From owner-lockmeter@oss.sgi.com Wed Nov 14 08:22:55 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.2/8.11.3) id fAEGMtt01421 for lockmeter-outgoing; Wed, 14 Nov 2001 08:22:55 -0800 Received: (from hawkes@localhost) by oss.sgi.com (8.11.2/8.11.3) id fAEGMsf01416 for lockmeter@oss.sgi.com; Wed, 14 Nov 2001 08:22:54 -0800 Date: Wed, 14 Nov 2001 08:22:54 -0800 From: John Hawkes Message-Id: <200111141622.fAEGMsf01416@oss.sgi.com> To: lockmeter@oss.sgi.com Subject: Lockmeter for 2.4.14 is available Sender: owner-lockmeter@oss.sgi.com Precedence: bulk FYI, the latest Lockmeter (v1.4.9) is now available as a patch against the 2.4.14 kernel: http:/oss.sgi.com/projects/lockmeter/download/lockmeter1.4.9-2.4.14.diff.gz John Hawkes hawkes@sgi.com From owner-lockmeter@oss.sgi.com Wed Nov 21 18:47:18 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.2/8.11.3) id fAM2lIR26664 for lockmeter-outgoing; Wed, 21 Nov 2001 18:47:18 -0800 Received: (from hawkes@localhost) by oss.sgi.com (8.11.2/8.11.3) id fAM2lHl26660 for lockmeter@oss.sgi.com; Wed, 21 Nov 2001 18:47:17 -0800 Date: Wed, 21 Nov 2001 18:47:17 -0800 From: John Hawkes Message-Id: <200111220247.fAM2lHl26660@oss.sgi.com> To: lockmeter@oss.sgi.com Subject: lockmeter v1.4.10 available for 2.4.14 Sender: owner-lockmeter@oss.sgi.com Precedence: bulk FYI, a new Lockmeter, v1.4.10, is now available as a patch against the 2.4.14 kernel. The difference between this and the previous v1.4.9 is that this version provides a clear look at who is using atomic_dec_and_lock() to acquire spinlocks. http://oss.sgi.com/projects/lockmeter/download/lockmeter1.4.10-2.4.14.diff.gz John Hawkes hawkes@sgi.com From owner-lockmeter@oss.sgi.com Tue Nov 27 02:02:27 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.2/8.11.3) id fARA2R211880 for lockmeter-outgoing; Tue, 27 Nov 2001 02:02:27 -0800 Received: from four.malevolentminds.com (four.malevolentminds.com [216.177.76.238]) by oss.sgi.com (8.11.2/8.11.3) with SMTP id fARA2Po11877 for ; Tue, 27 Nov 2001 02:02:25 -0800 Received: from localhost (khyron@localhost) by four.malevolentminds.com (8.11.6/8.11.6) with ESMTP id fAR92ca68055 for ; Tue, 27 Nov 2001 09:02:38 GMT Date: Tue, 27 Nov 2001 09:02:38 +0000 (GMT) From: Khyron To: Subject: Lockmetering on uniprocessor kernels? Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-lockmeter@oss.sgi.com Precedence: bulk Is there any stated reason that lockmeters can't/shouldn't be used in uniproc kernels? Is this by design or coincidence? I'm curious. "Everyone's got a story to tell, and everyone's got some pain. And so do you. Do you think you are invisble? And everyone's got a story to sell, and everyone is strange. And so are you. Did you think you were invincible?" - "Invisible", Majik Alex From owner-lockmeter@oss.sgi.com Tue Nov 27 02:15:55 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.2/8.11.3) id fARAFtF12209 for lockmeter-outgoing; Tue, 27 Nov 2001 02:15:55 -0800 Received: from mercury.Sun.COM (mercury.Sun.COM [192.9.25.1]) by oss.sgi.com (8.11.2/8.11.3) with SMTP id fARAFro12206 for ; Tue, 27 Nov 2001 02:15:53 -0800 Received: from hs-eber02-01.Germany.Sun.COM ([129.157.155.3]) by mercury.Sun.COM (8.9.3+Sun/8.9.3) with ESMTP id BAA20003 for ; Tue, 27 Nov 2001 01:15:51 -0800 (PST) Received: from ragnaroek (ragnaroek [129.157.155.118]) by hs-eber02-01.Germany.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id KAA17089 for ; Tue, 27 Nov 2001 10:15:46 +0100 (MET) Message-Id: <200111270915.KAA17089@hs-eber02-01.Germany.Sun.COM> Date: Tue, 27 Nov 2001 10:12:30 +0100 (MET) From: Frank Hofmann - European Solaris CTE-Sustaining Engineering Reply-To: Frank Hofmann - European Solaris CTE-Sustaining Engineering Subject: Re: Lockmetering on uniprocessor kernels? To: lockmeter@oss.sgi.com MIME-Version: 1.0 Content-Type: TEXT/plain; charset=us-ascii Content-MD5: f1K4GgGM10Tzhz7x3Ve1bA== X-Mailer: dtmail 1.3.0 @(#)CDE Version 1.4 SunOS 5.8 sun4u sparc Sender: owner-lockmeter@oss.sgi.com Precedence: bulk > Is there any stated reason that lockmeters can't/shouldn't > be used in uniproc kernels? Is this by design or coincidence? > I'm curious. It makes no sense. The default Linux kernel (which the lockmeter patches are against) does not do kernel-level preemption. Hence, all locking ops on uniprocessor machines are guaranteed to succeed under Linux. So what would you want to measure ? Bye, Frank From owner-lockmeter@oss.sgi.com Tue Nov 27 15:33:47 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.2/8.11.3) id fARNXlG02585 for lockmeter-outgoing; Tue, 27 Nov 2001 15:33:47 -0800 Received: from four.malevolentminds.com (four.malevolentminds.com [216.177.76.238]) by oss.sgi.com (8.11.2/8.11.3) with SMTP id fARNXho02572 for ; Tue, 27 Nov 2001 15:33:43 -0800 Received: from localhost (khyron@localhost) by four.malevolentminds.com (8.11.6/8.11.6) with ESMTP id fARMXtX69649; Tue, 27 Nov 2001 22:33:56 GMT Date: Tue, 27 Nov 2001 22:33:55 +0000 (GMT) From: Khyron To: Frank Hofmann - European Solaris CTE-Sustaining Engineering cc: Subject: Re: Lockmetering on uniprocessor kernels? In-Reply-To: <200111270915.KAA17089@hs-eber02-01.Germany.Sun.COM> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-lockmeter@oss.sgi.com Precedence: bulk But couldn't lockmetering still indicate which locks are being held for how long? Or is there a better way to go about it on uniproc kernels? Or are uniproc locks that finely tuned that its irrelevant? As well, its not technically impossible tho, from your last message? If memory serves, you said... > > > Is there any stated reason that lockmeters can't/shouldn't > > be used in uniproc kernels? Is this by design or coincidence? > > I'm curious. > > It makes no sense. The default Linux kernel (which the lockmeter patches > are against) does not do kernel-level preemption. Hence, all locking ops > on uniprocessor machines are guaranteed to succeed under Linux. > So what would you want to measure ? > > Bye, > Frank > "Everyone's got a story to tell, and everyone's got some pain. And so do you. Do you think you are invisble? And everyone's got a story to sell, and everyone is strange. And so are you. Did you think you were invincible?" - "Invisible", Majik Alex From owner-lockmeter@oss.sgi.com Tue Nov 27 16:11:49 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.2/8.11.3) id fAS0Bnx03687 for lockmeter-outgoing; Tue, 27 Nov 2001 16:11:49 -0800 Received: from pneumatic-tube.sgi.com (pneumatic-tube.sgi.com [204.94.214.22]) by oss.sgi.com (8.11.2/8.11.3) with SMTP id fAS0Bko03684 for ; Tue, 27 Nov 2001 16:11:46 -0800 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by pneumatic-tube.sgi.com (980327.SGI.8.8.8-aspam/980310.SGI-aspam) via ESMTP id PAA00475 for ; Tue, 27 Nov 2001 15:11:40 -0800 (PST) mail_from (hawkes@sgi.com) Received: from wrlarun (sshgate.corp.sgi.com [169.238.216.146]) by cthulhu.engr.sgi.com (SGI-8.9.3/8.9.3) with SMTP id PAA17281; Tue, 27 Nov 2001 15:10:28 -0800 (PST) Message-ID: <005301c17798$5a504970$6801a8c0@wrlarun> From: "John Hawkes" To: "Khyron" , "Frank Hofmann - European Solaris CTE-Sustaining Engineering" Cc: References: Subject: Re: Lockmetering on uniprocessor kernels? Date: Tue, 27 Nov 2001 15:07:57 -0800 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2919.6600 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600 Sender: owner-lockmeter@oss.sgi.com Precedence: bulk From: "Khyron" > But couldn't lockmetering still indicate which locks are > being held for how long? Or is there a better way to go about > it on uniproc kernels? Or are uniproc locks that finely tuned > that its irrelevant? But what's the point of how long the lock is held? There isn't going to be contention. Only the BKL (kernel_flag) can be held across a context switch, and even for that one the scheduler drops the lock and reacquires it when that thread gets the CPU back again. And if an interrupt-level piece of code acquires a lock, that lock better not be something that's held by a non-interrupt-level piece of code. I believe the only thing that's interesting is how long interrupts are disabled. Feel free to hack uniprocessor spinlocks to invoke the Lockmeter routines, if you want to experiment with them. John Hawkes