Re: [PATCH v2 0/3] migration/file: fix type mismatch and NULL deref in multifd_file_recv_data

[Peter Xu <peterx@redhat.com>]

On Tue, Mar 24, 2026 at 10:54:38AM +0000, Daniel P. Berrangé wrote:
> On Tue, Mar 24, 2026 at 04:27:38PM +0800, Junjie Cao wrote:
> > Hi Peter, Daniel,
> > 
> > I wanted to follow up on this series.  Looking back, I went
> > straight to adding new public API in io/channel.  I should 
> > have confirmed my approach and whether the Suggested-by
> > tag was appropriate here with you before sending.  Sorry for that.
> > 
> > I see two possible paths and would appreciate your guidance:
> > 
> >   (a) Keep it purely a fix in migration/file.c: add a local
> >       retry loop around qio_channel_pread() in
> >       multifd_file_recv_data(), fix the ssize_t/size_t
> >       mismatch, and use ERRP_GUARD() so error_prepend()
> >       is safe.
> > 
> >   (b) If pread_all() retry helpers are worth having as a
> >       general facility in io/channel, I'm happy to propose
> >       them as a separate RFC so we can discuss the API design
> >       on its own merits.
> > 
> > I'm leaning toward (a) and can send a v3 quickly.
> > Does that sound right?
> 
> Doing (b) is the right approach. We want these helpers in the
> general code, so we don't create technical debt that has to be
> fixed next time something else wants the same APIs.

Sorry for a late response, Junjie.  I agree with Dan.

-- 
Peter Xu