On Fri, Oct 16, 2015 at 11:15:23AM +0300, Michael S. Tsirkin wrote:
> On Fri, Oct 16, 2015 at 04:04:09PM +0800, Yuanhan Liu wrote:
> > On Fri, Oct 16, 2015 at 10:47:49AM +0300, Michael S. Tsirkin wrote:
> > > On Fri, Oct 16, 2015 at 03:32:56PM +0800, Yuanhan Liu wrote:
> > > > On Fri, Oct 16, 2015 at 10:21:40AM +0300, Michael S. Tsirkin wrote:
> > > > > On Fri, Oct 16, 2015 at 03:18:59PM +0800, Yuanhan Liu wrote:
> > > > > > This reverts commit d1f8b30ec8dde0318fd1b98d24a64926feae9625.
> > > > > >
> > > > > > It turned out that it breaks stuff, so revert it:
> > > > > >
> > > > > >
> > > > > > http://lists.nongnu.org/archive/html/qemu-devel/2015-10/msg00949.html
> > > > > >
> > > > > > CC: "Michael S. Tsirkin" <m...@redhat.com>
> > > > > > Reported-by: Paolo Bonzini <pbonz...@redhat.com>
> > > > > > Signed-off-by: Yuanhan Liu <yuanhan....@linux.intel.com>
> > > > >
> > > > > OK but that's not enough. We need to also
> > > > > - drop patch sending this on stop
> > > >
> > > > I could do that. But may I know why we have to drop it on stop?
> > >
> > > Because we need to get the state from remote after stop.
> > > RESET_OWNER discards that, so we can't resume the VM.
> >
> > Got it, and thanks for the explanation.
> >
> > > > If we don't send it on stop, when will we send it? The termination
> > > > of QEMU?
> > >
> > > If mq is negotiated, we don't need it. If mq isn't negotiated, maybe we
> > > should send it on reset/driver unload. It might be better than
> > > corrupting guest memory. And if we do, maybe we should keep
> > > the RESET_DEVICE name.
> >
> > Maybe we should just ignore it now, before I (we) have clear clue
> > on how to handle it correctly.
>
> Well if MQ is negotiated, then skipping this is clearly
> the right thing to do.
Let me put a bit more thoughts on this when get time.
> > >
> > > > > - send VHOST_USER_SET_VRING_ENABLE on stop and start instead
> > > >
> > > > That indeed looks more reasonable. And actually, Huawei asked whether
> > > > there is a message from QEMU to tell us that a vring is ready in
> > > > our weekly meeting this morning. And here it is :)
> > > >
> > > > --yliu
> > >
> > > Unfortunately that's only there if MQ is negotiated.
> >
> > Yes.
> >
> > >
> > > If MQ is not negotiated, then I think you can get away with
> > > starting processing after you get the 1st kick.
> >
> > We mark a vring is ready by checking the callfd and kickfd:
> > if both are valid (> 0), then it's ready.
> >
> > --yliu
>
> No, that's not right. If MQ is not there, you must not look at vring
> until you get at least 1 kick. In other words, until kickfd is readable.
> Whether callfd is set should not matter at all.
That makes sense to me. Huawei, what do you think?
--yliu
> If MQ is there, you should just go by VHOST_USER_SET_VRING_ENABLE.
>
>
> > >
> > >
> > > > >
> > > > > > ---
> > > > > > docs/specs/vhost-user.txt | 4 ++--
> > > > > > hw/net/vhost_net.c | 2 +-
> > > > > > hw/virtio/vhost-user.c | 8 ++++----
> > > > > > linux-headers/linux/vhost.h | 2 +-
> > > > > > tests/vhost-user-test.c | 2 +-
> > > > > > 5 files changed, 9 insertions(+), 9 deletions(-)
> > > > > >
> > > > > > diff --git a/docs/specs/vhost-user.txt b/docs/specs/vhost-user.txt
> > > > > > index 4eadad1..4bcd17d 100644
> > > > > > --- a/docs/specs/vhost-user.txt
> > > > > > +++ b/docs/specs/vhost-user.txt
> > > > > > @@ -211,10 +211,10 @@ Message types
> > > > > > as an owner of the session. This can be used on the Slave as
> > > > > > a
> > > > > > "session start" flag.
> > > > > >
> > > > > > - * VHOST_USER_RESET_DEVICE
> > > > > > + * VHOST_USER_RESET_OWNER
> > > > > >
> > > > > > Id: 4
> > > > > > - Equivalent ioctl: VHOST_RESET_DEVICE
> > > > > > + Equivalent ioctl: VHOST_RESET_OWNER
> > > > > > Master payload: N/A
> > > > > >
> > > > > > Issued when a new connection is about to be closed. The
> > > > > > Master will no
> > > > > > diff --git a/hw/net/vhost_net.c b/hw/net/vhost_net.c
> > > > > > index 2bce891..804f5c9 100644
> > > > > > --- a/hw/net/vhost_net.c
> > > > > > +++ b/hw/net/vhost_net.c
> > > > > > @@ -296,7 +296,7 @@ static void vhost_net_stop_one(struct vhost_net
> > > > > > *net,
> > > > > > } else if (net->nc->info->type ==
> > > > > > NET_CLIENT_OPTIONS_KIND_VHOST_USER) {
> > > > > > for (file.index = 0; file.index < net->dev.nvqs;
> > > > > > ++file.index) {
> > > > > > const VhostOps *vhost_ops = net->dev.vhost_ops;
> > > > > > - int r = vhost_ops->vhost_call(&net->dev,
> > > > > > VHOST_RESET_DEVICE,
> > > > > > + int r = vhost_ops->vhost_call(&net->dev,
> > > > > > VHOST_RESET_OWNER,
> > > > > > NULL);
> > > > > > assert(r >= 0);
> > > > > > }
> > > > > > diff --git a/hw/virtio/vhost-user.c b/hw/virtio/vhost-user.c
> > > > > > index b11c0d2..12a9104 100644
> > > > > > --- a/hw/virtio/vhost-user.c
> > > > > > +++ b/hw/virtio/vhost-user.c
> > > > > > @@ -34,7 +34,7 @@ typedef enum VhostUserRequest {
> > > > > > VHOST_USER_GET_FEATURES = 1,
> > > > > > VHOST_USER_SET_FEATURES = 2,
> > > > > > VHOST_USER_SET_OWNER = 3,
> > > > > > - VHOST_USER_RESET_DEVICE = 4,
> > > > > > + VHOST_USER_RESET_OWNER = 4,
> > > > > > VHOST_USER_SET_MEM_TABLE = 5,
> > > > > > VHOST_USER_SET_LOG_BASE = 6,
> > > > > > VHOST_USER_SET_LOG_FD = 7,
> > > > > > @@ -102,7 +102,7 @@ static unsigned long int
> > > > > > ioctl_to_vhost_user_request[VHOST_USER_MAX] = {
> > > > > > VHOST_GET_FEATURES, /* VHOST_USER_GET_FEATURES */
> > > > > > VHOST_SET_FEATURES, /* VHOST_USER_SET_FEATURES */
> > > > > > VHOST_SET_OWNER, /* VHOST_USER_SET_OWNER */
> > > > > > - VHOST_RESET_DEVICE, /* VHOST_USER_RESET_DEVICE */
> > > > > > + VHOST_RESET_OWNER, /* VHOST_USER_RESET_OWNER */
> > > > > > VHOST_SET_MEM_TABLE, /* VHOST_USER_SET_MEM_TABLE */
> > > > > > VHOST_SET_LOG_BASE, /* VHOST_USER_SET_LOG_BASE */
> > > > > > VHOST_SET_LOG_FD, /* VHOST_USER_SET_LOG_FD */
> > > > > > @@ -192,7 +192,7 @@ static bool
> > > > > > vhost_user_one_time_request(VhostUserRequest request)
> > > > > > {
> > > > > > switch (request) {
> > > > > > case VHOST_USER_SET_OWNER:
> > > > > > - case VHOST_USER_RESET_DEVICE:
> > > > > > + case VHOST_USER_RESET_OWNER:
> > > > > > case VHOST_USER_SET_MEM_TABLE:
> > > > > > case VHOST_USER_GET_QUEUE_NUM:
> > > > > > return true;
> > > > > > @@ -249,7 +249,7 @@ static int vhost_user_call(struct vhost_dev
> > > > > > *dev, unsigned long int request,
> > > > > > break;
> > > > > >
> > > > > > case VHOST_USER_SET_OWNER:
> > > > > > - case VHOST_USER_RESET_DEVICE:
> > > > > > + case VHOST_USER_RESET_OWNER:
> > > > > > break;
> > > > > >
> > > > > > case VHOST_USER_SET_MEM_TABLE:
> > > > > > diff --git a/linux-headers/linux/vhost.h
> > > > > > b/linux-headers/linux/vhost.h
> > > > > > index 14a0160..ead86db 100644
> > > > > > --- a/linux-headers/linux/vhost.h
> > > > > > +++ b/linux-headers/linux/vhost.h
> > > > > > @@ -78,7 +78,7 @@ struct vhost_memory {
> > > > > > #define VHOST_SET_OWNER _IO(VHOST_VIRTIO, 0x01)
> > > > > > /* Give up ownership, and reset the device to default values.
> > > > > > * Allows subsequent call to VHOST_OWNER_SET to succeed. */
> > > > > > -#define VHOST_RESET_DEVICE _IO(VHOST_VIRTIO, 0x02)
> > > > > > +#define VHOST_RESET_OWNER _IO(VHOST_VIRTIO, 0x02)
> > > > > >
> > > > > > /* Set up/modify memory layout */
> > > > > > #define VHOST_SET_MEM_TABLE _IOW(VHOST_VIRTIO, 0x03, struct
> > > > > > vhost_memory)
> > > > > > diff --git a/tests/vhost-user-test.c b/tests/vhost-user-test.c
> > > > > > index 56df5cc..f181391 100644
> > > > > > --- a/tests/vhost-user-test.c
> > > > > > +++ b/tests/vhost-user-test.c
> > > > > > @@ -53,7 +53,7 @@ typedef enum VhostUserRequest {
> > > > > > VHOST_USER_GET_FEATURES = 1,
> > > > > > VHOST_USER_SET_FEATURES = 2,
> > > > > > VHOST_USER_SET_OWNER = 3,
> > > > > > - VHOST_USER_RESET_DEVICE = 4,
> > > > > > + VHOST_USER_RESET_OWNER = 4,
> > > > > > VHOST_USER_SET_MEM_TABLE = 5,
> > > > > > VHOST_USER_SET_LOG_BASE = 6,
> > > > > > VHOST_USER_SET_LOG_FD = 7,
> > > > > > --
> > > > > > 1.9.0
