[PATCH 0/2] Print GMU version at boot on >A630

[PATCH 0/2] Print GMU version at boot on >A630

[Konrad Dybcio]

Might come in useful to try and track down otherwise inexplicable
failures in e.g. old CI logs. A630 uses a different mechanism that is
not implemented in this series.

Tested on 8280 and X1

Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
---
Konrad Dybcio (2):
      drm/msm: registers: Add GMU FW version register
      drm/msm/a6xx: Print GMU core firmware version at boot

 drivers/gpu/drm/msm/adreno/a6xx_gmu.c             | 7 +++++++
 drivers/gpu/drm/msm/registers/adreno/a6xx_gmu.xml | 5 +++++
 2 files changed, 12 insertions(+)
---
base-commit: 8503810115fbff903f626adc0788daa048302bc0
change-id: 20241219-topic-gmu_fw_ver-609e8b867802

Best regards,
-- 
Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

[PATCH 1/2] drm/msm: registers: Add GMU FW version register

[Konrad Dybcio]

From: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Add a register that contains the GMU core firmware version on non-
legacy (non-sdm845-family) SoCs.

The name is guesstimated based on what it does downstream, but it'll
do.

Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
---
 drivers/gpu/drm/msm/registers/adreno/a6xx_gmu.xml | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/gpu/drm/msm/registers/adreno/a6xx_gmu.xml b/drivers/gpu/drm/msm/registers/adreno/a6xx_gmu.xml
index 6531749d30f4e4e57ca4f7b43a28b7829504a9f3..3d2cc339b8f19c8d24b2c9144569b2364afc5ebc 100644
--- a/drivers/gpu/drm/msm/registers/adreno/a6xx_gmu.xml
+++ b/drivers/gpu/drm/msm/registers/adreno/a6xx_gmu.xml
@@ -52,6 +52,11 @@ xsi:schemaLocation="https://gitlab.freedesktop.org/freedreno/ rules-fd.xsd">
 	<reg32 offset="0x23fd" name="GMU_DCVS_PERF_SETTING"/>
 	<reg32 offset="0x23fe" name="GMU_DCVS_BW_SETTING"/>
 	<reg32 offset="0x23ff" name="GMU_DCVS_RETURN"/>
+	<reg32 offset="0x2bf8" name="GMU_CORE_FW_VERSION">
+		<bitfield name="MAJOR" low="28" high="31"/>
+		<bitfield name="MINOR" low="16" high="27"/>
+		<bitfield name="STEP" low="0" high="15"/>
+	</reg32>
 	<reg32 offset="0x4c00" name="GMU_ICACHE_CONFIG"/>
 	<reg32 offset="0x4c01" name="GMU_DCACHE_CONFIG"/>
 	<reg32 offset="0x4c0f" name="GMU_SYS_BUS_CONFIG"/>

-- 
2.47.1

[PATCH 2/2] drm/msm/a6xx: Print GMU core firmware version at boot

[Konrad Dybcio]

From: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Log the version for informational purposes, such as for keeping track
of possible GMU fw-related failures in crash / CI logs.

Intentionally not implemented on the if (gmu->legacy) codepath, as
these registers seem not to be used on there.

Downstream additionally warns if the firmware version is too old for
a given GPU, but we already pair the binary to a given GPU, so let's
not go there at the moment.

Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
---
 drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c
index 14db7376c712d19446b38152e480bd5a1e0a5198..a7ea2b2af1dc3816906236df929df36e37d8f606 100644
--- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c
+++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c
@@ -729,6 +729,7 @@ static int a6xx_gmu_fw_load(struct a6xx_gmu *gmu)
 	const struct firmware *fw_image = adreno_gpu->fw[ADRENO_FW_GMU];
 	const struct block_header *blk;
 	u32 reg_offset;
+	u32 ver;
 
 	u32 itcm_base = 0x00000000;
 	u32 dtcm_base = 0x00040000;
@@ -775,6 +776,12 @@ static int a6xx_gmu_fw_load(struct a6xx_gmu *gmu)
 		}
 	}
 
+	ver = gmu_read(gmu, REG_A6XX_GMU_CORE_FW_VERSION);
+	DRM_INFO("Loaded GMU firmware v%u.%u.%u\n",
+		 FIELD_GET(A6XX_GMU_CORE_FW_VERSION_MAJOR__MASK, ver),
+		 FIELD_GET(A6XX_GMU_CORE_FW_VERSION_MINOR__MASK, ver),
+		 FIELD_GET(A6XX_GMU_CORE_FW_VERSION_STEP__MASK, ver));
+
 	return 0;
 }
 

-- 
2.47.1

Re: [PATCH 0/2] Print GMU version at boot on >A630

[Akhil P Oommen]

On 12/20/2024 4:06 AM, Konrad Dybcio wrote:
> Might come in useful to try and track down otherwise inexplicable
> failures in e.g. old CI logs. A630 uses a different mechanism that is
> not implemented in this series.
> 
> Tested on 8280 and X1
> 
> Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

For the whole series:

Reviewed-by: Akhil P Oommen <quic_akhilpo@quicinc.com>

-Akhil

> ---
> Konrad Dybcio (2):
>       drm/msm: registers: Add GMU FW version register
>       drm/msm/a6xx: Print GMU core firmware version at boot
> 
>  drivers/gpu/drm/msm/adreno/a6xx_gmu.c             | 7 +++++++
>  drivers/gpu/drm/msm/registers/adreno/a6xx_gmu.xml | 5 +++++
>  2 files changed, 12 insertions(+)
> ---
> base-commit: 8503810115fbff903f626adc0788daa048302bc0
> change-id: 20241219-topic-gmu_fw_ver-609e8b867802
> 
> Best regards,

Re: [PATCH 2/2] drm/msm/a6xx: Print GMU core firmware version at boot

[Abel Vesa]

On 24-12-19 23:36:56, Konrad Dybcio wrote:
> From: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
> 
> Log the version for informational purposes, such as for keeping track
> of possible GMU fw-related failures in crash / CI logs.
> 
> Intentionally not implemented on the if (gmu->legacy) codepath, as
> these registers seem not to be used on there.
> 
> Downstream additionally warns if the firmware version is too old for
> a given GPU, but we already pair the binary to a given GPU, so let's
> not go there at the moment.
> 
> Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
> ---
>  drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c
> index 14db7376c712d19446b38152e480bd5a1e0a5198..a7ea2b2af1dc3816906236df929df36e37d8f606 100644
> --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c
> +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c
> @@ -729,6 +729,7 @@ static int a6xx_gmu_fw_load(struct a6xx_gmu *gmu)
>  	const struct firmware *fw_image = adreno_gpu->fw[ADRENO_FW_GMU];
>  	const struct block_header *blk;
>  	u32 reg_offset;
> +	u32 ver;
>  
>  	u32 itcm_base = 0x00000000;
>  	u32 dtcm_base = 0x00040000;
> @@ -775,6 +776,12 @@ static int a6xx_gmu_fw_load(struct a6xx_gmu *gmu)
>  		}
>  	}
>  
> +	ver = gmu_read(gmu, REG_A6XX_GMU_CORE_FW_VERSION);
> +	DRM_INFO("Loaded GMU firmware v%u.%u.%u\n",
> +		 FIELD_GET(A6XX_GMU_CORE_FW_VERSION_MAJOR__MASK, ver),
> +		 FIELD_GET(A6XX_GMU_CORE_FW_VERSION_MINOR__MASK, ver),
> +		 FIELD_GET(A6XX_GMU_CORE_FW_VERSION_STEP__MASK, ver));

I get the following spam:

[  109.928924] [drm] Loaded GMU firmware v3.1.10
[  110.301295] [drm] Loaded GMU firmware v3.1.10
[  110.472620] [drm] Loaded GMU firmware v3.1.10
[  111.164303] [drm] Loaded GMU firmware v3.1.10
[  111.864830] [drm] Loaded GMU firmware v3.1.10
...

Stacktrace shows this being called from runtime resume:

[   19.380390] Call trace:
[   19.380391]  show_stack+0x18/0x24 (C)
[   19.380399]  dump_stack_lvl+0x40/0x84
[   19.380403]  dump_stack+0x18/0x24
[   19.380405]  a6xx_gmu_resume+0x450/0xc44 [msm]
[   19.380426]  a6xx_gmu_pm_resume+0x34/0x220 [msm]
[   19.380437]  adreno_runtime_resume+0x28/0x34 [msm]
[   19.380446]  pm_generic_runtime_resume+0x28/0x3c
[   19.380451]  __rpm_callback+0x84/0x390
[   19.380453]  rpm_resume+0x3d0/0x5c0
[   19.380455]  __pm_runtime_resume+0x4c/0x94
[   19.380457]  adreno_get_param+0xdc/0x274 [msm]
[   19.380466]  msm_ioctl_get_param+0x5c/0x68 [msm]
[   19.380475]  drm_ioctl_kernel+0xd4/0x10c [drm]
[   19.380491]  drm_ioctl+0x26c/0x40c [drm]
[   19.380499]  __arm64_sys_ioctl+0x90/0xcc
[   19.380503]  invoke_syscall+0x40/0xf8
[   19.380505]  el0_svc_common+0xac/0xdc
[   19.380506]  do_el0_svc+0x1c/0x28
[   19.380508]  el0_svc+0x34/0x7c
[   19.380512]  el0t_64_sync_handler+0x84/0x108
[   19.380513]  el0t_64_sync+0x198/0x19c

So maybe DRM_INFO_ONCE instead ?

> +
>  	return 0;
>  }
>  
> 
> -- 
> 2.47.1
> 
> 

Re: [PATCH 2/2] drm/msm/a6xx: Print GMU core firmware version at boot

[Thomas Zimmermann]

Hi

Am 13.02.25 um 17:35 schrieb Abel Vesa:
> On 24-12-19 23:36:56, Konrad Dybcio wrote:
>> From: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
>>
>> Log the version for informational purposes, such as for keeping track
>> of possible GMU fw-related failures in crash / CI logs.
>>
>> Intentionally not implemented on the if (gmu->legacy) codepath, as
>> these registers seem not to be used on there.
>>
>> Downstream additionally warns if the firmware version is too old for
>> a given GPU, but we already pair the binary to a given GPU, so let's
>> not go there at the moment.
>>
>> Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
>> ---
>>   drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 7 +++++++
>>   1 file changed, 7 insertions(+)
>>
>> diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c
>> index 14db7376c712d19446b38152e480bd5a1e0a5198..a7ea2b2af1dc3816906236df929df36e37d8f606 100644
>> --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c
>> +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c
>> @@ -729,6 +729,7 @@ static int a6xx_gmu_fw_load(struct a6xx_gmu *gmu)
>>   	const struct firmware *fw_image = adreno_gpu->fw[ADRENO_FW_GMU];
>>   	const struct block_header *blk;
>>   	u32 reg_offset;
>> +	u32 ver;
>>   
>>   	u32 itcm_base = 0x00000000;
>>   	u32 dtcm_base = 0x00040000;
>> @@ -775,6 +776,12 @@ static int a6xx_gmu_fw_load(struct a6xx_gmu *gmu)
>>   		}
>>   	}
>>   
>> +	ver = gmu_read(gmu, REG_A6XX_GMU_CORE_FW_VERSION);
>> +	DRM_INFO("Loaded GMU firmware v%u.%u.%u\n",
>> +		 FIELD_GET(A6XX_GMU_CORE_FW_VERSION_MAJOR__MASK, ver),
>> +		 FIELD_GET(A6XX_GMU_CORE_FW_VERSION_MINOR__MASK, ver),
>> +		 FIELD_GET(A6XX_GMU_CORE_FW_VERSION_STEP__MASK, ver));
> I get the following spam:
>
> [  109.928924] [drm] Loaded GMU firmware v3.1.10
> [  110.301295] [drm] Loaded GMU firmware v3.1.10
> [  110.472620] [drm] Loaded GMU firmware v3.1.10
> [  111.164303] [drm] Loaded GMU firmware v3.1.10
> [  111.864830] [drm] Loaded GMU firmware v3.1.10
> ...
>
> Stacktrace shows this being called from runtime resume:
>
> [   19.380390] Call trace:
> [   19.380391]  show_stack+0x18/0x24 (C)
> [   19.380399]  dump_stack_lvl+0x40/0x84
> [   19.380403]  dump_stack+0x18/0x24
> [   19.380405]  a6xx_gmu_resume+0x450/0xc44 [msm]
> [   19.380426]  a6xx_gmu_pm_resume+0x34/0x220 [msm]
> [   19.380437]  adreno_runtime_resume+0x28/0x34 [msm]
> [   19.380446]  pm_generic_runtime_resume+0x28/0x3c
> [   19.380451]  __rpm_callback+0x84/0x390
> [   19.380453]  rpm_resume+0x3d0/0x5c0
> [   19.380455]  __pm_runtime_resume+0x4c/0x94
> [   19.380457]  adreno_get_param+0xdc/0x274 [msm]
> [   19.380466]  msm_ioctl_get_param+0x5c/0x68 [msm]
> [   19.380475]  drm_ioctl_kernel+0xd4/0x10c [drm]
> [   19.380491]  drm_ioctl+0x26c/0x40c [drm]
> [   19.380499]  __arm64_sys_ioctl+0x90/0xcc
> [   19.380503]  invoke_syscall+0x40/0xf8
> [   19.380505]  el0_svc_common+0xac/0xdc
> [   19.380506]  do_el0_svc+0x1c/0x28
> [   19.380508]  el0_svc+0x34/0x7c
> [   19.380512]  el0t_64_sync_handler+0x84/0x108
> [   19.380513]  el0t_64_sync+0x198/0x19c
>
> So maybe DRM_INFO_ONCE instead ?

drm_dbg() seems appropriate. Most of the time, firmware versions are not 
interesting.

Best regards
Thomas

>
>> +
>>   	return 0;
>>   }
>>   
>>
>> -- 
>> 2.47.1
>>
>>

-- 
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstrasse 146, 90461 Nuernberg, Germany
GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman
HRB 36809 (AG Nuernberg)

Re: [PATCH 2/2] drm/msm/a6xx: Print GMU core firmware version at boot

[Konrad Dybcio]

On 13.02.2025 5:41 PM, Thomas Zimmermann wrote:
> Hi
> 
> Am 13.02.25 um 17:35 schrieb Abel Vesa:
>> On 24-12-19 23:36:56, Konrad Dybcio wrote:
>>> From: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
>>>
>>> Log the version for informational purposes, such as for keeping track
>>> of possible GMU fw-related failures in crash / CI logs.
>>>
>>> Intentionally not implemented on the if (gmu->legacy) codepath, as
>>> these registers seem not to be used on there.
>>>
>>> Downstream additionally warns if the firmware version is too old for
>>> a given GPU, but we already pair the binary to a given GPU, so let's
>>> not go there at the moment.
>>>
>>> Signed-off-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
>>> ---

[...]

>> So maybe DRM_INFO_ONCE instead ?
> 
> drm_dbg() seems appropriate. Most of the time, firmware versions are not interesting.

Unfortunately, in our case they are, given users source them from all
kinds of places..

Konrad