[igt-dev] [PATCH i-g-t] tests/prime_self_import: Swap gtt mapping for cpu

[Chris Wilson]

Quoting Antonio Argenziano (2019-02-22 16:50:07)
> 
> 
> On 21/02/19 15:37, Chris Wilson wrote:
> > Quoting Antonio Argenziano (2019-02-21 22:38:59)
> >> The test uses gtt mapping to check two BOs have the same content, since
> >> it seems there is no contention on the BOs use cpu mapping to make it
> >> compatible with platforms that do not have a mappable aperture.
> >>
> >> v2:
> >>          - Need to have an explicit set_domain. (Chris)
> >>
> >> Signed-off-by: Antonio Argenziano <antonio.argenziano at intel.com>
> > 
> > I don't think has any semantic impact, so no need to verify every write
> > to one of the cloned fd is visible in the other, just once is enough.
> 
> Keep only one set_domain then?

No, I was talking about whether we needed to do this for all the
different mmap paths, and decided that we aren't doing any coherency
checks just verifying that we had one objects.

Whether to use one or two set-domain here is an interesting question
relating to how much we want to assume we only have a single bo here
(which is the object of the exercise ofc). I think I favour keeping it
as two distinct handles with the test being only that the memory
contents are the same. Otherwise we might as well just use one handle ;)
 
> Any objections to splitting this from the series and merge?

This can be applied right away.
-Chris