On Mon, Oct 2, 2023 at 8:56 AM Amir Goldstein <amir73il@xxxxxxxxx> wrote:
>
> On Mon, Oct 2, 2023 at 8:47 AM Amir Goldstein <amir73il@xxxxxxxxx> wrote:
> >
> > On Mon, Oct 2, 2023 at 5:36 AM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> > >
> > > d_inode_rcu() is right - we might be in rcu pathwalk;
> > > however, OVL_E() hides plain d_inode() on the same dentry...
> > >
> > > Signed-off-by: Al Viro <viro@xxxxxxxxxxxxxxxxxx>
> >
> > Reviewed-by: Amir Goldstein <amir73il@xxxxxxxxx>
> >
> > However, ovl_lowerstack(oe) does not appear to be stable in RCU walk...
> >
>
> Ah, you fixed that in another patch.
> If you are going to be sending this to Linus, please add
> Fixes: a6ff2bc0be17 ("ovl: use OVL_E() and OVL_E_FLAGS() accessors")
>
> I was going to send some fixes this week anyway, so I can
> pick those through the overlayfs tree if you like.
>
Al,
>From all your series, the two ovl fixes are for rather new regressions (v6.5)
so I queued your two regression fixes (13,14) and my own version of patch 15
to go into linux-next via overlayfs tree [1] and I will send them to Linus later
this week, so they can make their way to 6.5.y.
Thanks,
Amir.
[1] https://git.kernel.org/pub/scm/linux/kernel/git/overlayfs/vfs.git/log/?h=ovl-fixes
