[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20091112081317.GB25345@elte.hu>
Date: Thu, 12 Nov 2009 09:13:17 +0100
From: Ingo Molnar <mingo@...e.hu>
To: "H. Peter Anvin" <hpa@...or.com>
Cc: Bartlomiej Zolnierkiewicz <bzolnier@...il.com>, x86@...nel.org,
linux-kernel@...r.kernel.org, Dave Jones <davej@...emonkey.org.uk>
Subject: Re: [PATCH RFC] x86: fix confusing name of /proc/cpuinfo "ht" flag
* H. Peter Anvin <hpa@...or.com> wrote:
> On 11/11/2009 10:59 PM, Ingo Molnar wrote:
> >
> > * Bartlomiej Zolnierkiewicz <bzolnier@...il.com> wrote:
> >
> >> "ht" flag indicates only ability to detect siblings not HT presence
> >> itself.
> >>
> >> Inspired by:
> >> http://www.codemonkey.org.uk/2009/11/10/common-hyperthreading-misconception/
> >
> > i think instead of changing 'ht' to 'ht_detect', it would be even
> > more intuitive to only expose the 'ht' flag if the ht-detect
> > capability is there _and_ if the number of siblings is 2 or more.
> >
> > That way we dont change it - we just 'hide' the 'ht' string on the
> > category of systems that can enumerate HT via the CPUID but dont
> > actually have HyperThreading.
>
> /proc/cpuinfo is a user-space visible ABI. Changing it is bad
> chicken.
Almost ... something is only an ABI if an actual application relies on
the 'ht' flag being there on non-hyperthreading CPUs. We dont know
whether there's any, but the likelyhood looks pretty low. Could park it
a branch for some time to see whether anything trips up.
Ingo
--
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/
Powered by blists - more mailing lists