Skip to contentSkip to conversation

drm/amdgpu: fix mutex list null pointer reference

PATCH NAK Xiangliang Yu 4 participants 12 messages May 4, 2017 – May 5, 2017 9 messages not in the archive

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

PATCH 1/6 drm/amdgpu: fix mutex list null pointer reference

From: Pixel Ding <[email protected]>

Fix NULL pointer reference.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +2 -1
index e685ca9..aef2019 100644
@@ -2118,7 +2118,8 @@ void amdgpu_device_fini(struct amdgpu_device *adev)
 
 	DRM_INFO("amdgpu: finishing device.\n");
 	adev->shutdown = true;
-	drm_crtc_force_disable_all(adev->ddev);
+	if (adev->ddev->mode_config.funcs)
+		drm_crtc_force_disable_all(adev->ddev);
 	/* evict vram memory */
 	amdgpu_bo_evict_vram(adev);
 	amdgpu_ib_pool_fini(adev);
Signature · 6 lines
2.7.4

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

A message that is not in the archive (sent off-list or not yet received): 1493879668-26947-1-git-send-email-Xiangliang.Yu-5C7GfCeVMHo@public.gmane.org

#2 lore raw

PATCH 2/6 drm/amdgpu: reset GDW, GWS and OA software copy to update them

From: Pixel Ding <[email protected]>

Reset GDW, GWS and OA when SRIOV do reset.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +9 -1
index aef2019..f11241d 100644
@@ -2538,7 +2538,7 @@ static int amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev,
  */
 int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)
 {
-	int i, r = 0;
+	int i, j, r = 0;
 	int resched;
 	struct amdgpu_bo *bo, *tmp;
 	struct amdgpu_ring *ring;
@@ -2616,6 +2616,14 @@ int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)
 	}
 	fence_put(fence);
 
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) {
+		struct amdgpu_vm_id_manager *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		for (j = 1; j < id_mgr->num_ids; j++)
+			amdgpu_vm_reset_id(adev, i, j);
+	}
+
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
 		if (!ring || !ring->sched.thread)
Signature · 6 lines
2.7.4

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

A message that is not in the archive (sent off-list or not yet received): 1493879668-26947-2-git-send-email-Xiangliang.Yu-5C7GfCeVMHo@public.gmane.org

#4 lore raw

RE: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

NAK, this patch is not needed currently

1, Because TDR feature is still undergoing, so all patches related with TDR should pending now
2, I don't think this is needed, @Pixel can you get the latest amd-staging-4.9 driver and try it, see if the "vm fault" duplicated without this patch? Because I remember no such issue In staging driver

BR Monk

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Xiangliang Yu
Sent: Thursday, May 04, 2017 2:34 PM
To: [email protected]
Cc: Ding, Pixel <[email protected]>; Yu, Xiangliang <[email protected]>
Subject: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

From: Pixel Ding <[email protected]>

Reset GDW, GWS and OA when SRIOV do reset.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +9 -1
index aef2019..f11241d 100644
@@ -2538,7 +2538,7 @@ static int amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev,
  */
 int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)  {
-	int i, r = 0;
+	int i, j, r = 0;
 	int resched;
 	struct amdgpu_bo *bo, *tmp;
 	struct amdgpu_ring *ring;
@@ -2616,6 +2616,14 @@ int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)
 	}
 	fence_put(fence);
 
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) {
+		struct amdgpu_vm_id_manager *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		for (j = 1; j < id_mgr->num_ids; j++)
+			amdgpu_vm_reset_id(adev, i, j);
+	}
+
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
 		if (!ring || !ring->sched.thread)

--
2.7.4

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

A message that is not in the archive (sent off-list or not yet received): DM5PR12MB1610683638F53030A9648FED84EA0-2J9CzHegvk++jCVTvoAFKAdYzm3356FpvxpqHgZTriW3zl9H0oFU5g@public.gmane.org

#6 lore raw

Re: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

In theory this patch is mandatory. It fixes the VM fault issue after TDR on Tonga. You can remove it when you rewrite TDR, but currently Tonga GPU reset has problem without this patch if MCBP is enabled. We need to flush GDS after reset, it’s bypassed if the SW state is not cleaned.

—
Sincerely Yours,
Pixel

Quoted 54 lines · Monk

NAK, this patch is not needed currently

1, Because TDR feature is still undergoing, so all patches related with TDR should pending now
2, I don't think this is needed, @Pixel can you get the latest amd-staging-4.9 driver and try it, see if the "vm fault" duplicated without this patch? Because I remember no such issue In staging driver

BR Monk

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Xiangliang Yu
Sent: Thursday, May 04, 2017 2:34 PM
To: [email protected]
Cc: Ding, Pixel <[email protected]>; Yu, Xiangliang <[email protected]>
Subject: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

From: Pixel Ding <[email protected]>

Reset GDW, GWS and OA when SRIOV do reset.

---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 10 +++++++++-

1 file changed, 9 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +9 -2
index aef2019..f11241d 100644
@@ -2538,7 +2538,7 @@ static int amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev,
 */
 int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)  {
-	int i, r = 0;
+	int i, j, r = 0;
 	int resched;
 	struct amdgpu_bo *bo, *tmp;
 	struct amdgpu_ring *ring;
@@ -2616,6 +2616,14 @@ int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)
 	}
 	fence_put(fence);
 
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) {
+		struct amdgpu_vm_id_manager *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		for (j = 1; j < id_mgr->num_ids; j++)
+			amdgpu_vm_reset_id(adev, i, j);
+	}
+
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
 		if (!ring || !ring->sched.thread)
--

2.7.4

Quoted 4 lines

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

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

A message that is not in the archive (sent off-list or not yet received): 44D2C15E-B6C5-4E0F-94FC-BDCB00A02169-5C7GfCeVMHo@public.gmane.org

#8 lore raw

RE: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

Like I said, you need to test it on amd-staging-4.9 branch

I don't think this patch is necessary unless you prove it

-----Original Message-----
From: Ding, Pixel
Sent: Thursday, May 04, 2017 4:14 PM
To: Liu, Monk <[email protected]>; Yu, Xiangliang <[email protected]>; [email protected]
Subject: Re: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

In theory this patch is mandatory. It fixes the VM fault issue after TDR on Tonga. You can remove it when you rewrite TDR, but currently Tonga GPU reset has problem without this patch if MCBP is enabled. We need to flush GDS after reset, it’s bypassed if the SW state is not cleaned.

—
Sincerely Yours,
Pixel

Quoted 54 lines · Monk

NAK, this patch is not needed currently

1, Because TDR feature is still undergoing, so all patches related with TDR should pending now
2, I don't think this is needed, @Pixel can you get the latest amd-staging-4.9 driver and try it, see if the "vm fault" duplicated without this patch? Because I remember no such issue In staging driver

BR Monk

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Xiangliang Yu
Sent: Thursday, May 04, 2017 2:34 PM
To: [email protected]
Cc: Ding, Pixel <[email protected]>; Yu, Xiangliang <[email protected]>
Subject: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

From: Pixel Ding <[email protected]>

Reset GDW, GWS and OA when SRIOV do reset.

---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 10 +++++++++-

1 file changed, 9 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +9 -2
index aef2019..f11241d 100644
@@ -2538,7 +2538,7 @@ static int amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev,
 */
 int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)  {
-	int i, r = 0;
+	int i, j, r = 0;
 	int resched;
 	struct amdgpu_bo *bo, *tmp;
 	struct amdgpu_ring *ring;
@@ -2616,6 +2616,14 @@ int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)
 	}
 	fence_put(fence);
 
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) {
+		struct amdgpu_vm_id_manager *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		for (j = 1; j < id_mgr->num_ids; j++)
+			amdgpu_vm_reset_id(adev, i, j);
+	}
+
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
 		if (!ring || !ring->sched.thread)
--

2.7.4

Quoted 4 lines

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

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

A message that is not in the archive (sent off-list or not yet received): DM5PR12MB161010192C676A1BBAD92F2B84EA0-2J9CzHegvk++jCVTvoAFKAdYzm3356FpvxpqHgZTriW3zl9H0oFU5g@public.gmane.org

#10 lore raw

Re: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

Hi Xiangliang,

Please verify as Monk requested.

—
Sincerely Yours,
Pixel

Quoted 78 lines · Monk

Like I said, you need to test it on amd-staging-4.9 branch

I don't think this patch is necessary unless you prove it

-----Original Message-----
From: Ding, Pixel
Sent: Thursday, May 04, 2017 4:14 PM
To: Liu, Monk <[email protected]>; Yu, Xiangliang <[email protected]>; [email protected]
Subject: Re: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

In theory this patch is mandatory. It fixes the VM fault issue after TDR on Tonga. You can remove it when you rewrite TDR, but currently Tonga GPU reset has problem without this patch if MCBP is enabled. We need to flush GDS after reset, it’s bypassed if the SW state is not cleaned.

—
Sincerely Yours,
Pixel

Quoted 54 lines · Monk

NAK, this patch is not needed currently

1, Because TDR feature is still undergoing, so all patches related with TDR should pending now
2, I don't think this is needed, @Pixel can you get the latest amd-staging-4.9 driver and try it, see if the "vm fault" duplicated without this patch? Because I remember no such issue In staging driver

BR Monk

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Xiangliang Yu
Sent: Thursday, May 04, 2017 2:34 PM
To: [email protected]
Cc: Ding, Pixel <[email protected]>; Yu, Xiangliang <[email protected]>
Subject: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

From: Pixel Ding <[email protected]>

Reset GDW, GWS and OA when SRIOV do reset.

---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 10 +++++++++-

1 file changed, 9 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +9 -2
index aef2019..f11241d 100644
@@ -2538,7 +2538,7 @@ static int amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev,
 */
 int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)  {
-	int i, r = 0;
+	int i, j, r = 0;
 	int resched;
 	struct amdgpu_bo *bo, *tmp;
 	struct amdgpu_ring *ring;
@@ -2616,6 +2616,14 @@ int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool voluntary)
 	}
 	fence_put(fence);
 
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) {
+		struct amdgpu_vm_id_manager *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		for (j = 1; j < id_mgr->num_ids; j++)
+			amdgpu_vm_reset_id(adev, i, j);
+	}
+
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
 		if (!ring || !ring->sched.thread)
--

2.7.4

Quoted 4 lines
Quoted 4 lines

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

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

A message that is not in the archive (sent off-list or not yet received): 6EDD67DA-B477-479B-AB90-B59BE904D3A5-5C7GfCeVMHo@public.gmane.org

#12 lore raw

RE: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

@Pixel, @monk,

found out staging FLR has issue, I'll check it at first. Let's hold the patch.

Thanks!
Xiangliang Yu

Quoted 117 lines

-----Original Message-----
From: Ding, Pixel
Sent: Thursday, May 04, 2017 4:23 PM
To: Liu, Monk <[email protected]>; Yu, Xiangliang <[email protected]>; [email protected]
Subject: Re: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

Hi Xiangliang,

Please verify as Monk requested.

—
Sincerely Yours,
Pixel

Quoted 8 lines · Monk

Like I said, you need to test it on amd-staging-4.9 branch

I don't think this patch is necessary unless you prove it

-----Original Message-----
From: Ding, Pixel
Sent: Thursday, May 04, 2017 4:14 PM
To: Liu, Monk <[email protected]>; Yu, Xiangliang

<[email protected]>; [email protected]

Quoted 1 line

Subject: Re: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software

copy to update them

Quoted 2 lines

In theory this patch is mandatory. It fixes the VM fault issue after TDR on

Tonga. You can remove it when you rewrite TDR, but currently Tonga GPU reset has problem without this patch if MCBP is enabled. We need to flush GDS after reset, it’s bypassed if the SW state is not cleaned.

Quoted 11 lines

—
Sincerely Yours,
Pixel

Quoted 5 lines
Quoted 3 lines · Monk

NAK, this patch is not needed currently

1, Because TDR feature is still undergoing, so all patches related with TDR

should pending now

Quoted 1 line
Quoted 1 line

2, I don't think this is needed, @Pixel can you get the latest amd-staging-

4.9 driver and try it, see if the "vm fault" duplicated without this patch? Because I remember no such issue In staging driver

Quoted 5 lines
Quoted 5 lines

BR Monk

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On

Behalf Of Xiangliang Yu

Quoted 3 lines
Quoted 3 lines

Sent: Thursday, May 04, 2017 2:34 PM
To: [email protected]
Cc: Ding, Pixel <[email protected]>; Yu, Xiangliang

<[email protected]>

Quoted 1 line
Quoted 1 line

Subject: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software

copy to update them

Quoted 5 lines
Quoted 5 lines

From: Pixel Ding <[email protected]>

Reset GDW, GWS and OA when SRIOV do reset.

Quoted 7 lines

Review

Quoted 7 lines

---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 10 +++++++++-

1 file changed, 9 insertions(+), 1 deletion(-)
(quoted diff) +0 -0

b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

Quoted 4 lines

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c static int

Quoted 4 lines

index aef2019..f11241d 100644

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +0 -0
@@ -2538,7 +2538,7 @@ static int

amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev,

Quoted 2 lines
Quoted 2 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +0 -0
 */

int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool

voluntary) {

Quoted 6 lines

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c int amdgpu_sriov_gpu_reset(struct

Quoted 6 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +1 -1
-	int i, r = 0;
+	int i, j, r = 0;
	int resched;
	struct amdgpu_bo *bo, *tmp;
	struct amdgpu_ring *ring;
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +0 -0
@@ -2616,6 +2616,14 @@ int amdgpu_sriov_gpu_reset(struct

amdgpu_device *adev, bool voluntary)

Quoted 17 lines
Quoted 17 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +8 -1
 	}
 	fence_put(fence);
 
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) {
+		struct amdgpu_vm_id_manager *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		for (j = 1; j < id_mgr->num_ids; j++)
+			amdgpu_vm_reset_id(adev, i, j);
+	}
+
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
 		if (!ring || !ring->sched.thread)
--

2.7.4

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

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

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

A message that is not in the archive (sent off-list or not yet received): BY2PR1201MB093574C59B74EC96F45AA682EBEB0-O28G1zQ8oGkaqtME6NEo1mrFom/aUZj6nBOFsp37pqbUKgpGm//BTAC/[email protected]

#14 lore raw

RE: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

can you check if vega the same

-----Original Message-----
From: Yu, Xiangliang
Sent: Friday, May 5, 2017 5:18 PM
To: Ding, Pixel <[email protected]>; Liu, Monk <[email protected]>; [email protected]
Subject: RE: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

@Pixel, @monk,

found out staging FLR has issue, I'll check it at first. Let's hold the patch.

Thanks!
Xiangliang Yu

Quoted 120 lines

-----Original Message-----
From: Ding, Pixel
Sent: Thursday, May 04, 2017 4:23 PM
To: Liu, Monk <[email protected]>; Yu, Xiangliang <[email protected]>; [email protected]
Subject: Re: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software copy to update them

Hi Xiangliang,

Please verify as Monk requested.

—
Sincerely Yours,
Pixel

Quoted 8 lines · Monk

Like I said, you need to test it on amd-staging-4.9 branch

I don't think this patch is necessary unless you prove it

-----Original Message-----
From: Ding, Pixel
Sent: Thursday, May 04, 2017 4:14 PM
To: Liu, Monk <[email protected]>; Yu, Xiangliang

<[email protected]>; [email protected]

Quoted 1 line

Subject: Re: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software

copy to update them

Quoted 3 lines

In theory this patch is mandatory. It fixes the VM fault issue after TDR on

Tonga. You can remove it when you rewrite TDR, but currently Tonga GPU reset has problem without this patch if MCBP is enabled. We need to flush GDS after reset, it’s bypassed if the SW state is not cleaned.

Quoted 11 lines

—
Sincerely Yours,
Pixel

Quoted 6 lines
Quoted 4 lines · Monk

NAK, this patch is not needed currently

1, Because TDR feature is still undergoing, so all patches related with TDR

should pending now

Quoted 2 lines
Quoted 2 lines

2, I don't think this is needed, @Pixel can you get the latest amd-staging-

4.9 driver and try it, see if the "vm fault" duplicated without this patch? Because I remember no such issue In staging driver

Quoted 5 lines
Quoted 5 lines

BR Monk

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On

Behalf Of Xiangliang Yu

Quoted 3 lines
Quoted 3 lines

Sent: Thursday, May 04, 2017 2:34 PM
To: [email protected]
Cc: Ding, Pixel <[email protected]>; Yu, Xiangliang

<[email protected]>

Quoted 1 line
Quoted 1 line

Subject: [PATCH 2/6] drm/amdgpu: reset GDW, GWS and OA software

copy to update them

Quoted 5 lines
Quoted 5 lines

From: Pixel Ding <[email protected]>

Reset GDW, GWS and OA when SRIOV do reset.

Quoted 7 lines

Review

Quoted 7 lines

---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 10 +++++++++-

1 file changed, 9 insertions(+), 1 deletion(-)
(quoted diff) +0 -0

b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

Quoted 4 lines

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c static int

Quoted 4 lines

index aef2019..f11241d 100644

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +0 -0
@@ -2538,7 +2538,7 @@ static int

amdgpu_recover_vram_from_shadow(struct amdgpu_device *adev,

Quoted 2 lines
Quoted 2 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +0 -0
 */

int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, bool

voluntary) {

Quoted 6 lines

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c int amdgpu_sriov_gpu_reset(struct

Quoted 6 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +1 -1
-	int i, r = 0;
+	int i, j, r = 0;
	int resched;
	struct amdgpu_bo *bo, *tmp;
	struct amdgpu_ring *ring;
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +0 -0
@@ -2616,6 +2616,14 @@ int amdgpu_sriov_gpu_reset(struct

amdgpu_device *adev, bool voluntary)

Quoted 17 lines
Quoted 17 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +8 -1
 	}
 	fence_put(fence);
 
+	for (i = 0; i < AMDGPU_MAX_VMHUBS; i++) {
+		struct amdgpu_vm_id_manager *id_mgr =
+			&adev->vm_manager.id_mgr[i];
+
+		for (j = 1; j < id_mgr->num_ids; j++)
+			amdgpu_vm_reset_id(adev, i, j);
+	}
+
 	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
 		struct amdgpu_ring *ring = adev->rings[i];
 		if (!ring || !ring->sched.thread)
--

2.7.4

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

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

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

#15 lore raw

PATCH 3/6 drm/amdgpu: don't clean the framebuffer for VF

In reply to a message not in the archive

From: Pixel Ding <[email protected]>

The SRIOV host driver cleans framebuffer for each VF, guest driver needn't this action which costs much time on some virtualization platform, otherwise it might get timeout to initialize.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c +3 -0
index e869e60..6f5d994 100644
@@ -242,6 +242,9 @@ static int amdgpufb_create(struct drm_fb_helper *helper,
 	/* setup helper */
 	rfbdev->helper.fb = fb;
 
+	if (!amdgpu_sriov_vf(adev))
+		memset_io(abo->kptr, 0x0, amdgpu_bo_size(abo));
+
 	strcpy(info->fix.id, "amdgpudrmfb");
 
 	drm_fb_helper_fill_fix(info, fb->pitches[0], fb->depth);
Signature · 6 lines
2.7.4

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

A message that is not in the archive (sent off-list or not yet received): 1493879668-26947-3-git-send-email-Xiangliang.Yu-5C7GfCeVMHo@public.gmane.org

#17 lore raw

RE: [PATCH 3/6] drm/amdgpu: don't clean the framebuffer for VF

NAK

This behavior that Hypervisor cleans framebuffer for each VF is only on AWS, Regular GIM doesn't clean framebuffer after FLR, so guest driver still need this cleaning,

Please keep this patch in your AWS branch only.

BR Monk

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Xiangliang Yu
Sent: Thursday, May 04, 2017 2:34 PM
To: [email protected]
Cc: Ding, Pixel <[email protected]>; Yu, Xiangliang <[email protected]>
Subject: [PATCH 3/6] drm/amdgpu: don't clean the framebuffer for VF

From: Pixel Ding <[email protected]>

The SRIOV host driver cleans framebuffer for each VF, guest driver needn't this action which costs much time on some virtualization platform, otherwise it might get timeout to initialize.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c +3 -0
index e869e60..6f5d994 100644
@@ -242,6 +242,9 @@ static int amdgpufb_create(struct drm_fb_helper *helper,
 	/* setup helper */
 	rfbdev->helper.fb = fb;
 
+	if (!amdgpu_sriov_vf(adev))
+		memset_io(abo->kptr, 0x0, amdgpu_bo_size(abo));
+
 	strcpy(info->fix.id, "amdgpudrmfb");
 
 	drm_fb_helper_fill_fix(info, fb->pitches[0], fb->depth);

--
2.7.4

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

A message that is not in the archive (sent off-list or not yet received): DM5PR12MB1610A6123C5CDA271D09A5FB84EA0-2J9CzHegvk++jCVTvoAFKAdYzm3356FpvxpqHgZTriW3zl9H0oFU5g@public.gmane.org

#19 lore raw

Re: [PATCH 3/6] drm/amdgpu: don't clean the framebuffer for VF

This patch is not necessary since FB is already cleaned by GPU. —
Sincerely Yours,
Pixel

Quoted 43 lines · Monk

NAK

This behavior that Hypervisor cleans framebuffer for each VF is only on AWS, Regular GIM doesn't clean framebuffer after FLR, so guest driver still need this cleaning,

Please keep this patch in your AWS branch only.

BR Monk

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Xiangliang Yu
Sent: Thursday, May 04, 2017 2:34 PM
To: [email protected]
Cc: Ding, Pixel <[email protected]>; Yu, Xiangliang <[email protected]>
Subject: [PATCH 3/6] drm/amdgpu: don't clean the framebuffer for VF

From: Pixel Ding <[email protected]>

The SRIOV host driver cleans framebuffer for each VF, guest driver needn't this action which costs much time on some virtualization platform, otherwise it might get timeout to initialize.

---
drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c | 3 +++

1 file changed, 3 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c +3 -1
index e869e60..6f5d994 100644
@@ -242,6 +242,9 @@ static int amdgpufb_create(struct drm_fb_helper *helper,
 	/* setup helper */
 	rfbdev->helper.fb = fb;
 
+	if (!amdgpu_sriov_vf(adev))
+		memset_io(abo->kptr, 0x0, amdgpu_bo_size(abo));
+
 	strcpy(info->fix.id, "amdgpudrmfb");
 
 	drm_fb_helper_fill_fix(info, fb->pitches[0], fb->depth);
--

2.7.4

Quoted 4 lines

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

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

Reviewed-by Alex Deucher
#20 lore raw
In reply to a message not in the archive
Quoted 24 lines

-----Original Message-----
From: amd-gfx [mailto:[email protected]] On Behalf Of Xiangliang Yu
Sent: Thursday, May 04, 2017 2:34 AM
To: [email protected]
Cc: Ding, Pixel; Yu, Xiangliang
Subject: [PATCH 1/6] drm/amdgpu: fix mutex list null pointer reference

From: Pixel Ding <[email protected]>

Fix NULL pointer reference.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +0 -0
index e685ca9..aef2019 100644

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c void amdgpu_device_fini(struct amdgpu_device · line 2123

Quoted 7 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +2 -1
@@ -2118,7 +2118,8 @@ void amdgpu_device_fini(struct amdgpu_device
 *adev)
 
 	DRM_INFO("amdgpu: finishing device.\n");
 	adev->shutdown = true;
-	drm_crtc_force_disable_all(adev->ddev);
+	if (adev->ddev->mode_config.funcs)
+		drm_crtc_force_disable_all(adev->ddev);

For consistency, I think we generally check for adev->mode_info.mode_confg_initialized or adev->mode_info.num_crtc. Either way:

Quoted 10 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +0 -0
 	/* evict vram memory */
 	amdgpu_bo_evict_vram(adev);

amdgpu_ib_pool_fini(adev);
--
2.7.4

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

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