linux-kernel - Re: [PATCH v2 1/3] soc: qcom: qcom_stats: Add support to read DDR statistic

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: <d70da81f-29a6-487c-9781-c2fda6327a75@oss.qualcomm.com>
Date: Thu, 22 May 2025 11:18:29 +0530
From: "Maulik Shah (mkshah)" <maulik.shah@....qualcomm.com>
To: Konrad Dybcio <konrad.dybcio@....qualcomm.com>,
        Bjorn Andersson <andersson@...nel.org>,
        Konrad Dybcio <konradybcio@...nel.org>, Rob Herring <robh@...nel.org>,
        Krzysztof Kozlowski <krzk+dt@...nel.org>,
        Conor Dooley <conor+dt@...nel.org>
Cc: linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org,
        devicetree@...r.kernel.org,
        Marijn Suijten <marijn.suijten@...ainline.org>,
        Doug Anderson <dianders@...omium.org>
Subject: Re: [PATCH v2 1/3] soc: qcom: qcom_stats: Add support to read DDR
 statistic



On 5/21/2025 10:31 PM, Konrad Dybcio wrote:
> On 5/21/25 10:32 AM, Maulik Shah wrote:
>> DDR statistic provide different DDR LPM and DDR frequency statistic.
>> Add support to read from MSGRAM and display via debugfs.
>>
>> Signed-off-by: Maulik Shah <maulik.shah@....qualcomm.com>
>> ---
> 
> [...]
> 
>> +	case 0:
>> +		seq_printf(s, "DDR LPM Stat Name:0x%x\tcount:%u\tDuration (ticks):%llu\n",
>> +			   DDR_STATS_LPM_NAME(data->name), data->count, data->duration);
>> +		break;
>> +	case 1:
>> +		if (!data->count || !DDR_STATS_FREQ(data->name))
>> +			return;
>> +
>> +		cp_idx = DDR_STATS_CP_IDX(data->name);
>> +		seq_printf(s, "DDR Freq %uMhz:\tCP IDX:%u\tcount:%u\tDuration (ticks):%llu\n",
>> +			   DDR_STATS_FREQ(data->name), cp_idx, data->count, data->duration);
> 
> clang complains about both prints:
> 
> drivers/soc/qcom/qcom_stats.c:173:7: warning: format specifies type 'unsigned int' but the argument has type 'unsigned long' [-Wformat]
>   172 |                 seq_printf(s, "DDR LPM Stat Name:0x%x\tcount:%u\tDuration (ticks):%llu\n",
>       |                                                    ~~
>       |                                                    %lx
>   173 |                            DDR_STATS_LPM_NAME(data->name), data->count, data->duration);
>       |                            ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> drivers/soc/qcom/qcom_stats.c:181:7: warning: format specifies type 'unsigned int' but the argument has type 'unsigned long' [-Wformat]
>   180 |                 seq_printf(s, "DDR Freq %uMhz:\tCP IDX:%u\tcount:%u\tDuration (ticks):%llu\n",
>       |                                         ~~
>       |                                         %lu
>   181 |                            DDR_STATS_FREQ(data->name), cp_idx, data->count, data->duration);
>       |                            ^~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> 
> 

I will update correct format specifiers in v3.

> 
>> +
>> +	key = readl_relaxed(reg + config->ddr_stats_offset + DDR_STATS_MAGIC_KEY_ADDR);
>> +	if (key == DDR_STATS_MAGIC_KEY)
>> +		debugfs_create_file("ddr_stats", 0400, root,
>> +				    (__force void *)reg + config->ddr_stats_offset,
>> +				    &qcom_ddr_stats_fops);
> 
> else
> 	pr_err("Found invalid DDR stats magic\n");
> 
> (because through the compatible, we much expect it to be present)

The qcom,rpmh-stats compatible does not guarantee the DDR stats presence. DDR stats is only present if
magic value matches. The ddr stats was incrementally added over time so older SoCs like SM8150, QCS615
will not have the ddr stats and would end up printing this error during boot up but yes all almost all
rpmh targets do have the DDR stats present. If we are ok to print this error for older SoCs i can add it
or how about using pr_warn instead of pr_err?

Thanks,
Maulik

> 
> Konrad