Re: [PATCH 7/7] vfio/type1: Remove map_try_harder() code path

[Alex Williamson]

On Fri,  9 Nov 2018 12:07:12 +0100
Joerg Roedel <joro@xxxxxxxxxx> wrote:

> From: Joerg Roedel <jroedel@xxxxxxx>
> 
> The AMD IOMMU driver can now map a huge-page where smaller
> mappings existed before, so this code-path is no longer
> triggered.
> 
> Signed-off-by: Joerg Roedel <jroedel@xxxxxxx>
> ---
>  drivers/vfio/vfio_iommu_type1.c | 33 ++-------------------------------
>  1 file changed, 2 insertions(+), 31 deletions(-)

Cool, glad to see this finally fixed.  My "should be fixed soon"
comment turned out to be a little optimistic with the fix finally
coming 5 years later.  We could of course keep this code as it really
doesn't harm anything, but I'm in favor trying to remove it if we think
it's dead now.  In order to expedite into one pull:

Acked-by: Alex Williamson <alex.williamson@xxxxxxxxxx>

Thanks,
Alex
 
> diff --git a/drivers/vfio/vfio_iommu_type1.c b/drivers/vfio/vfio_iommu_type1.c
> index d9fd3188615d..7651cfb14836 100644
> --- a/drivers/vfio/vfio_iommu_type1.c
> +++ b/drivers/vfio/vfio_iommu_type1.c
> @@ -978,32 +978,6 @@ static int vfio_dma_do_unmap(struct vfio_iommu *iommu,
>  	return ret;
>  }
>  
> -/*
> - * Turns out AMD IOMMU has a page table bug where it won't map large pages
> - * to a region that previously mapped smaller pages.  This should be fixed
> - * soon, so this is just a temporary workaround to break mappings down into
> - * PAGE_SIZE.  Better to map smaller pages than nothing.
> - */
> -static int map_try_harder(struct vfio_domain *domain, dma_addr_t iova,
> -			  unsigned long pfn, long npage, int prot)
> -{
> -	long i;
> -	int ret = 0;
> -
> -	for (i = 0; i < npage; i++, pfn++, iova += PAGE_SIZE) {
> -		ret = iommu_map(domain->domain, iova,
> -				(phys_addr_t)pfn << PAGE_SHIFT,
> -				PAGE_SIZE, prot | domain->prot);
> -		if (ret)
> -			break;
> -	}
> -
> -	for (; i < npage && i > 0; i--, iova -= PAGE_SIZE)
> -		iommu_unmap(domain->domain, iova, PAGE_SIZE);
> -
> -	return ret;
> -}
> -
>  static int vfio_iommu_map(struct vfio_iommu *iommu, dma_addr_t iova,
>  			  unsigned long pfn, long npage, int prot)
>  {
> @@ -1013,11 +987,8 @@ static int vfio_iommu_map(struct vfio_iommu *iommu, dma_addr_t iova,
>  	list_for_each_entry(d, &iommu->domain_list, next) {
>  		ret = iommu_map(d->domain, iova, (phys_addr_t)pfn << PAGE_SHIFT,
>  				npage << PAGE_SHIFT, prot | d->prot);
> -		if (ret) {
> -			if (ret != -EBUSY ||
> -			    map_try_harder(d, iova, pfn, npage, prot))
> -				goto unwind;
> -		}
> +		if (ret)
> +			goto unwind;
>  
>  		cond_resched();
>  	}