Re: [Qemu-devel] [PATCH] pci: Error on PCI capability collisions

["Michael S. Tsirkin" <mst@redhat.com>]

On Wed, Aug 24, 2011 at 02:36:31PM +0200, Jan Kiszka wrote:
> On 2011-08-24 14:34, Michael S. Tsirkin wrote:
> > On Wed, Aug 24, 2011 at 02:29:36PM +0200, Jan Kiszka wrote:
> >> On 2011-08-24 13:58, Michael S. Tsirkin wrote:
> >>> On Wed, Aug 24, 2011 at 12:10:32PM +0200, Jan Kiszka wrote:
> >>>> On 2011-08-24 12:04, Michael S. Tsirkin wrote:
> >>>>> On Tue, Aug 23, 2011 at 07:28:08PM +0200, Jan Kiszka wrote:
> >>>>>> From: Alex Williamson <alex.williamson@redhat.com>
> >>>>>>
> >>>>>> Nothing good can happen when we overlap capabilities
> >>>>>>
> >>>>>> [ Jan: rebased over qemu, minor formatting ]
> >>>>>>
> >>>>>> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
> >>>>>
> >>>>> This doesn't build for me:
> >>>>>
> >>>>> /scm/qemu/hw/pci.c: In function ‘pci_add_capability’:
> >>>>> /scm/qemu/hw/pci.c:1970:45: error: ‘PCIDevice’ has no member named ‘config_map’
> >>>>
> >>>> Yeah, sorry, forgot to refresh the commit before posting.
> >>>>
> >>>>>
> >>>>> I think that what that includes is the capability including each given
> >>>>> offset, right?  It would be easy to write some code scanning the
> >>>>> capability list to figure this value out.
> >>>>> Something along the lines of (untested):
> >>>>>
> >>>>> static
> >>>>> uint8_t pci_find_capability_at_offset(PCIDevice *pdev, uint8_t offset)
> >>>>> {                                       
> >>>>>     uint8_t next, prev, found = 0;
> >>>>>
> >>>>>     if (!(pdev->config[PCI_STATUS] & PCI_STATUS_CAP_LIST))
> >>>>>         return 0;
> >>>>>
> >>>>>     for (prev = PCI_CAPABILITY_LIST; (next = pdev->config[prev]);
> >>>>>          prev = next + PCI_CAP_LIST_NEXT)
> >>>>>         if (next <= offset && next > found)
> >>>>>             found = next;
> >>>>>
> >>>>>     return found;
> >>>>> }
> >>>>
> >>>> Sounds useful, will enhance the patch.
> >>>>
> >>>> (Originally, I just wanted to reduce the qemu-kvm delta... :) )
> >>>>
> >>>> Jan
> >>>
> >>> Also, let's add a comment documenting the
> >>> reason for this check: device assignment
> >>> depends on this check to verify that the device
> >>> is not broken.
> >>
> >> Based on the previous discussion, I don't think this is accurate as it
> >> will also validate emulated devices.
> >>
> >> Jan
> > 
> > Something like the below is accurate, right?
> > 
> > /* Device assignment depends on this check to verify that the device
> >    is not broken. Should never trigger for emulated devices,
> >    but it's helpful for debugging these.
> >  */
> 
> I've expressed this in the commit message. Unless there is another
> reason to do v3, maybe you can merge the comment on commit.
> 
> Jan

Sure, I can do that, no need with v3. You are fine with the way
it's formulated?

> -- 
> Siemens AG, Corporate Technology, CT T DE IT 1
> Corporate Competence Center Embedded Linux