[Qemu-devel] [PATCH] virtio-blk: Release s->rq queue at system_reset

[Qemu-devel] [PATCH] virtio-blk: Release s->rq queue at system_reset

[Fam Zheng]

At system_reset, there is no point in retrying the queued request,
because the driver that issued the request won't be around any more.

Analyzed-by: Laszlo Ersek <lersek@redhat.com>
Reported-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
---
 hw/block/virtio-blk.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
index 475a822..89eca65 100644
--- a/hw/block/virtio-blk.c
+++ b/hw/block/virtio-blk.c
@@ -654,6 +654,7 @@ static void virtio_blk_reset(VirtIODevice *vdev)
 {
     VirtIOBlock *s = VIRTIO_BLK(vdev);
     AioContext *ctx;
+    VirtIOBlockReq *req;
 
     /*
      * This should cancel pending requests, but can't do nicely until there
@@ -661,6 +662,11 @@ static void virtio_blk_reset(VirtIODevice *vdev)
      */
     ctx = blk_get_aio_context(s->blk);
     aio_context_acquire(ctx);
+    while (s->rq) {
+        req = s->rq;
+        s->rq = req->next;
+        virtio_blk_free_request(req);
+    }
     blk_drain(s->blk);
 
     if (s->dataplane) {
-- 
2.7.4

Re: [Qemu-devel] [PATCH] virtio-blk: Release s->rq queue at system_reset

[Laszlo Ersek]

Fam,

On 07/29/16 12:22, Fam Zheng wrote:
> At system_reset, there is no point in retrying the queued request,
> because the driver that issued the request won't be around any more.
> 
> Analyzed-by: Laszlo Ersek <lersek@redhat.com>
> Reported-by: Laszlo Ersek <lersek@redhat.com>
> Signed-off-by: Fam Zheng <famz@redhat.com>
> ---
>  hw/block/virtio-blk.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
> index 475a822..89eca65 100644
> --- a/hw/block/virtio-blk.c
> +++ b/hw/block/virtio-blk.c
> @@ -654,6 +654,7 @@ static void virtio_blk_reset(VirtIODevice *vdev)
>  {
>      VirtIOBlock *s = VIRTIO_BLK(vdev);
>      AioContext *ctx;
> +    VirtIOBlockReq *req;
>  
>      /*
>       * This should cancel pending requests, but can't do nicely until there
> @@ -661,6 +662,11 @@ static void virtio_blk_reset(VirtIODevice *vdev)
>       */
>      ctx = blk_get_aio_context(s->blk);
>      aio_context_acquire(ctx);
> +    while (s->rq) {
> +        req = s->rq;
> +        s->rq = req->next;
> +        virtio_blk_free_request(req);
> +    }
>      blk_drain(s->blk);
>  
>      if (s->dataplane) {
> 

The comment

    /*
     * This should cancel pending requests, but can't do nicely until there
     * are per-device request lists.
     */

dates back to

commit 6e02c38dadfe4cf02b0da6135adfd8d9352b90e1
Author: aliguori <aliguori@c046a42c-6fe2-441c-8c8c-71466251a162>
Date:   Thu Dec 4 19:52:44 2008 +0000

    Add virtio-blk support
    
    Virtio-blk is a paravirtual block device based on VirtIO.  It can be used by
    specifying the if=virtio parameter to the -drive parameter.
    
    When using -enable-kvm, it can achieve very good performance compared to IDE or
    SCSI.
    
    Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
    
    
    
    git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5870 c046a42c-6fe2-441c-8c8c-71466251a162

I think the comment is obsolete now, the idea that it expresses was actually fixed in the following commit, in my opinion:

commit 6e40b3bfc7e82823cf4df5f0bf668f56db41e53a
Author: Alexander Yarygin <yarygin@linux.vnet.ibm.com>
Date:   Wed Jun 17 13:37:20 2015 +0300

    virtio-blk: Use blk_drain() to drain IO requests
    
    Each call of the virtio_blk_reset() function calls blk_drain_all(),
    which works for all existing BlockDriverStates, while draining only
    one is needed.
    
    This patch replaces blk_drain_all() by blk_drain() in
    virtio_blk_reset(). virtio_blk_data_plane_stop() should be called
    after draining because it restores vblk->complete_request.
    
    Cc: "Michael S. Tsirkin" <mst@redhat.com>
    Cc: Christian Borntraeger <borntraeger@de.ibm.com>
    Cc: Cornelia Huck <cornelia.huck@de.ibm.com>
    Cc: Kevin Wolf <kwolf@redhat.com>
    Cc: Paolo Bonzini <pbonzini@redhat.com>
    Cc: Stefan Hajnoczi <stefanha@redhat.com>
    Signed-off-by: Alexander Yarygin <yarygin@linux.vnet.ibm.com>
    Message-id: 1434537440-28236-3-git-send-email-yarygin@linux.vnet.ibm.com
    Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

So how about turning this patch into a two part series: the first patch should drop the comment, and reference 6e40b3bfc7e82823cf4df5f0bf668f56db41e53a, while the second patch should be this patch?

For this patch:

Reviewed-by: Laszlo Ersek <lersek@redhat.com>

I definitely recommend that Stefan review this patch too.

Thanks!
Laszlo

Re: [Qemu-devel] [PATCH] virtio-blk: Release s->rq queue at system_reset

[Fam Zheng]

On Fri, 07/29 12:41, Laszlo Ersek wrote:
> So how about turning this patch into a two part series: the first patch
> should drop the comment, and reference
> 6e40b3bfc7e82823cf4df5f0bf668f56db41e53a, while the second patch should be
> this patch?

Good idea, I'm sending a separate patch to remove the comment.

Fam

Re: [Qemu-devel] [PATCH] virtio-blk: Release s->rq queue at system_reset

[Paolo Bonzini]

On 29/07/2016 12:22, Fam Zheng wrote:
> At system_reset, there is no point in retrying the queued request,
> because the driver that issued the request won't be around any more.
> 
> Analyzed-by: Laszlo Ersek <lersek@redhat.com>
> Reported-by: Laszlo Ersek <lersek@redhat.com>
> Signed-off-by: Fam Zheng <famz@redhat.com>
> ---
>  hw/block/virtio-blk.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
> index 475a822..89eca65 100644
> --- a/hw/block/virtio-blk.c
> +++ b/hw/block/virtio-blk.c
> @@ -654,6 +654,7 @@ static void virtio_blk_reset(VirtIODevice *vdev)
>  {
>      VirtIOBlock *s = VIRTIO_BLK(vdev);
>      AioContext *ctx;
> +    VirtIOBlockReq *req;
>  
>      /*
>       * This should cancel pending requests, but can't do nicely until there
> @@ -661,6 +662,11 @@ static void virtio_blk_reset(VirtIODevice *vdev)
>       */
>      ctx = blk_get_aio_context(s->blk);
>      aio_context_acquire(ctx);
> +    while (s->rq) {
> +        req = s->rq;
> +        s->rq = req->next;
> +        virtio_blk_free_request(req);
> +    }
>      blk_drain(s->blk);

blk_drain can consume requests too, so I think it should be the other
way round: first drain, then drop any failed request that's been left in
s->rq.

Paolo

>      if (s->dataplane) {
> 

Re: [Qemu-devel] [PATCH] virtio-blk: Release s->rq queue at system_reset

[Fam Zheng]

On Tue, 08/02 08:46, Paolo Bonzini wrote:
> 
> 
> On 29/07/2016 12:22, Fam Zheng wrote:
> > At system_reset, there is no point in retrying the queued request,
> > because the driver that issued the request won't be around any more.
> > 
> > Analyzed-by: Laszlo Ersek <lersek@redhat.com>
> > Reported-by: Laszlo Ersek <lersek@redhat.com>
> > Signed-off-by: Fam Zheng <famz@redhat.com>
> > ---
> >  hw/block/virtio-blk.c | 6 ++++++
> >  1 file changed, 6 insertions(+)
> > 
> > diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
> > index 475a822..89eca65 100644
> > --- a/hw/block/virtio-blk.c
> > +++ b/hw/block/virtio-blk.c
> > @@ -654,6 +654,7 @@ static void virtio_blk_reset(VirtIODevice *vdev)
> >  {
> >      VirtIOBlock *s = VIRTIO_BLK(vdev);
> >      AioContext *ctx;
> > +    VirtIOBlockReq *req;
> >  
> >      /*
> >       * This should cancel pending requests, but can't do nicely until there
> > @@ -661,6 +662,11 @@ static void virtio_blk_reset(VirtIODevice *vdev)
> >       */
> >      ctx = blk_get_aio_context(s->blk);
> >      aio_context_acquire(ctx);
> > +    while (s->rq) {
> > +        req = s->rq;
> > +        s->rq = req->next;
> > +        virtio_blk_free_request(req);
> > +    }
> >      blk_drain(s->blk);
> 
> blk_drain can consume requests too, so I think it should be the other
> way round: first drain, then drop any failed request that's been left in
> s->rq.

I don't think there is any difference in this, blk_drain cannot trigger
virtio_blk_dma_restart_cb, because it is only hooked to vm state change.

Fam

Re: [Qemu-devel] [PATCH] virtio-blk: Release s->rq queue at system_reset

[Fam Zheng]

On Tue, 08/02 15:24, Fam Zheng wrote:
> On Tue, 08/02 08:46, Paolo Bonzini wrote:
> > 
> > 
> > On 29/07/2016 12:22, Fam Zheng wrote:
> > > At system_reset, there is no point in retrying the queued request,
> > > because the driver that issued the request won't be around any more.
> > > 
> > > Analyzed-by: Laszlo Ersek <lersek@redhat.com>
> > > Reported-by: Laszlo Ersek <lersek@redhat.com>
> > > Signed-off-by: Fam Zheng <famz@redhat.com>
> > > ---
> > >  hw/block/virtio-blk.c | 6 ++++++
> > >  1 file changed, 6 insertions(+)
> > > 
> > > diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
> > > index 475a822..89eca65 100644
> > > --- a/hw/block/virtio-blk.c
> > > +++ b/hw/block/virtio-blk.c
> > > @@ -654,6 +654,7 @@ static void virtio_blk_reset(VirtIODevice *vdev)
> > >  {
> > >      VirtIOBlock *s = VIRTIO_BLK(vdev);
> > >      AioContext *ctx;
> > > +    VirtIOBlockReq *req;
> > >  
> > >      /*
> > >       * This should cancel pending requests, but can't do nicely until there
> > > @@ -661,6 +662,11 @@ static void virtio_blk_reset(VirtIODevice *vdev)
> > >       */
> > >      ctx = blk_get_aio_context(s->blk);
> > >      aio_context_acquire(ctx);
> > > +    while (s->rq) {
> > > +        req = s->rq;
> > > +        s->rq = req->next;
> > > +        virtio_blk_free_request(req);
> > > +    }
> > >      blk_drain(s->blk);
> > 
> > blk_drain can consume requests too, so I think it should be the other
> > way round: first drain, then drop any failed request that's been left in
> > s->rq.
> 
> I don't think there is any difference in this, blk_drain cannot trigger
> virtio_blk_dma_restart_cb, because it is only hooked to vm state change.

Oh you mean that blk_drain can queue more requests to s->rq. That is a good
point, indeed.

Fam