[Pkg-fglrx-devel] Bug#601972: Bug#601972: 10-9: does not quiesce GPU bus-mastering before teardown on VT switch

[Henrique de Moraes Holschuh]

On Sun, 31 Oct 2010, Patrick Matthäi wrote:
> > Oct 31 11:14:36 khazad-dum2 kernel: DMAR:[DMA Write] Request device [03:00.0] fault addr 149598000 
> > Oct 31 11:14:36 khazad-dum2 kernel: DMAR:[fault reason 05] PTE Write access is not set
> > Oct 31 11:14:36 khazad-dum2 kernel: DMAR:[DMA Write] Request device [03:00.0] fault addr 149598000 
> > Oct 31 11:14:36 khazad-dum2 kernel: DMAR:[fault reason 05] PTE Write access is not set
> > Oct 31 11:14:36 khazad-dum2 kernel: DMAR:[DMA Write] Request device [03:00.0] fault addr 149598000 
> > Oct 31 11:14:36 khazad-dum2 kernel: DMAR:[fault reason 05] PTE Write access is not set
> > Oct 31 11:14:36 khazad-dum2 kernel: DMAR:[DMA Write] Request device [03:00.0] fault addr 149598000 
> > Oct 31 11:14:36 khazad-dum2 kernel: DMAR:[fault reason 05] PTE Write access is not set
> > Oct 31 11:14:36 khazad-dum2 kernel: DRHD: handling fault status reg 400
> > Oct 31 11:14:36 khazad-dum2 kernel: DRHD: handling fault status reg 402
> > Oct 31 11:14:36 khazad-dum2 kernel: DMAR:[DMA Write] Request device [03:00.0] fault addr 135c6e000 
> 
> Your xorg.conf does not look correct, please try out following:

Hmm?  What I am supposed to think when I complain about rogue PCI bus master
*writes* hitting the IOMMU shields, and you tell me my xorg.conf "does not
look correct"?

I will test with the Debian kernel, instead of upstream 2.6.32.25.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh