[Intel-gfx] [PATCH] tests/gem_userptr_blits: subtests for MAP_FIXED mappings of regular bo

[Chris Wilson]

On Mon, Jun 29, 2015 at 03:56:07PM +0100, Tvrtko Ursulin wrote:
> >Yes. But I would prefer MAP_FIXED to be the first invalidate, but that
> >looks like we then need to leak the ptr.
> 
> If it used mmap instead of posix_memalign and no free then what
> would it leak? MAP_FIXED would be guaranteed first invalidate and it
> would discard the mapping.

Nothing, good idea.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre