linux-arm-kernel.lists.infradead.org archive mirror
 help / color / mirror / Atom feed
* RE: Questions about /proc/cpuinfo for arch-ARM  - CAS-149046-Y1H3T8
       [not found] ` <VE1PR08MB47490C13405F61B3599D1314F0320@VE1PR08MB4749.eurprd08.prod.outlook.com>
@ 2020-01-20  9:24   ` Steven Miao
  2020-01-20  9:41     ` Marc Zyngier
  0 siblings, 1 reply; 2+ messages in thread
From: Steven Miao @ 2020-01-20  9:24 UTC (permalink / raw)
  To: Support-Software, Tao Zeng; +Cc: Mark Rutland, Robin Murphy, linux-arm-kernel

CC arm mailing list. Someone know the whole history may have a better answer.

Thanks,
Steven

From: Support-Software <mailto:support-software@arm.com>
Sent: Monday, January 20, 2020 5:17 PM
To: Tao Zeng <mailto:prime.zeng@hisilicon.com>
Cc: Mark Rutland <mailto:Mark.Rutland@arm.com>
Subject: Re: Questions about /proc/cpuinfo for arch-ARM - CAS-149046-Y1H3T8

From Steven Miao - Arm Partner Enablement Group
Please quote case reference number https://support.developer.arm.com/case-details/?id=f9beb263-333b-ea11-a813-000d3ab4f2f3 in any further emails to us on this subject.

Hi Tao,
>1. Since /proc/cpuinfo is treated as ABI, so where can I find the specification? Or is there any docs
>to describe the details? Or it is just some applications who have already token it?
 I think it is just de-facto ABI legacy application are relying on, there's no such specification.
>You know that, in x86, /proc/cpuinfo provide really good user experience, what we want to do is >providing the same or even better user experience.
We should add new stuff to /sys/, don't add new things to /proc. On x86 these strings are acquired from the CPU itself, via CPUID instructions, which means that it works for future CPUs. For ARM systems, we have no consistent way of acquiring a model name from a CPU itself. If users who want to decode MIDRs are going to have to use userspace tools.
Changing current cpuinfo:
- it breaks existing applications
- it is unmaintainable in the long run
- we can already get cpu information by dmidecode, lscpu, lshw...

Some discussion about this topic in mailing list:
https://patchwork.kernel.org/patch/10524949/
https://lkml.org/lkml/2019/12/6/485
http://lists.infradead.org/pipermail/linux-arm-kernel/2014-July/273022.html

Kind regards
Steven Miao
Arm Partner Enablement Group

For more support information, documentation, downloads and other useful resources see: https://developer.arm.com/support/

-----Original Message-----



From: Tao Zeng
Received: 20 Jan 2020 03:17 AM
Hi:

As discussed in the following mailing thread:
https://lore.kernel.org/patchwork/cover/834876/
I think the most important conclusion here is :
"The /proc/cpuinfo is already widely used and it's an in fact arch-specific ABI."

And I also agree mark's idea, but I still have got my questions here:
1. Since /proc/cpuinfo is treated as ABI, so where can I find the specification? Or is there any docs
to describe the details? Or it is just some applications who have already token it?

2. My idea here:
(1) If this is really an ABI, maybe we need to standardizing it in some way, and let more and more
people know about it other than asking "why not like X86?" again and again.
(2) If this is just used by some apps because of the historical reason, maybe we can help to improve
the apps?
(Some ideas like providing two kinds of /proc/cpuinfo, One is for server and the other is to keep the
 current /proc/cpuinfo, and controlled by kernel configs?)

You know that, in x86, /proc/cpuinfo provide really good user experience, what we want to do is providing
the same or even better user experience.
And I am not arguing what's more reasonable here, I just want to discuss for better solution.

Thanks.

Regards
Zengtao
IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: Questions about /proc/cpuinfo for arch-ARM  - CAS-149046-Y1H3T8
  2020-01-20  9:24   ` Questions about /proc/cpuinfo for arch-ARM - CAS-149046-Y1H3T8 Steven Miao
@ 2020-01-20  9:41     ` Marc Zyngier
  0 siblings, 0 replies; 2+ messages in thread
From: Marc Zyngier @ 2020-01-20  9:41 UTC (permalink / raw)
  To: Steven Miao, Tao Zeng
  Cc: Mark Rutland, Robin Murphy, linux-arm-kernel, Support-Software

Hi Steven,

On 2020-01-20 10:24, Steven Miao wrote:
> CC arm mailing list. Someone know the whole history may have a better 
> answer.
> 
> Thanks,
> Steven
> 
> From: Support-Software <mailto:support-software@arm.com>
> Sent: Monday, January 20, 2020 5:17 PM
> To: Tao Zeng <mailto:prime.zeng@hisilicon.com>
> Cc: Mark Rutland <mailto:Mark.Rutland@arm.com>
> Subject: Re: Questions about /proc/cpuinfo for arch-ARM - 
> CAS-149046-Y1H3T8
> 
> From Steven Miao - Arm Partner Enablement Group
> Please quote case reference number
> https://support.developer.arm.com/case-details/?id=f9beb263-333b-ea11-a813-000d3ab4f2f3
> in any further emails to us on this subject.
> 
> Hi Tao,
>> 1. Since /proc/cpuinfo is treated as ABI, so where can I find the 
>> specification? Or is there any docs
>> to describe the details? Or it is just some applications who have 
>> already token it?
>  I think it is just de-facto ABI legacy application are relying on,
> there's no such specification.
>> You know that, in x86, /proc/cpuinfo provide really good user 
>> experience,
>> what we want to do is providing the same or even better user 
>> experience.

Tao: And by "better user experience", you mean breaking existing 
software?
I'm sure users will enjoy the "improvement".

> We should add new stuff to /sys/, don't add new things to /proc. On
> x86 these strings are acquired from the CPU itself, via CPUID
> instructions, which means that it works for future CPUs. For ARM
> systems, we have no consistent way of acquiring a model name from a
> CPU itself. If users who want to decode MIDRs are going to have to use
> userspace tools.
> Changing current cpuinfo:
> - it breaks existing applications
> - it is unmaintainable in the long run
> - we can already get cpu information by dmidecode, lscpu, lshw...

Exactly. The subject has been rehashed to death, and is not up for
discussion anymore. /proc/cpuinfo is ABI, which means its format is
forever immutable. Userspace has all the required tools at its disposal.

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

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2020-01-20  9:42 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
     [not found] <FFCDED37A161475FA3F5B19E21DECA0D1D5CF72567E6@SUPPORT-SOFTWARE.ARM.COM>
     [not found] ` <VE1PR08MB47490C13405F61B3599D1314F0320@VE1PR08MB4749.eurprd08.prod.outlook.com>
2020-01-20  9:24   ` Questions about /proc/cpuinfo for arch-ARM - CAS-149046-Y1H3T8 Steven Miao
2020-01-20  9:41     ` Marc Zyngier

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).