On Wed, Feb 19, 2025 at 06:35:44AM +0800, Anand Jain wrote: > From: Qu Wenruo <wqu@xxxxxxxx> > > Update comments that were previously missed. > > Signed-off-by: Anand Jain <anand.jain@xxxxxxxxxx> > --- > tests/btrfs/226 | 6 ++---- > 1 file changed, 2 insertions(+), 4 deletions(-) > > diff --git a/tests/btrfs/226 b/tests/btrfs/226 > index 359813c4f394..ce53b7d48c49 100755 > --- a/tests/btrfs/226 > +++ b/tests/btrfs/226 > @@ -22,10 +22,8 @@ _require_xfs_io_command fpunch > > _scratch_mkfs >>$seqres.full 2>&1 > > -# This test involves RWF_NOWAIT direct IOs, but for inodes with data checksum, > -# btrfs will fall back to buffered IO unconditionally to prevent data checksum > -# mimsatch, and that will break RWF_NOWAIT with -EAGAIN. > -# So here we have to go with nodatasum mount option. > +# RWF_NOWAIT works only with direct I/O and requires an inode with nodatasum > +# to avoid checksum mismatches. Otherwise, it falls back to buffered I/O. As a supplement patch, this is good to me. Reviewed-by: Zorro Lang <zlang@xxxxxxxxxx> > _scratch_mount -o nodatasum > > # Test a write against COW file/extent - should fail with -EAGAIN. Disable the > -- > 2.47.0 > >
