On (Wed) 23 Mar 2011 [14:33:25], Juan Quintela wrote:
> Amit Shah <[email protected]> wrote:
> > After a port unplug operation, the port->info->have_data() pointer was
> > set to NULL. The problem is, the ->info struct is shared by all ports,
> > effectively disabling writes to other ports.
> >
> > Reported-by: juzhang <[email protected]>
> > Signed-off-by: Amit Shah <[email protected]>
> > ---
> > hw/virtio-console.c | 1 -
> > 1 files changed, 0 insertions(+), 1 deletions(-)
> >
> > diff --git a/hw/virtio-console.c b/hw/virtio-console.c
> > index 4440784..be59558 100644
> > --- a/hw/virtio-console.c
> > +++ b/hw/virtio-console.c
> > @@ -82,7 +82,6 @@ static int virtconsole_exitfn(VirtIOSerialPort *port)
> > VirtConsole *vcon = DO_UPCAST(VirtConsole, port, port);
> >
> > if (vcon->chr) {
> > - port->info->have_data = NULL;
> > qemu_chr_close(vcon->chr);
> > }
>
> Discussed with Amit over irc, I think that we are missing setup of
> have_data for non console devices, but that is a different bug that the
> one being fixed here.
Actually other virtio_serial devices will provide their own have_data,
like spice did earlier (now it's a chardev, so it uses this code
path).
I think the bug is that we should set have_data regardless of a chardev
backend and call qemu_chr_write() in have_data only if a chardev
exists.
Amit
