linux-kernel - Re: [PATCH 3/5] drm: msm: add support for A750 GPU

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <bcad544c-7ca2-4b4f-805b-4ccaedbd091c@linaro.org>
Date: Mon, 12 Feb 2024 15:45:56 +0100
From: Neil Armstrong <neil.armstrong@...aro.org>
To: Konrad Dybcio <konrad.dybcio@...aro.org>, Rob Clark
 <robdclark@...il.com>, Abhinav Kumar <quic_abhinavk@...cinc.com>,
 Dmitry Baryshkov <dmitry.baryshkov@...aro.org>, Sean Paul <sean@...rly.run>,
 Marijn Suijten <marijn.suijten@...ainline.org>,
 David Airlie <airlied@...il.com>, Daniel Vetter <daniel@...ll.ch>,
 Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
 Maxime Ripard <mripard@...nel.org>, Thomas Zimmermann <tzimmermann@...e.de>,
 Rob Herring <robh@...nel.org>,
 Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
 Conor Dooley <conor+dt@...nel.org>, Will Deacon <will@...nel.org>,
 Robin Murphy <robin.murphy@....com>, Joerg Roedel <joro@...tes.org>,
 Bjorn Andersson <andersson@...nel.org>
Cc: linux-arm-msm@...r.kernel.org, dri-devel@...ts.freedesktop.org,
 freedreno@...ts.freedesktop.org, devicetree@...r.kernel.org,
 linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
 iommu@...ts.linux.dev
Subject: Re: [PATCH 3/5] drm: msm: add support for A750 GPU

On 12/02/2024 11:46, Konrad Dybcio wrote:
> On 12.02.2024 11:37, Neil Armstrong wrote:
>> Add support for the A750 GPU found on the SM8650 platform
>>
>> Unlike the the very close A740 GPU on the SM8550 SoC, the A750 GPU
>> doesn't have an HWCFG block but a separate register set.
>>
>> The missing registers are added in the a6xx.xml.h file that would
>> require a subsequent sync and the non-existent hwcfg is handled
>> in a6xx_set_hwcg().
> 
> These should also be submitted to mesa to make sure the next header sync
> doesn't wipe them

Ack submitting them right now: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27576

> 
> [...]
> 
>> --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
>> +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu.c
>> @@ -958,10 +958,11 @@ static void a6xx_set_hwcg(struct msm_gpu *gpu, bool state)
>>   	struct a6xx_gpu *a6xx_gpu = to_a6xx_gpu(adreno_gpu);
>>   	struct a6xx_gmu *gmu = &a6xx_gpu->gmu;
>>   	const struct adreno_reglist *reg;
>> +	bool skip_programming = !(adreno_gpu->info->hwcg || adreno_is_a7xx(adreno_gpu));
> 
> is_a750?

OK right, I was thinking of the next gpu which will probably also miss an hwcfg

> 
>>   	unsigned int i;
>>   	u32 val, clock_cntl_on, cgc_mode;
>>   
>> -	if (!adreno_gpu->info->hwcg)
>> +	if (skip_programming)
>>   		return;
>>   
>>   	if (adreno_is_a630(adreno_gpu))
>> @@ -982,6 +983,25 @@ static void a6xx_set_hwcg(struct msm_gpu *gpu, bool state)
>>   			  state ? 0x5555 : 0);
>>   	}
>>   
>> +	if (!adreno_gpu->info->hwcg) {
> 
> I don't think this block of code is reachable now, no?

It is because we didn't skip when adreno_is_a7xx(adreno_gpu)

> 
> Maybe remove the skip_programming and if_a750 here?
This would require:
 >> -	if (!adreno_gpu->info->hwcg || )
 >> +	if (!(adreno_gpu->info->hwcg || adreno_is_a750(adreno_gpu)))

and:

 >> +	if (adreno_is_a750(adreno_gpu)) {

But if the next gpu also doesn't have an hwcfg, we will need to use
the current design...

I just tried with:
====================><===============================
@@ -961,7 +961,7 @@ static void a6xx_set_hwcg(struct msm_gpu *gpu, bool state)
         unsigned int i;
         u32 val, clock_cntl_on, cgc_mode;

-       if (!adreno_gpu->info->hwcg)
+       if (!(adreno_gpu->info->hwcg || adreno_is_a750(adreno_gpu)))
                 return;

         if (adreno_is_a630(adreno_gpu))
@@ -982,6 +982,25 @@ static void a6xx_set_hwcg(struct msm_gpu *gpu, bool state)
                           state ? 0x5555 : 0);
         }

+       if (adreno_is_a750(adreno_gpu)) {
+               gpu_write(gpu, REG_A7XX_RBBM_CLOCK_CNTL_GLOBAL, 1);
+               gpu_write(gpu, REG_A7XX_RBBM_CGC_GLOBAL_LOAD_CMD, state ? 1 : 0);
+
+               if (state) {
+                       gpu_write(gpu, REG_A7XX_RBBM_CGC_P2S_TRIG_CMD, 1);
+
+                       if (gpu_poll_timeout(gpu, REG_A7XX_RBBM_CGC_P2S_STATUS, val,
+                                            val & A7XX_RBBM_CGC_P2S_STATUS_TXDONE, 1, 10)) {
+                               dev_err(&gpu->pdev->dev, "RBBM_CGC_P2S_STATUS TXDONE Poll failed\n");
+                               return;
+                       }
+
+                       gpu_write(gpu, REG_A7XX_RBBM_CLOCK_CNTL_GLOBAL, 0);
+               }
+
+               return;
+       }
+
         val = gpu_read(gpu, REG_A6XX_RBBM_CLOCK_CNTL);

         /* Don't re-program the registers if they are already correct */
====================><===============================

And it works fine, does it work it for you ?

> 
>> +		gpu_write(gpu, REG_A7XX_RBBM_CLOCK_CNTL_GLOBAL, 1);
>> +		gpu_write(gpu, REG_A7XX_RBBM_CGC_GLOBAL_LOAD_CMD, state ? 1 : 0);
>> +
>> +		if (state) {
>> +			gpu_write(gpu, REG_A7XX_RBBM_CGC_P2S_TRIG_CMD, 1);
>> +
>> +			if (gpu_poll_timeout(gpu, REG_A7XX_RBBM_CGC_P2S_STATUS, val,
>> +					     val & BIT(0), 1, 10)) {
> 
> We should define that bit name (the err suggests it's
> REG_A7XX_RBBM_GCC_P2S_STATUS_TXDONE or so)
> 
> [...]
> 
>> +static inline int adreno_is_a750(struct adreno_gpu *gpu)
>> +{
>> +	return gpu->info->chip_ids[0] == 0x43051401;
>> +}
>> +
>>   /* Placeholder to make future diffs smaller */
> 
> Please also remove this comment now that it's invalid

Ack

> 
> Konrad

Thanks,
Neil