Skip to contentSkip to conversation

drm/i915: Free resources correctly if we cannot map status page during ctx create

PATCH In review Arun Siluvery 6 participants 8 messages Nov 17, 2014 – Nov 18, 2014

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

PATCH 1/1 drm/i915: Free resources correctly if we cannot map status page during ctx create

We are not freeing memory allocated for ringbuf and ctx if we fail to map status page so release all resources correctly.

---
 drivers/gpu/drm/i915/intel_lrc.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_lrc.c +4 -2
index f3efdbd..a84d24b 100644
@@ -1777,8 +1777,10 @@ int intel_lr_context_deferred_create(struct intel_context *ctx,
 		ring->status_page.gfx_addr = i915_gem_obj_ggtt_offset(ctx_obj);
 		ring->status_page.page_addr =
 				kmap(sg_page(ctx_obj->pages->sgl));
-		if (ring->status_page.page_addr == NULL)
-			return -ENOMEM;
+		if (ring->status_page.page_addr == NULL) {
+			ret = -ENOMEM;
+			goto error;
+		}
 		ring->status_page.obj = ctx_obj;
 	}
 
Signature · 6 lines
2.1.2

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw
Quoted 59 lines

-----Original Message-----

From: Intel-gfx [mailto:[email protected]] On Behalf

Of Arun Siluvery

Sent: Monday, November 17, 2014 3:48 PM

To: [email protected]

Subject: [Intel-gfx] [PATCH] drm/i915: Free resources correctly if we cannot

map status page during ctx create

We are not freeing memory allocated for ringbuf and ctx if we fail to map

status page so release all resources correctly.

---

 drivers/gpu/drm/i915/intel_lrc.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
(quoted diff) +0 -0

b/drivers/gpu/drm/i915/intel_lrc.c

index f3efdbd..a84d24b 100644

--- a/drivers/gpu/drm/i915/intel_lrc.c

+++ b/drivers/gpu/drm/i915/intel_lrc.c

(quoted diff) +1 -2
@@ -1777,8 +1777,10 @@ int intel_lr_context_deferred_create(struct
 
 intel_context *ctx,
 
                            ring->status_page.gfx_addr =
 
 i915_gem_obj_ggtt_offset(ctx_obj);
 
                            ring->status_page.page_addr =
 
                                                            kmap(sg_page(ctx_obj->pages->sgl));
 
-                           if (ring->status_page.page_addr == NULL)
 
-                                           return -ENOMEM;
 
+                          if (ring->status_page.page_addr == NULL) {

Review int intel_lr_context_deferred_create(struct

Quoted 10 lines
(quoted diff) +3 -0
 
+                                          ret = -ENOMEM;
 
+                                          goto error;
 
+                          }
 
                            ring->status_page.obj = ctx_obj;
 
            }

Hi Arun,

I think your tree is out of date. See this patch:

http://patchwork.freedesktop.org/patch/35828/

Cheers,

Thomas.

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

#2 lore raw
Quoted 80 lines · Thomas
Quoted 59 lines

-----Original Message-----

From: Intel-gfx [mailto:[email protected]] On Behalf

Of Arun Siluvery

Sent: Monday, November 17, 2014 3:48 PM

To: [email protected]

Subject: [Intel-gfx] [PATCH] drm/i915: Free resources correctly if we cannot

map status page during ctx create

We are not freeing memory allocated for ringbuf and ctx if we fail to map

status page so release all resources correctly.

---

  drivers/gpu/drm/i915/intel_lrc.c | 6 ++++--
  1 file changed, 4 insertions(+), 2 deletions(-)
(quoted diff) +0 -0

b/drivers/gpu/drm/i915/intel_lrc.c

index f3efdbd..a84d24b 100644

--- a/drivers/gpu/drm/i915/intel_lrc.c

+++ b/drivers/gpu/drm/i915/intel_lrc.c

(quoted diff) +1 -2
@@ -1777,8 +1777,10 @@ int intel_lr_context_deferred_create(struct
 
 intel_context *ctx,
 
                             ring->status_page.gfx_addr =
 
 i915_gem_obj_ggtt_offset(ctx_obj);
 
                             ring->status_page.page_addr =
 
                                                             kmap(sg_page(ctx_obj->pages->sgl));
 
-                           if (ring->status_page.page_addr == NULL)
 
-                                           return -ENOMEM;
 
+                          if (ring->status_page.page_addr == NULL) {

Review int intel_lr_context_deferred_create(struct

Quoted 10 lines
(quoted diff) +3 -0
 
+                                          ret = -ENOMEM;
 
+                                          goto error;
 
+                          }
 
                             ring->status_page.obj = ctx_obj;
 
             }

Hi Arun,

I think your tree is out of date. See this patch:

Quoted 9 lines

http://patchwork.freedesktop.org/patch/35828/

Cheers,

Thomas.

You are right, I don't have latest changes.
This patch can be ignored.

regards
Arun

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

#3 lore raw
In reply to Arun Siluvery
Quoted 13 lines · Arun Siluvery

We are not freeing memory allocated for ringbuf and ctx if we fail to map status page so release all resources correctly.

---
 drivers/gpu/drm/i915/intel_lrc.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_lrc.c +0 -0
index f3efdbd..a84d24b 100644

drivers/gpu/drm/i915/intel_lrc.c int intel_lr_context_deferred_create(struct intel_context *ctx, · line 1783

Quoted 9 lines
drivers/gpu/drm/i915/intel_lrc.c +4 -2
@@ -1777,8 +1777,10 @@ int intel_lr_context_deferred_create(struct intel_context *ctx,
 		ring->status_page.gfx_addr = i915_gem_obj_ggtt_offset(ctx_obj);
 		ring->status_page.page_addr =
 				kmap(sg_page(ctx_obj->pages->sgl));
-		if (ring->status_page.page_addr == NULL)
-			return -ENOMEM;
+		if (ring->status_page.page_addr == NULL) {
+			ret = -ENOMEM;
+			goto error;
+		}

Since this popped up: Do we have an automated igt testcase to exercise this corner-case?

We do have it for legacy (gen6/7) contexts, including exercising the shrinker logic to get rid of misplaced ctx objects to avoid ggtt fragmentation hassles ...
-Daniel

Quoted 10 lines
drivers/gpu/drm/i915/intel_lrc.c +0 -0
 		ring->status_page.obj = ctx_obj;
 	}
 
Signature · 6 lines
2.1.2

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#4 lore raw
Quoted 18 lines · Daniel Vetter
Quoted 17 lines · Arun Siluvery

We are not freeing memory allocated for ringbuf and ctx if we fail to map status page so release all resources correctly.

---
 drivers/gpu/drm/i915/intel_lrc.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_lrc.c +0 -1
index f3efdbd..a84d24b 100644
@@ -1777,8 +1777,10 @@ int intel_lr_context_deferred_create(struct intel_context *ctx,
 		ring->status_page.gfx_addr = i915_gem_obj_ggtt_offset(ctx_obj);
 		ring->status_page.page_addr =
 				kmap(sg_page(ctx_obj->pages->sgl));
-		if (ring->status_page.page_addr == NULL)
Quoted 8 lines

drivers/gpu/drm/i915/intel_lrc.c int intel_lr_context_deferred_create(struct intel_context *ctx, · line 1783

Quoted 5 lines
drivers/gpu/drm/i915/intel_lrc.c +4 -1
-			return -ENOMEM;
+		if (ring->status_page.page_addr == NULL) {
+			ret = -ENOMEM;
+			goto error;
+		}

Since this popped up: Do we have an automated igt testcase to exercise this corner-case?

How? kmap() never returns NULL.
-Chris

Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#5 lore raw
Quoted 28 lines · Chris Wilson
Quoted 18 lines · Daniel Vetter
Quoted 17 lines · Arun Siluvery

We are not freeing memory allocated for ringbuf and ctx if we fail to map status page so release all resources correctly.

---
 drivers/gpu/drm/i915/intel_lrc.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/intel_lrc.c +0 -1
index f3efdbd..a84d24b 100644
@@ -1777,8 +1777,10 @@ int intel_lr_context_deferred_create(struct intel_context *ctx,
 		ring->status_page.gfx_addr = i915_gem_obj_ggtt_offset(ctx_obj);
 		ring->status_page.page_addr =
 				kmap(sg_page(ctx_obj->pages->sgl));
-		if (ring->status_page.page_addr == NULL)
Quoted 8 lines

drivers/gpu/drm/i915/intel_lrc.c int intel_lr_context_deferred_create(struct intel_context *ctx, · line 1783

Quoted 5 lines
drivers/gpu/drm/i915/intel_lrc.c +4 -1
-			return -ENOMEM;
+		if (ring->status_page.page_addr == NULL) {
+			ret = -ENOMEM;
+			goto error;
+		}

Since this popped up: Do we have an automated igt testcase to exercise this corner-case?

Quoted 1 line

How? kmap() never returns NULL.

Oh dear.

/me hangs head in shame

I'll do a patch to correct this.
-Daniel

Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
#6 lore raw

PATCH 1/1 drm/i915: Drop return value from lrc_setup_hardware_status_page

In reply to Chris Wilson

kmap never fails.

---
 drivers/gpu/drm/i915/intel_lrc.c | 15 +++------------
 1 file changed, 3 insertions(+), 12 deletions(-)
drivers/gpu/drm/i915/intel_lrc.c +3 -12
index b7c4c9ab9012..3cf15c4da0e8 100644
@@ -1706,7 +1706,7 @@ static uint32_t get_lr_context_size(struct intel_engine_cs *ring)
 	return ret;
 }
 
-static int lrc_setup_hardware_status_page(struct intel_engine_cs *ring,
+static void lrc_setup_hardware_status_page(struct intel_engine_cs *ring,
 		struct drm_i915_gem_object *default_ctx_obj)
 {
 	struct drm_i915_private *dev_priv = ring->dev->dev_private;
@@ -1716,15 +1716,11 @@ static int lrc_setup_hardware_status_page(struct intel_engine_cs *ring,
 	ring->status_page.gfx_addr = i915_gem_obj_ggtt_offset(default_ctx_obj);
 	ring->status_page.page_addr =
 			kmap(sg_page(default_ctx_obj->pages->sgl));
-	if (ring->status_page.page_addr == NULL)
-		return -ENOMEM;
 	ring->status_page.obj = default_ctx_obj;
 
 	I915_WRITE(RING_HWS_PGA(ring->mmio_base),
 			(u32)ring->status_page.gfx_addr);
 	POSTING_READ(RING_HWS_PGA(ring->mmio_base));
-
-	return 0;
 }
 
 /**
@@ -1811,13 +1807,8 @@ int intel_lr_context_deferred_create(struct intel_context *ctx,
 	ctx->engine[ring->id].ringbuf = ringbuf;
 	ctx->engine[ring->id].state = ctx_obj;
 
-	if (ctx == ring->default_context) {
-		ret = lrc_setup_hardware_status_page(ring, ctx_obj);
-		if (ret) {
-			DRM_ERROR("Failed to setup hardware status page\n");
-			goto error;
-		}
-	}
+	if (ctx == ring->default_context)
+		lrc_setup_hardware_status_page(ring, ctx_obj);
 
 	if (ring->id == RCS && !ctx->rcs_initialized) {
 		if (ring->init_context) {
Signature · 6 lines
2.1.1

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Tested-by on v1 1/1
#7 lore raw

Re: [PATCH] drm/i915: Drop return value from

-------------------------------------Summary-------------------------------------
Platform: baseline_drm_intel_nightly_pass_rate->patch_applied_pass_rate
BYT: pass/total=290/290->290/290
PNV: pass/total=362/362->362/362
ILK: pass/total=381/381->374/381
IVB: pass/total=522/559->522/559
SNB: pass/total=444/444->444/444
HSW: pass/total=595/595->594/595
BDW: pass/total=436/436->436/436
-------------------------------------Detailed-------------------------------------
test_platform: test_suite, test_case, result_with_drm_intel_nightly(count, machine_id...)...->result_with_patch_applied(count, machine_id)...
ILK: Intel_gpu_tools, igt_kms_flip_bcs-wf_vblank-vs-modeset-interruptible, PASS(4, M37M26) -> DMESG_WARN(1, M26)PASS(3, M26)
ILK: Intel_gpu_tools, igt_kms_flip_flip-vs-rmfb-interruptible, PASS(4, M37M26) -> DMESG_WARN(2, M26)PASS(2, M26)
ILK: Intel_gpu_tools, igt_kms_flip_plain-flip, PASS(4, M37M26) -> DMESG_WARN(1, M26)PASS(3, M26)
ILK: Intel_gpu_tools, igt_kms_flip_rcs-wf_vblank-vs-dpms-interruptible, PASS(4, M37M26) -> DMESG_WARN(3, M26)PASS(1, M26)
ILK: Intel_gpu_tools, igt_kms_flip_vblank-vs-hang-interruptible, PASS(4, M37M26) -> DMESG_WARN(3, M26)PASS(1, M26)
ILK: Intel_gpu_tools, igt_kms_flip_wf_vblank-ts-check, PASS(4, M37M26) -> DMESG_WARN(1, M26)PASS(3, M26)
ILK: Intel_gpu_tools, igt_kms_flip_wf_vblank-vs-dpms, DMESG_WARN(1, M26)PASS(3, M6M26) -> DMESG_WARN(2, M26)PASS(2, M26)
HSW: Intel_gpu_tools, igt_gem_reset_stats_close-pending-render, PASS(3, M19M40) -> NO_RESULT(1, M40)PASS(3, M40)
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx