From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:46979) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YZO85-00029P-9k for qemu-devel@nongnu.org; Sat, 21 Mar 2015 14:28:14 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YZO82-0001Wg-3k for qemu-devel@nongnu.org; Sat, 21 Mar 2015 14:28:13 -0400 Received: from mx1.redhat.com ([209.132.183.28]:59422) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YZO81-0001WR-SK for qemu-devel@nongnu.org; Sat, 21 Mar 2015 14:28:10 -0400 Date: Sat, 21 Mar 2015 19:27:49 +0100 From: "Michael S. Tsirkin" Message-ID: <20150321192636-mutt-send-email-mst@redhat.com> References: <1426671309-13645-1-git-send-email-jasowang@redhat.com> <1426671309-13645-7-git-send-email-jasowang@redhat.com> <20150318140614-mutt-send-email-mst@redhat.com> <20150320083924.7267dfa2.cornelia.huck@de.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20150320083924.7267dfa2.cornelia.huck@de.ibm.com> Subject: Re: [Qemu-devel] [PATCH V4 06/19] virtio-ccw: using VIRTIO_NO_VECTOR instead of 0 for invalid virtqueue List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Cornelia Huck Cc: Christian Borntraeger , Jason Wang , qemu-devel@nongnu.org, Alexander Graf , Richard Henderson On Fri, Mar 20, 2015 at 08:39:24AM +0100, Cornelia Huck wrote: > On Wed, 18 Mar 2015 14:08:56 +0100 > "Michael S. Tsirkin" wrote: > > > On Wed, Mar 18, 2015 at 05:34:56PM +0800, Jason Wang wrote: > > > There's no need to use vector 0 for invalid virtqueue. So this patch > > > changes to use VIRTIO_NO_VECTOR instead. > > > > > > Cc: Michael S. Tsirkin > > > Cc: Cornelia Huck > > > CC: Christian Borntraeger > > > Cc: Richard Henderson > > > Cc: Alexander Graf > > > Signed-off-by: Jason Wang > > > > I don't know what does this actually do. > > Cornelia? > > I actually have the same patch somewhere in my queue. The point here is > that 0 is plain wrong (it's a valid queue), while VIRTIO_NO_VECTOR is > most certainly no valid queue. > > > > > > --- > > > hw/s390x/virtio-ccw.c | 2 +- > > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > > > diff --git a/hw/s390x/virtio-ccw.c b/hw/s390x/virtio-ccw.c > > > index 130535c..c8b87aa 100644 > > > --- a/hw/s390x/virtio-ccw.c > > > +++ b/hw/s390x/virtio-ccw.c > > > @@ -281,7 +281,7 @@ static int virtio_ccw_set_vqs(SubchDev *sch, uint64_t addr, uint32_t align, > > > > > > virtio_queue_set_addr(vdev, index, addr); > > > if (!addr) { > > > - virtio_queue_set_vector(vdev, index, 0); > > > + virtio_queue_set_vector(vdev, index, VIRTIO_NO_VECTOR); > > > } else { > > > /* Fail if we don't have a big enough queue. */ > > > /* TODO: Add interface to handle vring.num changing */ > > > > Right below this, we have > > /* tell notify handler in case of config change */ > > vdev->config_vector = VIRTIO_PCI_QUEUE_MAX; > > > > which also does not seem to make sense. > > Basically we have: > > - at most 64 virtqueues with their own indicators (always 64 indicator > bits when using classic I/O interrupts, up to 64 indicator bits when > using adapter interrupts) > - another indicator bit for configuration changes (bit 0 of the > secondary indicator bits) > > That way, the configuration change indicator is always one bit behind > the last possible queue indicator. But VIRTIO_PCI_QUEUE_MAX only makes sense as a VQ number. Why does it make sense as a vector number? Jason's patches actually change VIRTIO_PCI_QUEUE_MAX so we need to figure our what to do for this code. > > > > These changes need some testing though. > > My identical patch seemed to work for me.