2022-01-26 20:44:50

by Liu Chao

[permalink] [raw]
Subject: [PATCH v2] add cpu frequency to cpuinfo in arm64

There is cpu frequency in /proc/cpuinfo in x86 but not in arm64

Signed-off-by: Chao Liu <[email protected]>
---
Change in v2:
- fix spelling errors and remove remove redundant blank line

arch/arm64/kernel/cpuinfo.c | 6 ++++++
1 file changed, 6 insertions(+)

diff --git a/arch/arm64/kernel/cpuinfo.c b/arch/arm64/kernel/cpuinfo.c
index 591c18a889a5..39a36e6e8079 100644
--- a/arch/arm64/kernel/cpuinfo.c
+++ b/arch/arm64/kernel/cpuinfo.c
@@ -24,6 +24,7 @@
#include <linux/sched.h>
#include <linux/smp.h>
#include <linux/delay.h>
+#include <linux/cpufreq.h>

/*
* In case the boot CPU is hotpluggable, we record its initial state and
@@ -144,6 +145,7 @@ static int c_show(struct seq_file *m, void *v)
for_each_online_cpu(i) {
struct cpuinfo_arm64 *cpuinfo = &per_cpu(cpu_data, i);
u32 midr = cpuinfo->reg_midr;
+ unsigned int freq = cpufreq_quick_get(i);

/*
* glibc reads /proc/cpuinfo to determine the number of
@@ -159,6 +161,10 @@ static int c_show(struct seq_file *m, void *v)
loops_per_jiffy / (500000UL/HZ),
loops_per_jiffy / (5000UL/HZ) % 100);

+ if (freq)
+ seq_printf(m, "cpu MHz\t\t: %u.%03u\n",
+ freq / 1000, (freq % 1000));
+
/*
* Dump out the common processor features in a single line.
* Userspace should read the hwcaps with getauxval(AT_HWCAP)
--
2.23.0


2022-01-26 20:57:44

by Sudeep Holla

[permalink] [raw]
Subject: Re: [PATCH v2] add cpu frequency to cpuinfo in arm64

On Wed, Jan 26, 2022 at 09:23:49AM +0000, Chao Liu wrote:
> There is cpu frequency in /proc/cpuinfo in x86 but not in arm64
>

This has been repeatedly rejected in the past for valid reasons. I
could find this link[1] quickly, but I am sure you can find a handful of
such attempts in the past.

--
Regards,
Sudeep

[1] https://patchwork.kernel.org/project/linux-arm-kernel/patch/[email protected]/

2022-01-26 21:12:31

by Robin Murphy

[permalink] [raw]
Subject: Re: [PATCH v2] add cpu frequency to cpuinfo in arm64

On 2022-01-26 09:23, Chao Liu wrote:
> There is cpu frequency in /proc/cpuinfo in x86 but not in arm64

arm64 cpuinfo also doesn't show the APIC ID or microcode version.
Different architectures are different.

Besides, x86 cpuinfo only shows MHz if the CPU has X86_FEATURE_TSC, so
for compatibility we should only do this on AArch64 CPUs which have
X86_FEATURE_TSC.

Robin.

> Signed-off-by: Chao Liu <[email protected]>
> ---
> Change in v2:
> - fix spelling errors and remove remove redundant blank line
>
> arch/arm64/kernel/cpuinfo.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/arch/arm64/kernel/cpuinfo.c b/arch/arm64/kernel/cpuinfo.c
> index 591c18a889a5..39a36e6e8079 100644
> --- a/arch/arm64/kernel/cpuinfo.c
> +++ b/arch/arm64/kernel/cpuinfo.c
> @@ -24,6 +24,7 @@
> #include <linux/sched.h>
> #include <linux/smp.h>
> #include <linux/delay.h>
> +#include <linux/cpufreq.h>
>
> /*
> * In case the boot CPU is hotpluggable, we record its initial state and
> @@ -144,6 +145,7 @@ static int c_show(struct seq_file *m, void *v)
> for_each_online_cpu(i) {
> struct cpuinfo_arm64 *cpuinfo = &per_cpu(cpu_data, i);
> u32 midr = cpuinfo->reg_midr;
> + unsigned int freq = cpufreq_quick_get(i);
>
> /*
> * glibc reads /proc/cpuinfo to determine the number of
> @@ -159,6 +161,10 @@ static int c_show(struct seq_file *m, void *v)
> loops_per_jiffy / (500000UL/HZ),
> loops_per_jiffy / (5000UL/HZ) % 100);
>
> + if (freq)
> + seq_printf(m, "cpu MHz\t\t: %u.%03u\n",
> + freq / 1000, (freq % 1000));
> +
> /*
> * Dump out the common processor features in a single line.
> * Userspace should read the hwcaps with getauxval(AT_HWCAP)

2022-01-26 21:14:06

by Marc Zyngier

[permalink] [raw]
Subject: Re: [PATCH v2] add cpu frequency to cpuinfo in arm64

On 2022-01-26 12:05, Robin Murphy wrote:
> On 2022-01-26 09:23, Chao Liu wrote:
>> There is cpu frequency in /proc/cpuinfo in x86 but not in arm64
>
> arm64 cpuinfo also doesn't show the APIC ID or microcode version.
> Different architectures are different.
>
> Besides, x86 cpuinfo only shows MHz if the CPU has X86_FEATURE_TSC, so
> for compatibility we should only do this on AArch64 CPUs which have
> X86_FEATURE_TSC.

/me prepares a patch to be sent in just over 2 months...

M.
--
Jazz is not dead. It just smells funny...