linux-kernel - Re: [tip:perf/core] perf/x86/intel: Use Intel family macros for core perf events

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: <alpine.DEB.2.20.1606081609360.5993@macbook-air>
Date:	Wed, 8 Jun 2016 16:15:25 -0400 (EDT)
From:	Vince Weaver <vincent.weaver@...ne.edu>
To:	Borislav Petkov <bp@...en8.de>
cc:	Ingo Molnar <mingo@...nel.org>, tglx@...utronix.de,
	brgerst@...il.com, dave@...1.net, kan.liang@...el.com,
	torvalds@...ux-foundation.org, acme@...hat.com,
	alexander.shishkin@...ux.intel.com, jolsa@...hat.com,
	linux-kernel@...r.kernel.org, eranian@...gle.com,
	peterz@...radead.org, dvlasenk@...hat.com, hpa@...or.com,
	dave.hansen@...ux.intel.com, luto@...capital.net,
	linux-tip-commits@...r.kernel.org
Subject: Re: [tip:perf/core] perf/x86/intel: Use Intel family macros for core
 perf events

On Wed, 8 Jun 2016, Borislav Petkov wrote:

> On Wed, Jun 08, 2016 at 06:25:29PM +0200, Ingo Molnar wrote:
> >  model string    : IvyBridge_X
> > 
> > ... or something like that. The string could even be auto-generated from the list 
> > in intel-family.h?
> 
> Yap, that sounds cool too. And then we should issue that too during boot:
> 
> [    0.542398] smpboot: CPU0: AMD FX(tm)-8350 Eight-Core Processor (family: 0x15, model: 0x2, stepping: 0x0; Piledriver)
> 
> so that we don't need to ask for dmesg *and* /proc/cpuinfo.
> 
> Hohumm, this should be very useful.

However you do it, you want to make the strings match intel-family.h for 
maximum grepability.

It's also unclear what internal names matter.  For example is AMD 15h/60h
"Carrizo" or "Excavator" or both?

I just spent the afternoon making a huge RAPL/APM kernel support matrix 
for the PAPI people.

Lots of fun since
	 arch/x86/events/intel/rapl.c
and
	 drivers/powercap/intel_rapl.c

support widely different subsets of Intel hardware, and one uses decimal 
while one uses hex for the model numbers.  So I reluctantly admit unifying 
this all somehow would be nice.

(why we need two different interfaces for RAPL is another thing, not to 
mention the fact that powercap you can read the values as a normal user 
but with perf you need to be root to read the exact same thing).

Vince