| Message ID | 20240411234743.3464688-1-mcgrof@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [kdevops] xfs: update expunges with two new crashes one related to btrfs sparse files | expand |
diff --git a/workflows/fstests/expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_4k.txt b/workflows/fstests/expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_4k.txt index 5b12210a6194..12d6702668ab 100644 --- a/workflows/fstests/expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_4k.txt +++ b/workflows/fstests/expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_4k.txt @@ -6,6 +6,7 @@ generic/269 # F:1/11 generic/297 generic/298 generic/388 # Hangs. SOAK_DURATION=9900 +generic/447 # korg#218227 fsstress + compaction generic/455 # F:1/8 generic/471 generic/482 # F:1/4 diff --git a/workflows/fstests/expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt b/workflows/fstests/expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt index 1fb67ccacd81..aee7b603c9f2 100644 --- a/workflows/fstests/expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt +++ b/workflows/fstests/expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt @@ -8,6 +8,7 @@ generic/388 # Hangs. SOAK_DURATION=9900 generic/455 # F:1/12 generic/457 # F:1/6 generic/471 +generic/476 # Bad page state in workqueue btrfs-endio-write btrfs_work_helper SOAK_DURATION=9900 https://gist.github.com/mcgrof/76e026ca48b95922a365be3502fddf45 generic/482 # metadata corruption: https://gist.github.com/dagmcr/9fc650327b2dafd89fabd45ede0a932d generic/648 # https://gist.githubusercontent.com/dagmcr/447a5030dde1f91206604f1b8ee9d2a5/raw/6c8568284b6c5c05920a132c73c67fd86dcbe6f8/gistfile1.txt generic/670 diff --git a/workflows/fstests/expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_4k.txt b/workflows/fstests/expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_4k.txt index 5b12210a6194..12d6702668ab 100644 --- a/workflows/fstests/expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_4k.txt +++ b/workflows/fstests/expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_4k.txt @@ -6,6 +6,7 @@ generic/269 # F:1/11 generic/297 generic/298 generic/388 # Hangs. SOAK_DURATION=9900 +generic/447 # korg#218227 fsstress + compaction generic/455 # F:1/8 generic/471 generic/482 # F:1/4 diff --git a/workflows/fstests/expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt b/workflows/fstests/expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt index 1fb67ccacd81..aee7b603c9f2 100644 --- a/workflows/fstests/expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt +++ b/workflows/fstests/expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt @@ -8,6 +8,7 @@ generic/388 # Hangs. SOAK_DURATION=9900 generic/455 # F:1/12 generic/457 # F:1/6 generic/471 +generic/476 # Bad page state in workqueue btrfs-endio-write btrfs_work_helper SOAK_DURATION=9900 https://gist.github.com/mcgrof/76e026ca48b95922a365be3502fddf45 generic/482 # metadata corruption: https://gist.github.com/dagmcr/9fc650327b2dafd89fabd45ede0a932d generic/648 # https://gist.githubusercontent.com/dagmcr/447a5030dde1f91206604f1b8ee9d2a5/raw/6c8568284b6c5c05920a132c73c67fd86dcbe6f8/gistfile1.txt generic/670
One is already known, the other one seems new, and it is triggerable by using sparse files on a btrfs partition for testing xfs on test generic/476 (no SOAK_DURATION needed), so the bug is specific to btrfs. The bug is triggered on the btrfs-endio-write btrfs_work_helper workqueue and hits BUG_ON() on a dad page state on the call btrfs_release_extent_buffer_pages() [0]. We've done extensive testing of XFS on v6.6-rc5 and recently we moved to v6.8, and so this could be an issue introduced on v6.7 or v6.8. [0] https://gist.github.com/mcgrof/76e026ca48b95922a365be3502fddf45 Signed-off-by: Luis Chamberlain <mcgrof@kernel.org> --- This is a kdevops specific expunge patch, but I figured btrfs folks would be interested about hearing about this btrfs crash. Hopefully the above sumarizes enough to enable to folks who may want to reproduce to give it a shot. I am not yet sure of the failure rate as I just ran into it on my first shot on v6.8. A fix for the xfs specifically triggered fsstress + compaction crash is already on my radar but documenting it for posterity. .../fstests/expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_4k.txt | 1 + .../expunges/6.8.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt | 1 + .../fstests/expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_4k.txt | 1 + .../expunges/6.9.0-rc2/xfs/unassigned/xfs_reflink_stripe_len.txt | 1 + 4 files changed, 4 insertions(+)