linux-kernel - RE: [PATCH] sched: Support current clocksource handling in fallback sched

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, 27 May 2009 02:04:05 +0200 (CEST)
From:	Thomas Gleixner <tglx@...utronix.de>
To:	"Mangalampalli, JayantX" <jayantx.mangalampalli@...el.com>
cc:	john stultz <johnstul@...ibm.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Linus Walleij <linus.ml.walleij@...il.com>,
	Paul Mundt <lethal@...ux-sh.org>, Ingo Molnar <mingo@...e.hu>,
	Andrew Victor <linux@...im.org.za>,
	Haavard Skinnemoen <hskinnemoen@...el.com>,
	Andrew Morton <akpm@...ux-foundation.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"linux-sh@...r.kernel.org" <linux-sh@...r.kernel.org>,
	"linux-arm-kernel@...ts.arm.linux.org.uk" 
	<linux-arm-kernel@...ts.arm.linux.org.uk>,
	John Stultz <johnstul@...ux.vnet.ibm.com>
Subject: RE: [PATCH] sched: Support current clocksource handling in fallback
 sched_clock().

On Tue, 26 May 2009, Mangalampalli, JayantX wrote:

1) please do not top post.
2) please fix your mail client to do proper line breaks at ~78 chars

(see http://www.tux.org/lkml/)

> Isn't rdtsc inherently dependent on CPU tick rate, which is somewhat
> variable? Shouldn't HPET be more reliable? Or can constant_tsc be
> relied on for important things like scheduler?

1) Please understand that there is an universe which has useful timer
implementations. i.e. almost everything except arch/x86

2) rdtsc is not inherently dependent on CPU frequency. It just depends
on the CPU frequency for most of the CPU implementations which have
the ability to change the CPU core frequency. Newer CPUs have core
frequency invariant TSC implementations, but they are not perfect yet
as the TSC stops in deeper C-states.

3) HPET is reliable but the hardware access to HPET is way more
expensive than the access to TSC. Also it does not scale on SMP
machines as the HPET access is serialized and worse than a global
lock. So there is a damned good reason why we went there to add the
horror of sched_clock.c to utilize TSC for hot code pathes.

Thanks,

	tglx
--
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/