On Mon, May 12, 2025 at 11:20:50PM +0200, Stefano Brivio wrote:
> On Mon, 12 May 2025 18:47:00 +0200
> Laurent Vivier <lvivier@redhat.com> wrote:
> 
> > In eea8a76caf85 ("flow: fix podman issue #26073"), we unregister
> > the fd from epoll_ctl() in case of error, but we also need to close it.
> > 
> > As flowside_sock_l4() also calls sock_l4_sa() via flowside_sock_splice()
> > we can do it unconditionally.
> 
> The previous PIF_HOST check made sense to me because it excluded
> PIF_TAP, not so much because it excluded PIF_SPLICE.

Right.  pif_is_socket() exists for exactly that case..

> But anyway, we'll never reach here with anything that's not PIF_HOST
> nor PIF_TAP (kind of by definition), and a spurious epoll_del() is
> surely better than the alternative in any case...

..but in this case it's not relevant because you have to already have
a socket by this point.

> 
> > Fixes: eea8a76caf85 ("flow: fix podman issue #26073")
> > Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> 
> Applied, thanks. New release coming soon.
> 

-- 
David Gibson (he or they)	| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you, not the other way
				| around.
http://www.ozlabs.org/~dgibson