Messages in this thread | | | Subject | Re: 2.6.32-rc4: Reported regressions from 2.6.31 | From | David Woodhouse <> | Date | Mon, 12 Oct 2009 15:26:38 +0100 |
| |
On Mon, 2009-10-12 at 11:19 +0100, David Woodhouse wrote: > So the better fix is probably just to fix that problem -- move the final > PCI quirks so they happen a little earlier. If we move them to > fs_initcall_sync() and then move the IOMMU init to rootfs_initcall(), > then everything ought to work, I think...
I've sent that series of patches for review. Assuming you like it, please pull those patches and a couple of other fixes -- one cosmetic and one serious crash fix (another BIOS workaround, in fact), from
git://git.infradead.org/~dwmw2/iommu-2.6.32.git
David Woodhouse (5): intel-iommu: Yet another BIOS workaround: Isoch DMAR unit with no TLB space Rename pci_init() to pci_apply_final_quirks(), move it to quirks.c Mark pci_apply_final_quirks() __init rather than __devinit Run pci_apply_final_quirks() sooner. x86: Move pci_iommu_init to rootfs_initcall()
Roland Dreier (2): intel-iommu: Make "Unknown DMAR structure" message more informative intel-iommu: Decode (and ignore) RHSA entries
arch/x86/kernel/pci-dma.c | 2 +- drivers/pci/dmar.c | 13 ++++++- drivers/pci/intel-iommu.c | 82 ++++++++++++++++++++++++++++++++++++++++++--- drivers/pci/pci.c | 13 ------- drivers/pci/quirks.c | 13 +++++++ 5 files changed, 103 insertions(+), 20 deletions(-)
-- David Woodhouse Open Source Technology Centre David.Woodhouse@intel.com Intel Corporation
| |