Jan Kiszka <jan.kis...@siemens.com> writes: > Otherwise we may start processing sockets in slirp_pollfds_poll that > were created past slirp_pollfds_fill. > > Signed-off-by: Jan Kiszka <jan.kis...@siemens.com>
I think there's still something wrong... Somehow, my char flow series breaks slirp. Prior to the g_poll conversion, this was not the case. I'll look into it more later and post a git tree but FYI. I suspect moving the chardevs to GIOChannel is uncovering a latent bug in the slirp main loop interaction. Regards, Anthony Liguori > --- > > Not sure if this pattern also applies to other users besides slirp. > Worth checking I suppose. > > slirp/socket.c | 1 + > 1 files changed, 1 insertions(+), 0 deletions(-) > > diff --git a/slirp/socket.c b/slirp/socket.c > index a7ab933..bb639ae 100644 > --- a/slirp/socket.c > +++ b/slirp/socket.c > @@ -51,6 +51,7 @@ socreate(Slirp *slirp) > so->so_state = SS_NOFDREF; > so->s = -1; > so->slirp = slirp; > + so->pollfds_idx = -1; > } > return(so); > } > -- > 1.7.3.4
