Alistair Popple

Commit 6be3e21d25ca ("fs/dax: don't skip locked entries when scanning entries") introduced a new function, wait_entry_unlocked_exclusive(), which waits for the current entry to become unlocked without advancing the XArray iterator state.

Waiting for the entry to become unlocked requires dropping the XArray lock. This requires calling xas_pause() prior to dropping the lock which leaves the xas in a suitable state for the next iteration. However this has the side-effect of advancing the xas state to the next index. Normally this isn't an issue because xas_for_each() contains code to detect this state and thus avoid advancing the index a second time on the next loop iteration.

Quoted 1 line

[...]

Applied to the vfs.fixes branch of the vfs/vfs.git tree. Patches in the vfs.fixes branch should appear in linux-next soon.

Please report any outstanding bugs that were missed during review in a new review to the original patch series allowing us to drop it.

It's encouraged to provide Acked-bys and Reviewed-bys even though the patch has now been applied. If possible patch trailers will be updated.

Note that commit hashes shown below are subject to change due to rebase, trailer updates or similar. If in doubt, please check the listed branch.

tree:   https://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs.git
branch: vfs.fixes

[1/1] fs/dax: Fix "don't skip locked entries when scanning entries"

      https://git.kernel.org/vfs/vfs/c/dd59137bfe70