| Message ID | 20220627073311.2800330-1-amir73il@gmail.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | xfs stable candidate patches for 5.10.y (from v5.13) | expand |
On Mon, Jun 27, 2022 at 10:33:04AM +0300, Amir Goldstein wrote: > Hi all, > > This is a resend of the series that was posted 3 weeks ago [v1]. > The backports in this series are from circa v5.12..v5.13. > The remaining queue of tested 5.10 backports [1] contains 25 more patches > from v5.13..v5.19-rc1. > > There have been no comments on the first post except for Dave's request > to collaborate the backports review process with Leah who had earlier > sent out another series of backports for 5.15.y. > > Following Dave's request, I had put this series a side to collaborate > the shared review of 5.15/5.10 series with Leah and now that the shared > series has been posted to stable, I am re-posting to request ACKs on this > 5.10.y specific series. > > There are four user visible fixes in this series, one patch for dependency > ("rename variable mp") and two patches to improve testability of LTS. Aha, I had wondered why the journal_info thing was in this branch, and if that would even fit under the usual stable rules... > Specifically, I selected the fix ("use current->journal_info for > detecting transaction recursion") after I got a false positive assert > while testing LTS kernel with XFS_DEBUG and at another incident, it > helped me triage a regression that would have been harder to trace > back to the offending code otherwise. ...but clearly maintainers have been hitting this, so that's ok by /me/ to have it. If nothing else, XFS doesn't support nested transactions, so any weird stuff that falls out was already a dangerous bug. Acked-by: Darrick J. Wong <djwong@kernel.org> --D > This series has been looping in kdevops for a long while, with and > without the shared 5.15 backport with no regressions observed. > > Thanks, > Amir. > > Changes since [v1]: > - Rebased and tested on top of the v5.15+ ACKed backports [2] > > [1] https://github.com/amir73il/linux/commits/xfs-5.10.y > [2] https://lore.kernel.org/linux-xfs/20220624063702.2380990-1-amir73il@gmail.com/ > [v1] https://lore.kernel.org/linux-xfs/20220606160537.689915-1-amir73il@gmail.com/ > > Anthony Iliopoulos (1): > xfs: fix xfs_trans slab cache name > > Darrick J. Wong (1): > xfs: fix xfs_reflink_unshare usage of filemap_write_and_wait_range > > Dave Chinner (2): > xfs: use current->journal_info for detecting transaction recursion > xfs: update superblock counters correctly for !lazysbcount > > Gao Xiang (1): > xfs: ensure xfs_errortag_random_default matches XFS_ERRTAG_MAX > > Pavel Reichl (2): > xfs: rename variable mp to parsing_mp > xfs: Skip repetitive warnings about mount options > > fs/iomap/buffered-io.c | 7 --- > fs/xfs/libxfs/xfs_btree.c | 12 +++- > fs/xfs/libxfs/xfs_sb.c | 16 ++++- > fs/xfs/xfs_aops.c | 17 +++++- > fs/xfs/xfs_error.c | 2 + > fs/xfs/xfs_reflink.c | 3 +- > fs/xfs/xfs_super.c | 120 +++++++++++++++++++++----------------- > fs/xfs/xfs_trans.c | 23 +++----- > fs/xfs/xfs_trans.h | 30 ++++++++++ > 9 files changed, 148 insertions(+), 82 deletions(-) > > -- > 2.25.1 >
On Wed, Jun 29, 2022 at 7:20 PM Darrick J. Wong <djwong@kernel.org> wrote: > > On Mon, Jun 27, 2022 at 10:33:04AM +0300, Amir Goldstein wrote: > > Hi all, > > > > This is a resend of the series that was posted 3 weeks ago [v1]. > > The backports in this series are from circa v5.12..v5.13. > > The remaining queue of tested 5.10 backports [1] contains 25 more patches > > from v5.13..v5.19-rc1. > > > > There have been no comments on the first post except for Dave's request > > to collaborate the backports review process with Leah who had earlier > > sent out another series of backports for 5.15.y. > > > > Following Dave's request, I had put this series a side to collaborate > > the shared review of 5.15/5.10 series with Leah and now that the shared > > series has been posted to stable, I am re-posting to request ACKs on this > > 5.10.y specific series. > > > > There are four user visible fixes in this series, one patch for dependency > > ("rename variable mp") and two patches to improve testability of LTS. > > Aha, I had wondered why the journal_info thing was in this branch, and > if that would even fit under the usual stable rules... > > > Specifically, I selected the fix ("use current->journal_info for > > detecting transaction recursion") after I got a false positive assert > > while testing LTS kernel with XFS_DEBUG and at another incident, it > > helped me triage a regression that would have been harder to trace > > back to the offending code otherwise. > > ...but clearly maintainers have been hitting this, so that's ok by /me/ to > have it. If nothing else, XFS doesn't support nested transactions, so any > weird stuff that falls out was already a dangerous bug. Exactly. > > Acked-by: Darrick J. Wong <djwong@kernel.org> > I am not going to post this to xfs list again with Acked-by before posting to stable, because this is the second posting already with no changes since v1. I am going to wait until Greg picks up the already posted series for 5.10 and 5.15 - it looks like he is also on vacation... Thanks! Amir.
Hi all, This is a resend of the series that was posted 3 weeks ago [v1]. The backports in this series are from circa v5.12..v5.13. The remaining queue of tested 5.10 backports [1] contains 25 more patches from v5.13..v5.19-rc1. There have been no comments on the first post except for Dave's request to collaborate the backports review process with Leah who had earlier sent out another series of backports for 5.15.y. Following Dave's request, I had put this series a side to collaborate the shared review of 5.15/5.10 series with Leah and now that the shared series has been posted to stable, I am re-posting to request ACKs on this 5.10.y specific series. There are four user visible fixes in this series, one patch for dependency ("rename variable mp") and two patches to improve testability of LTS. Specifically, I selected the fix ("use current->journal_info for detecting transaction recursion") after I got a false positive assert while testing LTS kernel with XFS_DEBUG and at another incident, it helped me triage a regression that would have been harder to trace back to the offending code otherwise. This series has been looping in kdevops for a long while, with and without the shared 5.15 backport with no regressions observed. Thanks, Amir. Changes since [v1]: - Rebased and tested on top of the v5.15+ ACKed backports [2] [1] https://github.com/amir73il/linux/commits/xfs-5.10.y [2] https://lore.kernel.org/linux-xfs/20220624063702.2380990-1-amir73il@gmail.com/ [v1] https://lore.kernel.org/linux-xfs/20220606160537.689915-1-amir73il@gmail.com/ Anthony Iliopoulos (1): xfs: fix xfs_trans slab cache name Darrick J. Wong (1): xfs: fix xfs_reflink_unshare usage of filemap_write_and_wait_range Dave Chinner (2): xfs: use current->journal_info for detecting transaction recursion xfs: update superblock counters correctly for !lazysbcount Gao Xiang (1): xfs: ensure xfs_errortag_random_default matches XFS_ERRTAG_MAX Pavel Reichl (2): xfs: rename variable mp to parsing_mp xfs: Skip repetitive warnings about mount options fs/iomap/buffered-io.c | 7 --- fs/xfs/libxfs/xfs_btree.c | 12 +++- fs/xfs/libxfs/xfs_sb.c | 16 ++++- fs/xfs/xfs_aops.c | 17 +++++- fs/xfs/xfs_error.c | 2 + fs/xfs/xfs_reflink.c | 3 +- fs/xfs/xfs_super.c | 120 +++++++++++++++++++++----------------- fs/xfs/xfs_trans.c | 23 +++----- fs/xfs/xfs_trans.h | 30 ++++++++++ 9 files changed, 148 insertions(+), 82 deletions(-)