Hide Forgot
XFS has many post-4.5.0 patches in RHEL7.3 which are not present in the kernel-aarch64 package for RHELSA. To be feature-and-bug compatible with RHEL7.3, we should pull in the following patches from RHEL7.3 (which are all also upstream, other than one rhel7-only change related to structure packing): 11dd85c [fs] revert "libxfs: pack the agfl header structure so XFS_AGFL_SIZE is correct" 9d5bb1c [fs] xfs: track and serialize in-flight async buffers against unmount 7249189 [fs] xfs: exclude never-released buffers from buftarg I/O accounting 0cc5a4e [fs] xfs: don't reset b_retries to 0 on every failure 8de549c [fs] xfs: remove extraneous buffer flag changes 2464a5a [fs] xfs: fix xfs_error_get_cfg for negative errnos 699b410 [fs] xfs: cancel eofblocks background trimming on remount read-only f3c91b3 [fs] xfs: add "fail at unmount" error handling configuration 93db2e3 [fs] xfs: add configuration handlers for specific errors eea7c0e [fs] xfs: add configuration of error failure speed 665896f5 [fs] xfs: introduce table-based init for error behaviors ec77d49 [fs] xfs: add configurable error support to metadata buffers 7bae874 [fs] xfs: introduce metadata IO error class 060eb6b [fs] xfs: configurable error behavior via sysfs d60c507 [fs] xfs: disallow rw remount on fs with unknown ro-compat features ef1bb0b [fs] xfs: fix xfs_log_ticket leak in xfs_end_io() after fs shutdown 183451a [fs] xfs: clean up unwritten buffers on write failure 2629608 [fs] xfs: Split default quota limits by quota type 5249c75 [fs] quota: Fixup comments about return value of Q_[X]GETNEXTQUOTA d43e09e [fs] xfs: wire up Q_XGETNEXTQUOTA / get_nextdqblk 65a0e4b [fs] xfs: Factor xfs_seek_hole_data into helper 4684c1e [fs] xfs: get quota inode from mp & flags rather than dqp e603d41 [fs] xfs: don't overflow quota ID when initializing dqblk 66f264c [fs] quota: add new quotactl Q_GETNEXTQUOTA bbd97f0 [fs] quota: add new quotactl Q_XGETNEXTQUOTA b1d0d66 [fs] quota: remove unused cmd argument from quota_quotaon() This series adds 2 features: * configurable error handling, needed for proper behavior on thinly provisioned storage * "getnextquota" interface for efficiently querying quota with large user databases and several bugfixes cherry-picked from upstream to avoid oopses and corruptions. I have this patch series prepared and tested via xfstests.
Eric et al. Thank you for what you have done with these xfs bugs. Does this patchset fix the following: bz1365979 XFS: Metadata corruption detected at xfs_dir3_data_write_verify bz1368200 aarch64: XFS free block count corruption They are both blockers right now. There are a number of other xfs issues too.
(In reply to John Feeney from comment #1) > Eric et al. > > Thank you for what you have done with these xfs bugs. > > Does this patchset fix the following: > bz1365979 XFS: Metadata corruption detected at xfs_dir3_data_write_verify Not certain on this one, with the gcc bug and everything, I didn't really dig into it. But based on Brian's comment on a similar bug at https://bugzilla.redhat.com/show_bug.cgi?id=1362730#c27 I think we are still working this out upstream. > bz1368200 aarch64: XFS free block count corruption Yes, this one should be fixed. -Eric > They are both blockers right now. > > There are a number of other xfs issues too. Understood; triaging everything twice is tough. Perhaps a scratch build and a request for a retest of outstanding issues would be in order, and if resolved, those bugs can be dup'd to this one, or whatever bookkeeping needs to be done...
I suppose it's possible that bz1365979 XFS: Metadata corruption detected at xfs_dir3_data_write_verify could also be related to the gcc miscompilation problem, but I can't say for sure.
*** Bug 1368622 has been marked as a duplicate of this bug. ***
*** Bug 1368681 has been marked as a duplicate of this bug. ***
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHBA-2016-2145.html