[Intel-gfx] [PATCH 0/8] Convert the intel iommu driver to the dma-iommu api

[Tvrtko Ursulin]

Hi,

On 27/08/2020 22:36, Logan Gunthorpe wrote:
> On 2020-08-23 6:04 p.m., Tom Murphy wrote:
>> I have added a check for the sg_dma_len == 0 :
>> """
>>   } __sgt_iter(struct scatterlist *sgl, bool dma) {
>>          struct sgt_iter s = { .sgp = sgl };
>>
>> +       if (sgl && sg_dma_len(sgl) == 0)
>> +           s.sgp = NULL;
>>
>>          if (s.sgp) {
>>              .....
>> """
>> at location [1].
>> but it doens't fix the problem.
> 
> Based on my read of the code, it looks like we also need to change usage
> of sgl->length... Something like the rough patch below, maybe?

This thread was brought to my attention and I initially missed this 
reply. Essentially I came to the same conclusion about the need to use 
sg_dma_len. One small correction below:

> Also, Tom, do you have an updated version of the patchset to convert the
> Intel IOMMU to dma-iommu available? The last one I've found doesn't
> apply cleanly (I'm assuming parts of it have been merged in slightly
> modified forms).
> 
> Thanks,
> 
> Logan
> 
> --
> 
> diff --git a/drivers/gpu/drm/i915/i915_scatterlist.h
> b/drivers/gpu/drm/i915/i915
> index b7b59328cb76..9367ac801f0c 100644
> --- a/drivers/gpu/drm/i915/i915_scatterlist.h
> +++ b/drivers/gpu/drm/i915/i915_scatterlist.h
> @@ -27,13 +27,19 @@ static __always_inline struct sgt_iter {
>   } __sgt_iter(struct scatterlist *sgl, bool dma) {
>          struct sgt_iter s = { .sgp = sgl };
> 
> +       if (sgl && !sg_dma_len(s.sgp))

I'd extend the condition to be, just to be safe:

	if (dma && sgl && !sg_dma_len(s.sgp))

> +               s.sgp = NULL;
> +
>          if (s.sgp) {
>                  s.max = s.curr = s.sgp->offset;
> -               s.max += s.sgp->length;
> -               if (dma)
> +
> +               if (dma) {
> +                       s.max += sg_dma_len(s.sgp);
>                          s.dma = sg_dma_address(s.sgp);
> -               else
> +               } else {
> +                       s.max += s.sgp->length;
>                          s.pfn = page_to_pfn(sg_page(s.sgp));
> +               }

Otherwise has this been tested or alternatively how to test it? (How to 
repro the issue.)

Regards,

Tvrtko