2020-09-08 16:34:21

by Logan Gunthorpe

[permalink] [raw]
Subject: Re: [Intel-gfx] [PATCH 0/8] Convert the intel iommu driver to the dma-iommu api



On 2020-09-08 9:28 a.m., Tvrtko Ursulin wrote:
>>
>> 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))
>

Right, good catch, that's definitely necessary.

>> +               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.)

It has not been tested. To test it, you need Tom's patch set without the
last "DO NOT MERGE" patch:

https://lkml.kernel.org/lkml/[email protected]/T/

Thanks,

Logan