Back merge tag 'v4.4-rc4' into drm-next

We've picked up a few conflicts and it would be nice
to resolve them before we move onwards.

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -539,6 +539,7 @@ struct amdgpu_bo {
 	/* Constant after initialization */
 	struct amdgpu_device		*adev;
 	struct drm_gem_object		gem_base;
+	struct amdgpu_bo		*parent;
 
 	struct ttm_bo_kmap_obj		dma_buf_vmap;
 	pid_t				pid;
@@ -955,6 +956,8 @@ struct amdgpu_vm {
 	struct amdgpu_vm_id	ids[AMDGPU_MAX_RINGS];
 	/* for interval tree */
 	spinlock_t		it_lock;
+	/* protecting freed */
+	spinlock_t		freed_lock;
 };
 
 struct amdgpu_vm_manager {

drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c

@@ -222,6 +222,8 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
 				}
 
 				p->uf.bo = gem_to_amdgpu_bo(gobj);
+				amdgpu_bo_ref(p->uf.bo);
+				drm_gem_object_unreference_unlocked(gobj);
 				p->uf.offset = fence_data->offset;
 			} else {
 				ret = -EINVAL;
@@ -487,7 +489,7 @@ static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error, bo
 			amdgpu_ib_free(parser->adev, &parser->ibs[i]);
 	kfree(parser->ibs);
 	if (parser->uf.bo)
-		drm_gem_object_unreference_unlocked(&parser->uf.bo->gem_base);
+		amdgpu_bo_unref(&parser->uf.bo);
 }
 
 static int amdgpu_bo_vm_update_pte(struct amdgpu_cs_parser *p,
@@ -776,7 +778,7 @@ static int amdgpu_cs_free_job(struct amdgpu_job *job)
 			amdgpu_ib_free(job->adev, &job->ibs[i]);
 	kfree(job->ibs);
 	if (job->uf.bo)
-		drm_gem_object_unreference_unlocked(&job->uf.bo->gem_base);
+		amdgpu_bo_unref(&job->uf.bo);
 	return 0;
 }
 

drivers/gpu/drm/amd/amdgpu/amdgpu_display.c

@@ -73,6 +73,8 @@ static void amdgpu_flip_work_func(struct work_struct *__work)
 	struct drm_crtc *crtc = &amdgpuCrtc->base;
 	unsigned long flags;
 	unsigned i;
+	int vpos, hpos, stat, min_udelay;
+	struct drm_vblank_crtc *vblank = &crtc->dev->vblank[work->crtc_id];
 
 	amdgpu_flip_wait_fence(adev, &work->excl);
 	for (i = 0; i < work->shared_count; ++i)
@@ -81,6 +83,41 @@ static void amdgpu_flip_work_func(struct work_struct *__work)
 	/* We borrow the event spin lock for protecting flip_status */
 	spin_lock_irqsave(&crtc->dev->event_lock, flags);
 
+	/* If this happens to execute within the "virtually extended" vblank
+	 * interval before the start of the real vblank interval then it needs
+	 * to delay programming the mmio flip until the real vblank is entered.
+	 * This prevents completing a flip too early due to the way we fudge
+	 * our vblank counter and vblank timestamps in order to work around the
+	 * problem that the hw fires vblank interrupts before actual start of
+	 * vblank (when line buffer refilling is done for a frame). It
+	 * complements the fudging logic in amdgpu_get_crtc_scanoutpos() for
+	 * timestamping and amdgpu_get_vblank_counter_kms() for vblank counts.
+	 *
+	 * In practice this won't execute very often unless on very fast
+	 * machines because the time window for this to happen is very small.
+	 */
+	for (;;) {
+		/* GET_DISTANCE_TO_VBLANKSTART returns distance to real vblank
+		 * start in hpos, and to the "fudged earlier" vblank start in
+		 * vpos.
+		 */
+		stat = amdgpu_get_crtc_scanoutpos(adev->ddev, work->crtc_id,
+						  GET_DISTANCE_TO_VBLANKSTART,
+						  &vpos, &hpos, NULL, NULL,
+						  &crtc->hwmode);
+
+		if ((stat & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE)) !=
+		    (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE) ||
+		    !(vpos >= 0 && hpos <= 0))
+			break;
+
+		/* Sleep at least until estimated real start of hw vblank */
+		spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
+		min_udelay = (-hpos + 1) * max(vblank->linedur_ns / 1000, 5);
+		usleep_range(min_udelay, 2 * min_udelay);
+		spin_lock_irqsave(&crtc->dev->event_lock, flags);
+	};
+
 	/* do the flip (mmio) */
 	adev->mode_info.funcs->page_flip(adev, work->crtc_id, work->base);
 	/* set the flip status */
@@ -109,7 +146,7 @@ static void amdgpu_unpin_work_func(struct work_struct *__work)
 	} else
 		DRM_ERROR("failed to reserve buffer after flip\n");
 
-	drm_gem_object_unreference_unlocked(&work->old_rbo->gem_base);
+	amdgpu_bo_unref(&work->old_rbo);
 	kfree(work->shared);
 	kfree(work);
 }
@@ -148,8 +185,8 @@ int amdgpu_crtc_page_flip(struct drm_crtc *crtc,
 	obj = old_amdgpu_fb->obj;
 
 	/* take a reference to the old object */
-	drm_gem_object_reference(obj);
 	work->old_rbo = gem_to_amdgpu_bo(obj);
+	amdgpu_bo_ref(work->old_rbo);
 
 	new_amdgpu_fb = to_amdgpu_framebuffer(fb);
 	obj = new_amdgpu_fb->obj;
@@ -222,7 +259,7 @@ pflip_cleanup:
 	amdgpu_bo_unreserve(new_rbo);
 
 cleanup:
-	drm_gem_object_unreference_unlocked(&work->old_rbo->gem_base);
+	amdgpu_bo_unref(&work->old_rbo);
 	fence_put(work->excl);
 	for (i = 0; i < work->shared_count; ++i)
 		fence_put(work->shared[i]);
@@ -712,6 +749,15 @@ bool amdgpu_crtc_scaling_mode_fixup(struct drm_crtc *crtc,
  * \param dev Device to query.
  * \param pipe Crtc to query.
  * \param flags Flags from caller (DRM_CALLED_FROM_VBLIRQ or 0).
+ *              For driver internal use only also supports these flags:
+ *
+ *              USE_REAL_VBLANKSTART to use the real start of vblank instead
+ *              of a fudged earlier start of vblank.
+ *
+ *              GET_DISTANCE_TO_VBLANKSTART to return distance to the
+ *              fudged earlier start of vblank in *vpos and the distance
+ *              to true start of vblank in *hpos.
+ *
  * \param *vpos Location where vertical scanout position should be stored.
  * \param *hpos Location where horizontal scanout position should go.
  * \param *stime Target location for timestamp taken immediately before
@@ -776,10 +822,40 @@ int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
 		vbl_end = 0;
 	}
 
+	/* Called from driver internal vblank counter query code? */
+	if (flags & GET_DISTANCE_TO_VBLANKSTART) {
+	    /* Caller wants distance from real vbl_start in *hpos */
+	    *hpos = *vpos - vbl_start;
+	}
+
+	/* Fudge vblank to start a few scanlines earlier to handle the
+	 * problem that vblank irqs fire a few scanlines before start
+	 * of vblank. Some driver internal callers need the true vblank
+	 * start to be used and signal this via the USE_REAL_VBLANKSTART flag.
+	 *
+	 * The cause of the "early" vblank irq is that the irq is triggered
+	 * by the line buffer logic when the line buffer read position enters
+	 * the vblank, whereas our crtc scanout position naturally lags the
+	 * line buffer read position.
+	 */
+	if (!(flags & USE_REAL_VBLANKSTART))
+		vbl_start -= adev->mode_info.crtcs[pipe]->lb_vblank_lead_lines;
+
 	/* Test scanout position against vblank region. */
 	if ((*vpos < vbl_start) && (*vpos >= vbl_end))
 		in_vbl = false;
 
+	/* In vblank? */
+	if (in_vbl)
+	    ret |= DRM_SCANOUTPOS_IN_VBLANK;
+
+	/* Called from driver internal vblank counter query code? */
+	if (flags & GET_DISTANCE_TO_VBLANKSTART) {
+		/* Caller wants distance from fudged earlier vbl_start */
+		*vpos -= vbl_start;
+		return ret;
+	}
+
 	/* Check if inside vblank area and apply corrective offsets:
 	 * vpos will then be >=0 in video scanout area, but negative
 	 * within vblank area, counting down the number of lines until
@@ -795,32 +871,6 @@ int amdgpu_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
 	/* Correct for shifted end of vbl at vbl_end. */
 	*vpos = *vpos - vbl_end;
 
-	/* In vblank? */
-	if (in_vbl)
-		ret |= DRM_SCANOUTPOS_IN_VBLANK;
-
-	/* Is vpos outside nominal vblank area, but less than
-	 * 1/100 of a frame height away from start of vblank?
-	 * If so, assume this isn't a massively delayed vblank
-	 * interrupt, but a vblank interrupt that fired a few
-	 * microseconds before true start of vblank. Compensate
-	 * by adding a full frame duration to the final timestamp.
-	 * Happens, e.g., on ATI R500, R600.
-	 *
-	 * We only do this if DRM_CALLED_FROM_VBLIRQ.
-	 */
-	if ((flags & DRM_CALLED_FROM_VBLIRQ) && !in_vbl) {
-		vbl_start = mode->crtc_vdisplay;
-		vtotal = mode->crtc_vtotal;
-
-		if (vbl_start - *vpos < vtotal / 100) {
-			*vpos -= vtotal;
-
-			/* Signal this correction as "applied". */
-			ret |= 0x8;
-		}
-	}
-
 	return ret;
 }
 

drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c

@@ -235,8 +235,9 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 	    AMDGPU_GEM_USERPTR_REGISTER))
 		return -EINVAL;
 
-	if (!(args->flags & AMDGPU_GEM_USERPTR_ANONONLY) ||
-		   !(args->flags & AMDGPU_GEM_USERPTR_REGISTER)) {
+	if (!(args->flags & AMDGPU_GEM_USERPTR_READONLY) && (
+	     !(args->flags & AMDGPU_GEM_USERPTR_ANONONLY) ||
+	     !(args->flags & AMDGPU_GEM_USERPTR_REGISTER))) {
 
 		/* if we want to write to it we must require anonymous
 		   memory and install a MMU notifier */

drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c

@@ -611,13 +611,59 @@ void amdgpu_driver_preclose_kms(struct drm_device *dev,
 u32 amdgpu_get_vblank_counter_kms(struct drm_device *dev, unsigned int pipe)
 {
 	struct amdgpu_device *adev = dev->dev_private;
+	int vpos, hpos, stat;
+	u32 count;
 
 	if (pipe >= adev->mode_info.num_crtc) {
 		DRM_ERROR("Invalid crtc %u\n", pipe);
 		return -EINVAL;
 	}
 
-	return amdgpu_display_vblank_get_counter(adev, pipe);
+	/* The hw increments its frame counter at start of vsync, not at start
+	 * of vblank, as is required by DRM core vblank counter handling.
+	 * Cook the hw count here to make it appear to the caller as if it
+	 * incremented at start of vblank. We measure distance to start of
+	 * vblank in vpos. vpos therefore will be >= 0 between start of vblank
+	 * and start of vsync, so vpos >= 0 means to bump the hw frame counter
+	 * result by 1 to give the proper appearance to caller.
+	 */
+	if (adev->mode_info.crtcs[pipe]) {
+		/* Repeat readout if needed to provide stable result if
+		 * we cross start of vsync during the queries.
+		 */
+		do {
+			count = amdgpu_display_vblank_get_counter(adev, pipe);
+			/* Ask amdgpu_get_crtc_scanoutpos to return vpos as
+			 * distance to start of vblank, instead of regular
+			 * vertical scanout pos.
+			 */
+			stat = amdgpu_get_crtc_scanoutpos(
+				dev, pipe, GET_DISTANCE_TO_VBLANKSTART,
+				&vpos, &hpos, NULL, NULL,
+				&adev->mode_info.crtcs[pipe]->base.hwmode);
+		} while (count != amdgpu_display_vblank_get_counter(adev, pipe));
+
+		if (((stat & (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE)) !=
+		    (DRM_SCANOUTPOS_VALID | DRM_SCANOUTPOS_ACCURATE))) {
+			DRM_DEBUG_VBL("Query failed! stat %d\n", stat);
+		} else {
+			DRM_DEBUG_VBL("crtc %d: dist from vblank start %d\n",
+				      pipe, vpos);
+
+			/* Bump counter if we are at >= leading edge of vblank,
+			 * but before vsync where vpos would turn negative and
+			 * the hw counter really increments.
+			 */
+			if (vpos >= 0)
+				count++;
+		}
+	} else {
+		/* Fallback to use value as is. */
+		count = amdgpu_display_vblank_get_counter(adev, pipe);
+		DRM_DEBUG_VBL("NULL mode info! Returned count may be wrong.\n");
+	}
+
+	return count;
 }
 
 /**

drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h

@@ -407,6 +407,7 @@ struct amdgpu_crtc {
 	u32 line_time;
 	u32 wm_low;
 	u32 wm_high;
+	u32 lb_vblank_lead_lines;
 	struct drm_display_mode hw_mode;
 };
 
@@ -528,6 +529,10 @@ struct amdgpu_framebuffer {
 #define ENCODER_MODE_IS_DP(em) (((em) == ATOM_ENCODER_MODE_DP) || \
 				((em) == ATOM_ENCODER_MODE_DP_MST))
 
+/* Driver internal use only flags of amdgpu_get_crtc_scanoutpos() */
+#define USE_REAL_VBLANKSTART 		(1 << 30)
+#define GET_DISTANCE_TO_VBLANKSTART	(1 << 31)
+
 void amdgpu_link_encoder_connector(struct drm_device *dev);
 
 struct drm_connector *

drivers/gpu/drm/amd/amdgpu/amdgpu_object.c

@@ -100,6 +100,7 @@ static void amdgpu_ttm_bo_destroy(struct ttm_buffer_object *tbo)
 	list_del_init(&bo->list);
 	mutex_unlock(&bo->adev->gem.mutex);
 	drm_gem_object_release(&bo->gem_base);
+	amdgpu_bo_unref(&bo->parent);
 	kfree(bo->metadata);
 	kfree(bo);
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c

@@ -587,9 +587,13 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
 	uint32_t flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, bo_mem);
 	int r;
 
-	if (gtt->userptr)
-		amdgpu_ttm_tt_pin_userptr(ttm);
-
+	if (gtt->userptr) {
+		r = amdgpu_ttm_tt_pin_userptr(ttm);
+		if (r) {
+			DRM_ERROR("failed to pin userptr\n");
+			return r;
+		}
+	}
 	gtt->offset = (unsigned long)(bo_mem->start << PAGE_SHIFT);
 	if (!ttm->num_pages) {
 		WARN(1, "nothing to bind %lu pages for mreg %p back %p!\n",
@@ -797,11 +801,12 @@ uint32_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
 	if (mem && mem->mem_type != TTM_PL_SYSTEM)
 		flags |= AMDGPU_PTE_VALID;
 
-	if (mem && mem->mem_type == TTM_PL_TT)
+	if (mem && mem->mem_type == TTM_PL_TT) {
 		flags |= AMDGPU_PTE_SYSTEM;
 
-	if (!ttm || ttm->caching_state == tt_cached)
-		flags |= AMDGPU_PTE_SNOOPED;
+		if (ttm->caching_state == tt_cached)
+			flags |= AMDGPU_PTE_SNOOPED;
+	}
 
 	if (adev->asic_type >= CHIP_TOPAZ)
 		flags |= AMDGPU_PTE_EXECUTABLE;

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c

@@ -885,17 +885,21 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 	struct amdgpu_bo_va_mapping *mapping;
 	int r;
 
+	spin_lock(&vm->freed_lock);
 	while (!list_empty(&vm->freed)) {
 		mapping = list_first_entry(&vm->freed,
 			struct amdgpu_bo_va_mapping, list);
 		list_del(&mapping->list);
-
+		spin_unlock(&vm->freed_lock);
 		r = amdgpu_vm_bo_update_mapping(adev, vm, mapping, 0, 0, NULL);
 		kfree(mapping);
 		if (r)
 			return r;
 
+		spin_lock(&vm->freed_lock);
 	}
+	spin_unlock(&vm->freed_lock);
+
 	return 0;
 
 }
@@ -1079,6 +1083,11 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
 		if (r)
 			goto error_free;
 
+		/* Keep a reference to the page table to avoid freeing
+		 * them up in the wrong order.
+		 */
+		pt->parent = amdgpu_bo_ref(vm->page_directory);
+
 		r = amdgpu_vm_clear_bo(adev, pt);
 		if (r) {
 			amdgpu_bo_unref(&pt);
@@ -1150,10 +1159,13 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev,
 	spin_unlock(&vm->it_lock);
 	trace_amdgpu_vm_bo_unmap(bo_va, mapping);
 
-	if (valid)
+	if (valid) {
+		spin_lock(&vm->freed_lock);
 		list_add(&mapping->list, &vm->freed);
-	else
+		spin_unlock(&vm->freed_lock);
+	} else {
 		kfree(mapping);
+	}
 
 	return 0;
 }
@@ -1186,7 +1198,9 @@ void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
 		interval_tree_remove(&mapping->it, &vm->va);
 		spin_unlock(&vm->it_lock);
 		trace_amdgpu_vm_bo_unmap(bo_va, mapping);
+		spin_lock(&vm->freed_lock);
 		list_add(&mapping->list, &vm->freed);
+		spin_unlock(&vm->freed_lock);
 	}
 	list_for_each_entry_safe(mapping, next, &bo_va->invalids, list) {
 		list_del(&mapping->list);
@@ -1247,6 +1261,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	INIT_LIST_HEAD(&vm->cleared);
 	INIT_LIST_HEAD(&vm->freed);
 	spin_lock_init(&vm->it_lock);
+	spin_lock_init(&vm->freed_lock);
 	pd_size = amdgpu_vm_directory_size(adev);
 	pd_entries = amdgpu_vm_num_pdes(adev);
 

drivers/gpu/drm/amd/amdgpu/dce_v10_0.c

@@ -1250,7 +1250,7 @@ static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
 	u32 pixel_period;
 	u32 line_time = 0;
 	u32 latency_watermark_a = 0, latency_watermark_b = 0;
-	u32 tmp, wm_mask;
+	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
 
 	if (amdgpu_crtc->base.enabled && num_heads && mode) {
 		pixel_period = 1000000 / (u32)mode->clock;
@@ -1333,6 +1333,7 @@ static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
 		    (adev->mode_info.disp_priority == 2)) {
 			DRM_DEBUG_KMS("force priority to high\n");
 		}
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
 	}
 
 	/* select wm A */
@@ -1357,6 +1358,8 @@ static void dce_v10_0_program_watermarks(struct amdgpu_device *adev,
 	amdgpu_crtc->line_time = line_time;
 	amdgpu_crtc->wm_high = latency_watermark_a;
 	amdgpu_crtc->wm_low = latency_watermark_b;
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
 }
 
 /**

drivers/gpu/drm/amd/amdgpu/dce_v11_0.c

@@ -1238,7 +1238,7 @@ static void dce_v11_0_program_watermarks(struct amdgpu_device *adev,
 	u32 pixel_period;
 	u32 line_time = 0;
 	u32 latency_watermark_a = 0, latency_watermark_b = 0;
-	u32 tmp, wm_mask;
+	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
 
 	if (amdgpu_crtc->base.enabled && num_heads && mode) {
 		pixel_period = 1000000 / (u32)mode->clock;
@@ -1321,6 +1321,7 @@ static void dce_v11_0_program_watermarks(struct amdgpu_device *adev,
 		    (adev->mode_info.disp_priority == 2)) {
 			DRM_DEBUG_KMS("force priority to high\n");
 		}
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
 	}
 
 	/* select wm A */
@@ -1345,6 +1346,8 @@ static void dce_v11_0_program_watermarks(struct amdgpu_device *adev,
 	amdgpu_crtc->line_time = line_time;
 	amdgpu_crtc->wm_high = latency_watermark_a;
 	amdgpu_crtc->wm_low = latency_watermark_b;
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
 }
 
 /**

drivers/gpu/drm/amd/amdgpu/dce_v8_0.c

@@ -1193,7 +1193,7 @@ static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
 	u32 pixel_period;
 	u32 line_time = 0;
 	u32 latency_watermark_a = 0, latency_watermark_b = 0;
-	u32 tmp, wm_mask;
+	u32 tmp, wm_mask, lb_vblank_lead_lines = 0;
 
 	if (amdgpu_crtc->base.enabled && num_heads && mode) {
 		pixel_period = 1000000 / (u32)mode->clock;
@@ -1276,6 +1276,7 @@ static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
 		    (adev->mode_info.disp_priority == 2)) {
 			DRM_DEBUG_KMS("force priority to high\n");
 		}
+		lb_vblank_lead_lines = DIV_ROUND_UP(lb_size, mode->crtc_hdisplay);
 	}
 
 	/* select wm A */
@@ -1302,6 +1303,8 @@ static void dce_v8_0_program_watermarks(struct amdgpu_device *adev,
 	amdgpu_crtc->line_time = line_time;
 	amdgpu_crtc->wm_high = latency_watermark_a;
 	amdgpu_crtc->wm_low = latency_watermark_b;
+	/* Save number of lines the linebuffer leads before the scanout */
+	amdgpu_crtc->lb_vblank_lead_lines = lb_vblank_lead_lines;
 }
 
 /**

drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c

@@ -513,7 +513,7 @@ static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
 	WREG32(mmVM_L2_CNTL3, tmp);
 	/* setup context0 */
 	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gtt_start >> 12);
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, (adev->mc.gtt_end >> 12) - 1);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gtt_end >> 12);
 	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
 	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
 			(u32)(adev->dummy_page.addr >> 12));

drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c

@@ -657,7 +657,7 @@ static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
 	WREG32(mmVM_L2_CNTL4, tmp);
 	/* setup context0 */
 	WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gtt_start >> 12);
-	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, (adev->mc.gtt_end >> 12) - 1);
+	WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gtt_end >> 12);
 	WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
 	WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
 			(u32)(adev->dummy_page.addr >> 12));

drivers/gpu/drm/amd/scheduler/gpu_scheduler.c

@@ -288,6 +288,7 @@ amd_sched_entity_pop_job(struct amd_sched_entity *entity)
  */
 static bool amd_sched_entity_in(struct amd_sched_job *sched_job)
 {
+	struct amd_gpu_scheduler *sched = sched_job->sched;
 	struct amd_sched_entity *entity = sched_job->s_entity;
 	bool added, first = false;
 
@@ -302,7 +303,7 @@ static bool amd_sched_entity_in(struct amd_sched_job *sched_job)
 
 	/* first job wakes up scheduler */
 	if (first)
-		amd_sched_wakeup(sched_job->sched);
+		amd_sched_wakeup(sched);
 
 	return added;
 }
@@ -318,9 +319,9 @@ void amd_sched_entity_push_job(struct amd_sched_job *sched_job)
 {
 	struct amd_sched_entity *entity = sched_job->s_entity;
 
+	trace_amd_sched_job(sched_job);
 	wait_event(entity->sched->job_scheduled,
 		   amd_sched_entity_in(sched_job));
-	trace_amd_sched_job(sched_job);
 }
 
 /**