| Message ID | cover.1708797432.git.fdmanana@suse.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | btrfs: some fiemap fixes | expand |
On Sun, Feb 25, 2024 at 07:51:23PM +0000, fdmanana@kernel.org wrote: > From: Filipe Manana <fdmanana@suse.com> > > There's a recent regression with fiemap due to a fix for a deadlock between > fiemap and memory mapped writes when the fiemap buffer is memory mapped to > the same file range, which leads to a race triggering a warning and making > fiemap fail. Plus one more long standing race when using FIEMAP_FLAG_SYNC. > Details in the change logs. > > V4: Updated patch 1/2, added a lot more comments about that's going on, > how each case is dealt with and why, added a missing handling for > a delalloc case that could result in emmiting overlapping ranges. > > V3: Deal with the case where offset == cache->offset which is also > possible if we had delalloc in the range of a hole or prealloc extent. > > V2: Updated patch 1/2 to deal with the case of a hole/prealloc extent > with multiple delalloc ranges inside it. > > Filipe Manana (2): > btrfs: fix race between ordered extent completion and fiemap > btrfs: ensure fiemap doesn't race with writes when FIEMAP_FLAG_SYNC is given > Eesh sorry about that, I must have missed the other test failure when I ran my fix through CI. Reviewed-by: Josef Bacik <josef@toxicpanda.com> Thanks, Josef
From: Filipe Manana <fdmanana@suse.com> There's a recent regression with fiemap due to a fix for a deadlock between fiemap and memory mapped writes when the fiemap buffer is memory mapped to the same file range, which leads to a race triggering a warning and making fiemap fail. Plus one more long standing race when using FIEMAP_FLAG_SYNC. Details in the change logs. V4: Updated patch 1/2, added a lot more comments about that's going on, how each case is dealt with and why, added a missing handling for a delalloc case that could result in emmiting overlapping ranges. V3: Deal with the case where offset == cache->offset which is also possible if we had delalloc in the range of a hole or prealloc extent. V2: Updated patch 1/2 to deal with the case of a hole/prealloc extent with multiple delalloc ranges inside it. Filipe Manana (2): btrfs: fix race between ordered extent completion and fiemap btrfs: ensure fiemap doesn't race with writes when FIEMAP_FLAG_SYNC is given fs/btrfs/extent_io.c | 124 ++++++++++++++++++++++++++++++++++++------- fs/btrfs/inode.c | 22 +++++++- 2 files changed, 125 insertions(+), 21 deletions(-)