On Wed, Jan 27, 2021 at 4:58 AM Eryu Guan <eguan@xxxxxxxxxxxxxxxxx> wrote:
>
> On Mon, Jan 25, 2021 at 03:24:07PM +0200, Amir Goldstein wrote:
> > On Sat, Jan 16, 2021 at 6:56 PM Amir Goldstein <amir73il@xxxxxxxxx> wrote:
> > >
> > > Chengguang Xu reported [1] that append-only flag is lost on copy-up.
> > > I had noticed that for directories, immutable flag can also be lost on
> > > copy up (when parent is copied up). That's an old overlayfs bug.
> > >
> > > Overlayfs added the ability to set inode flags (e.g. chattr +i) in
> > > kernel 5.10 by commit 61536bed2149 ("ovl: support [S|G]ETFLAGS and
> > > FS[S|G]ETXATTR ioctls for directories").
> > > Icenowy Zheng reported [2] a regression in that commit that causes
> > > a deadlock when setting inode flags on lower dir.
> > >
> > > There is a commented line in the test that triggers this deadlock,
> > > but it has been left commented out until a fix is merged upstream.
> > >
> >
> > Re-iterate in correct thread:
> >
> > The fix for above is in overlayfs-next:
> >
> > * 147ec02b8705 - ovl: avoid deadlock on directory ioctl
> >
> > But I wouldn't uncomment that line in the test just yet.
>
> Then I'd prefer wait for the deadlock fix land in upstream first, and
> merge the test with the deadlock trigger in place.
>
> Or as you mentioned in previous thread, we could seperate the deadlock
> case as a new test (also remove it from current overlay/075), so we
> could merge the [s,g]etxflags case first, then the deadlock case only
> when the fix is upstreamd.
>
> Either way works for me, I just want to avoid merging the test without
> the deadlock trigger, then uncomment it when the fix is available.
>
So let's wait for the deadlock fix to land and I will re-post with split
tests and address your minor comment.
Thanks,
Amir.
