From: zhengzucheng <zhengzucheng@huawei.com>
To: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Viresh Kumar <viresh.kumar@linaro.org>,
Thomas Gleixner <tglx@linutronix.de>,
Len Brown <len.brown@intel.com>,
Linux PM <linux-pm@vger.kernel.org>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
Stable <stable@vger.kernel.org>
Subject: Re: [PATCH] cpufreq: fix cpufreq_get() can't get correct CPU frequency
Date: Tue, 15 Mar 2022 10:30:23 +0800 [thread overview]
Message-ID: <100a5228-a941-0ff7-8133-14273472b6f5@huawei.com> (raw)
In-Reply-To: <CAJZ5v0jponp=ijVx6W=eNEGrfTKh0KbGmOQG_V0P-Mq366559g@mail.gmail.com>
On 2022/3/11 23:52, Rafael J. Wysocki wrote:
> On Fri, Mar 11, 2022 at 9:11 AM z00314508 <zhengzucheng@huawei.com> wrote:
>> From: Zucheng Zheng <zhengzucheng@huawei.com>
>>
>> On some specific platforms, the cpufreq driver does not define
>> cpufreq_driver.get() routine (eg:x86 intel_pstate driver), as a
> I guess you mean the cpufreq driver ->get callback.
>
> No, intel_pstate doesn't implement it, because it cannot reliably
> return the current CPU frequency.
>
>> result, the cpufreq_get() can't get the correct CPU frequency.
> No, it can't, if intel_pstate is the driver, but what's the problem?
> This function is only called in one place in the kernel and not on x8
> even.
>
>> Modern x86 processors include the hardware needed to accurately
>> calculate frequency over an interval -- APERF, MPERF and the TSC.
> You can compute the average frequency over an interval, but ->get is
> expected to return the actual current frequency at the time call time.
>
>> Here we use arch_freq_get_on_cpu() in preference to any driver
>> driver-specific cpufreq_driver.get() routine to get CPU frequency.
>>
>> Fixes: f8475cef9008 ("x86: use common aperfmperf_khz_on_cpu() to calculate KHz using APERF/MPERF")
> No kidding.
>
>> Signed-off-by: Zucheng Zheng <zhengzucheng@huawei.com>
>> ---
>> drivers/cpufreq/cpufreq.c | 6 +++++-
>> 1 file changed, 5 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
>> index 80f535cc8a75..d777257b4454 100644
>> --- a/drivers/cpufreq/cpufreq.c
>> +++ b/drivers/cpufreq/cpufreq.c
>> @@ -1806,10 +1806,14 @@ unsigned int cpufreq_get(unsigned int cpu)
>> {
>> struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
>> unsigned int ret_freq = 0;
>> + unsigned int freq;
>>
>> if (policy) {
>> down_read(&policy->rwsem);
>> - if (cpufreq_driver->get)
>> + freq = arch_freq_get_on_cpu(policy->cpu);
>> + if (freq)
>> + ret_freq = freq;
>> + else if (cpufreq_driver->get)
> Again, what problem exactly does this address?
Thank you for review.
In some scenarios, cpufreq driver ->get is not defined,
some driver get the CPU frequency by calling cpufreq_get() will return 0.
The modification to this problem is inspired by the implementation of
the show_scaling_cur_freq().
>
>> ret_freq = __cpufreq_get(policy);
>> up_read(&policy->rwsem);
>>
>> --
> .
>
next prev parent reply other threads:[~2022-03-15 2:30 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-03-11 8:11 [PATCH] cpufreq: fix cpufreq_get() can't get correct CPU frequency z00314508
2022-03-11 8:33 ` Greg KH
2022-03-11 15:52 ` Rafael J. Wysocki
2022-03-15 2:30 ` zhengzucheng [this message]
2022-03-15 10:47 ` Rafael J. Wysocki
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=100a5228-a941-0ff7-8133-14273472b6f5@huawei.com \
--to=zhengzucheng@huawei.com \
--cc=len.brown@intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pm@vger.kernel.org \
--cc=rafael@kernel.org \
--cc=stable@vger.kernel.org \
--cc=tglx@linutronix.de \
--cc=viresh.kumar@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).