Re: [PATCH] Intel IOMMU Pass Through Support

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 15 Apr 2009 17:19:57 -0700
Fenghua Yu <fenghua.yu@intel.com> wrote:

> The patch adds kernel parameter intel_iommu=pt to set up pass through mode in
> context mapping entry. This disables DMAR in linux kernel; but KVM still runs on
> VT-d and interrupt remapping still works.
> 
> In this mode, kernel uses swiotlb for DMA API functions but other VT-d
> functionalities are enabled for KVM. KVM always uses multi level translation
> page table in VT-d. By default, pass though mode is disabled in kernel.
> 
> This is useful when people don't want to enable VT-d DMAR in kernel but still
> want to use KVM and interrupt remapping for reasons like DMAR performance
> concern or debug purpose.
> 

The patch is now in linux-next and broke my build.

arch/x86/built-in.o: In function `iommu_setup':
arch/x86/kernel/pci-dma.c:215: undefined reference to `iommu_pass_through'
arch/x86/built-in.o: In function `pci_swiotlb_init':
arch/x86/kernel/pci-swiotlb.c:74: undefined reference to `iommu_pass_through'

Because iommu_pass_through is defined in drivers/pci/intel-iommu.c and

# CONFIG_DMAR is not set

I'll need to cook up some local hack to work around that.



Also, the patch in linux-next (but not the one which I'm replying to
here does:

: --- a/arch/x86/kernel/pci-dma.c
: +++ b/arch/x86/kernel/pci-dma.c
: @@ -160,6 +160,8 @@ again:
:         return page_address(page);
:  }
:  
: +extern int iommu_pass_through;
: +

Which is wrong and unnecessary - the variable was already declared in a
header file.

scripts/checkpatch.pl would have warned you about this error, but
apparently neither you nor the people who reviewed or merged the patch
bother to use it.