linux-kernel - Re: [PATCH] cpufreq: amd-pstate: add check for cpufreq_cpu

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] [day] [month] [year] [list]

Message-ID: <d112c64c-b7d6-4aa3-ba78-8031bb7d1e1a@amd.com>
Date: Fri, 23 Aug 2024 11:19:19 -0500
From: Mario Limonciello <mario.limonciello@....com>
To: Anastasia Belova <abelova@...ralinux.ru>,
 "Gautham R. Shenoy" <gautham.shenoy@....com>
Cc: Huang Rui <ray.huang@....com>, Perry Yuan <perry.yuan@....com>,
 "Rafael J. Wysocki" <rafael@...nel.org>,
 Viresh Kumar <viresh.kumar@...aro.org>, linux-pm@...r.kernel.org,
 linux-kernel@...r.kernel.org, lvc-project@...uxtesting.org
Subject: Re: [PATCH] cpufreq: amd-pstate: add check for cpufreq_cpu_get's
 return value

On 8/8/2024 08:07, Anastasia Belova wrote:
> Hello,
> 
> 06/06/24 12:55, Gautham R. Shenoy пишет:
>> Hello,
>>
>> On Mon, Jun 03, 2024 at 02:07:41PM +0300, Anastasia Belova wrote:
>>> cpufreq_cpu_get may return NULL. To avoid NULL-dereference check it
>>> and return in case of error.
>>>
>>> Found by Linux Verification Center (linuxtesting.org) with SVACE.
>>>
>>> Signed-off-by: Anastasia Belova <abelova@...ralinux.ru>
>> Thank you for the patch. Indeed we should be checking if the policy is
>> valid before dereferencing it.
>>
>>> ---
>>>   drivers/cpufreq/amd-pstate.c | 4 ++++
>>>   1 file changed, 4 insertions(+)
>>>
>>> diff --git a/drivers/cpufreq/amd-pstate.c b/drivers/cpufreq/amd-pstate.c
>>> index 1b7e82a0ad2e..672cb6c280a4 100644
>>> --- a/drivers/cpufreq/amd-pstate.c
>>> +++ b/drivers/cpufreq/amd-pstate.c
>>> @@ -621,6 +621,8 @@ static void amd_pstate_adjust_perf(unsigned int cpu,
>>>       unsigned long max_perf, min_perf, des_perf,
>>>                 cap_perf, lowest_nonlinear_perf, max_freq;
>>>       struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
>>> +    if (!policy)
>>> +        return;
>> This patch mixes code and declarations. While I personally don't
>> prefer that, since we have moved to using C99, the compiler does
>> not complain, nor does checkpatch complain.
>>
>> So is this ok for cpufreq, Rafael?
> 
> Should I form the second version without mixing code and declarations?
> Or it is better to wait for Rafael's answer?

FWIW, I don't really like it either.  As it's amd-pstate code I'd say 
Gautham and I should make the call.

Can you please change it to avoid mixing code and declarations?

> 
>>
>> Or would you prefer something like:
>>
>>     unsigned long cap_perf, lowest_nonlinear_perf;
>>     unsigned long max_perf, min_perf, des_perf;
>>     struct cpufreq_policy *policy;
>>     struct amd_cpudata *cpudata;
>>     unsigned int target_freq;
>>     unsigned long max_freq;
>>
>>     policy = cpufreq_cpu_get(cpu);
>>     if (!policy)
>>         return;
>>
>>     cpudata = policy->driver_data;
>>
>>
>>
>>>       struct amd_cpudata *cpudata = policy->driver_data;
>>>       unsigned int target_freq;
>>> @@ -777,6 +779,8 @@ static void amd_pstate_init_prefcore(struct 
>>> amd_cpudata *cpudata)
>>>   static void amd_pstate_update_limits(unsigned int cpu)
>>>   {
>>>       struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
>>> +    if (!policy)
>>> +        return;
>> Ditto.
>>
>>>       struct amd_cpudata *cpudata = policy->driver_data;
>>>       u32 prev_high = 0, cur_high = 0;
>>>       int ret;
>>> -- 
>>> 2.30.2
>>>
>> -- 
>> Thanks and Regards
>> gautham.
> 
> Thanks,
> Anastasia Belova