linux-kernel - Re: [PATCH V8 18/19] virtio_ring: factor out split detaching logic

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] [day] [month] [year] [list]

Message-ID: <20251022014109-mutt-send-email-mst@kernel.org>
Date: Wed, 22 Oct 2025 01:44:22 -0400
From: "Michael S. Tsirkin" <mst@...hat.com>
To: Jason Wang <jasowang@...hat.com>
Cc: xuanzhuo@...ux.alibaba.com, eperezma@...hat.com,
	virtualization@...ts.linux.dev, linux-kernel@...r.kernel.org
Subject: Re: [PATCH V8 18/19] virtio_ring: factor out split detaching logic

On Wed, Oct 22, 2025 at 12:00:53PM +0800, Jason Wang wrote:
> On Tue, Oct 21, 2025 at 4:27 PM Michael S. Tsirkin <mst@...hat.com> wrote:
> >
> > On Tue, Oct 21, 2025 at 11:36:12AM +0800, Jason Wang wrote:
> > > On Mon, Oct 20, 2025 at 11:18 PM Michael S. Tsirkin <mst@...hat.com> wrote:
> > > >
> > > > On Mon, Oct 20, 2025 at 03:10:02PM +0800, Jason Wang wrote:
> > > > > This patch factors out the split core detaching logic that could be
> > > > > reused by in order feature into a dedicated function.
> > > > >
> > > > > Acked-by: Eugenio Pérez <eperezma@...hat.com>
> > > > > Reviewed-by: Xuan Zhuo <xuanzhuo@...ux.alibaba.com>
> > > > > Signed-off-by: Jason Wang <jasowang@...hat.com>
> > > > > ---
> > > > >  drivers/virtio/virtio_ring.c | 18 ++++++++++++++----
> > > > >  1 file changed, 14 insertions(+), 4 deletions(-)
> > > > >
> > > > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
> > > > > index 0f07a6637acb..96d7f165ec88 100644
> > > > > --- a/drivers/virtio/virtio_ring.c
> > > > > +++ b/drivers/virtio/virtio_ring.c
> > > > > @@ -802,8 +802,9 @@ static void detach_indirect_split(struct vring_virtqueue *vq,
> > > > >       vq->split.desc_state[head].indir_desc = NULL;
> > > > >  }
> > > > >
> > > > > -static void detach_buf_split(struct vring_virtqueue *vq, unsigned int head,
> > > > > -                          void **ctx)
> > > > > +static unsigned detach_buf_split_in_order(struct vring_virtqueue *vq,
> > > > > +                                       unsigned int head,
> > > > > +                                       void **ctx)
> > > >
> > > >
> > > > Well not really _inorder, right? This is a common function.
> > >
> > > Yes, but inorder is a subset for ooo so I use this name.
> >
> > Can't say it is consistent. I suggest for example:
> >         _in_order -> specific to in order
> >         _ooo -> specific to ooo
> >         no suffix - common
> >
> > or some other scheme where it's clear which is which.
> 
> Will do that.
> 
> >
> >
> >
> > > > You want to call it __detach_buf_split or something maybe.
> > > >
> > > > Additionally the very first line in there is:
> > > >
> > > >         __virtio16 nextflag = cpu_to_virtio16(vq->vq.vdev, VRING_DESC_F_NEXT);
> > > >
> > > > and the byte swap is not needed for inorder.
> > >
> > > I don't see why?
> >
> > To be more precise we do need a swap we do not need it
> > conditional.
> >
> >
> > No, I mean inorder is a modern only feature. So we do not
> > need a branch in the inorder path,
> > you can use __cpu_to_virtio16 with true flag,
> > not cpu_to_virtio16.
> 
> The problem is that the core logic will be reused by the ooo as well.
> I'm not sure it's worthwhile to introduce a new flag parameter for the
> logic like:
> 
> detach_buf_split_in_order()
> {
>         __virtio16 nextflag = __cpu_to_virtio16(true, VRING_DESC_F_NEXT);
>         detach_buf_split(..., nextflag);
> }
> 
> ?

If it's common code then no.


> >
> > > > you could just do __cpu_to_virtio16(true, VRING_DESC_F_NEXT)
> > >
> > > Probably you mean a leftover for hardening? E.g should we check
> > > desc_extra.flag instead of desc.flag here?
> > >
> > > while (vq->split.vring.desc[i].flags & nextflag) {
> > >                 vring_unmap_one_split(vq, &extra[i]);
> > >         i = vq->split.desc_extra[i].next;
> > >                 vq->vq.num_free++;
> > >         }
> > >
> > > Thanks
> >
> > If it is not exploitable we do not care.
> 
> It looks like it can be triggered by the device as the descriptor ring
> is writable. Will post a fix.
> 
> Thanks

question is if the guest is exploitable as a result.

> >
> > --
> > MST
> >
> >