qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Fam Zheng <famz@redhat.com>
To: Ed Swierk <eswierk@skyportsystems.com>
Cc: qemu-devel@nongnu.org, stefanha@redhat.com
Subject: Re: [Qemu-devel] virtio-scsi-pci iothread spins at 100%
Date: Thu, 9 Feb 2017 14:43:09 +0800	[thread overview]
Message-ID: <20170209064309.GI22807@lemon.lan> (raw)
In-Reply-To: <20170209053917.GH22807@lemon.lan>

On Thu, 02/09 13:39, Fam Zheng wrote:
> On Wed, 02/08 19:44, Ed Swierk wrote:
> > On Wed, Feb 8, 2017 at 6:52 PM, Fam Zheng <famz@redhat.com> wrote:
> > > This means virtio-scsi event vq handler is returning true but actually no
> > > progress is made. Can you try the following patch to see if it's because a
> > > stalled cache of VQ index?
> > >
> > > diff --git a/hw/virtio/virtio.c b/hw/virtio/virtio.c
> > > index 6365706..7f7ab57 100644
> > > --- a/hw/virtio/virtio.c
> > > +++ b/hw/virtio/virtio.c
> > > @@ -2126,7 +2126,7 @@ static bool virtio_queue_host_notifier_aio_poll(void *opaque)
> > >      EventNotifier *n = opaque;
> > >      VirtQueue *vq = container_of(n, VirtQueue, host_notifier);
> > >
> > > -    if (virtio_queue_empty(vq)) {
> > > +    if (vring_avail_idx(vq) == vq->last_avail_idx) {
> > >          return false;
> > >      }
> > 
> > I tried this change but the behavior is the same, unfortunately.
> 
> Hmm, maybe I'm missing something. How about this:
> 
> diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c
> index ce19eff..8b23e80 100644
> --- a/hw/scsi/virtio-scsi.c
> +++ b/hw/scsi/virtio-scsi.c
> @@ -754,9 +754,7 @@ out:
>  
>  void virtio_scsi_handle_event_vq(VirtIOSCSI *s, VirtQueue *vq)
>  {
> -    if (s->events_dropped) {
> -        virtio_scsi_push_event(s, NULL, VIRTIO_SCSI_T_NO_EVENT, 0);
> -    }
> +    virtio_scsi_push_event(s, NULL, VIRTIO_SCSI_T_NO_EVENT, 0);
>  }
>  
>  static void virtio_scsi_handle_event(VirtIODevice *vdev, VirtQueue *vq)
> 

Ignore this. It will break the "assert(event == VIRTIO_SCSI_T_EVENTS_MISSED)" in
the called function. 

Anyway I think the cause is virtio_queue_host_notifier_aio_poll() only checks if
there is an available req in the virtqueue, but in virtio-scsi, eventq actually
contain buffers that are only useful when an event needs to be pushed.

It is reproducible with ubuntu LTS 14.04.5. I'll see if I can sort out a proper
fix.

Fam

  reply	other threads:[~2017-02-09  6:43 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-02-08 16:33 [Qemu-devel] virtio-scsi-pci iothread spins at 100% Ed Swierk
2017-02-08 16:54 ` Ed Swierk
2017-02-09  1:47 ` Fam Zheng
2017-02-09  2:11   ` Ed Swierk
2017-02-09  2:52     ` Fam Zheng
2017-02-09  3:44       ` Ed Swierk
2017-02-09  5:39         ` Fam Zheng
2017-02-09  6:43           ` Fam Zheng [this message]
2017-02-09 10:12         ` Fam Zheng
2017-02-09 13:52           ` Ed Swierk
2017-02-09 14:16             ` Fam Zheng

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20170209064309.GI22807@lemon.lan \
    --to=famz@redhat.com \
    --cc=eswierk@skyportsystems.com \
    --cc=qemu-devel@nongnu.org \
    --cc=stefanha@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).