linux-kernel - Re: [PATCH -tip] x86: cpu/proc.c adding extended_cpuid

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]

Date:	Wed, 10 Jun 2009 19:42:23 +0200
From:	Borislav Petkov <borislav.petkov@....com>
To:	"H. Peter Anvin" <hpa@...or.com>
CC:	Jaswinder Singh Rajput <jaswinder@...nel.org>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Ingo Molnar <mingo@...e.hu>, x86 maintainers <x86@...nel.org>,
	LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH -tip] x86: cpu/proc.c adding extended_cpuid_level for
	/proc/cpuinfo

On Wed, Jun 10, 2009 at 10:14:09AM -0700, H. Peter Anvin wrote:
> >>>>
> >>> The more I'm thinking about this I think it was a mistake to put cpuid
> >>> level: there in the first place, too.  My opinion is increasingly to
> >>> leave this to x86info or other user-space tools.
> >>>
> >> cpuid level is as important as cpu family, model and stepping.
> >>
> >> For Intel, in some cases cpuid level is more important then cpu family,
> >> model and stepping. Like you cannot tell by looking at cpu family, model
> >> and stepping which model is new and which is old like 05_01 or 06_1A or
> >> 0F_03H ?
> >>
> >> But by looking at cpuid level and extended cpuid level you can tell
> >> which is new and which is old and which supports more features.
> >>
> >> So cpuid level and extended cpuid level is better scale than cpu family,
> >> model and stepping. So I think hiding this valuable information is a
> >> crime.
> > 
> > cpuid level and extended cpuid level tells the information about Intel
> > processor model.  Do you still think it is useless and should not be present
> > in /proc/cpuinfo .
> > 
> 
> I think it's pointless, because if you're doing this kind of stuff you
> might as well talk to CPUID directly.  We have a CPUID interface
> already.  The kernel isn't meant to replicate x86info or any of those
> tools -- it really can't, and at that point why stop at x86info... we
> could replicate arbitrary applications at that point.
> 
> As far as extended CPUID, why only the 0x0000 and 0x8000 ranges?
> Transmeta used 0x8086 and VIA uses 0xC000, but we don't even cache those
> internally, nevermind display them.
> 
> I'm not really all that doctrinal about this, but I'd like to get a
> decent answer to the question "where does it stop, *and why*".  It's a
> slippery slope, and without a target, it goes on forever.

I still fail to see a real use case which requires adding that info to
the kernel and querying CPUID directly is _absolutely_ not an option.
This is what the decisive argument should be and not some "but we have
already this and that in there." Otherwise its like painting a broken
car pink - it's still broken.

-- 
Regards/Gruss,
Boris.

Operating | Advanced Micro Devices GmbH
  System  | Karl-Hammerschmidt-Str. 34, 85609 Dornach b. München, Germany
 Research | Geschäftsführer: Thomas M. McCoy, Giuliano Meroni
  Center  | Sitz: Dornach, Gemeinde Aschheim, Landkreis München
  (OSRC)  | Registergericht München, HRB Nr. 43632

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/