Re: [Qemu-devel] [PATCH RFC for-2.2] virtio-blk: force 1st s/g to match header

[Michael S. Tsirkin]

On Fri, Nov 28, 2014 at 04:14:35PM +0000, Peter Maydell wrote:
> On 28 November 2014 at 11:43, Stefan Hajnoczi <address@hidden> wrote:
> > Right, the test case explicitly tests different descriptor layouts,
> > even though virtio-blk-pci does not set the ANY_LAYOUT feature bit.
> >
> > Either the test case needs to check ANY_LAYOUT before using the
> > 2-descriptor layout or it needs to expect QEMU to refuse (in this case
> > exit(1), which is not very graceful).
> >
> > The quick fix is to skip the 2-descriptor layout tests and re-enable
> > them once virtio-blk actually supports ANY_LAYOUT.  Any objections?
> 
> So what do we want to do with this for 2.2? We have I think
> two choices:
>  (1) say that this isn't causing problems in practice, and defer all
>  this to 2.3
>  (2) add something like this patch plus fix the 'make check' tests
>  (but turning "maybe something misbehaves" into "qemu definitely
>  blows up and exits" doesn't seem like a great improvement to me)
> 
> I started looking at virtio-blk initially because I wasn't sure
> if we should fix the virtio-net issue in the core virtio code.
> But since we've decided not to do that, whether virtio-blk's
> problems are release-blockers or not is something that we can
> decide on their own merits.
> 
> My current thought is that we don't need to address this for 2.2;
> is there something I'm missing that means we shouldn't defer to 2.3?
> 
> thanks
> -- PMM

The result of this is host mapping leak.
What effect does this have? Can this DOS host?
If not, I agree.