[PATCH 3/9] drm/amdgpu: Don't force BOs into visible VRAM for page faults

From: John Brooks <john@fastquake.com>
To: amd-gfx@lists.freedesktop.org,
	"Michel Dänzer" <michel@daenzer.net>,
	"Marek Olšák" <maraeo@gmail.com>
Cc: dri-devel@lists.freedesktop.org
Subject: [PATCH 3/9] drm/amdgpu: Don't force BOs into visible VRAM for page faults
Date: Fri, 23 Jun 2017 13:39:34 -0400	[thread overview]
Message-ID: <1498239580-17360-4-git-send-email-john@fastquake.com> (raw)
In-Reply-To: <1498239580-17360-1-git-send-email-john@fastquake.com>

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.

Signed-off-by: John Brooks <john@fastquake.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 244a7d3..751bc05 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -981,10 +981,11 @@ int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
 
 	/* 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);
 	lpfn =	adev->mc.visible_vram_size >> PAGE_SHIFT;
 	for (i = 0; i < abo->placement.num_placement; i++) {
-		/* Force into visible VRAM */
+		/* Try to move the BO into visible VRAM */
 		if ((abo->placements[i].flags & TTM_PL_FLAG_VRAM) &&
 		    (!abo->placements[i].lpfn ||
 		     abo->placements[i].lpfn > lpfn))
@@ -993,16 +994,13 @@ int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
 	abo->placement.busy_placement = abo->placement.placement;
 	abo->placement.num_busy_placement = abo->placement.num_placement;
 	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;
-- 
2.7.4

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel
