Re: [PATCH v3 3/3] vduse: add F_QUEUE_READY feature

[Michael S. Tsirkin]

On Tue, Mar 24, 2026 at 03:01:47PM +0100, Eugenio Perez Martin wrote:
> On Fri, Mar 13, 2026 at 8:08 AM Eugenio Perez Martin
> <eperezma@xxxxxxxxxx> wrote:
> >
> > On Fri, Mar 13, 2026 at 7:05 AM Jason Wang <jasowang@xxxxxxxxxx> wrote:
> > >
> > > On Thu, Mar 12, 2026 at 2:24 PM Eugenio Perez Martin
> > > <eperezma@xxxxxxxxxx> wrote:
> > > >
> > > > On Thu, Mar 12, 2026 at 4:58 AM Jason Wang <jasowang@xxxxxxxxxx> wrote:
> > > > >
> > > > > On Wed, Mar 11, 2026 at 3:08 AM Eugenio Pérez <eperezma@xxxxxxxxxx> wrote:
> > > > > >
> > > > > > Add the VDUSE_F_QUEUE_READY feature flag. This allows the kernel module
> > > > > > to explicitly signal userspace when a specific virtqueue has been
> > > > > > enabled.
> > > > > >
> > > > > > In scenarios like Live Migration of VirtIO net devices, the dataplane
> > > > > > starts after the control virtqueue allowing QEMU to apply configuration
> > > > > > in the destination device.
> > > > > >
> > > > > > Signed-off-by: Eugenio Pérez <eperezma@xxxxxxxxxx>
> > > > > > ---
> > > > > > v2:
> > > > > > * Fix comment of vduse_dev_request.vq_ready
> > > > > > * Set vq_ready before sending the message to the VDUSE userland
> > > > > >   instance, avoiding the need for SMP sync after receiving the message.
> > > > > > ---
> > > > > >  drivers/vdpa/vdpa_user/vduse_dev.c | 28 +++++++++++++++++++++++++++-
> > > > > >  include/uapi/linux/vduse.h         | 18 ++++++++++++++++++
> > > > > >  2 files changed, 45 insertions(+), 1 deletion(-)
> > > > > >
> > > > > > diff --git a/drivers/vdpa/vdpa_user/vduse_dev.c b/drivers/vdpa/vdpa_user/vduse_dev.c
> > > > > > index 17e0358d3a68..4f642b95a7cb 100644
> > > > > > --- a/drivers/vdpa/vdpa_user/vduse_dev.c
> > > > > > +++ b/drivers/vdpa/vdpa_user/vduse_dev.c
> > > > > > @@ -9,6 +9,7 @@
> > > > > >   */
> > > > > >
> > > > > >  #include "linux/virtio_net.h"
> > > > > > +#include <linux/bits.h>
> > > > > >  #include <linux/cleanup.h>
> > > > > >  #include <linux/init.h>
> > > > > >  #include <linux/module.h>
> > > > > > @@ -53,7 +54,7 @@
> > > > > >  #define IRQ_UNBOUND -1
> > > > > >
> > > > > >  /* Supported VDUSE features */
> > > > > > -static const uint64_t vduse_features;
> > > > > > +static const uint64_t vduse_features = BIT_U64(VDUSE_F_QUEUE_READY);
> > > > > >
> > > > > >  /*
> > > > > >   * VDUSE instance have not asked the vduse API version, so assume 0.
> > > > > > @@ -120,6 +121,7 @@ struct vduse_dev {
> > > > > >         char *name;
> > > > > >         struct mutex lock;
> > > > > >         spinlock_t msg_lock;
> > > > > > +       u64 vduse_features;
> > > > > >         u64 msg_unique;
> > > > > >         u32 msg_timeout;
> > > > > >         wait_queue_head_t waitq;
> > > > > > @@ -601,8 +603,29 @@ static void vduse_vdpa_set_vq_ready(struct vdpa_device *vdpa,
> > > > > >  {
> > > > > >         struct vduse_dev *dev = vdpa_to_vduse(vdpa);
> > > > > >         struct vduse_virtqueue *vq = dev->vqs[idx];
> > > > > > +       struct vduse_dev_msg msg = { 0 };
> > > > > > +       int r;
> > > > > >
> > > > > >         vq->ready = ready;
> > > > > > +
> > > > > > +       if (!(dev->vduse_features & BIT_U64(VDUSE_F_QUEUE_READY)))
> > > > > > +               return;
> > > > > > +
> > > > > > +       msg.req.type = VDUSE_SET_VQ_READY;
> > > > > > +       msg.req.vq_ready.num = idx;
> > > > > > +       msg.req.vq_ready.ready = !!ready;
> > > > > > +
> > > > > > +       r = vduse_dev_msg_sync(dev, &msg);
> > > > > > +
> > > > > > +       if (r < 0) {
> > > > > > +               dev_dbg(&vdpa->dev, "device refuses to set vq %u ready %u",
> > > > > > +                       idx, ready);
> > > > > > +
> > > > > > +               /* We can't do better than break the device in this case */
> > > > >
> > > > > It's better to explain why we can't depend on vduse_dev_msg_sync() here.
> > > > >
> > > > > For example it did:
> > > > >
> > > > >         if (unlikely(dev->broken))
> > > > >                 return -EIO;
> > > > >
> > > > >         init_waitqueue_head(&msg->waitq);
> > > > >         spin_lock(&dev->msg_lock);
> > > > >         if (unlikely(dev->broken)) {
> > > > >                 spin_unlock(&dev->msg_lock);
> > > > >                 return -EIO;
> > > > >         }
> > > > >
> > > > > and
> > > > >
> > > > >         if (!msg->completed) {
> > > > >                 list_del(&msg->list);
> > > > >                 msg->resp.result = VDUSE_REQ_RESULT_FAILED;
> > > > >                 /* Mark the device as malfunction when there is a timeout */
> > > > >                 if (!ret)
> > > > >                         vduse_dev_broken(dev);
> > > > >         }
> > > > >
> > > >
> > > > I'm not sure I follow you here.
> > > >
> > > > We can't do better than breaking the device because the function
> > > > returns no error state, and the caller does not expect an error code.
> > > > Do you mean we can't depend on vduse_dev_msg_sync to call
> > > > vduse_dev_broken(dev) by itself?
> > >
> > > I think I meant, reset seems to be more heavyweight than suspend.
> > >
> > > So if reset can fail, I don't see reason ot break device only for
> > > suspend failure.
> > >
> >
> > Sorry I still don't get you.
> >
> > This series does not implement suspend at all. It doesn't modify the
> > VDUSE device reset or the virtio reset behavior. It only implements
> > the vq ready message for the device. If the device returns an error
> > from that operation, what is your proposal for when the driver sends
> > new messages like resume?
> >
> 
> Friendly ping.

Jason, more comments? If this is to go in it has to go into linux next.

-- 
MST