Virtualization - Oct 2021 - [PATCH v2] vduse: Fix race condition between resetting and irq injecting

On Wed, Sep 29, 2021 at 04:50:24PM +0800, Yongji Xie
wrote:
> On Wed, Sep 29, 2021 at 4:41 PM Jason Wang <jasowang at redhat.com>
wrote:
> >
> > On Wed, Sep 29, 2021 at 4:32 PM Xie Yongji <xieyongji at
bytedance.com> wrote:
> > >
> > > The interrupt might be triggered after a reset since there is
> > > no synchronization between resetting and irq injecting. And it
> > > might break something if the interrupt is delayed until a new
> > > round of device initialization.
> > >
> > > Fixes: c8a6153b6c59 ("vduse: Introduce VDUSE - vDPA Device
in Userspace")
> > > Signed-off-by: Xie Yongji <xieyongji at bytedance.com>
> > > ---
> > >  drivers/vdpa/vdpa_user/vduse_dev.c | 37
+++++++++++++++++++++++++------------
> > >  1 file changed, 25 insertions(+), 12 deletions(-)
> > >
> > > diff --git a/drivers/vdpa/vdpa_user/vduse_dev.c
b/drivers/vdpa/vdpa_user/vduse_dev.c
> > > index cefb301b2ee4..841667a896dd 100644
> > > --- a/drivers/vdpa/vdpa_user/vduse_dev.c
> > > +++ b/drivers/vdpa/vdpa_user/vduse_dev.c
> > > @@ -80,6 +80,7 @@ struct vduse_dev {
> > >         struct vdpa_callback config_cb;
> > >         struct work_struct inject;
> > >         spinlock_t irq_lock;
> > > +       struct rw_semaphore rwsem;
> > >         int minor;
> > >         bool broken;
> > >         bool connected;
> > > @@ -410,6 +411,8 @@ static void vduse_dev_reset(struct vduse_dev
*dev)
> > >         if (domain->bounce_map)
> > >                 vduse_domain_reset_bounce_map(domain);
> > >
> > > +       down_write(&dev->rwsem);
> > > +
> > >         dev->status = 0;
> > >         dev->driver_features = 0;
> > >         dev->generation++;
> > > @@ -443,6 +446,8 @@ static void vduse_dev_reset(struct vduse_dev
*dev)
> > >                 flush_work(&vq->inject);
> > >                 flush_work(&vq->kick);
> > >         }
> > > +
> > > +       up_write(&dev->rwsem);
> >
> > Rethink about this, do we need to sync set_status() as well?
> >
> 
> I'm not sure. But I don't see any case that needs synchronization.
> 
> Thanks,
> Yongji
Well you are testing under a lock but it can change at any time...

-- 
MST