On Thu, Sep 21, 2023 at 5:24 PM Zorro Lang <zlang@xxxxxxxxxx> wrote: > > On Thu, Sep 21, 2023 at 11:00:37AM +0300, Amir Goldstein wrote: > > On Thu, Sep 21, 2023 at 9:26 AM Zorro Lang <zlang@xxxxxxxxxx> wrote: > > > > > > On Wed, Sep 20, 2023 at 06:34:21PM +0300, Amir Goldstein wrote: > > > > On Wed, Sep 20, 2023 at 6:14 PM Zorro Lang <zlang@xxxxxxxxxx> wrote: > > > > > > > > > > On Wed, Sep 20, 2023 at 04:03:55PM +0300, Amir Goldstein wrote: > > > > > > A test for a regression from v5.15 reported by Ruiwen Zhao: > > > > > > https://lore.kernel.org/linux-unionfs/CAKd=y5Hpg7J2gxrFT02F94o=FM9QvGp=kcH1Grctx8HzFYvpiA@xxxxxxxxxxxxxx/ > > > > > > Could you give one more sentence to tell what kind of regression > > > does this case test for? Not only a link address. > > > > > > > > > > > > > > > Signed-off-by: Amir Goldstein <amir73il@xxxxxxxxx> > > > > > > --- > > > > > > > > > > > > Zorro, > > > > > > > > > > > > This is a test for a regression in kernel v5.15. > > > > > > The fix was merged for 6.6-rc2 and has been picked for > > > > > > the upcoming LTS releases 5.15, 6.1, 6.5. > > > > > > > > > > > > The reproducer only manifests the bug in fs that inherit noatime flag, > > > > > > namely ext4, btrfs, ... but not xfs. > > > > FYI, I made a mistake in the statement above. > > xfs does support inherit of noatime flag, but > > it does not inherit noatime for *symlinks*. > > > > I added the _require_chattr_inherit helper that you suggested > > in v2, but it only checks for inherit of noatime flag (the 2nd _notrun). > > I did not add a helper for _require_chattr_inherit_symlink > > because it was too specific and so I left the 3rd _notrun > > open coded in the test in v2. > > OK, if xfs thinks it's an expected behavior which won't be changed :) > I find it quite strange that symlink and special inodes inherit those attributes, but then the attributes cannot be queried and not changed on those inodes. It seems to me like all the fs expect xfs have it wrong, but I don't think that is likely to change. Thanks, Amir.
