Skip to contentSkip to conversation

drm/amd/powerplay: do proper cleanups on hw_fini

PATCH In review Quan, Evan 4 participants 6 messages Sep 3, 2019 – Nov 11, 2019 3 messages not in the archive

Subsystems (2)
lore
Share & tools
lore

Quan, Evan

author
#0 lore raw

PATCH 1/1 drm/amd/powerplay: do proper cleanups on hw_fini

These are needed for smu_reset support.

Change-Id If29ede4b99758adb08fd4e16665f44fd893ec99b
---
 drivers/gpu/drm/amd/powerplay/amdgpu_smu.c     | 17 +++++++++++++++++
 drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h |  3 +++
 drivers/gpu/drm/amd/powerplay/smu_v11_0.c      | 10 ++++++++++
 3 files changed, 30 insertions(+)
drivers/gpu/drm/amd/powerplay/amdgpu_smu.c +17 -0
index d5ee13a78eb7..3cf8d944f890 100644
@@ -1286,6 +1286,11 @@ static int smu_hw_init(void *handle)
 	return ret;
 }
 
+static int smu_stop_dpms(struct smu_context *smu)
+{
+	return smu_send_smc_msg(smu, SMU_MSG_DisableAllSmuFeatures);
+}
+
 static int smu_hw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -1298,6 +1303,18 @@ static int smu_hw_fini(void *handle)
 		smu_powergate_vcn(&adev->smu, true);
 	}
 
+	ret = smu_stop_thermal_control(smu);
+	if (ret) {
+		pr_warn("Fail to stop thermal control!\n");
+		return ret;
+	}
+
+	ret = smu_stop_dpms(smu);
+	if (ret) {
+		pr_warn("Fail to stop Dpms!\n");
+		return ret;
+	}
+
 	kfree(table_context->driver_pptable);
 	table_context->driver_pptable = NULL;
 
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +3 -0
index b19224cb6d6d..8e4b0ad24712 100644
@@ -498,6 +498,7 @@ struct smu_funcs
 	int (*get_current_clk_freq)(struct smu_context *smu, enum smu_clk_type clk_id, uint32_t *value);
 	int (*init_max_sustainable_clocks)(struct smu_context *smu);
 	int (*start_thermal_control)(struct smu_context *smu);
+	int (*stop_thermal_control)(struct smu_context *smu);
 	int (*read_sensor)(struct smu_context *smu, enum amd_pp_sensors sensor,
 			   void *data, uint32_t *size);
 	int (*set_deep_sleep_dcefclk)(struct smu_context *smu, uint32_t clk);
@@ -647,6 +648,8 @@ struct smu_funcs
 	((smu)->ppt_funcs->set_thermal_fan_table ? (smu)->ppt_funcs->set_thermal_fan_table((smu)) : 0)
 #define smu_start_thermal_control(smu) \
 	((smu)->funcs->start_thermal_control? (smu)->funcs->start_thermal_control((smu)) : 0)
+#define smu_stop_thermal_control(smu) \
+	((smu)->funcs->stop_thermal_control? (smu)->funcs->stop_thermal_control((smu)) : 0)
 #define smu_read_sensor(smu, sensor, data, size) \
 	((smu)->ppt_funcs->read_sensor? (smu)->ppt_funcs->read_sensor((smu), (sensor), (data), (size)) : 0)
 #define smu_smc_read_sensor(smu, sensor, data, size) \
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +10 -0
index db5e94ce54af..1a38af84394e 100644
@@ -1209,6 +1209,15 @@ static int smu_v11_0_start_thermal_control(struct smu_context *smu)
 	return ret;
 }
 
+static int smu_v11_0_stop_thermal_control(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+
+	WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_ENA, 0);
+
+	return 0;
+}
+
 static uint16_t convert_to_vddc(uint8_t vid)
 {
 	return (uint16_t) ((6200 - (vid * 25)) / SMU11_VOLTAGE_SCALE);
@@ -1783,6 +1792,7 @@ static const struct smu_funcs smu_v11_0_funcs = {
 	.get_current_clk_freq = smu_v11_0_get_current_clk_freq,
 	.init_max_sustainable_clocks = smu_v11_0_init_max_sustainable_clocks,
 	.start_thermal_control = smu_v11_0_start_thermal_control,
+	.stop_thermal_control = smu_v11_0_stop_thermal_control,
 	.read_sensor = smu_v11_0_read_sensor,
 	.set_deep_sleep_dcefclk = smu_v11_0_set_deep_sleep_dcefclk,
 	.display_clock_voltage_request = smu_v11_0_display_clock_voltage_request,
Signature · 6 lines
2.23.0

_______________________________________________
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): [email protected]

Reviewed-by Jack Gui
#2 lore raw

-----Original Message-----
From: Quan, Evan <[email protected]>
Sent: Tuesday, September 3, 2019 11:44 AM
To: [email protected]
Cc: Li, Candice <[email protected]>; Gui, Jack <[email protected]>; Quan, Evan <[email protected]>
Subject: [PATCH] drm/amd/powerplay: do proper cleanups on hw_fini

These are needed for smu_reset support.

Change-Id If29ede4b99758adb08fd4e16665f44fd893ec99b
---
 drivers/gpu/drm/amd/powerplay/amdgpu_smu.c     | 17 +++++++++++++++++
 drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h |  3 +++
 drivers/gpu/drm/amd/powerplay/smu_v11_0.c      | 10 ++++++++++
 3 files changed, 30 insertions(+)
drivers/gpu/drm/amd/powerplay/amdgpu_smu.c +3 -0
index d5ee13a78eb7..3cf8d944f890 100644
@@ -1286,6 +1286,11 @@ static int smu_hw_init(void *handle)
 	return ret;
 }
 
+static int smu_stop_dpms(struct smu_context *smu) {
+	return smu_send_smc_msg(smu, SMU_MSG_DisableAllSmuFeatures); }
+
 static int smu_hw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -1298,6 +1303,18 @@ static int smu_hw_fini(void *handle)

smu_powergate_vcn(&adev->smu, true);
}

+ ret = smu_stop_thermal_control(smu);
+ if (ret) {
+ pr_warn("Fail to stop thermal control!\n");
+ return ret;
+ }
+
+ ret = smu_stop_dpms(smu);
+ if (ret) {
+ pr_warn("Fail to stop Dpms!\n");
+ return ret;
+ }
+
kfree(table_context->driver_pptable);
table_context->driver_pptable = NULL;

drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +1 -0
index b19224cb6d6d..8e4b0ad24712 100644
@@ -498,6 +498,7 @@ struct smu_funcs
 	int (*get_current_clk_freq)(struct smu_context *smu, enum smu_clk_type clk_id, uint32_t *value);
 	int (*init_max_sustainable_clocks)(struct smu_context *smu);
 	int (*start_thermal_control)(struct smu_context *smu);
+	int (*stop_thermal_control)(struct smu_context *smu);
 	int (*read_sensor)(struct smu_context *smu, enum amd_pp_sensors sensor,
 			   void *data, uint32_t *size);
 	int (*set_deep_sleep_dcefclk)(struct smu_context *smu, uint32_t clk); @@ -647,6 +648,8 @@ struct smu_funcs

((smu)->ppt_funcs->set_thermal_fan_table ? (smu)->ppt_funcs->set_thermal_fan_table((smu)) : 0) #define smu_start_thermal_control(smu) \
((smu)->funcs->start_thermal_control? (smu)->funcs->start_thermal_control((smu)) : 0)
+#define smu_stop_thermal_control(smu) \
+ ((smu)->funcs->stop_thermal_control?
+(smu)->funcs->stop_thermal_control((smu)) : 0)
#define smu_read_sensor(smu, sensor, data, size) \
((smu)->ppt_funcs->read_sensor? (smu)->ppt_funcs->read_sensor((smu), (sensor), (data), (size)) : 0) #define smu_smc_read_sensor(smu, sensor, data, size) \ diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index db5e94ce54af..1a38af84394e 100644

drivers/gpu/drm/amd/powerplay/smu_v11_0.c +8 -0
@@ -1209,6 +1209,15 @@ static int smu_v11_0_start_thermal_control(struct smu_context *smu)
 	return ret;
 }
 
+static int smu_v11_0_stop_thermal_control(struct smu_context *smu) {
+	struct amdgpu_device *adev = smu->adev;
+
+	WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_ENA, 0);
+
+	return 0;
+}
+
 static uint16_t convert_to_vddc(uint8_t vid)  {
 	return (uint16_t) ((6200 - (vid * 25)) / SMU11_VOLTAGE_SCALE); @@ -1783,6 +1792,7 @@ static const struct smu_funcs smu_v11_0_funcs = {
 	.get_current_clk_freq = smu_v11_0_get_current_clk_freq,

.init_max_sustainable_clocks = smu_v11_0_init_max_sustainable_clocks, .start_thermal_control = smu_v11_0_start_thermal_control,
+ .stop_thermal_control = smu_v11_0_stop_thermal_control,
.read_sensor = smu_v11_0_read_sensor, .set_deep_sleep_dcefclk = smu_v11_0_set_deep_sleep_dcefclk, .display_clock_voltage_request = smu_v11_0_display_clock_voltage_request,
--
2.23.0

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

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

We should probably also add the callbacks for set_mp1_state like we have for the older powerplay code so we properly send the SMU shutdown or reset message on GPU reset and driver unload.

Alex

________________________________
From: amd-gfx <amd-gfx-bounces-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org> on behalf of Quan, Evan <[email protected]>
Sent: Monday, September 2, 2019 11:44 PM
To: [email protected] <[email protected]>
Cc: Li, Candice <[email protected]>; Gui, Jack <[email protected]>; Quan, Evan <[email protected]>
Subject: [PATCH] drm/amd/powerplay: do proper cleanups on hw_fini

These are needed for smu_reset support.

Change-Id If29ede4b99758adb08fd4e16665f44fd893ec99b
---
 drivers/gpu/drm/amd/powerplay/amdgpu_smu.c     | 17 +++++++++++++++++
 drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h |  3 +++
 drivers/gpu/drm/amd/powerplay/smu_v11_0.c      | 10 ++++++++++
 3 files changed, 30 insertions(+)
drivers/gpu/drm/amd/powerplay/amdgpu_smu.c +17 -0
index d5ee13a78eb7..3cf8d944f890 100644
@@ -1286,6 +1286,11 @@ static int smu_hw_init(void *handle)
         return ret;
 }
 
+static int smu_stop_dpms(struct smu_context *smu)
+{
+       return smu_send_smc_msg(smu, SMU_MSG_DisableAllSmuFeatures);
+}
+
 static int smu_hw_fini(void *handle)
 {
         struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -1298,6 +1303,18 @@ static int smu_hw_fini(void *handle)
                 smu_powergate_vcn(&adev->smu, true);
         }
 
+       ret = smu_stop_thermal_control(smu);
+       if (ret) {
+               pr_warn("Fail to stop thermal control!\n");
+               return ret;
+       }
+
+       ret = smu_stop_dpms(smu);
+       if (ret) {
+               pr_warn("Fail to stop Dpms!\n");
+               return ret;
+       }
+
         kfree(table_context->driver_pptable);
         table_context->driver_pptable = NULL;
 
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +3 -0
index b19224cb6d6d..8e4b0ad24712 100644
@@ -498,6 +498,7 @@ struct smu_funcs
         int (*get_current_clk_freq)(struct smu_context *smu, enum smu_clk_type clk_id, uint32_t *value);
         int (*init_max_sustainable_clocks)(struct smu_context *smu);
         int (*start_thermal_control)(struct smu_context *smu);
+       int (*stop_thermal_control)(struct smu_context *smu);
         int (*read_sensor)(struct smu_context *smu, enum amd_pp_sensors sensor,
                            void *data, uint32_t *size);
         int (*set_deep_sleep_dcefclk)(struct smu_context *smu, uint32_t clk);
@@ -647,6 +648,8 @@ struct smu_funcs
         ((smu)->ppt_funcs->set_thermal_fan_table ? (smu)->ppt_funcs->set_thermal_fan_table((smu)) : 0)
 #define smu_start_thermal_control(smu) \
         ((smu)->funcs->start_thermal_control? (smu)->funcs->start_thermal_control((smu)) : 0)
+#define smu_stop_thermal_control(smu) \
+       ((smu)->funcs->stop_thermal_control? (smu)->funcs->stop_thermal_control((smu)) : 0)
 #define smu_read_sensor(smu, sensor, data, size) \
         ((smu)->ppt_funcs->read_sensor? (smu)->ppt_funcs->read_sensor((smu), (sensor), (data), (size)) : 0)
 #define smu_smc_read_sensor(smu, sensor, data, size) \
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +10 -0
index db5e94ce54af..1a38af84394e 100644
@@ -1209,6 +1209,15 @@ static int smu_v11_0_start_thermal_control(struct smu_context *smu)
         return ret;
 }
 
+static int smu_v11_0_stop_thermal_control(struct smu_context *smu)
+{
+       struct amdgpu_device *adev = smu->adev;
+
+       WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_ENA, 0);
+
+       return 0;
+}
+
 static uint16_t convert_to_vddc(uint8_t vid)
 {
         return (uint16_t) ((6200 - (vid * 25)) / SMU11_VOLTAGE_SCALE);
@@ -1783,6 +1792,7 @@ static const struct smu_funcs smu_v11_0_funcs = {
         .get_current_clk_freq = smu_v11_0_get_current_clk_freq,
         .init_max_sustainable_clocks = smu_v11_0_init_max_sustainable_clocks,
         .start_thermal_control = smu_v11_0_start_thermal_control,
+       .stop_thermal_control = smu_v11_0_stop_thermal_control,
         .read_sensor = smu_v11_0_read_sensor,
         .set_deep_sleep_dcefclk = smu_v11_0_set_deep_sleep_dcefclk,
         .display_clock_voltage_request = smu_v11_0_display_clock_voltage_request,

--
2.23.0

_______________________________________________
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

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

Hey guys,

This patch caused some kind of reversion with smu_reset on Navi10. I'm no expert since everything I know comes from just reading through the code, so this could be some kind of intended behavior, but after this patch, if you write a pptable to the sysfs pp_table interface on navi10, then the SMU will fail to reset successfully, and the result is seemingly an unrecoverable situation.

I put in a report on bugzilla with dmesg logs
:
https://bugs.freedesktop.org/show_bug.cgi?id=112234

Finding this change was the result of a bisect to find where the issue started, and reverting the changes to smu_hw_fini resolved the issue. Any advice on possible proper fixes?

Thanks in advance,

Matt

Quoted 88 lines · Evan

These are needed for smu_reset support.

Change-Id If29ede4b99758adb08fd4e16665f44fd893ec99b
---
 drivers/gpu/drm/amd/powerplay/amdgpu_smu.c     | 17 +++++++++++++++++
 drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h |  3 +++
 drivers/gpu/drm/amd/powerplay/smu_v11_0.c      | 10 ++++++++++
 3 files changed, 30 insertions(+)
drivers/gpu/drm/amd/powerplay/amdgpu_smu.c +17 -0
index d5ee13a78eb7..3cf8d944f890 100644
@@ -1286,6 +1286,11 @@ static int smu_hw_init(void *handle)
 	return ret;
 }
 
+static int smu_stop_dpms(struct smu_context *smu)
+{
+	return smu_send_smc_msg(smu, SMU_MSG_DisableAllSmuFeatures);
+}
+
 static int smu_hw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@@ -1298,6 +1303,18 @@ static int smu_hw_fini(void *handle)
 		smu_powergate_vcn(&adev->smu, true);
 	}
 
+	ret = smu_stop_thermal_control(smu);
+	if (ret) {
+		pr_warn("Fail to stop thermal control!\n");
+		return ret;
+	}
+
+	ret = smu_stop_dpms(smu);
+	if (ret) {
+		pr_warn("Fail to stop Dpms!\n");
+		return ret;
+	}
+
 	kfree(table_context->driver_pptable);
 	table_context->driver_pptable = NULL;
 
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +3 -0
index b19224cb6d6d..8e4b0ad24712 100644
@@ -498,6 +498,7 @@ struct smu_funcs
 	int (*get_current_clk_freq)(struct smu_context *smu, enum smu_clk_type clk_id, uint32_t *value);
 	int (*init_max_sustainable_clocks)(struct smu_context *smu);
 	int (*start_thermal_control)(struct smu_context *smu);
+	int (*stop_thermal_control)(struct smu_context *smu);
 	int (*read_sensor)(struct smu_context *smu, enum amd_pp_sensors sensor,
 			   void *data, uint32_t *size);
 	int (*set_deep_sleep_dcefclk)(struct smu_context *smu, uint32_t clk);
@@ -647,6 +648,8 @@ struct smu_funcs
 	((smu)->ppt_funcs->set_thermal_fan_table ? (smu)->ppt_funcs->set_thermal_fan_table((smu)) : 0)
 #define smu_start_thermal_control(smu) \
 	((smu)->funcs->start_thermal_control? (smu)->funcs->start_thermal_control((smu)) : 0)
+#define smu_stop_thermal_control(smu) \
+	((smu)->funcs->stop_thermal_control? (smu)->funcs->stop_thermal_control((smu)) : 0)
 #define smu_read_sensor(smu, sensor, data, size) \
 	((smu)->ppt_funcs->read_sensor? (smu)->ppt_funcs->read_sensor((smu), (sensor), (data), (size)) : 0)
 #define smu_smc_read_sensor(smu, sensor, data, size) \
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +9 -0
index db5e94ce54af..1a38af84394e 100644
@@ -1209,6 +1209,15 @@ static int smu_v11_0_start_thermal_control(struct smu_context *smu)
 	return ret;
 }
 
+static int smu_v11_0_stop_thermal_control(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+
+	WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_ENA, 0);
+
+	return 0;
+}
+
 static uint16_t convert_to_vddc(uint8_t vid)
 {
 	return (uint16_t) ((6200 - (vid * 25)) / SMU11_VOLTAGE_SCALE);

drivers/gpu/drm/amd/powerplay/smu_v11_0.c static const struct smu_funcs smu_v11_0_funcs = { · line 1798

Quoted 7 lines
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +1 -0
@@ -1783,6 +1792,7 @@ static const struct smu_funcs smu_v11_0_funcs = {
 	.get_current_clk_freq = smu_v11_0_get_current_clk_freq,
 	.init_max_sustainable_clocks = smu_v11_0_init_max_sustainable_clocks,
 	.start_thermal_control = smu_v11_0_start_thermal_control,
+	.stop_thermal_control = smu_v11_0_stop_thermal_control,
 	.read_sensor = smu_v11_0_read_sensor,
 	.set_deep_sleep_dcefclk = smu_v11_0_set_deep_sleep_dcefclk,
 	.display_clock_voltage_request = smu_v11_0_display_clock_voltage_request,

_______________________________________________
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): 50104e7b-e38f-6750-7aae-dfef399790f3-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

author
#6 lore raw

Just sent out a patch which should be able to address this issue.
https://lists.freedesktop.org/archives/amd-gfx/2019-November/042458.html

Regards,
Evan

Quoted 132 lines

-----Original Message-----
From: Matt Coffin <[email protected]>
Sent: Saturday, November 9, 2019 4:50 AM
To: Quan, Evan <[email protected]>; [email protected]
Cc: Li, Candice <[email protected]>; Gui, Jack <[email protected]>; Alex Deucher <[email protected]>
Subject: Re: [PATCH] drm/amd/powerplay: do proper cleanups on hw_fini

Hey guys,

This patch caused some kind of reversion with smu_reset on Navi10. I'm no expert since everything I know comes from just reading through the code, so this could be some kind of intended behavior, but after this patch, if you write a pptable to the sysfs pp_table interface on navi10, then the SMU will fail to reset successfully, and the result is seemingly an unrecoverable situation.

I put in a report on bugzilla with dmesg logs
:
https://bugs.freedesktop.org/show_bug.cgi?id=112234

Finding this change was the result of a bisect to find where the issue started, and reverting the changes to smu_hw_fini resolved the issue.
Any advice on possible proper fixes?

Thanks in advance,

Matt

Quoted 6 lines · Evan

These are needed for smu_reset support.

Change-Id If29ede4b99758adb08fd4e16665f44fd893ec99b
---
 drivers/gpu/drm/amd/powerplay/amdgpu_smu.c     | 17

+++++++++++++++++

Quoted 44 lines
 drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h |  3 +++
 drivers/gpu/drm/amd/powerplay/smu_v11_0.c      | 10 ++++++++++
 3 files changed, 30 insertions(+)
drivers/gpu/drm/amd/powerplay/amdgpu_smu.c +15 -1
index d5ee13a78eb7..3cf8d944f890 100644
@@ -1286,6 +1286,11 @@ static int smu_hw_init(void *handle)
 	return ret;
 }
 
+static int smu_stop_dpms(struct smu_context *smu) {
+	return smu_send_smc_msg(smu, SMU_MSG_DisableAllSmuFeatures); }
+
 static int smu_hw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@
-1298,6 +1303,18 @@ static int smu_hw_fini(void *handle)
 		smu_powergate_vcn(&adev->smu, true);
 	}
 
+	ret = smu_stop_thermal_control(smu);
+	if (ret) {
+		pr_warn("Fail to stop thermal control!\n");
+		return ret;
+	}
+
+	ret = smu_stop_dpms(smu);
+	if (ret) {
+		pr_warn("Fail to stop Dpms!\n");
+		return ret;
+	}
+
 	kfree(table_context->driver_pptable);
 	table_context->driver_pptable = NULL;
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +0 -0
index b19224cb6d6d..8e4b0ad24712 100644

drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h struct smu_funcs · line 498

Quoted 1 line
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +0 -0
@@ -498,6 +498,7 @@ struct smu_funcs
 	int (*get_current_clk_freq)(struct smu_context *smu, enum

smu_clk_type clk_id, uint32_t *value);

drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h struct smu_funcs

Quoted 4 lines
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +1 -0
 	int (*init_max_sustainable_clocks)(struct smu_context *smu);
 	int (*start_thermal_control)(struct smu_context *smu);
+	int (*stop_thermal_control)(struct smu_context *smu);
 	int (*read_sensor)(struct smu_context *smu, enum amd_pp_sensors

sensor,

Quoted 5 lines
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +0 -0
 			   void *data, uint32_t *size);
 	int (*set_deep_sleep_dcefclk)(struct smu_context *smu, uint32_t

clk); @@ -647,6 +648,8 @@ struct smu_funcs
((smu)->ppt_funcs->set_thermal_fan_table ?
(smu)->ppt_funcs->set_thermal_fan_table((smu)) : 0) #define

smu_start_thermal_control(smu) \

drivers/gpu/drm/amd/powerplay/smu_v11_0.c static int

Quoted 16 lines

((smu)->funcs->start_thermal_control?
(smu)->funcs->start_thermal_control((smu)) : 0)

drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +3 -0
+#define smu_stop_thermal_control(smu) \
+	((smu)->funcs->stop_thermal_control?
+(smu)->funcs->stop_thermal_control((smu)) : 0)
 #define smu_read_sensor(smu, sensor, data, size) \
 	((smu)->ppt_funcs->read_sensor? (smu)->ppt_funcs-

read_sensor((smu),
(sensor), (data), (size)) : 0) #define smu_smc_read_sensor(smu, sensor, data, size) \ diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index db5e94ce54af..1a38af84394e 100644

drivers/gpu/drm/amd/powerplay/smu_v11_0.c +0 -0
@@ -1209,6 +1209,15 @@ static int

smu_v11_0_start_thermal_control(struct smu_context *smu)

Quoted 6 lines
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +3 -0
 	return ret;
 }
 
+static int smu_v11_0_stop_thermal_control(struct smu_context *smu) {
+	struct amdgpu_device *adev = smu->adev;
+

drivers/gpu/drm/amd/powerplay/smu_v11_0.c static int smu_v11_0_start_thermal_control(struct smu_context *smu)

Quoted 10 lines
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +5 -1
+	WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_ENA, 0);
+
+	return 0;
+}
+
 static uint16_t convert_to_vddc(uint8_t vid)  {
 	return (uint16_t) ((6200 - (vid * 25)) / SMU11_VOLTAGE_SCALE); @@
-1783,6 +1792,7 @@ static const struct smu_funcs smu_v11_0_funcs = {
 	.get_current_clk_freq = smu_v11_0_get_current_clk_freq,
 	.init_max_sustainable_clocks =

smu_v11_0_init_max_sustainable_clocks,

Quoted 7 lines
Quoted 7 lines
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +1 -0
 	.start_thermal_control = smu_v11_0_start_thermal_control,
+	.stop_thermal_control = smu_v11_0_stop_thermal_control,
 	.read_sensor = smu_v11_0_read_sensor,
 	.set_deep_sleep_dcefclk = smu_v11_0_set_deep_sleep_dcefclk,
 	.display_clock_voltage_request =

smu_v11_0_display_clock_voltage_request,

_______________________________________________
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): MN2PR12MB3344AE3F4B75313B425FAB9EE4740-rweVpJHSKToDMgCC8P//[email protected]

#8 lore raw

Thanks Evan. I can confirm that the linked patch resolves the issue for me.

I commented and resolved the bug as well in case other people find it.

Cheers,
Matt

Quoted 137 lines · Evan

Just sent out a patch which should be able to address this issue.
https://lists.freedesktop.org/archives/amd-gfx/2019-November/042458.html

Regards,
Evan

Quoted 132 lines

-----Original Message-----
From: Matt Coffin <[email protected]>
Sent: Saturday, November 9, 2019 4:50 AM
To: Quan, Evan <[email protected]>; [email protected]
Cc: Li, Candice <[email protected]>; Gui, Jack <[email protected]>; Alex Deucher <[email protected]>
Subject: Re: [PATCH] drm/amd/powerplay: do proper cleanups on hw_fini

Hey guys,

This patch caused some kind of reversion with smu_reset on Navi10. I'm no expert since everything I know comes from just reading through the code, so this could be some kind of intended behavior, but after this patch, if you write a pptable to the sysfs pp_table interface on navi10, then the SMU will fail to reset successfully, and the result is seemingly an unrecoverable situation.

I put in a report on bugzilla with dmesg logs
:
https://bugs.freedesktop.org/show_bug.cgi?id=112234

Finding this change was the result of a bisect to find where the issue started, and reverting the changes to smu_hw_fini resolved the issue.
Any advice on possible proper fixes?

Thanks in advance,

Matt

Quoted 6 lines · Evan

These are needed for smu_reset support.

Change-Id If29ede4b99758adb08fd4e16665f44fd893ec99b
---
 drivers/gpu/drm/amd/powerplay/amdgpu_smu.c     | 17

+++++++++++++++++

Quoted 44 lines
 drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h |  3 +++
 drivers/gpu/drm/amd/powerplay/smu_v11_0.c      | 10 ++++++++++
 3 files changed, 30 insertions(+)
drivers/gpu/drm/amd/powerplay/amdgpu_smu.c +15 -1
index d5ee13a78eb7..3cf8d944f890 100644
@@ -1286,6 +1286,11 @@ static int smu_hw_init(void *handle)
 	return ret;
 }
 
+static int smu_stop_dpms(struct smu_context *smu) {
+	return smu_send_smc_msg(smu, SMU_MSG_DisableAllSmuFeatures); }
+
 static int smu_hw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@
-1298,6 +1303,18 @@ static int smu_hw_fini(void *handle)
 		smu_powergate_vcn(&adev->smu, true);
 	}
 
+	ret = smu_stop_thermal_control(smu);
+	if (ret) {
+		pr_warn("Fail to stop thermal control!\n");
+		return ret;
+	}
+
+	ret = smu_stop_dpms(smu);
+	if (ret) {
+		pr_warn("Fail to stop Dpms!\n");
+		return ret;
+	}
+
 	kfree(table_context->driver_pptable);
 	table_context->driver_pptable = NULL;
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +0 -0
index b19224cb6d6d..8e4b0ad24712 100644

drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h struct smu_funcs · line 498

Quoted 1 line
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +0 -0
@@ -498,6 +498,7 @@ struct smu_funcs
 	int (*get_current_clk_freq)(struct smu_context *smu, enum

smu_clk_type clk_id, uint32_t *value);

drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h struct smu_funcs

Quoted 4 lines
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +1 -0
 	int (*init_max_sustainable_clocks)(struct smu_context *smu);
 	int (*start_thermal_control)(struct smu_context *smu);
+	int (*stop_thermal_control)(struct smu_context *smu);
 	int (*read_sensor)(struct smu_context *smu, enum amd_pp_sensors

sensor,

Quoted 5 lines
drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +0 -0
 			   void *data, uint32_t *size);
 	int (*set_deep_sleep_dcefclk)(struct smu_context *smu, uint32_t

clk); @@ -647,6 +648,8 @@ struct smu_funcs
((smu)->ppt_funcs->set_thermal_fan_table ?
(smu)->ppt_funcs->set_thermal_fan_table((smu)) : 0) #define

smu_start_thermal_control(smu) \

drivers/gpu/drm/amd/powerplay/smu_v11_0.c static int

Quoted 16 lines

((smu)->funcs->start_thermal_control?
(smu)->funcs->start_thermal_control((smu)) : 0)

drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h +3 -0
+#define smu_stop_thermal_control(smu) \
+	((smu)->funcs->stop_thermal_control?
+(smu)->funcs->stop_thermal_control((smu)) : 0)
 #define smu_read_sensor(smu, sensor, data, size) \
 	((smu)->ppt_funcs->read_sensor? (smu)->ppt_funcs-

read_sensor((smu),
(sensor), (data), (size)) : 0) #define smu_smc_read_sensor(smu, sensor, data, size) \ diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index db5e94ce54af..1a38af84394e 100644

drivers/gpu/drm/amd/powerplay/smu_v11_0.c +0 -0
@@ -1209,6 +1209,15 @@ static int

smu_v11_0_start_thermal_control(struct smu_context *smu)

Quoted 6 lines
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +3 -0
 	return ret;
 }
 
+static int smu_v11_0_stop_thermal_control(struct smu_context *smu) {
+	struct amdgpu_device *adev = smu->adev;
+

drivers/gpu/drm/amd/powerplay/smu_v11_0.c static int smu_v11_0_start_thermal_control(struct smu_context *smu)

Quoted 10 lines
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +5 -1
+	WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_ENA, 0);
+
+	return 0;
+}
+
 static uint16_t convert_to_vddc(uint8_t vid)  {
 	return (uint16_t) ((6200 - (vid * 25)) / SMU11_VOLTAGE_SCALE); @@
-1783,6 +1792,7 @@ static const struct smu_funcs smu_v11_0_funcs = {
 	.get_current_clk_freq = smu_v11_0_get_current_clk_freq,
 	.init_max_sustainable_clocks =

smu_v11_0_init_max_sustainable_clocks,

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
drivers/gpu/drm/amd/powerplay/smu_v11_0.c +1 -0
 	.start_thermal_control = smu_v11_0_start_thermal_control,
+	.stop_thermal_control = smu_v11_0_stop_thermal_control,
 	.read_sensor = smu_v11_0_read_sensor,
 	.set_deep_sleep_dcefclk = smu_v11_0_set_deep_sleep_dcefclk,
 	.display_clock_voltage_request =

smu_v11_0_display_clock_voltage_request,

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