[igt-dev] ✓ Fi.CI.IGT: success for Introduce IGT allocator (rev14)

[Chris Wilson]

Quoting Patchwork (2021-01-11 13:33:28)
> Patch Details
> 
> Series:  Introduce IGT allocator (rev14)
> URL:     https://patchwork.freedesktop.org/series/82954/
> State:   success
> Details: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_5377/index.html
> 
> CI Bug Log - changes from CI_DRM_9577_full -> IGTPW_5377_full
> 
> Summary
> 
> SUCCESS
> 
> No regressions found.
> 
> External URL: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_5377/index.html
> 
> New tests
> 
> New tests have been introduced between CI_DRM_9577_full and IGTPW_5377_full:
> 
> New IGT tests (32)
> 
>   • igt at api_intel_allocator@alloc-simple:

One request in addition to api_intel_allocator, is to put it to use in a
few gem_softpin subtests.

Goal of the test would be to try and make sure the allocations are
valid, so just a random bunch of objects with EXEC_OBJECT_PINNED and
expect no EINVAL/ENOSPC. I don't think we need to build up anything too
complicated, just smoketest that the allocator produces valid offsets
for execbuf.

Another idea is given a full-ppgtt, to pass in some random allocations
without EXEC_OBJECT_PINNED but the same size/alignment constraints and
check that the kernel agrees and does not relocate them.

That should be sufficient to have a basic test that bridges the gap
between the library API tests and kernel feature tests.
-Chris