On Wed, Feb 12, 2025 at 10:45 AM Jordi Pujol <jordipujolp@xxxxxxxxx> wrote:
>
> Hello,
>
> I allways work on a Live system that uses overlayfs, and also remaster
> the pristine filesystem using overlayfs.
> After upgrading kernel to version 6.12 have experienced several
> filesystem problems, thus have compared the overlayfs code of previous
> versions.
Thank you for the report, but we can do very little with the information
"experienced several filesystem problems"
Can you elaborate?
> By intuition, have found that these lines have been removed from
> version 6.11. This difference is key:
>
> overlayfs-sync-upper.patch
> --- linux-6.13.2/fs/overlayfs/super.c
> +++ linux-6.11.11/fs/overlayfs/super.c
> @@ -202,9 +202,15 @@
> int ret;
>
> ret = ovl_sync_status(ofs);
> -
> - if (ret < 0)
> + /*
> + * We have to always set the err, because the return value isn't
> + * checked in syncfs, and instead indirectly return an error via
> + * the sb's writeback errseq, which VFS inspects after this call.
> + */
> + if (ret < 0) {
> + errseq_set(&sb->s_wb_err, -EIO);
> return -EIO;
> + }
>
> if (!ret)
> return ret;
>
> In latest versions the filesystems work like a charm when applying
> previous patch
Please provide an objective comparison between the behavior of
"filesystem problems" vs. "filesystem work like a charm".
I am assuming that you are using the "volatile" overlayfs feature?
otherwise, unless I am missing something, the removal of code
above should not have had any effect.
The removed code, would have propagate the s_wb_err state from
the upper fs sb to overlayfs sb, but the only code that checks
s_wb_err state is syncfs() should be returning -EIO in this case anyway,
so I am not seeing where the change of behavior you are observing
is coming from.
Are you using a patched kernel or an out of tree filesystem
underneath overlayfs?
Thanks,
Amir.
