Red Hat Bugzilla – Bug 499352
Last modified: 2015-08-31 23:53:33 EDT
I've just noticed this:
* Tue Mar 24 2009 Kyle McMartin <email@example.com>
- Disable DMAR by default until suspend & resume is fixed.
Later, we had this:
* Fri Apr 10 2009 David Woodhouse <David.Woodhouse@intel.com>
- Fix suspend/resume with Intel IOMMU, handle devices behind PCI-PCI
bridges, cope with BIOS claiming IOMMU is at address zero.
Should we re-enable it be default again?
For reference, this is needed for https://fedoraproject.org/wiki/Features/KVM_PCI_Device_Assignment - without CONFIG_DMAR_DEFAULT_ON people need to boot with "intel_iommu=on" in order to use device assignment
I don't think we have nearly all the DMAR fixes that have gone into 2.6.30. (A few just got merged in the last few days.)
Adding a known intel_iommu lockup as a dependency
(In reply to comment #1)
> I don't think we have nearly all the DMAR fixes that have gone into 2.6.30. (A
> few just got merged in the last few days.)
We should have everything that's needed. What do you think is missing?
This bug appears to have been reported against 'rawhide' during the Fedora 11 development cycle.
Changing version to '11'.
More information and reason for this action is here:
How are we looking for enabling this in F-12?
Good, I think. I have some more BIOS workarounds which I can drop in now and re-enable it...
OK, it's enabled. I'd quite like to drop the BROKEN_GFX_WA option too, but let's take things one step at a time...
* Mon Aug 10 2009 David Woodhouse <David.Woodhouse@intel.com> 2.6.31-0.145.rc5.git3
- Merge latest Intel IOMMU fixes and BIOS workarounds, re-enable by default.