linux-kernel - RE: [PATCH 4/7] cpufreq: amd-pstate: fix max_perf calculation for amd_get_max

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:
 <CYYPR12MB8655EB0CDA7F893AABE198AD9C7E2@CYYPR12MB8655.namprd12.prod.outlook.com>
Date: Mon, 29 Jan 2024 05:16:37 +0000
From: "Yuan, Perry" <Perry.Yuan@....com>
To: "Limonciello, Mario" <Mario.Limonciello@....com>,
	"rafael.j.wysocki@...el.com" <rafael.j.wysocki@...el.com>,
	"viresh.kumar@...aro.org" <viresh.kumar@...aro.org>, "Huang, Ray"
	<Ray.Huang@....com>, "Shenoy, Gautham Ranjal" <gautham.shenoy@....com>,
	"Petkov, Borislav" <Borislav.Petkov@....com>
CC: "Deucher, Alexander" <Alexander.Deucher@....com>, "Huang, Shimmer"
	<Shimmer.Huang@....com>, "Du, Xiaojian" <Xiaojian.Du@....com>, "Meng, Li
 (Jassmine)" <Li.Meng@....com>, "linux-pm@...r.kernel.org"
	<linux-pm@...r.kernel.org>, "linux-kernel@...r.kernel.org"
	<linux-kernel@...r.kernel.org>
Subject: RE: [PATCH 4/7] cpufreq: amd-pstate: fix max_perf calculation for
 amd_get_max_freq()

[AMD Official Use Only - General]

Hi  Mario,

> -----Original Message-----
> From: Limonciello, Mario <Mario.Limonciello@....com>
> Sent: Friday, January 26, 2024 11:54 PM
> To: Yuan, Perry <Perry.Yuan@....com>; rafael.j.wysocki@...el.com;
> viresh.kumar@...aro.org; Huang, Ray <Ray.Huang@....com>; Shenoy,
> Gautham Ranjal <gautham.shenoy@....com>; Petkov, Borislav
> <Borislav.Petkov@....com>
> Cc: Deucher, Alexander <Alexander.Deucher@....com>; Huang, Shimmer
> <Shimmer.Huang@....com>; Du, Xiaojian <Xiaojian.Du@....com>; Meng,
> Li (Jassmine) <Li.Meng@....com>; linux-pm@...r.kernel.org; linux-
> kernel@...r.kernel.org
> Subject: Re: [PATCH 4/7] cpufreq: amd-pstate: fix max_perf calculation for
> amd_get_max_freq()
>
> On 1/26/2024 02:08, Perry Yuan wrote:
> > From: Perry Yuan <Perry.Yuan@....com>
> >
> > When CPU core Precision Boost state changed, the max frequency will
> > also need to be updated according to the current boost state, if boost
> > is disabled now, the max perf will be limited to nominal perf values.
> > otherwise the max frequency will be showed wrongly.
>
> What about the previous cppc_req?  Shouldn't it be explicitly updated as a
> result of this too?

The CPPC REQ value has been updated in this function.
When boost state changed, the MSR will be updated firstly.

+static int amd_cpu_boost_update(struct amd_cpudata *cpudata, u32 on)
+{
+       struct cpufreq_policy *policy = cpufreq_cpu_acquire(cpudata->cpu);
+       struct cppc_perf_ctrls perf_ctrls;
+       u32 highest_perf, nominal_perf;
+       int ret;
+
+       if (!policy)
+               return -ENODATA;
+
+       highest_perf = READ_ONCE(cpudata->highest_perf);
+       nominal_perf = READ_ONCE(cpudata->nominal_perf);
+
+       if (boot_cpu_has(X86_FEATURE_CPPC)) {
+               u64 value = READ_ONCE(cpudata->cppc_req_cached);
+
+               value &= ~GENMASK_ULL(7, 0);
+               value |= on ? highest_perf : nominal_perf;
+               WRITE_ONCE(cpudata->cppc_req_cached, value);
+
+               wrmsrl_on_cpu(cpudata->cpu, MSR_AMD_CPPC_REQ, value);




>
> >
> > Signed-off-by: Perry Yuan <Perry.Yuan@....com>
> > ---
> >   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 b37bea7440b9..3286d72f375e
> 100644
> > --- a/drivers/cpufreq/amd-pstate.c
> > +++ b/drivers/cpufreq/amd-pstate.c
> > @@ -599,6 +599,10 @@ static int amd_get_max_freq(struct amd_cpudata
> *cpudata)
> >     nominal_perf = READ_ONCE(cpudata->nominal_perf);
> >     max_perf = READ_ONCE(cpudata->highest_perf);
> >
> > +   /* when boost is off, the highest perf will be limited to nominal_perf */
> > +   if (!global.cpb_boost)
> > +           max_perf = nominal_perf;
> > +
> >     boost_ratio = div_u64(max_perf << SCHED_CAPACITY_SHIFT,
> >                           nominal_perf);
> >