Skip to contentSkip to conversation

drm/amdgpu/virt: skip VM fault handler for VF (v2)

PATCH In review Pixel Ding 3 participants 4 messages Feb 7, 2017 2 messages not in the archive

Subsystems (2)
lore
Share & tools
lore

Pixel Ding

author
#0 lore raw

PATCH 1/1 drm/amdgpu/virt: skip VM fault handler for VF (v2)

VF uses KIQ to access registers. When VM fault occurs, the driver can't get back the fence of KIQ submission and runs into CPU soft lockup.

v2: print IV entry info

---
 drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +7 -0
index 7669b32..6502508 100644
@@ -1237,6 +1237,13 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
 {
 	u32 addr, status, mc_client;
 
+	if (amdgpu_sriov_vf(adev)) {
+		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
+			entry->src_id, entry->src_data);
+		dev_err(adev->dev, " Can't decode VM fault info here on SRIOV VF\n");
+		return 0;
+	}
+
 	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
 	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
 	mc_client = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
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): 1486447619-22549-1-git-send-email-Pixel.Ding-5C7GfCeVMHo@public.gmane.org

Reviewed-by Junwei Zhang
#2 lore raw

If that info is enough for vf.

Regards,
Jerry (Junwei Zhang)

Linux Base Graphics
SRDC Software Development
_____________________________________

Quoted 36 lines

-----Original Message-----
From: Pixel Ding [mailto:[email protected]]
Sent: Tuesday, February 07, 2017 14:07
To: Zhou, David(ChunMing); Koenig, Christian; Zhang, Jerry
Cc: [email protected]; Ding, Pixel
Subject: [PATCH] drm/amdgpu/virt: skip VM fault handler for VF (v2)

VF uses KIQ to access registers. When VM fault occurs, the driver can't get back the fence of KIQ submission and runs into CPU soft lockup.

v2: print IV entry info

---
 drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +7 -0
index 7669b32..6502508 100644
@@ -1237,6 +1237,13 @@ static int gmc_v8_0_process_interrupt(struct
 amdgpu_device *adev,  {
 	u32 addr, status, mc_client;
 
+	if (amdgpu_sriov_vf(adev)) {
+		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
+			entry->src_id, entry->src_data);
+		dev_err(adev->dev, " Can't decode VM fault info here on SRIOV
 VF\n");
+		return 0;
+	}
+
 	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
 	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
Quoted 4 lines

mc_client =
RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
--
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): DM5PR12MB18183F0066F10FD5AF0041DFFF430-2J9CzHegvk+QhrfEZJlvtAdYzm3356FpvxpqHgZTriW3zl9H0oFU5g@public.gmane.org

#4 lore raw

Thank you, Jerry. You can find more detailed discussion in previous thread. —
Sincerely Yours,
Pixel

Quoted 37 lines · Jerry

If that info is enough for vf.

Regards,
Jerry (Junwei Zhang)

Linux Base Graphics
SRDC Software Development
_____________________________________

Quoted 26 lines

-----Original Message-----
From: Pixel Ding [mailto:[email protected]]
Sent: Tuesday, February 07, 2017 14:07
To: Zhou, David(ChunMing); Koenig, Christian; Zhang, Jerry
Cc: [email protected]; Ding, Pixel
Subject: [PATCH] drm/amdgpu/virt: skip VM fault handler for VF (v2)

VF uses KIQ to access registers. When VM fault occurs, the driver can't get back the fence of KIQ submission and runs into CPU soft lockup.

v2: print IV entry info

---
 drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +0 -0
index 7669b32..6502508 100644
@@ -1237,6 +1237,13 @@ static int gmc_v8_0_process_interrupt(struct
 amdgpu_device *adev,  {
 	u32 addr, status, mc_client;
 
Quoted 15 lines
Quoted 10 lines
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +7 -0
+	if (amdgpu_sriov_vf(adev)) {
+		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
+			entry->src_id, entry->src_data);
+		dev_err(adev->dev, " Can't decode VM fault info here on SRIOV
 VF\n");
+		return 0;
+	}
+
 	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
 	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
Quoted 4 lines

mc_client =
RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
--
2.7.4

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

#5 lore raw
In reply to a message not in the archive

Yeah, the src_data is unfortunately not completed (you simply can't have all the bits of a 40bit address+flags in a 32bit dw).

But it's the best we can do, so the patch is Reviewed-by: Christian König <[email protected]> as well.

Regards,
Christian.

Am 07.02.2017 um 08:03 schrieb Zhang, Jerry:

Quoted 47 lines

If that info is enough for vf.

Regards,
Jerry (Junwei Zhang)

Linux Base Graphics
SRDC Software Development
_____________________________________

Quoted 36 lines

-----Original Message-----
From: Pixel Ding [mailto:[email protected]]
Sent: Tuesday, February 07, 2017 14:07
To: Zhou, David(ChunMing); Koenig, Christian; Zhang, Jerry
Cc: [email protected]; Ding, Pixel
Subject: [PATCH] drm/amdgpu/virt: skip VM fault handler for VF (v2)

VF uses KIQ to access registers. When VM fault occurs, the driver can't get back the fence of KIQ submission and runs into CPU soft lockup.

v2: print IV entry info

---
  drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c | 7 +++++++
  1 file changed, 7 insertions(+)
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +7 -0
index 7669b32..6502508 100644
@@ -1237,6 +1237,13 @@ static int gmc_v8_0_process_interrupt(struct
 amdgpu_device *adev,  {
  	u32 addr, status, mc_client;
 
+	if (amdgpu_sriov_vf(adev)) {
+		dev_err(adev->dev, "GPU fault detected: %d 0x%08x\n",
+			entry->src_id, entry->src_data);
+		dev_err(adev->dev, " Can't decode VM fault info here on SRIOV
 VF\n");
+		return 0;
+	}
+
  	addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR);
  	status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS);
Quoted 8 lines
Quoted 4 lines

mc_client =
RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT);
--
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