Skip to contentSkip to conversation

runtime pm fix

PATCH In review Kenneth Feng 3 participants 6 messages Nov 17, 2020 – Nov 18, 2020

Subsystems (3)
lore
Share & tools
lore

Kenneth Feng

author maintainer
#0 lore raw

cover runtime pm fix

this 2 patches fix the runtime pm problems on SIENNA_CICHLID.

Kenneth Feng (2):
drm/amd/amdgpu: fix null pointer in runtime pm drm/amd/pm: fix the crash after runtime pm resume

 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 4 ++--
 drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c  | 7 +++++--
 2 files changed, 7 insertions(+), 4 deletions(-)
Signature · 6 lines
2.17.1

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

PATCH 1/2 drm/amd/amdgpu: fix null pointer in runtime pm

fix the null pointer issue when runtime pm is triggered.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +2 -2
index 1fe850e0a94d..0ebba4380979 100644
@@ -4859,7 +4859,7 @@ int amdgpu_device_baco_enter(struct drm_device *dev)
 	if (!amdgpu_device_supports_baco(adev_to_drm(adev)))
 		return -ENOTSUPP;
 
-	if (ras && ras->supported)
+	if (ras && ras->supported && adev->nbio.funcs->enable_doorbell_interrupt)
 		adev->nbio.funcs->enable_doorbell_interrupt(adev, false);
 
 	return amdgpu_dpm_baco_enter(adev);
@@ -4878,7 +4878,7 @@ int amdgpu_device_baco_exit(struct drm_device *dev)
 	if (ret)
 		return ret;
 
-	if (ras && ras->supported)
+	if (ras && ras->supported && adev->nbio.funcs->enable_doorbell_interrupt)
 		adev->nbio.funcs->enable_doorbell_interrupt(adev, true);
 
 	return 0;
Signature · 6 lines
2.17.1

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

PATCH 2/2 drm/amd/pm: fix the crash after runtime pm resume

In reply to Kenneth Feng

Some features are still disabled after runtime pm resume. This can take the hardware back. Unlike other projects, this doesn't need pptable retransfer.

---
 drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +5 -2
index 39990790ed67..ebd50f144c5d 100644
@@ -914,11 +914,14 @@ static int smu_smc_hw_setup(struct smu_context *smu)
 {
 	struct amdgpu_device *adev = smu->adev;
 	uint32_t pcie_gen = 0, pcie_width = 0;
-	int ret;
+	int ret = 0;
 
 	if (adev->in_suspend && smu_is_dpm_running(smu)) {
 		dev_info(adev->dev, "dpm has been enabled\n");
-		return 0;
+		/* this is needed specifically */
+		if (adev->asic_type == CHIP_SIENNA_CICHLID)
+			ret = smu_system_features_control(smu, true);
+		return ret;
 	}
 
 	ret = smu_init_display_count(smu, 0);
Signature · 6 lines
2.17.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
maintainer Reviewed-by on v1 2/2
#3 lore raw

Re: [PATCH 2/2] drm/amd/pm: fix the crash after runtime pm resume

[AMD Official Use Only - Internal Distribution Only]

Series is:

Does the same thing need to be applied to navy flounder and dimgrey cavefish?

Alex

________________________________
From: amd-gfx <[email protected]> on behalf of Kenneth Feng <[email protected]>
Sent: Tuesday, November 17, 2020 8:30 AM
To: [email protected] <[email protected]>
Cc: Feng, Kenneth <[email protected]>
Subject: [PATCH 2/2] drm/amd/pm: fix the crash after runtime pm resume

Some features are still disabled after runtime pm resume. This can take the hardware back. Unlike other projects, this doesn't need pptable retransfer.

---
 drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +5 -2
index 39990790ed67..ebd50f144c5d 100644
@@ -914,11 +914,14 @@ static int smu_smc_hw_setup(struct smu_context *smu)
 {
         struct amdgpu_device *adev = smu->adev;
         uint32_t pcie_gen = 0, pcie_width = 0;
-       int ret;
+       int ret = 0;
 
         if (adev->in_suspend && smu_is_dpm_running(smu)) {
                 dev_info(adev->dev, "dpm has been enabled\n");
-               return 0;
+               /* this is needed specifically */
+               if (adev->asic_type == CHIP_SIENNA_CICHLID)
+                       ret = smu_system_features_control(smu, true);
+               return ret;
         }
 
         ret = smu_init_display_count(smu, 0);

--
2.17.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Famd-gfx&amp;data=04%7C01%7Calexander.deucher%40amd.com%7Cdd553fd04f0f44ebf3ca08d88afd11ad%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637412166838877500%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=swZeNMubEey64ZdWb7W5sTg88as3VNS9ILwmLb08GHY%3D&amp;reserved=0

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

#4 lore raw

RE: [PATCH 2/2] drm/amd/pm: fix the crash after runtime pm resume

In reply to Kenneth Feng

[AMD Official Use Only - Internal Distribution Only]

I suspect "smu_is_dpm_running" reports a wrong result due to "smu_system_features_control(smu, false)" was skipped for SIENNA_CICHLID in smu_disable_dpms(). That is outdated cached values were used for "smu_is_dpm_running".
If that's true, a fix in smu_disable_dpms() may be more proper.

Evan
-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Kenneth Feng
Sent: Tuesday, November 17, 2020 9:31 PM
To: [email protected]
Cc: Feng, Kenneth <[email protected]>
Subject: [PATCH 2/2] drm/amd/pm: fix the crash after runtime pm resume

Some features are still disabled after runtime pm resume. This can take the hardware back. Unlike other projects, this doesn't need pptable retransfer.

---
 drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +5 -3
index 39990790ed67..ebd50f144c5d 100644
@@ -914,11 +914,14 @@ static int smu_smc_hw_setup(struct smu_context *smu)  {
 struct amdgpu_device *adev = smu->adev;
 uint32_t pcie_gen = 0, pcie_width = 0;
-int ret;
+int ret = 0;
 
 if (adev->in_suspend && smu_is_dpm_running(smu)) {
 dev_info(adev->dev, "dpm has been enabled\n");
-return 0;
+/* this is needed specifically */
+if (adev->asic_type == CHIP_SIENNA_CICHLID)
+ret = smu_system_features_control(smu, true);
+return ret;
 }
 
 ret = smu_init_display_count(smu, 0);
--

2.17.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Famd-gfx&amp;data=04%7C01%7Cevan.quan%40amd.com%7Cdd553fd04f0f44ebf3ca08d88afd11ad%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637412166811942772%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=SgtA0lq6Sx5aiQ5PSLvf9GdXUU50cFLE5ELMY0e8k5g%3D&amp;reserved=0
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

author maintainer
#5 lore raw

RE: [PATCH 2/2] drm/amd/pm: fix the crash after runtime pm resume

[AMD Official Use Only - Internal Distribution Only]

[AMD Official Use Only - Internal Distribution Only]

Hi Evan/Alex,
In the process of baco enter/exit, smu disables all features, so smu_disable_dpms is not needed.
And the sequential steps in smu_smc_hw_setup after function smu_is_dpm_running are not needed since baco doesn't clean up the smu status. This fix is specific to SIENNA_CICHLID. For the other ASICS, need to confirm since the firmware might be different. Thanks.

Best Regards
Kenneth

-----Original Message-----
From: Quan, Evan <[email protected]>
Sent: Wednesday, November 18, 2020 8:46 AM
To: Feng, Kenneth <[email protected]>; [email protected]
Cc: Feng, Kenneth <[email protected]>
Subject: RE: [PATCH 2/2] drm/amd/pm: fix the crash after runtime pm resume

[AMD Official Use Only - Internal Distribution Only]

I suspect "smu_is_dpm_running" reports a wrong result due to "smu_system_features_control(smu, false)" was skipped for SIENNA_CICHLID in smu_disable_dpms(). That is outdated cached values were used for "smu_is_dpm_running".
If that's true, a fix in smu_disable_dpms() may be more proper.

Evan
-----Original Message-----
From: amd-gfx <[email protected]> On Behalf Of Kenneth Feng
Sent: Tuesday, November 17, 2020 9:31 PM
To: [email protected]
Cc: Feng, Kenneth <[email protected]>
Subject: [PATCH 2/2] drm/amd/pm: fix the crash after runtime pm resume

Some features are still disabled after runtime pm resume. This can take the hardware back. Unlike other projects, this doesn't need pptable retransfer.

---
 drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +4 -1
index 39990790ed67..ebd50f144c5d 100644
@@ -914,11 +914,14 @@ static int smu_smc_hw_setup(struct smu_context *smu)  {  struct amdgpu_device *adev = smu->adev;  uint32_t pcie_gen = 0, pcie_width = 0; -int ret;
+int ret = 0;
 
 if (adev->in_suspend && smu_is_dpm_running(smu)) {  dev_info(adev->dev, "dpm has been enabled\n"); -return 0;
+/* this is needed specifically */
+if (adev->asic_type == CHIP_SIENNA_CICHLID) ret =
+smu_system_features_control(smu, true); return ret;
 }
 
 ret = smu_init_display_count(smu, 0);
--

2.17.1

_______________________________________________
amd-gfx mailing list
[email protected]
https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Famd-gfx&amp;data=04%7C01%7Cevan.quan%40amd.com%7Cdd553fd04f0f44ebf3ca08d88afd11ad%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637412166811942772%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=SgtA0lq6Sx5aiQ5PSLvf9GdXUU50cFLE5ELMY0e8k5g%3D&amp;reserved=0
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx