From owner-lockmeter@oss.sgi.com Fri Mar 9 16:25:43 2001 Received: by oss.sgi.com id ; Fri, 9 Mar 2001 16:25:24 -0800 Received: from deliverator.sgi.com ([204.94.214.10]:33132 "EHLO deliverator.sgi.com") by oss.sgi.com with ESMTP id ; Fri, 9 Mar 2001 16:25:12 -0800 Received: from cthulhu.engr.sgi.com (gate3-relay.engr.sgi.com [130.62.1.234]) by deliverator.sgi.com (980309.SGI.8.8.8-aspam-6.2/980310.SGI-aspam) via ESMTP id QAA11701 for ; Fri, 9 Mar 2001 16:24:04 -0800 (PST) mail_from (hawkes@engr.sgi.com) Received: from pchawkes (sshgate.corp.sgi.com [169.238.216.146]) by cthulhu.engr.sgi.com (SGI-8.9.3/8.9.3) with SMTP id QAA83118 for ; Fri, 9 Mar 2001 16:23:54 -0800 (PST) Message-ID: <00a901c0a8f8$51405420$6701a8c0@marin1.sfba.home.com> From: "John Hawkes" To: Subject: new lockstat cmd for linux lockmeter'ed kernels Date: Fri, 9 Mar 2001 16:23:22 -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 5.50.4522.1200 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200 Sender: owner-lockmeter@oss.sgi.com Precedence: bulk Return-Path: X-Orcpt: rfc822;lockmeter-outgoing FYI, a new lockstat command, version 1.4.4, is now available on http://oss.sgi.com/projects/lockmeter/download, which fixes a bug in the printed output -- the mean "hold-time" values were occasionally misreported (typically, under-reported). The latest version of lockmeter is a patch against the 2.4.1 kernel. This version should also patch cleanly against 2.4.2. John Hawkes hawkes@engr.sgi.com From owner-lockmeter@oss.sgi.com Fri Mar 16 17:00:19 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f2H10Jp19038 for lockmeter-outgoing; Fri, 16 Mar 2001 17:00:19 -0800 Received: from pneumatic-tube.sgi.com (pneumatic-tube.sgi.com [204.94.214.22]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f2H10IM19035 for ; Fri, 16 Mar 2001 17:00:18 -0800 Received: from cthulhu.engr.sgi.com (gate3-relay.engr.sgi.com [130.62.1.234]) by pneumatic-tube.sgi.com (980327.SGI.8.8.8-aspam/980310.SGI-aspam) via ESMTP id RAA05384 for ; Fri, 16 Mar 2001 17:10:12 -0800 (PST) mail_from (hawkes@engr.sgi.com) Received: from pchawkes (sshgate.corp.sgi.com [169.238.216.146]) by cthulhu.engr.sgi.com (SGI-8.9.3/8.9.3) with SMTP id QAA95972 for ; Fri, 16 Mar 2001 16:59:01 -0800 (PST) Message-ID: <021101c0ae7d$6d2bb5a0$6701a8c0@marin1.sfba.home.com> From: "John Hawkes" To: Subject: Lockmeter: new 'lockstat' command available Date: Fri, 16 Mar 2001 16:58:48 -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 5.50.4522.1200 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200 Sender: owner-lockmeter@oss.sgi.com Precedence: bulk FYI, a new 'lockstat' command tarball is available: http://oss.sgi.com/projects/lockmeter/download/lockstat-1.4.5.tar.gz Some differences from the previous version: 1) Many output fields are narrower, e.g., using fewer digits to the right of a decimal point. 2) The TOTAL lock usage field remains as an integer count, but the NOWAIT, SPIN, and REJECT fields are now percentages of the TOTAL, rather than simple counters. Not only does this consume fewer output columns, but I believe it improves comprehension when the counts are high (since the numeric outputs weren't pretty-printed with commas). 3) A new WAIT field is added to the existing MEAN and MAX hold times: "%CPU". This value reports the percentage of available CPU cycles that are being consumed by waiting on a lock. This turns out to be a very useful way to identify important hot locks. The CON (CONtention) field identifies the percentage of time that an attempt to acquire a lock finds the lock already owned. While this CON value is interesting, you previously had to multiply the SPIN value (the number of actual WAITs) by the mean wait-time, then divide by the length of the sampling time and divide by the number of CPUs, in order to understand exactly how detrimental the contention was. Now this calculation is done for you. And, as previously stated, the Lockmeter kernel patch is available against 2.4.2, 2.4.1, and 2.4.0 sources. Enjoy! John Hawkes hawkes@engr.sgi.com From owner-lockmeter@oss.sgi.com Sun Mar 18 14:44:00 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f2IMi0L18787 for lockmeter-outgoing; Sun, 18 Mar 2001 14:44:00 -0800 Received: from pneumatic-tube.sgi.com (pneumatic-tube.sgi.com [204.94.214.22]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f2IMhxM18784 for ; Sun, 18 Mar 2001 14:43:59 -0800 Received: from babylon.engr.sgi.com (babylon.engr.sgi.com [163.154.10.144]) by pneumatic-tube.sgi.com (980327.SGI.8.8.8-aspam/980310.SGI-aspam) via ESMTP id OAA05148 for ; Sun, 18 Mar 2001 14:53:56 -0800 (PST) mail_from (hawkes@babylon.engr.sgi.com) Received: (from hawkes@localhost) by babylon.engr.sgi.com (SGI-8.9.3/8.9.3) id OAA93390 for lockmeter@oss.sgi.com; Sun, 18 Mar 2001 14:40:23 -0800 (PST) Date: Sun, 18 Mar 2001 14:40:23 -0800 (PST) From: John Hawkes Message-Id: <200103182240.OAA93390@babylon.engr.sgi.com> To: lockmeter@oss.sgi.com Subject: Lockmeter: more 'lockstat' fixes Sender: owner-lockmeter@oss.sgi.com Precedence: bulk FYI, a new 'lockstat' command, version 1.4.6, is now available at: http://oss.sgi.com/projects/lockmeter/download that contains a few more fixes to the reporting of write-locks (the "write" side of rwlock_t). Earlier versions of lockstat misreported the wait-time data for the write_lock behavior of rwlock_t's -- miscalculating the mean wait-times, and miscalculating the percentage of waits relative to the total number of write-lock requests. Lockstat tries to be clever and to differentiate waiting because a lock is owned by another write_lock() owner, vs. waiting because a lock is owned by a write_lock() or a read_lock() owner, and this clever calculation wasn't clever enough to be correct. Some other small fixes were done to clean up the lockstat output when using "-c" (to see cpu-specific lock statistics). Specifically, some NaN values were being calculated (using a divisor of zero) and printed. Spinlock_t reporting is unchanged, and read_lock() for rwlock_t is unchanged. Those appear to be correct. John Hawkes hawkes@engr.sgi.com From owner-lockmeter@oss.sgi.com Fri Mar 23 17:07:09 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f2O179c15520 for lockmeter-outgoing; Fri, 23 Mar 2001 17:07:09 -0800 Received: from yog-sothoth.sgi.com (eugate.sgi.com [192.48.160.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f2O178M15517 for ; Fri, 23 Mar 2001 17:07:08 -0800 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by yog-sothoth.sgi.com (980305.SGI.8.8.8-aspam-6.2/980304.SGI-aspam-europe) via ESMTP id CAA6133006 for ; Sat, 24 Mar 2001 02:07:06 +0100 (CET) mail_from (hawkes@engr.sgi.com) Received: from pchawkes (sshgate.corp.sgi.com [169.238.216.146]) by cthulhu.engr.sgi.com (SGI-8.9.3/8.9.3) with SMTP id RAA06187 for ; Fri, 23 Mar 2001 17:05:48 -0800 (PST) Message-ID: <002101c0b3fe$786194e0$6501a8c0@marin1.sfba.home.com> From: "John Hawkes" To: Subject: New Lockmeter/Lockstat available Date: Fri, 23 Mar 2001 17:05:07 -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 5.50.4522.1200 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200 Sender: owner-lockmeter@oss.sgi.com Precedence: bulk Content-Length: 815 Lines: 20 A new Lockmeter patch (lockmeter1.4.8-2.4.2.diff.gz, i.e., version 1.4.8 which patches Linux kernel v2.4.2) and Lockstat command (lockstat-1.4.7.tar.gz) are now available at: http://oss.sgi.com/projects/lockmeter/download This new version of lockmeter fixes a small timing bug that occasionally produced a read-lock max-busy value that was absurdly high. This new version of lockstat fixes an occasional bogus lock name (or procedure name). If you haven't downloaded a new lockstat command in the last week or two, you ought to download this version. There have been recent improvements in the lockstat output -- to use slightly fewer columns, and to display a new field "% CPU" that tells you the percentage of total available CPU cycles that were consumed by a lock wait. John Hawkes hawkes@engr.sgi.com From owner-lockmeter@oss.sgi.com Fri Mar 23 17:33:35 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f2O1XZ916269 for lockmeter-outgoing; Fri, 23 Mar 2001 17:33:35 -0800 Received: from vortex.undoo.com ([209.19.65.227]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f2O1XZM16266 for ; Fri, 23 Mar 2001 17:33:35 -0800 Received: from undoo.com (coral.undoo.com [192.168.100.43]) by vortex.undoo.com (8.11.3/8.11.3) with ESMTP id f2O1XUi10193 for ; Fri, 23 Mar 2001 17:33:30 -0800 (PST) Message-ID: <3ABBF969.AE4728F0@undoo.com> Date: Fri, 23 Mar 2001 17:33:29 -0800 From: Claude Johnson X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.15 i686) X-Accept-Language: en MIME-Version: 1.0 To: lockmeter@oss.sgi.com Subject: lockmeter on 2.2.x Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-lockmeter@oss.sgi.com Precedence: bulk Content-Length: 576 Lines: 17 I see that the most recent version is for 2.2.14. However, when I patch a stock 2.2.14 kernel, I don't get a "make config" or "make oldconfig" option to enable lock metering. I just looked at the presentation slides at http://oss.sgi.com/projects/lockmeter/als2000/Slide2.html. Would I be incorrect in saying that attempting to apply lockmetering on a UP kernel will fail? -- Claude Johnson claude@undoo.com Undoo Technologies 949.743.5145 (D) *undoo everything* 301.806.7399 (M) Key fingerprint = F00A 34E2 C141 F29D 7E41 6DF6 DB04 0769 592D 14AD From owner-lockmeter@oss.sgi.com Fri Mar 23 17:47:37 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f2O1lbN16490 for lockmeter-outgoing; Fri, 23 Mar 2001 17:47:37 -0800 Received: from yog-sothoth.sgi.com (eugate.sgi.com [192.48.160.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f2O1lRM16487 for ; Fri, 23 Mar 2001 17:47:27 -0800 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by yog-sothoth.sgi.com (980305.SGI.8.8.8-aspam-6.2/980304.SGI-aspam-europe) via ESMTP id CAA6340658 for ; Sat, 24 Mar 2001 02:47:25 +0100 (CET) mail_from (hawkes@engr.sgi.com) Received: from pchawkes (sshgate.corp.sgi.com [169.238.216.146]) by cthulhu.engr.sgi.com (SGI-8.9.3/8.9.3) with SMTP id RAA89015; Fri, 23 Mar 2001 17:46:07 -0800 (PST) Message-ID: <003801c0b404$1492ae80$6501a8c0@marin1.sfba.home.com> From: "John Hawkes" To: "Claude Johnson" , References: <3ABBF969.AE4728F0@undoo.com> Subject: Re: lockmeter on 2.2.x Date: Fri, 23 Mar 2001 17:45:17 -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 5.50.4522.1200 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200 Sender: owner-lockmeter@oss.sgi.com Precedence: bulk Content-Length: 1196 Lines: 30 From: "Claude Johnson" > I see that the most recent version is for 2.2.14. The most recent version of Lockmeter is v1.4.8, which patches the 2.4.2 kernel. While it is true that http://oss.sgi.com/projects/lockmeter/download contains a version of Lockmeter that patches the 2.2.14 kernel, please understand that that Lockmeter is *very* old and bears little resemblence to the Lockmeters in 2.4.x. > Would I be incorrect in saying that attempting to apply > lockmetering on a UP kernel will fail? Lockmeter only makes sense in an SMP kernel. A UP kernel implements locks by disabling/enabling interrupts, and Lockmeter doesn't instrument this form of locking. The "make config" only shows you the Lockmeter option if you have previously specified that you are building an SMP kernel. My rationale for essentially abandoning Lockmeter for 2.2.x is an assumption that the vast majority of developers who are serious about analyzing spinlock behavior in SMP kernels will be using a 2.4 (or 2.5) kernel. Of course, if anyone wishes to back-port Lockmeter v1.4.8 against a 2.2.x kernel, I'd be happy to put the patch on oss.sgi.com. John Hawkes hawkes@engr.sgi.com From owner-lockmeter@oss.sgi.com Mon Mar 26 04:01:53 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f2QC1rc18604 for lockmeter-outgoing; Mon, 26 Mar 2001 04:01:53 -0800 Received: from giasmd01.vsnl.net.in (giasmd01.vsnl.net.in [202.54.6.1]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f2QC1oM18591 for ; Mon, 26 Mar 2001 04:01:52 -0800 Received: from hotpop.com (unknown [203.197.130.99]) by giasmd01.vsnl.net.in (Postfix) with SMTP id 15CB3D6FB for ; Mon, 26 Mar 2001 17:41:45 +0530 (IST) From: "Murali Kumar" To: Subject: New LINUX Web Site Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1" Date: Mon, 26 Mar 2001 17:35:20 +0530 Content-Transfer-Encoding: 8bit Message-Id: <20010326121145.15CB3D6FB@giasmd01.vsnl.net.in> Sender: owner-lockmeter@oss.sgi.com Precedence: bulk Content-Length: 579 Lines: 23 Dear Sir, We have launched Linux Section http://www.softlandindia.com/Linux/ on March 12, 2001. India's Website for Linux OS, Freeware and Shareware Downloads. All the Best Softwares are listed and organized in Categories with the latest versions. New Softwares are Added Every Day. Please Visit us. http://www.softlandindia.com/Linux/ Software Submit Format can be had from http://www.softlandindia.com/Linux/Submit.htm Sincerely yours, K.Murali Kumar Webmaster www.softlandindia.com Email Ids: linux.softlandindia@hotpop.com and linuxsoftlandindia@hotpop.com