Skip to contentSkip to conversation

drm/i915: use dev_priv directly in gen8_ppgtt_notify_vgt

PATCH In review Matthew Auld 5 participants 14 messages Apr 12, 2016 – Apr 19, 2016

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/5 drm/i915: use dev_priv directly in gen8_ppgtt_notify_vgt

Remove dev local and use to_i915() in gen8_ppgtt_notify_vgt.

v2: use dev_priv directly for QUESTION_MACROS (Joonas Lahtinen)

Cc Joonas Lahtinen <[email protected]>
---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -3
index c5cb049..5ed713d 100644
@@ -905,11 +905,10 @@ static int gen8_init_scratch(struct i915_address_space *vm)
 static int gen8_ppgtt_notify_vgt(struct i915_hw_ppgtt *ppgtt, bool create)
 {
 	enum vgt_g2v_type msg;
-	struct drm_device *dev = ppgtt->base.dev;
-	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct drm_i915_private *dev_priv = to_i915(ppgtt->base.dev);
 	int i;
 
-	if (USES_FULL_48BIT_PPGTT(dev)) {
+	if (USES_FULL_48BIT_PPGTT(dev_priv)) {
 		u64 daddr = px_dma(&ppgtt->pml4);
 
 		I915_WRITE(vgtif_reg(pdp[0].lo), lower_32_bits(daddr));
Signature · 6 lines
2.4.11

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

PATCH 2/5 drm/i915: harden allocation paths in allocate_va_range

For the gen6/gen8_allocate_va_range functions remove the WARN_ON range sanity checks in favour of simply hardening the allocation paths. This also fixes the inconsistency in which we don't always handle the potential overflow in our checks.

Cc Chris Wilson <[email protected]>
Cc Joonas Lahtinen <[email protected]>
---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 17 ++++-------------
 1 file changed, 4 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +4 -13
index 5ed713d..fa583d5 100644
@@ -1220,15 +1220,6 @@ static int gen8_alloc_va_range_3lvl(struct i915_address_space *vm,
 	uint32_t pdpes = I915_PDPES_PER_PDP(dev);
 	int ret;
 
-	/* Wrap is never okay since we can only represent 48b, and we don't
-	 * actually use the other side of the canonical address space.
-	 */
-	if (WARN_ON(start + length < start))
-		return -ENODEV;
-
-	if (WARN_ON(start + length > vm->total))
-		return -ENODEV;
-
 	ret = alloc_gen8_temp_bitmaps(&new_page_dirs, &new_page_tables, pdpes);
 	if (ret)
 		return ret;
@@ -1370,6 +1361,8 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,
 {
 	struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);
 
+	length = min_t(uint64_t, length, vm->total);
+
 	if (USES_FULL_48BIT_PPGTT(vm->dev))
 		return gen8_alloc_va_range_4lvl(vm, &ppgtt->pml4, start, length);
 	else
@@ -1860,11 +1853,9 @@ static int gen6_alloc_va_range(struct i915_address_space *vm,
 	uint32_t pde, temp;
 	int ret;
 
-	if (WARN_ON(start_in + length_in > ppgtt->base.total))
-		return -ENODEV;
-
 	start = start_save = start_in;
-	length = length_save = length_in;
+	length = length_save = length_in = min_t(uint64_t, length_in,
+						 ppgtt->base.total);
 
 	bitmap_zero(new_page_tables, I915_PDES);
 
Signature · 6 lines
2.4.11

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer
#2 lore raw

Re: [PATCH 2/5] drm/i915: harden allocation paths in allocate_va_range

Quoted 5 lines · Matthew Auld

For the gen6/gen8_allocate_va_range functions remove the WARN_ON range sanity checks in favour of simply hardening the allocation paths. This also fixes the inconsistency in which we don't always handle the potential overflow in our checks.

Not sure this is the right thing to do. I think I do prefer erroring out on invalid arguments as we're talking of internal programmer interface.

Regards, Joonas

Quoted 50 lines
Cc Chris Wilson <[email protected]>
Cc Joonas Lahtinen <[email protected]>

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 17 ++++-------------

 1 file changed, 4 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 5ed713d..fa583d5 100644
@@ -1220,15 +1220,6 @@ static int gen8_alloc_va_range_3lvl(struct i915_address_space *vm,
  	uint32_t pdpes = I915_PDPES_PER_PDP(dev);
  	int ret;

drivers/gpu/drm/i915/i915_gem_gtt.c +0 -9
-	/* Wrap is never okay since we can only represent 48b, and we don't
-	 * actually use the other side of the canonical address space.
-	 */
-	if (WARN_ON(start + length < start))
-		return -ENODEV;
-
-	if (WARN_ON(start + length > vm->total))
-		return -ENODEV;
-

  ret = alloc_gen8_temp_bitmaps(&new_page_dirs, &new_page_tables, pdpes); if (ret)
  return ret;

drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
@@ -1370,6 +1361,8 @@ static int gen8_alloc_va_range(struct i915_address_space *vm,
  {
  	struct i915_hw_ppgtt *ppgtt = i915_vm_to_ppgtt(vm);

drivers/gpu/drm/i915/i915_gem_gtt.c +2 -0
+	length = min_t(uint64_t, length, vm->total);
+

  if (USES_FULL_48BIT_PPGTT(vm->dev))
  return gen8_alloc_va_range_4lvl(vm, &ppgtt->pml4, start, length); else

drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
@@ -1860,11 +1853,9 @@ static int gen6_alloc_va_range(struct i915_address_space *vm,
  	uint32_t pde, temp;
  	int ret;

drivers/gpu/drm/i915/i915_gem_gtt.c +0 -3
-	if (WARN_ON(start_in + length_in > ppgtt->base.total))
-		return -ENODEV;
-

  start = start_save = start_in;

drivers/gpu/drm/i915/i915_gem_gtt.c +2 -1
-	length = length_save = length_in;
+	length = length_save = length_in = min_t(uint64_t, length_in,
+						 ppgtt->base.total);


  bitmap_zero(new_page_tables, I915_PDES);

Signature · 7 lines
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#3 lore raw

PATCH 3/5 drm/i915: bail in alloc_pdp when !FULL_48BIT_PPGTT

In reply to Matthew Auld

If we are not in FULL_48BIT_PPGTT mode then we really shouldn't continue on with our allocations, given that the call to free_dpd would bail early without freeing everything, thus leaking memory.

Cc Chris Wilson <[email protected]>
Cc Joonas Lahtinen <[email protected]>
---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -1
index fa583d5..6601b11 100644
@@ -569,7 +569,8 @@ i915_page_directory_pointer *alloc_pdp(struct drm_device *dev)
 	struct i915_page_directory_pointer *pdp;
 	int ret = -ENOMEM;
 
-	WARN_ON(!USES_FULL_48BIT_PPGTT(dev));
+	if (WARN_ON(!USES_FULL_48BIT_PPGTT(dev)))
+		return ERR_PTR(-EINVAL);
 
 	pdp = kzalloc(sizeof(*pdp), GFP_KERNEL);
 	if (!pdp)
Signature · 6 lines
2.4.11

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer
#4 lore raw

Re: [PATCH 3/5] drm/i915: bail in alloc_pdp when !FULL_48BIT_PPGTT

Quoted 16 lines · Matthew Auld

If we are not in FULL_48BIT_PPGTT mode then we really shouldn't continue on with our allocations, given that the call to free_dpd would bail early without freeing everything, thus leaking memory.

Cc Chris Wilson <[email protected]>
Cc Joonas Lahtinen <[email protected]>

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 3 ++-

 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index fa583d5..6601b11 100644

drivers/gpu/drm/i915/i915_gem_gtt.c i915_page_directory_pointer *alloc_pdp(struct drm_device *dev) · line 570

Quoted 2 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
@@ -569,7 +569,8 @@ i915_page_directory_pointer *alloc_pdp(struct drm_device *dev)
  	struct i915_page_directory_pointer *pdp;
  	int ret = -ENOMEM;

This default value is unused.

While touching the function, I'd rather make it ret = -EINVAL and add goto fail; to clean it up further.

drivers/gpu/drm/i915/i915_gem_gtt.c i915_page_directory_pointer *alloc_pdp(struct drm_device *dev) · line 573

Quoted 3 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -1
-	WARN_ON(!USES_FULL_48BIT_PPGTT(dev));
+	if (WARN_ON(!USES_FULL_48BIT_PPGTT(dev)))
+		return ERR_PTR(-EINVAL);

goto fail here.

Quoted 3 lines


  pdp = kzalloc(sizeof(*pdp), GFP_KERNEL);
  if (!pdp)

ret = -ENOMEM and goto fail here?

Regards, Joonas

Signature · 7 lines
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#5 lore raw

PATCH 4/5 drm/i915: call kunmap_px on pt_vaddr

In reply to Matthew Auld

We need to kunmap pt_vaddr and not pt itself, otherwise we end up mapping a bunch of pages without ever unmapping them.

Cc Chris Wilson <[email protected]>
Cc Joonas Lahtinen <[email protected]>
---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
index 6601b11..7aef7bf 100644
@@ -746,7 +746,7 @@ static void gen8_ppgtt_clear_pte_range(struct i915_address_space *vm,
 			num_entries--;
 		}
 
-		kunmap_px(ppgtt, pt);
+		kunmap_px(ppgtt, pt_vaddr);
 
 		pte = 0;
 		if (++pde == I915_PDES) {
Signature · 6 lines
2.4.11

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer Reviewed-by on v1 4/5
#6 lore raw

Re: [PATCH 4/5] drm/i915: call kunmap_px on pt_vaddr

Quoted 3 lines · Matthew Auld

We need to kunmap pt_vaddr and not pt itself, otherwise we end up mapping a bunch of pages without ever unmapping them.

It's always a good idea to use 'git blame' to figure out the bug inducing commit, and include that information + CC'ing the author (Mika here):

Fixes d1c54acd67dc ("drm/i915/gtt: Introduce kmap|kunmap for dma page")

With that added,

Mika to comment on too, looks like  fairly straightforward error when doing a mass replace.

Regards, Joonas

Quoted 20 lines
Cc Chris Wilson <[email protected]>
Cc Joonas Lahtinen <[email protected]>

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 2 +-

 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 6601b11..7aef7bf 100644
@@ -746,7 +746,7 @@ static void gen8_ppgtt_clear_pte_range(struct i915_address_space *vm,
  			num_entries--;
  		}

drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
-		kunmap_px(ppgtt, pt);
+		kunmap_px(ppgtt, pt_vaddr);


  pte = 0;
  if (++pde == I915_PDES) {

Signature · 7 lines
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#7 lore raw

Re: [PATCH 4/5] drm/i915: call kunmap_px on pt_vaddr

Quoted 10 lines · Joonas Lahtinen

[ text/plain ]

Quoted 3 lines · Matthew Auld

We need to kunmap pt_vaddr and not pt itself, otherwise we end up mapping a bunch of pages without ever unmapping them.

It's always a good idea to use 'git blame' to figure out the bug inducing commit, and include that information + CC'ing the author (Mika here):

Quoted 9 lines
Fixes d1c54acd67dc ("drm/i915/gtt: Introduce kmap|kunmap for dma page")

With that added,

Mika to comment on too, looks like  fairly straightforward error when doing a mass replace.

Fixes tag added and pushed to dinq. Thanks for patch and review. -Mika

Quoted 9 lines

Regards, Joonas

Quoted 7 lines
Cc Chris Wilson <[email protected]>
Cc Joonas Lahtinen <[email protected]>

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 2 +-

 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 17 lines
Quoted 13 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -0
index 6601b11..7aef7bf 100644
@@ -746,7 +746,7 @@ static void gen8_ppgtt_clear_pte_range(struct i915_address_space *vm,
  			num_entries--;
  		}

drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
-		kunmap_px(ppgtt, pt);
+		kunmap_px(ppgtt, pt_vaddr);


  pte = 0;
  if (++pde == I915_PDES) {

Signature · 3 lines
Joonas Lahtinen
Open Source Technology Center
Intel Corporation

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#8 lore raw

PATCH 5/5 drm/i915: Propagate error from drm_gem_object_init()

In reply to Matthew Auld

From: Chris Wilson <[email protected]>

From: Chris Wilson <[email protected]>

Propagate the real error from drm_gem_object_init(). Note this also fixes some confusion in the error return from i915_gem_alloc_object...

v2:
(Matthew Auld)
- updated new users of gem_alloc_object from latest drm-nightly
- replaced occurrences of IS_ERR_OR_NULL() with IS_ERR()

---
 drivers/gpu/drm/i915/i915_gem.c              | 14 ++++++++------
 drivers/gpu/drm/i915/i915_gem_batch_pool.c   |  4 ++--
 drivers/gpu/drm/i915/i915_gem_context.c      |  4 ++--
 drivers/gpu/drm/i915/i915_gem_render_state.c |  7 +++++--
 drivers/gpu/drm/i915/i915_guc_submission.c   |  2 +-
 drivers/gpu/drm/i915/intel_display.c         |  4 ++--
 drivers/gpu/drm/i915/intel_fbdev.c           |  4 ++--
 drivers/gpu/drm/i915/intel_lrc.c             | 10 ++++++----
 drivers/gpu/drm/i915/intel_overlay.c         |  2 +-
 drivers/gpu/drm/i915/intel_ringbuffer.c      | 19 ++++++++++---------
 10 files changed, 39 insertions(+), 31 deletions(-)
drivers/gpu/drm/i915/i915_gem.c +8 -6
index b37ffea..6a6998c 100644
@@ -387,8 +387,8 @@ i915_gem_create(struct drm_file *file,
 
 	/* Allocate the new object */
 	obj = i915_gem_alloc_object(dev, size);
-	if (obj == NULL)
-		return -ENOMEM;
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
 
 	ret = drm_gem_handle_create(file, &obj->base, &handle);
 	/* drop reference from allocate - handle holds it now */
@@ -4527,14 +4527,16 @@ struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev,
 	struct drm_i915_gem_object *obj;
 	struct address_space *mapping;
 	gfp_t mask;
+	int ret;
 
 	obj = i915_gem_object_alloc(dev);
 	if (obj == NULL)
-		return NULL;
+		return ERR_PTR(-ENOMEM);
 
-	if (drm_gem_object_init(dev, &obj->base, size) != 0) {
+	ret = drm_gem_object_init(dev, &obj->base, size);
+	if (ret) {
 		i915_gem_object_free(obj);
-		return NULL;
+		return ERR_PTR(ret);
 	}
 
 	mask = GFP_HIGHUSER | __GFP_RECLAIMABLE;
@@ -5390,7 +5392,7 @@ i915_gem_object_create_from_data(struct drm_device *dev,
 	int ret;
 
 	obj = i915_gem_alloc_object(dev, round_up(size, PAGE_SIZE));
-	if (IS_ERR_OR_NULL(obj))
+	if (IS_ERR(obj))
 		return obj;
 
 	ret = i915_gem_object_set_to_cpu_domain(obj, true);
drivers/gpu/drm/i915/i915_gem_batch_pool.c +2 -2
index 7bf2f3f..d79caa2 100644
@@ -135,8 +135,8 @@ i915_gem_batch_pool_get(struct i915_gem_batch_pool *pool,
 		int ret;
 
 		obj = i915_gem_alloc_object(pool->dev, size);
-		if (obj == NULL)
-			return ERR_PTR(-ENOMEM);
+		if (IS_ERR(obj))
+			return obj;
 
 		ret = i915_gem_object_get_pages(obj);
 		if (ret)
drivers/gpu/drm/i915/i915_gem_context.c +2 -2
index fe580cb..a5adc66 100644
@@ -179,8 +179,8 @@ i915_gem_alloc_context_obj(struct drm_device *dev, size_t size)
 	int ret;
 
 	obj = i915_gem_alloc_object(dev, size);
-	if (obj == NULL)
-		return ERR_PTR(-ENOMEM);
+	if (IS_ERR(obj))
+		return obj;
 
 	/*
 	 * Try to make the context utilize L3 as well as LLC.
drivers/gpu/drm/i915/i915_gem_render_state.c +5 -2
index 71611bf..071c11e 100644
@@ -58,8 +58,11 @@ static int render_state_init(struct render_state *so, struct drm_device *dev)
 		return -EINVAL;
 
 	so->obj = i915_gem_alloc_object(dev, 4096);
-	if (so->obj == NULL)
-		return -ENOMEM;
+	if (IS_ERR(so->obj)) {
+		ret = PTR_ERR(so->obj);
+		so->obj = NULL;
+		return ret;
+	}
 
 	ret = i915_gem_obj_ggtt_pin(so->obj, 4096, 0);
 	if (ret)
drivers/gpu/drm/i915/i915_guc_submission.c +1 -1
index da86bdb..004f8f7 100644
@@ -617,7 +617,7 @@ static struct drm_i915_gem_object *gem_allocate_guc_obj(struct drm_device *dev,
 	struct drm_i915_gem_object *obj;
 
 	obj = i915_gem_alloc_object(dev, size);
-	if (!obj)
+	if (IS_ERR(obj))
 		return NULL;
 
 	if (i915_gem_object_get_pages(obj)) {
drivers/gpu/drm/i915/intel_display.c +2 -2
index 551541b303..2247bdb 100644
@@ -10283,8 +10283,8 @@ intel_framebuffer_create_for_mode(struct drm_device *dev,
 
 	obj = i915_gem_alloc_object(dev,
 				    intel_framebuffer_size_for_mode(mode, bpp));
-	if (obj == NULL)
-		return ERR_PTR(-ENOMEM);
+	if (IS_ERR(obj))
+		return ERR_CAST(obj);
 
 	mode_cmd.width = mode->hdisplay;
 	mode_cmd.height = mode->vdisplay;
drivers/gpu/drm/i915/intel_fbdev.c +2 -2
index 79ac202..7047b38 100644
@@ -151,9 +151,9 @@ static int intelfb_alloc(struct drm_fb_helper *helper,
 		obj = i915_gem_object_create_stolen(dev, size);
 	if (obj == NULL)
 		obj = i915_gem_alloc_object(dev, size);
-	if (!obj) {
+	if (IS_ERR(obj)) {
 		DRM_ERROR("failed to allocate framebuffer\n");
-		ret = -ENOMEM;
+		ret = PTR_ERR(obj);
 		goto out;
 	}
 
drivers/gpu/drm/i915/intel_lrc.c +6 -4
index 0d6dc5e..c0543bb 100644
@@ -1483,9 +1483,11 @@ static int lrc_setup_wa_ctx_obj(struct intel_engine_cs *engine, u32 size)
 
 	engine->wa_ctx.obj = i915_gem_alloc_object(engine->dev,
 						   PAGE_ALIGN(size));
-	if (!engine->wa_ctx.obj) {
+	if (IS_ERR(engine->wa_ctx.obj)) {
 		DRM_DEBUG_DRIVER("alloc LRC WA ctx backing obj failed.\n");
-		return -ENOMEM;
+		ret = PTR_ERR(engine->wa_ctx.obj);
+		engine->wa_ctx.obj = NULL;
+		return ret;
 	}
 
 	ret = i915_gem_obj_ggtt_pin(engine->wa_ctx.obj, PAGE_SIZE, 0);
@@ -2638,9 +2640,9 @@ int intel_lr_context_deferred_alloc(struct intel_context *ctx,
 	context_size += PAGE_SIZE * LRC_PPHWSP_PN;
 
 	ctx_obj = i915_gem_alloc_object(dev, context_size);
-	if (!ctx_obj) {
+	if (IS_ERR(ctx_obj)) {
 		DRM_DEBUG_DRIVER("Alloc LRC backing obj failed.\n");
-		return -ENOMEM;
+		return PTR_ERR(ctx_obj);
 	}
 
 	ringbuf = intel_engine_create_ringbuffer(engine, 4 * PAGE_SIZE);
drivers/gpu/drm/i915/intel_overlay.c +1 -1
index 6694e92..77ee12f 100644
@@ -1397,7 +1397,7 @@ void intel_setup_overlay(struct drm_device *dev)
 		reg_bo = i915_gem_object_create_stolen(dev, PAGE_SIZE);
 	if (reg_bo == NULL)
 		reg_bo = i915_gem_alloc_object(dev, PAGE_SIZE);
-	if (reg_bo == NULL)
+	if (IS_ERR(reg_bo))
 		goto out_free;
 	overlay->reg_bo = reg_bo;
 
drivers/gpu/drm/i915/intel_ringbuffer.c +10 -9
index 41b604e..b20cf91 100644
@@ -672,9 +672,10 @@ intel_init_pipe_control(struct intel_engine_cs *engine)
 	WARN_ON(engine->scratch.obj);
 
 	engine->scratch.obj = i915_gem_alloc_object(engine->dev, 4096);
-	if (engine->scratch.obj == NULL) {
+	if (IS_ERR(engine->scratch.obj)) {
 		DRM_ERROR("Failed to allocate seqno page\n");
-		ret = -ENOMEM;
+		ret = PTR_ERR(engine->scratch.obj);
+		engine->scratch.obj = NULL;
 		goto err;
 	}
 
@@ -2021,9 +2022,9 @@ static int init_status_page(struct intel_engine_cs *engine)
 		int ret;
 
 		obj = i915_gem_alloc_object(engine->dev, 4096);
-		if (obj == NULL) {
+		if (IS_ERR(obj)) {
 			DRM_ERROR("Failed to allocate status page\n");
-			return -ENOMEM;
+			return PTR_ERR(obj);
 		}
 
 		ret = i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
@@ -2156,8 +2157,8 @@ static int intel_alloc_ringbuffer_obj(struct drm_device *dev,
 		obj = i915_gem_object_create_stolen(dev, ringbuf->size);
 	if (obj == NULL)
 		obj = i915_gem_alloc_object(dev, ringbuf->size);
-	if (obj == NULL)
-		return -ENOMEM;
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
 
 	/* mark ring buffers as read-only from GPU side by default */
 	obj->gt_ro = 1;
@@ -2780,7 +2781,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
 	if (INTEL_INFO(dev)->gen >= 8) {
 		if (i915_semaphore_is_enabled(dev)) {
 			obj = i915_gem_alloc_object(dev, 4096);
-			if (obj == NULL) {
+			if (IS_ERR(obj)) {
 				DRM_ERROR("Failed to allocate semaphore bo. Disabling semaphores\n");
 				i915.semaphores = 0;
 			} else {
@@ -2889,9 +2890,9 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
 	/* Workaround batchbuffer to combat CS tlb bug. */
 	if (HAS_BROKEN_CS_TLB(dev)) {
 		obj = i915_gem_alloc_object(dev, I830_WA_SIZE);
-		if (obj == NULL) {
+		if (IS_ERR(obj)) {
 			DRM_ERROR("Failed to allocate batch bo\n");
-			return -ENOMEM;
+			return PTR_ERR(obj);
 		}
 
 		ret = i915_gem_obj_ggtt_pin(obj, 0, 0);
Signature · 6 lines
2.4.11

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer Reviewed-by on v1 5/5
#9 lore raw

Re: [PATCH 5/5] drm/i915: Propagate error from drm_gem_object_init()

Quoted 3 lines · Matthew Auld

From: Chris Wilson <[email protected]>

From: Chris Wilson <[email protected]>

Double "From: ", one should be enough.

Quoted 54 lines

Propagate the real error from drm_gem_object_init(). Note this also fixes some confusion in the error return from i915_gem_alloc_object...

v2:
(Matthew Auld)
  - updated new users of gem_alloc_object from latest drm-nightly
  - replaced occurrences of IS_ERR_OR_NULL() with IS_ERR()

---
 drivers/gpu/drm/i915/i915_gem.c              | 14 ++++++++------ drivers/gpu/drm/i915/i915_gem_batch_pool.c   |  4 ++-- drivers/gpu/drm/i915/i915_gem_context.c      |  4 ++-- drivers/gpu/drm/i915/i915_gem_render_state.c |  7 +++++-- drivers/gpu/drm/i915/i915_guc_submission.c   |  2 +- drivers/gpu/drm/i915/intel_display.c         |  4 ++-- drivers/gpu/drm/i915/intel_fbdev.c           |  4 ++-- drivers/gpu/drm/i915/intel_lrc.c             | 10 ++++++---- drivers/gpu/drm/i915/intel_overlay.c         |  2 +- drivers/gpu/drm/i915/intel_ringbuffer.c      | 19 ++++++++++---------

 10 files changed, 39 insertions(+), 31 deletions(-)
drivers/gpu/drm/i915/i915_gem.c +0 -0
index b37ffea..6a6998c 100644
@@ -387,8 +387,8 @@ i915_gem_create(struct drm_file *file,


  /* Allocate the new object */
  obj = i915_gem_alloc_object(dev, size);

drivers/gpu/drm/i915/i915_gem.c +2 -2
-	if (obj == NULL)
-		return -ENOMEM;
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);


  ret = drm_gem_handle_create(file, &obj->base, &handle); /* drop reference from allocate - handle holds it now */

drivers/gpu/drm/i915/i915_gem.c +1 -0
@@ -4527,14 +4527,16 @@ struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev,
  	struct drm_i915_gem_object *obj;
  	struct address_space *mapping;
  	gfp_t mask;
+	int ret;


  obj = i915_gem_object_alloc(dev);
  if (obj == NULL)

drivers/gpu/drm/i915/i915_gem.c +1 -1
-		return NULL;
+		return ERR_PTR(-ENOMEM);

drivers/gpu/drm/i915/i915_gem.c +2 -1
-	if (drm_gem_object_init(dev, &obj->base, size) != 0) {
+	ret = drm_gem_object_init(dev, &obj->base, size);
+	if (ret) {

  i915_gem_object_free(obj);

drivers/gpu/drm/i915/i915_gem.c

Quoted 2 lines
drivers/gpu/drm/i915/i915_gem.c +1 -1
-		return NULL;
+		return ERR_PTR(ret);

While at it, I'd make this have a goto teardown path.

Quoted 50 lines

  }

  mask = GFP_HIGHUSER | __GFP_RECLAIMABLE;

drivers/gpu/drm/i915/i915_gem.c +0 -0
@@ -5390,7 +5392,7 @@ i915_gem_object_create_from_data(struct drm_device *dev,
  	int ret;


  obj = i915_gem_alloc_object(dev, round_up(size, PAGE_SIZE));

drivers/gpu/drm/i915/i915_gem.c +1 -1
-	if (IS_ERR_OR_NULL(obj))
+	if (IS_ERR(obj))

  return obj;

  ret = i915_gem_object_set_to_cpu_domain(obj, true);

drivers/gpu/drm/i915/i915_gem_batch_pool.c +0 -0
index 7bf2f3f..d79caa2 100644
@@ -135,8 +135,8 @@ i915_gem_batch_pool_get(struct i915_gem_batch_pool *pool,
  		int ret;


  obj = i915_gem_alloc_object(pool->dev, size);

drivers/gpu/drm/i915/i915_gem_batch_pool.c +2 -2
-		if (obj == NULL)
-			return ERR_PTR(-ENOMEM);
+		if (IS_ERR(obj))
+			return obj;


  ret = i915_gem_object_get_pages(obj); if (ret)

drivers/gpu/drm/i915/i915_gem_context.c +0 -0
index fe580cb..a5adc66 100644
@@ -179,8 +179,8 @@ i915_gem_alloc_context_obj(struct drm_device *dev, size_t size)
  	int ret;


  obj = i915_gem_alloc_object(dev, size);

drivers/gpu/drm/i915/i915_gem_context.c +2 -2
-	if (obj == NULL)
-		return ERR_PTR(-ENOMEM);
+	if (IS_ERR(obj))
+		return obj;


  /*
   * Try to make the context utilize L3 as well as LLC.

drivers/gpu/drm/i915/i915_gem_render_state.c +0 -0
index 71611bf..071c11e 100644
@@ -58,8 +58,11 @@ static int render_state_init(struct render_state *so, struct drm_device *dev)
  		return -EINVAL;


  so->obj = i915_gem_alloc_object(dev, 4096);

drivers/gpu/drm/i915/i915_gem_render_state.c static int render_state_init(struct render_state *so, struct drm_device *dev) · line 65

Quoted 7 lines
drivers/gpu/drm/i915/i915_gem_render_state.c +5 -2
-	if (so->obj == NULL)
-		return -ENOMEM;
+	if (IS_ERR(so->obj)) {
+		ret = PTR_ERR(so->obj);
+		so->obj = NULL;
+		return ret;
+	}

The teardown path in this function leaves so->obj != NULL in later failure path. Also i915_gem_alloc_object has only drm_gem_object_unreference as a counterpart so it'll leak too. Should be fixed in a follow-up patch.

Quoted 149 lines


  ret = i915_gem_obj_ggtt_pin(so->obj, 4096, 0); if (ret)

drivers/gpu/drm/i915/i915_guc_submission.c +0 -0
index da86bdb..004f8f7 100644
@@ -617,7 +617,7 @@ static struct drm_i915_gem_object *gem_allocate_guc_obj(struct drm_device *dev,
  	struct drm_i915_gem_object *obj;


  obj = i915_gem_alloc_object(dev, size);

drivers/gpu/drm/i915/i915_guc_submission.c +1 -1
-	if (!obj)
+	if (IS_ERR(obj))

  return NULL;

  if (i915_gem_object_get_pages(obj)) {

drivers/gpu/drm/i915/intel_display.c +0 -0
index 551541b303..2247bdb 100644
@@ -10283,8 +10283,8 @@ intel_framebuffer_create_for_mode(struct drm_device *dev,


  obj = i915_gem_alloc_object(dev, intel_framebuffer_size_for_mode(mode, bpp));

drivers/gpu/drm/i915/intel_display.c +2 -2
-	if (obj == NULL)
-		return ERR_PTR(-ENOMEM);
+	if (IS_ERR(obj))
+		return ERR_CAST(obj);


  mode_cmd.width = mode->hdisplay;
  mode_cmd.height = mode->vdisplay;

drivers/gpu/drm/i915/intel_fbdev.c +2 -2
index 79ac202..7047b38 100644
@@ -151,9 +151,9 @@ static int intelfb_alloc(struct drm_fb_helper *helper,
  		obj = i915_gem_object_create_stolen(dev, size);
  	if (obj == NULL)
  		obj = i915_gem_alloc_object(dev, size);
-	if (!obj) {
+	if (IS_ERR(obj)) {
  		DRM_ERROR("failed to allocate framebuffer\n");
-		ret = -ENOMEM;
+		ret = PTR_ERR(obj);
  		goto out;
  	}

drivers/gpu/drm/i915/intel_lrc.c +0 -0
index 0d6dc5e..c0543bb 100644
@@ -1483,9 +1483,11 @@ static int lrc_setup_wa_ctx_obj(struct intel_engine_cs *engine, u32 size)


  engine->wa_ctx.obj = i915_gem_alloc_object(engine->dev, PAGE_ALIGN(size));

drivers/gpu/drm/i915/intel_lrc.c +1 -1
-	if (!engine->wa_ctx.obj) {
+	if (IS_ERR(engine->wa_ctx.obj)) {

  DRM_DEBUG_DRIVER("alloc LRC WA ctx backing obj failed.\n");

drivers/gpu/drm/i915/intel_lrc.c +3 -1
-		return -ENOMEM;
+		ret = PTR_ERR(engine->wa_ctx.obj);
+		engine->wa_ctx.obj = NULL;
+		return ret;

  }

  ret = i915_gem_obj_ggtt_pin(engine->wa_ctx.obj, PAGE_SIZE, 0);

drivers/gpu/drm/i915/intel_lrc.c +0 -0
@@ -2638,9 +2640,9 @@ int intel_lr_context_deferred_alloc(struct intel_context *ctx,
  	context_size += PAGE_SIZE * LRC_PPHWSP_PN;


  ctx_obj = i915_gem_alloc_object(dev, context_size);

drivers/gpu/drm/i915/intel_lrc.c +1 -1
-	if (!ctx_obj) {
+	if (IS_ERR(ctx_obj)) {

  DRM_DEBUG_DRIVER("Alloc LRC backing obj failed.\n");

drivers/gpu/drm/i915/intel_lrc.c +1 -1
-		return -ENOMEM;
+		return PTR_ERR(ctx_obj);

  }

  ringbuf = intel_engine_create_ringbuffer(engine, 4 * PAGE_SIZE);

drivers/gpu/drm/i915/intel_overlay.c +1 -1
index 6694e92..77ee12f 100644
@@ -1397,7 +1397,7 @@ void intel_setup_overlay(struct drm_device *dev)
  		reg_bo = i915_gem_object_create_stolen(dev, PAGE_SIZE);
  	if (reg_bo == NULL)
  		reg_bo = i915_gem_alloc_object(dev, PAGE_SIZE);
-	if (reg_bo == NULL)
+	if (IS_ERR(reg_bo))
  		goto out_free;
  	overlay->reg_bo = reg_bo;

drivers/gpu/drm/i915/intel_ringbuffer.c +0 -0
index 41b604e..b20cf91 100644
@@ -672,9 +672,10 @@ intel_init_pipe_control(struct intel_engine_cs *engine)
  	WARN_ON(engine->scratch.obj);


  engine->scratch.obj = i915_gem_alloc_object(engine->dev, 4096);

drivers/gpu/drm/i915/intel_ringbuffer.c +1 -1
-	if (engine->scratch.obj == NULL) {
+	if (IS_ERR(engine->scratch.obj)) {

  DRM_ERROR("Failed to allocate seqno page\n");

drivers/gpu/drm/i915/intel_ringbuffer.c +2 -1
-		ret = -ENOMEM;
+		ret = PTR_ERR(engine->scratch.obj);
+		engine->scratch.obj = NULL;

  goto err;
  }

drivers/gpu/drm/i915/intel_ringbuffer.c +0 -0
@@ -2021,9 +2022,9 @@ static int init_status_page(struct intel_engine_cs *engine)
  		int ret;


  obj = i915_gem_alloc_object(engine->dev, 4096);

drivers/gpu/drm/i915/intel_ringbuffer.c +1 -1
-		if (obj == NULL) {
+		if (IS_ERR(obj)) {

  DRM_ERROR("Failed to allocate status page\n");

drivers/gpu/drm/i915/intel_ringbuffer.c +1 -1
-			return -ENOMEM;
+			return PTR_ERR(obj);

  }

  ret = i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);

drivers/gpu/drm/i915/intel_ringbuffer.c +2 -2
@@ -2156,8 +2157,8 @@ static int intel_alloc_ringbuffer_obj(struct drm_device *dev,
  		obj = i915_gem_object_create_stolen(dev, ringbuf->size);
  	if (obj == NULL)
  		obj = i915_gem_alloc_object(dev, ringbuf->size);
-	if (obj == NULL)
-		return -ENOMEM;
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);


  /* mark ring buffers as read-only from GPU side by default */ obj->gt_ro = 1;

drivers/gpu/drm/i915/intel_ringbuffer.c +3 -3
@@ -2780,7 +2781,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
  	if (INTEL_INFO(dev)->gen >= 8) {
  		if (i915_semaphore_is_enabled(dev)) {
  			obj = i915_gem_alloc_object(dev, 4096);
-			if (obj == NULL) {
+			if (IS_ERR(obj)) {
  				DRM_ERROR("Failed to allocate semaphore bo. Disabling semaphores\n");
  				i915.semaphores = 0;
  			} else {
@@ -2889,9 +2890,9 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
  	/* Workaround batchbuffer to combat CS tlb bug. */
  	if (HAS_BROKEN_CS_TLB(dev)) {
  		obj = i915_gem_alloc_object(dev, I830_WA_SIZE);
-		if (obj == NULL) {
+		if (IS_ERR(obj)) {
  			DRM_ERROR("Failed to allocate batch bo\n");
-			return -ENOMEM;
+			return PTR_ERR(obj);
  		}
Quoted 2 lines


  ret = i915_gem_obj_ggtt_pin(obj, 0, 0);

Some comments above.

Regards, Joonas

Signature · 7 lines
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#10 lore raw

✗ Fi.CI.BAT: failure for series starting with [1/5] drm/i915: use dev_priv directly in gen8_ppgtt_notify_vgt

In reply to Matthew Auld
== Series Details ==

== Series Details ==

Series: series starting with [1/5] drm/i915: use dev_priv directly in gen8_ppgtt_notify_vgt
URL : https://patchwork.freedesktop.org/series/5601/
State : failure

== Summary ==

Series 5601v1 Series without cover letter
http://patchwork.freedesktop.org/api/1.0/series/5601/revisions/1/mbox/

Test gem_exec_suspend:

        Subgroup basic-s3:
                incomplete -> PASS       (hsw-gt2)
Test gem_sync:
        Subgroup basic-vebox:
                pass       -> INCOMPLETE (hsw-gt2)
Test kms_flip:
        Subgroup basic-flip-vs-wf_vblank:
                pass       -> FAIL       (bsw-nuc-2)
bdw-nuci7        total:203  pass:191  dwarn:0   dfail:0   fail:0   skip:12 
bdw-ultra        total:203  pass:180  dwarn:0   dfail:0   fail:0   skip:23 
bsw-nuc-2        total:202  pass:162  dwarn:0   dfail:0   fail:1   skip:39 
byt-nuc          total:202  pass:164  dwarn:0   dfail:0   fail:0   skip:38 
hsw-brixbox      total:203  pass:179  dwarn:0   dfail:0   fail:0   skip:24 
hsw-gt2          total:142  pass:128  dwarn:0   dfail:0   fail:0   skip:13 
ilk-hp8440p      total:203  pass:135  dwarn:0   dfail:0   fail:0   skip:68 
ivb-t430s        total:203  pass:175  dwarn:0   dfail:0   fail:0   skip:28 
skl-i7k-2        total:203  pass:178  dwarn:0   dfail:0   fail:0   skip:25 
skl-nuci5        total:203  pass:192  dwarn:0   dfail:0   fail:0   skip:11 
snb-dellxps      total:203  pass:165  dwarn:0   dfail:0   fail:0   skip:38 
snb-x220t        total:203  pass:165  dwarn:0   dfail:0   fail:1   skip:37 

Results at /archive/results/CI_IGT_test/Patchwork_1878/

631ffd2f45bb43964f729e8661532fb115f5eeec drm-intel-nightly: 2016y-04m-13d-13h-00m-18s UTC integration manifest 9a1e64a drm/i915: Propagate error from drm_gem_object_init()
b2d1dcb drm/i915: call kunmap_px on pt_vaddr
4c75324 drm/i915: bail in alloc_pdp when !FULL_48BIT_PPGTT
b6ba9cf drm/i915: harden allocation paths in allocate_va_range
472e351 drm/i915: use dev_priv directly in gen8_ppgtt_notify_vgt

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

maintainer Reviewed-by on v1 1/5
#11 lore raw
In reply to Matthew Auld
Quoted 5 lines · Matthew Auld

Remove dev local and use to_i915() in gen8_ppgtt_notify_vgt.

v2: use dev_priv directly for QUESTION_MACROS (Joonas Lahtinen)

Cc Joonas Lahtinen <[email protected]>
Quoted 23 lines

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 5 ++---

 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -2
index c5cb049..5ed713d 100644
@@ -905,11 +905,10 @@ static int gen8_init_scratch(struct i915_address_space *vm)
  static int gen8_ppgtt_notify_vgt(struct i915_hw_ppgtt *ppgtt, bool create)
  {
  	enum vgt_g2v_type msg;
-	struct drm_device *dev = ppgtt->base.dev;
-	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct drm_i915_private *dev_priv = to_i915(ppgtt->base.dev);
  	int i;

drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
-	if (USES_FULL_48BIT_PPGTT(dev)) {
+	if (USES_FULL_48BIT_PPGTT(dev_priv)) {

  u64 daddr = px_dma(&ppgtt->pml4);

  I915_WRITE(vgtif_reg(pdp[0].lo), lower_32_bits(daddr));

Signature · 7 lines
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#12 lore raw
Quoted 14 lines · Joonas Lahtinen
Quoted 5 lines · Matthew Auld

Remove dev local and use to_i915() in gen8_ppgtt_notify_vgt.

v2: use dev_priv directly for QUESTION_MACROS (Joonas Lahtinen)

Cc Joonas Lahtinen <[email protected]>
Quoted 5 lines

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 5 ++---

 1 file changed, 2 insertions(+), 3 deletions(-)
Quoted 15 lines
Quoted 13 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -2
index c5cb049..5ed713d 100644
@@ -905,11 +905,10 @@ static int gen8_init_scratch(struct i915_address_space *vm)
  static int gen8_ppgtt_notify_vgt(struct i915_hw_ppgtt *ppgtt, bool create)
  {
  	enum vgt_g2v_type msg;
-	struct drm_device *dev = ppgtt->base.dev;
-	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct drm_i915_private *dev_priv = to_i915(ppgtt->base.dev);
  	int i;

drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_init_scratch(struct i915_address_space *vm) · line 911

Quoted 2 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
-	if (USES_FULL_48BIT_PPGTT(dev)) {
+	if (USES_FULL_48BIT_PPGTT(dev_priv)) {

The irony of this macro, is that it doesn't use its argument, and should it ever do so, it will use information stored in the ppgtt->vm. -Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
maintainer
#13 lore raw
Quoted 26 lines · Chris Wilson
Quoted 25 lines · Joonas Lahtinen
Quoted 6 lines · Matthew Auld

Remove dev local and use to_i915() in gen8_ppgtt_notify_vgt.

v2: use dev_priv directly for QUESTION_MACROS (Joonas Lahtinen)

Cc Joonas Lahtinen <[email protected]>
Quoted 15 lines

---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 5 ++---

 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +0 -1
index c5cb049..5ed713d 100644
@@ -905,11 +905,10 @@ static int gen8_init_scratch(struct i915_address_space *vm)
  static int gen8_ppgtt_notify_vgt(struct i915_hw_ppgtt *ppgtt, bool create)
  {
  	enum vgt_g2v_type msg;
-	struct drm_device *dev = ppgtt->base.dev;
Quoted 8 lines
Quoted 6 lines

drivers/gpu/drm/i915/i915_gem_gtt.c static int gen8_init_scratch(struct i915_address_space *vm) · line 911

Quoted 6 lines
drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
-	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct drm_i915_private *dev_priv = to_i915(ppgtt->base.dev);
  	int i;

drivers/gpu/drm/i915/i915_gem_gtt.c +1 -1
-	if (USES_FULL_48BIT_PPGTT(dev)) {
+	if (USES_FULL_48BIT_PPGTT(dev_priv)) {

The irony of this macro, is that it doesn't use its argument, and should it ever do so, it will use information stored in the ppgtt->vm.

Oh dear, only now looked at it. Pretty horrible macro :P

Regards, Joonas

Quoted 2 lines

-Chris

Signature · 7 lines
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx