On Wed, Mar 23, 2022 at 06:46:38PM +0100, Zbigniew Kempczyński wrote:
This reverts commit 904ebf2ba89edaeba5c7c10540e43dba63541dc6. Failures on dg2 tests were caused by invalid alignment when local memory was in use. Changes which adopt alignment according to gen were already merged in IGT so lets revert relocation temporary enabler for dg2. Keeping it is a little bit problematic for IGT because on premerge we would see results with kernel which supports relocation. To see no-relocation results we need to send disabler (like this revert), point IGT with "Test-with" tag what is cumbersome and time consuming so lets do this permanently. If we will see some failures they need to be fixed instead of keeping relocation enabler. Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@xxxxxxxxx> Cc: Lucas De Marchi <lucas.demarchi@xxxxxxxxx> Cc: Dave Airlie <airlied@xxxxxxxxxx> Cc: Daniel Vetter <daniel.vetter@xxxxxxxxx> Cc: Jason Ekstrand <jason@xxxxxxxxxxxxxx> Cc: Rodrigo Vivi <rodrigo.vivi@xxxxxxxxx>
thanks. I double checked BAT and things seem to be equivalent now without that hack. I removed it from topic/core-for-CI. How are we with the igt tests executed in full run? thanks Lucas De Marchi
--- drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 42a49fd2f2ab..8b0b4aeb6716 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -501,7 +501,7 @@ static bool platform_has_relocs_enabled(const struct i915_execbuffer *eb) */ if (GRAPHICS_VER(eb->i915) < 12 || IS_TIGERLAKE(eb->i915) || IS_ROCKETLAKE(eb->i915) || IS_ALDERLAKE_S(eb->i915) || - IS_ALDERLAKE_P(eb->i915) || IS_DG2(eb->i915)) + IS_ALDERLAKE_P(eb->i915)) return true; return false; -- 2.32.0
