In some deployments, an empty lowerdir is not considered an error.
The current implementation logs this scenario as an error during boot,
which can be misleading and cause unnecessary concern for users. This
commit changes the log level from pr_err to pr_info to reflect the
non-error nature of an empty lowerdir in these cases.
Reported-by: Wei Wang <[email protected]>
Signed-off-by: Eric Curtin <[email protected]>
---
fs/overlayfs/params.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/overlayfs/params.c b/fs/overlayfs/params.c
index 4860fcc4611bb..53170d73bb79c 100644
--- a/fs/overlayfs/params.c
+++ b/fs/overlayfs/params.c
@@ -231,7 +231,7 @@ static int ovl_mount_dir_noesc(const char *name, struct path *path)
int err = -EINVAL;
if (!*name) {
- pr_err("empty lowerdir\n");
+ pr_info("empty lowerdir\n");
goto out;
}
err = kern_path(name, LOOKUP_FOLLOW, path);
--
2.45.0
On Tue, 28 May 2024 at 11:03, Eric Curtin <[email protected]> wrote:
>
> In some deployments, an empty lowerdir is not considered an error.
I don't think this can be triggered in upstream kernel and can be
removed completely.
Or do you have a reproducer?
Thanks,
Miklos
On Tue, 28 May 2024 at 11:34, Miklos Szeredi <[email protected]> wrote:
>
> On Tue, 28 May 2024 at 11:03, Eric Curtin <[email protected]> wrote:
> >
> > In some deployments, an empty lowerdir is not considered an error.
>
> I don't think this can be triggered in upstream kernel and can be
> removed completely.
True... Just switched to Fedora Rawhide and instead we just see this one:
pr_err("cannot append lower layer");
>
> Or do you have a reproducer?
Run one of these vms:
https://github.com/osbuild/bootc-image-builder
And on boot:
sudo bootc switch quay.io/fedora/fedora-bootc:rawhide && sudo reboot
is the reproducer that is closest to upstream
Is mise le meas/Regards,
Eric Curtin
>
> Thanks,
> Miklos
>
On Tue, 28 May 2024 at 12:55, Eric Curtin <[email protected]> wrote:
>
> On Tue, 28 May 2024 at 11:34, Miklos Szeredi <[email protected]> wrote:
> >
> > On Tue, 28 May 2024 at 11:03, Eric Curtin <[email protected]> wrote:
> > >
> > > In some deployments, an empty lowerdir is not considered an error.
> >
> > I don't think this can be triggered in upstream kernel and can be
> > removed completely.
>
> True... Just switched to Fedora Rawhide and instead we just see this one:
>
> pr_err("cannot append lower layer");
>
> >
> > Or do you have a reproducer?
>
> Run one of these vms:
>
> https://github.com/osbuild/bootc-image-builder
Apparently it is using the legacy lowerdir append mode
"lowerdir=:foo". This works only on 6.5.
In 6.6 and later the same can be achieved with "lowerdir+=foo".
It's strange that there are not side effects other then the error message.
Thanks,
Miklos
On Tue, 28 May 2024 at 13:23, Alexander Larsson <[email protected]> wrote:
>
>
>
> On Tue, May 28, 2024 at 1:34 PM Miklos Szeredi <[email protected]> wrote:
>>
>> On Tue, 28 May 2024 at 12:55, Eric Curtin <[email protected]> wrote:
>> >
>> > On Tue, 28 May 2024 at 11:34, Miklos Szeredi <[email protected]> wrote:
>> > >
>> > > On Tue, 28 May 2024 at 11:03, Eric Curtin <[email protected]> wrote:
>> > > >
>> > > > In some deployments, an empty lowerdir is not considered an error.
>> > >
>> > > I don't think this can be triggered in upstream kernel and can be
>> > > removed completely.
>> >
>> > True... Just switched to Fedora Rawhide and instead we just see this one:
>> >
>> > pr_err("cannot append lower layer");
>> >
>> > >
>> > > Or do you have a reproducer?
>> >
>> > Run one of these vms:
>> >
>> > https://github.com/osbuild/bootc-image-builder
>>
>> Apparently it is using the legacy lowerdir append mode
>> "lowerdir=:foo". This works only on 6.5.
>>
>> In 6.6 and later the same can be achieved with "lowerdir+=foo".
>>
>> It's strange that there are not side effects other then the error message.
>
>
> The code tries to use the new mode, but then falls back on ENOSYS:
>
> https://github.com/containers/composefs/blob/main/libcomposefs/lcfs-mountc#L431
>
> So, I guess with a more recent kernel it will not print the warning.
>
> --
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
> Alexander Larsson Red Hat, Inc
> [email protected] [email protected]
I realized I never posted the version of the kernel producing this
error message:
pr_err("cannot append lower layer");
So just for clarity in general, that was this one:
Linux fedora 6.9.0-64.fc41.x86_64 #1 SMP PREEMPT_DYNAMIC Mon May 13
11:58:46 UTC 2024 x86_64 GNU/Linux
Is mise le meas/Regards,
Eric Curtin