From: John Brooks <john@fastquake.com>
Date: Tue, 27 Jun 2017 22:33:21 -0400
Subject: drm/amdgpu: Don't force BOs into visible VRAM for page faults
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Git-commit: 41d9a6a728ea34d697dc74bc5fea1b6f6c5fce80
Patch-mainline: v4.14-rc1
References: FATE#326289 FATE#326079 FATE#326049 FATE#322398 FATE#326166

There is no need for page faults to force BOs into visible VRAM if it's
full, and the time it takes to do so is great enough to cause noticeable
stuttering. Add GTT as a possible placement so that if visible VRAM is
full, page faults move BOs to GTT instead of evicting other BOs from VRAM.

Suggested-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: John Brooks <john@fastquake.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Petr Tesarik <ptesarik@suse.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.c |   17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -974,18 +974,21 @@ int amdgpu_bo_fault_reserve_notify(struc
 
 	/* hurrah the memory is not visible ! */
 	atomic64_inc(&adev->num_vram_cpu_page_faults);
-	amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM);
+	amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
+					 AMDGPU_GEM_DOMAIN_GTT);
+
+	/* Avoid costly evictions; only set GTT as a busy placement */
+	abo->placement.num_busy_placement = 1;
+	abo->placement.busy_placement = &abo->placements[1];
+
 	r = ttm_bo_validate(bo, &abo->placement, false, false);
-	if (unlikely(r == -ENOMEM)) {
-		amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT);
-		return ttm_bo_validate(bo, &abo->placement, false, false);
-	} else if (unlikely(r != 0)) {
+	if (unlikely(r != 0))
 		return r;
-	}
 
 	offset = bo->mem.start << PAGE_SHIFT;
 	/* this should never happen */
-	if ((offset + size) > adev->mc.visible_vram_size)
+	if (bo->mem.mem_type == TTM_PL_VRAM &&
+	    (offset + size) > adev->mc.visible_vram_size)
 		return -EINVAL;
 
 	return 0;