On Mon, Oct 21, 2024 at 2:22 PM Christian Brauner <brauner@xxxxxxxxxx> wrote:
>
> On Mon, Oct 21, 2024 at 01:58:16PM +0200, Amir Goldstein wrote:
> > On Mon, Oct 21, 2024 at 12:33 PM Miklos Szeredi <mszeredi@xxxxxxxxxx> wrote:
> > >
> > > - Pass iocb to ctx->end_write() instead of file + pos
> > >
> > > - Get rid of ctx->user_file, which is redundant most of the time
> > >
> > > - Instead pass iocb to backing_file_splice_read and
> > > backing_file_splice_write
> > >
> > > Signed-off-by: Miklos Szeredi <mszeredi@xxxxxxxxxx>
> > > ---
> > > v2:
> > > Pass ioctb to backing_file_splice_{read|write}()
> > >
> > > Applies on fuse.git#for-next.
> >
> > This looks good to me.
> > you may add
> > Reviewed-by: Amir Goldstein <amir73il@xxxxxxxxx>
> >
> > However, this conflicts with ovl_real_file() changes on overlayfs-next
> > AND on the fixes in fuse.git#for-next, so we will need to collaborate.
> >
> > Were you planning to send the fuse fixes for the 6.12 cycle?
> > If so, I could rebase overlayfs-next over 6.12-rcX after fuse fixes
> > are merged and then apply your patch to overlayfs-next and resolve conflicts.
>
> Wouldn't you be able to use a shared branch?
>
> If you're able to factor out the backing file changes I could e.g.,
> provide you with a base branch that I'll merge into vfs.file, you can
> use either as base to overlayfs and fuse or merge into overlayfs and
> fuse and fix any potential conflicts. Both works and my PRs all go out
> earlier than yours anyway.
Yes, but the question remains, whether Miklos wants to send the fuse
fixes to 6.12-rcX or to 6.13.
I was under the impression that he was going to send them to 6.12-rcX
and this patch depends on them.
I suggested to Miklos to squash this cleanup patch with the API change
for the fuse passthrough fix, which is a stable backport candidate, but
he preferred to not include this cleanup in the backport candidate patch.
Thanks,
Amir.
