linux-kernel - Re: [RFC PATCH] virtio: do not reset stateful devices on resume

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 15 Dec 2021 18:27:40 +0100
From:   Mikhail Golubev <mgo@...nsynergy.com>
To:     "Michael S. Tsirkin" <mst@...hat.com>
Cc:     Mikhail Golubev <Mikhail.Golubev@...nsynergy.com>,
        Jason Wang <jasowang@...hat.com>,
        virtualization@...ts.linux-foundation.org,
        linux-kernel@...r.kernel.org,
        Anton Yakovlev <Anton.Yakovlev@...nsynergy.com>
Subject: Re: [RFC PATCH] virtio: do not reset stateful devices on resume

The 12/14/2021 18:26, Michael S. Tsirkin wrote:
> On Tue, Dec 14, 2021 at 05:33:05PM +0100, Mikhail Golubev wrote:
> > From: Anton Yakovlev <Anton.Yakovlev@...nsynergy.com>
> > 
> > We assume that stateful devices can maintain their state while
> > suspended. And for this reason they don't have a freeze callback. If
> > such a device is reset during resume, the device state/context will be
> > lost on the device side. And the virtual device will stop working.
> > 
> > Signed-off-by: Anton Yakovlev <Anton.Yakovlev@...nsynergy.com>
> > Signed-off-by: Mikhail Golubev <mikhail.golubev@...nsynergy.com>
> 
> A bit more specific? Which configs does this patch fix?

We had encountered an issue related to 'stateful' GPU 3d (virglrenderer) and
video (gstreamer) devices.

BR,
Mikhail

> 
> > ---
> >  drivers/virtio/virtio.c | 8 ++++++++
> >  1 file changed, 8 insertions(+)
> > 
> > diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c
> > index 236081afe9a2..defa15b56eb8 100644
> > --- a/drivers/virtio/virtio.c
> > +++ b/drivers/virtio/virtio.c
> > @@ -472,6 +472,13 @@ int virtio_device_restore(struct virtio_device *dev)
> >  	struct virtio_driver *drv = drv_to_virtio(dev->dev.driver);
> >  	int ret;
> >  
> > +	/* Short path for stateful devices. Here we assume that if the device
> > +	 * does not have a freeze callback, its state was not changed when
> > +	 * suspended.
> > +	 */
> > +	if (drv && !drv->freeze)
> > +		goto on_config_enable;
> > +
> >  	/* We always start by resetting the device, in case a previous
> >  	 * driver messed it up. */
> >  	dev->config->reset(dev);
> > @@ -503,6 +510,7 @@ int virtio_device_restore(struct virtio_device *dev)
> >  	/* Finally, tell the device we're all set */
> >  	virtio_add_status(dev, VIRTIO_CONFIG_S_DRIVER_OK);
> >  
> > +on_config_enable:
> >  	virtio_config_enable(dev);
> >  
> >  	return 0;
> > -- 
> > 2.34.1
> > 
> > 
> > -- 
> 

--