linux-kernel - Re: [PATCH v3 1/1] cpufreq: qcom-nvmem: add compatible fallback for ipq806x for no SMEM

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: <3c5a4d02-7391-48ec-bd45-d2ccacca1df6@kernel.org>
Date: Wed, 5 Nov 2025 10:28:25 +0100
From: Krzysztof Kozlowski <krzk@...nel.org>
To: Konrad Dybcio <konrad.dybcio@....qualcomm.com>,
 Christian Marangi <ansuelsmth@...il.com>, Ilia Lin <ilia.lin@...nel.org>,
 "Rafael J. Wysocki" <rafael@...nel.org>,
 Viresh Kumar <viresh.kumar@...aro.org>, linux-pm@...r.kernel.org,
 linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org
Cc: Dmitry Baryshkov <dmitry.baryshkov@....qualcomm.com>
Subject: Re: [PATCH v3 1/1] cpufreq: qcom-nvmem: add compatible fallback for
 ipq806x for no SMEM

On 05/11/2025 10:12, Konrad Dybcio wrote:
> On 11/5/25 9:05 AM, Krzysztof Kozlowski wrote:
>> On 04/11/2025 15:06, Christian Marangi wrote:
>>> On some IPQ806x SoC SMEM might be not initialized by SBL. This is the
>>> case for some Google devices (the OnHub family) that can't make use of
>>> SMEM to detect the SoC ID.
>>>
>>> To handle these specific case, check if the SMEM is not initialized (by
>>> checking if the qcom_smem_get_soc_id returns -ENODEV) and fallback to
>>> OF machine compatible checking to identify the SoC variant.
>>>
>>> Suggested-by: Dmitry Baryshkov <dmitry.baryshkov@....qualcomm.com>
>>> Signed-off-by: Christian Marangi <ansuelsmth@...il.com>
>>> ---
>>>  drivers/cpufreq/qcom-cpufreq-nvmem.c | 35 ++++++++++++++++++++++++++--
>>>  1 file changed, 33 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/drivers/cpufreq/qcom-cpufreq-nvmem.c b/drivers/cpufreq/qcom-cpufreq-nvmem.c
>>> index 3a8ed723a23e..17c79955ff2f 100644
>>> --- a/drivers/cpufreq/qcom-cpufreq-nvmem.c
>>> +++ b/drivers/cpufreq/qcom-cpufreq-nvmem.c
>>> @@ -252,13 +252,22 @@ static int qcom_cpufreq_krait_name_version(struct device *cpu_dev,
>>>  	return ret;
>>>  }
>>>  
>>> +static const struct of_device_id qcom_cpufreq_ipq806x_match_list[] = {
>>> +	{ .compatible = "qcom,ipq8062", .data = (const void *)QCOM_ID_IPQ8062 },
>>> +	{ .compatible = "qcom,ipq8064", .data = (const void *)QCOM_ID_IPQ8064 },
>>> +	{ .compatible = "qcom,ipq8065", .data = (const void *)QCOM_ID_IPQ8065 },
>>> +	{ .compatible = "qcom,ipq8066", .data = (const void *)QCOM_ID_IPQ8066 },
>>> +	{ .compatible = "qcom,ipq8068", .data = (const void *)QCOM_ID_IPQ8068 },
>>> +	{ .compatible = "qcom,ipq8069", .data = (const void *)QCOM_ID_IPQ8069 },
>>> +};
>>> +
>>>  static int qcom_cpufreq_ipq8064_name_version(struct device *cpu_dev,
>>>  					     struct nvmem_cell *speedbin_nvmem,
>>>  					     char **pvs_name,
>>>  					     struct qcom_cpufreq_drv *drv)
>>>  {
>>> +	int msm_id = -1, ret = 0;
>>>  	int speed = 0, pvs = 0;
>>> -	int msm_id, ret = 0;
>>>  	u8 *speedbin;
>>>  	size_t len;
>>>  
>>> @@ -275,8 +284,30 @@ static int qcom_cpufreq_ipq8064_name_version(struct device *cpu_dev,
>>>  	get_krait_bin_format_a(cpu_dev, &speed, &pvs, speedbin);
>>>  
>>>  	ret = qcom_smem_get_soc_id(&msm_id);
>>> -	if (ret)
>>> +	if (ret == -ENODEV) {
>>> +		const struct of_device_id *match;
>>> +		struct device_node *root;
>>> +
>>> +		root = of_find_node_by_path("/");
>>> +		if (!root) {
>>> +			ret = -ENODEV;
>>> +			goto exit;
>>> +		}
>>> +
>>> +		/* Fallback to compatible match with no SMEM initialized */
>>> +		match = of_match_node(qcom_cpufreq_ipq806x_match_list, root);
>>
>> Aren't you re-implementing matching machine? Or actually - the socinfo
>> driver? You are doing the matching of compatible into SOC ID second
>> time. Just do it once - via socinfo driver.
> 
> The issue here is that if SMEM is absent, the socinfo driver can
> not function and this is a contained workaround

OK, some short paragraph about this should be in commit msg though.

Best regards,
Krzysztof