Re: [PATCH 2/2] AMD IOMMU: use ioremap()

[Suravee Suthikulanit <suravee.suthikulpanit@amd.com>]

On 7/10/2013 5:49 AM, Jan Beulich wrote:
> There's no point in using the fixmap here, and it gets
> map_iommu_mmio_region() in line with unmap_iommu_mmio_region(), which
> was already using iounmap() (thus crashing if actually used).
>
> Signed-off-by: Jan Beulich<jbeulich@suse.com>
>
> --- a/xen/drivers/passthrough/amd/iommu_init.c
> +++ b/xen/drivers/passthrough/amd/iommu_init.c
> @@ -48,19 +48,10 @@ static int iommu_has_ht_flag(struct amd_
>   
>   static int __init map_iommu_mmio_region(struct amd_iommu *iommu)
>   {
> -    unsigned long mfn;
> -
> -    if ( nr_amd_iommus > MAX_AMD_IOMMUS )
> -    {
> -        AMD_IOMMU_DEBUG("nr_amd_iommus %d > MAX_IOMMUS\n", nr_amd_iommus);
> +    iommu->mmio_base = ioremap(iommu->mmio_base_phys,
> +                               IOMMU_MMIO_REGION_LENGTH);
> +    if ( iommu->mmio_base )
This should have been "if ( !iommu->mmio_base )".

>           return -ENOMEM;
> -    }
> -
> -    iommu->mmio_base = (void *)fix_to_virt(
> -        FIX_IOMMU_MMIO_BASE_0 + nr_amd_iommus * MMIO_PAGES_PER_IOMMU);
> -    mfn = (unsigned long)(iommu->mmio_base_phys >> PAGE_SHIFT);
> -    map_pages_to_xen((unsigned long)iommu->mmio_base, mfn,
> -                     MMIO_PAGES_PER_IOMMU, PAGE_HYPERVISOR_NOCACHE);
>   
>       memset(iommu->mmio_base, 0, IOMMU_MMIO_REGION_LENGTH);
Once I changed the above logic, IOMMU is initialized correctly. I have 
tested pass-though a NIC and thing looks fine.

Suravee