Re: [PATCH v2] vhsot-user: Do not wait for replay for not sent VHOST_USER_SET_LOG_BASE

[Michael S. Tsirkin]

On Sun, Jul 21, 2024 at 11:20:56AM +0800, BillXiang wrote:
> 
> > From: "Michael S. Tsirkin"<mst@redhat.com>
> > Date:  Sun, Jul 21, 2024, 03:01
> > Subject:  Re: [PATCH v2] vhsot-user: Do not wait for replay for not sent 
> > VHOST_USER_SET_LOG_BASE
> > To: "BillXiang"<xiangwencheng@dayudpu.com>
> > Cc: <qemu-devel@nongnu.org>
> > typos in subject do not inspire confidence.
> 
> Sorry
> 
> > 
> > On Thu, Jul 18, 2024 at 04:31:03PM +0800, BillXiang wrote:
> > > From: BillXiang <xiangwencheng@dayudpu.com>
> > > 
> > > We have added VHOST_USER_SET_LOG_BASE to vhost_user_per_device_request
> > > in https://lists.nongnu.org/archive/html/qemu-devel/2024-06/msg02559.html
> > 
> > Pls put commit here not ML link:
> > commit ABCDEFGHIKLM ("subject")
> > 
> > > and will send this message only for vq 0.
> > 
> > you mean "should send"?
> > 
> 
> What I mean is the commit 7c211eb078c4 ("vhost-user: Skip unnecessary 
> duplicated VHOST_USER_SET_LOG_BASE requests") 
> will cause VHOST_USER_SET_LOG_BASE to be sent only when 'vq_index == 0' in 
> vhost_user_write, 
> so that we can not use vhost_user_read to get reply when 'vq_index != 0'


do we try to do it now? in which configurations?


> > 
> > > 
> > > Signed-off-by: BillXiang <xiangwencheng@dayudpu.com>
> > > ---
> > > V1[1] -> V2:
> > >  - Refrain from appending flags to messages that could 
> > > precipitate validation failures upon execution of the 
> > > `vhost_user_read_header` function. 
> > > 
> > > [1]https://lists.nongnu.org/archive/html/qemu-devel/2024-07/msg01923.html
> > > ---
> > >  hw/virtio/vhost-user.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/hw/virtio/vhost-user.c b/hw/virtio/vhost-user.c
> > > index 00561daa06..fd12992d15 100644
> > > --- a/hw/virtio/vhost-user.c
> > > +++ b/hw/virtio/vhost-user.c
> > > @@ -460,7 +460,7 @@ static int vhost_user_set_log_base(struct vhost_dev 
> > > *dev, uint64_t base,
> > >          return ret;
> > >      }
> > >  
> > > -    if (shmfd) {
> > > +    if (shmfd && (dev->vq_index == 0)) {
> > 
> > extra () not needed here.
> > 
> > >          msg.hdr.size = 0;
> > >          ret = vhost_user_read(dev, &msg);
> > >          if (ret < 0) {
> > > -- 
> > > 2.30.0