Re: [Qemu-devel] [PATCH 1/2] libvhost-user: fix vu_queue_get_avail_bytes() crash

[Philippe Mathieu-Daudé]

> > On 05/03/2017 09:11 AM, Marc-André Lureau wrote:
> > > Calling vu_queue_get_avail_bytes() when the queue doesn't yet have
> > > addresses will result in the following crash:
> > >
> > > Program received signal SIGSEGV, Segmentation fault.
> > > 0x000055c414112ce4 in vring_avail_idx (vq=0x55c41582fd68,
> > > vq=0x55c41582fd68)
> > >     at /home/dgilbert/git/qemu/contrib/libvhost-user/libvhost-user.c:940
> > > 940            vq->shadow_avail_idx = vq->vring.avail->idx;
> > > (gdb) p vq
> > > $1 = (VuVirtq *) 0x55c41582fd68
> > > (gdb) p vq->vring
> > > $2 = {num = 0, desc = 0x0, avail = 0x0, used = 0x0, log_guest_addr = 0,
> > > flags = 0}
> > >
> > >     at /home/dgilbert/git/qemu/contrib/libvhost-user/libvhost-user.c:940
> > > No locals.
> > >     at /home/dgilbert/git/qemu/contrib/libvhost-user/libvhost-user.c:960
> > >         num_heads = <optimized out>
> > >     address@hidden,
> > >     address@hidden,
> > >     address@hidden) at
> > >     /home/dgilbert/git/qemu/contrib/libvhost-user/libvhost-user.c:1034
> > >
> > > Check if vring.avail != null before accessing it.
> > >
> > > Signed-off-by: Marc-André Lureau <address@hidden>
> > > ---
> > >  contrib/libvhost-user/libvhost-user.c | 4 ++++
> > >  1 file changed, 4 insertions(+)
> > >
> > > diff --git a/contrib/libvhost-user/libvhost-user.c
> > > b/contrib/libvhost-user/libvhost-user.c
> > > index af4faad60b..f9680b6279 100644
> > > --- a/contrib/libvhost-user/libvhost-user.c
> > > +++ b/contrib/libvhost-user/libvhost-user.c
> > > @@ -1031,6 +1031,10 @@ vu_queue_get_avail_bytes(VuDev *dev, VuVirtq *vq,
> > > unsigned int *in_bytes,
> > >      idx = vq->last_avail_idx;
> > >
> > >      total_bufs = in_total = out_total = 0;
> > > +    if (!vq->vring.avail) {
> > > +        goto done;
> > > +    }
> > > +
> > >      while ((rc = virtqueue_num_heads(dev, vq, idx)) > 0) {
> > >          unsigned int max, num_bufs, indirect = 0;
> > >          struct vring_desc *desc;
> >
> > It seems to me safer to fix instead vring_avail_ring(), and fix
> > neighbours vring_avail_flags() and vring_avail_idx() while here.
>
> Those are internal/static functions, possibly on hot paths. So I would rather 
> keep the check on the external/public functions only.

Fair enough.

Reviewed-by: Philippe Mathieu-Daudé <address@hidden>