Attachment 144140 Details for Bug 217134 – Determine number of logical processors active

[patch] Determine number of logical processors active

oprofile-kernel-cpu_type.patch.eml (text/plain), 7.13 KB, created by William Cohen on 2006-12-20 21:17:04 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: William Cohen

Created: 2006-12-20 21:17:04 UTC

Size: 7.13 KB

patch

obsolete

>Return-Path: <oprofile-list-bounces@lists.sourceforge.net>
>Received: from pobox.corp.redhat.com ([unix socket])
>	 by pobox.corp.redhat.com (Cyrus v2.2.12-Invoca-RPM-2.2.12-3.RHEL4.1) with LMTPA;
>	 Tue, 19 Dec 2006 15:26:33 -0500
>X-Sieve: CMU Sieve 2.2
>Received: from int-mx1.corp.redhat.com (int-mx1.corp.redhat.com [172.16.52.254])
>	by pobox.corp.redhat.com (8.13.1/8.12.8) with ESMTP id kBJKQXEm019768
>	for <wcohen@pobox.corp.redhat.com>; Tue, 19 Dec 2006 15:26:33 -0500
>Received: from mx1.redhat.com (mx1.redhat.com [172.16.48.31])
>	by int-mx1.corp.redhat.com (8.13.1/8.13.1) with ESMTP id kBJKQWbn023020
>	for <wcohen@redhat.com>; Tue, 19 Dec 2006 15:26:32 -0500
>Received: from lists-outbound.sourceforge.net (lists-outbound.sourceforge.net [66.35.250.225])
>	by mx1.redhat.com (8.12.11.20060308/8.12.11) with ESMTP id kBJKQWVi003590
>	for <wcohen@redhat.com>; Tue, 19 Dec 2006 15:26:32 -0500
>Received: from sc8-sf-list2-new.sourceforge.net (sc8-sf-list2-new-b.sourceforge.net [10.3.1.94])
>	by sc8-sf-spam2.sourceforge.net (Postfix) with ESMTP
>	id 53FD0160B0; Tue, 19 Dec 2006 12:26:26 -0800 (PST)
>Received: from sc8-sf-mx1-b.sourceforge.net ([10.3.1.91]
>	helo=mail.sourceforge.net)
>	by sc8-sf-list2-new.sourceforge.net with esmtp (Exim 4.43)
>	id 1GwlXZ-0004L7-82
>	for oprofile-list@lists.sourceforge.net; Tue, 19 Dec 2006 12:26:21 -0800
>Received: from ccerelbas04.cce.hp.com ([161.114.21.107])
>	by mail.sourceforge.net with esmtps (TLSv1:AES256-SHA:256)
>	(Exim 4.44) id 1GwlXY-0001Ti-CY
>	for oprofile-list@lists.sourceforge.net; Tue, 19 Dec 2006 12:26:22 -0800
>Received: from cceexg11.americas.cpqcorp.net (cceexg11.americas.cpqcorp.net
>	[16.81.1.59])
>	by ccerelbas04.cce.hp.com (Postfix) with ESMTP id AA78134486;
>	Tue, 19 Dec 2006 14:26:14 -0600 (CST)
>Received: from G3W0072.americas.hpqcorp.net ([16.232.1.11]) by
>	cceexg11.americas.cpqcorp.net with Microsoft SMTPSVC(6.0.3790.1830); 
>	Tue, 19 Dec 2006 14:26:14 -0600
>X-MimeOLE: Produced By Microsoft Exchange V6.5
>Content-class: urn:content-classes:message
>MIME-Version: 1.0
>Subject: RE: How to make OProfile use cpu model P4 when HT is disabled by BIOS
>Date: Tue, 19 Dec 2006 14:26:12 -0600
>Message-ID: <08CA2245AFCF444DB3AC415E47CC40AF592261@G3W0072.americas.hpqcorp.net>
>In-Reply-To: <4587E880.7060607@nc.rr.com>
>X-MS-Has-Attach: 
>X-MS-TNEF-Correlator: 
>Thread-Topic: How to make OProfile use cpu model P4 when HT is disabled by BIOS
>Thread-Index: AccjcU8+bFm4gPZRQc2AOjvFILbV5wAOGMsg
>From: "Santos, Jose Renato G" <joserenato.santos@hp.com>
>To: "William Cohen" <wcohen@nc.rr.com>
>X-OriginalArrivalTime: 19 Dec 2006 20:26:14.0481 (UTC)
>	FILETIME=[EE201C10:01C723AB]
>X-Spam-Report: Spam Filtering performed by sourceforge.net.
>	See http://spamassassin.org/tag/ for more details.
>	Report problems to
>	http://sf.net/tracker/?func=add&group_id=1&atid=200001
>Cc: oprofile-list@lists.sourceforge.net
>X-BeenThere: oprofile-list@lists.sourceforge.net
>X-Mailman-Version: 2.1.8
>Precedence: list
>List-Id: "For discussion of the profiler oprofile."
>	<oprofile-list.lists.sourceforge.net>
>List-Unsubscribe: <https://lists.sourceforge.net/lists/listinfo/oprofile-list>, 
>	<mailto:oprofile-list-request@lists.sourceforge.net?subject=unsubscribe>
>List-Archive: <http://sourceforge.net/mailarchive/forum.php?forum=oprofile-list>
>List-Post: <mailto:oprofile-list@lists.sourceforge.net>
>List-Help: <mailto:oprofile-list-request@lists.sourceforge.net?subject=help>
>List-Subscribe: <https://lists.sourceforge.net/lists/listinfo/oprofile-list>, 
>	<mailto:oprofile-list-request@lists.sourceforge.net?subject=subscribe>
>Content-Type: text/plain; charset="us-ascii"
>Content-Transfer-Encoding: 7bit
>Sender: oprofile-list-bounces@lists.sourceforge.net
>Errors-To: oprofile-list-bounces@lists.sourceforge.net
>X-RedHat-Spam-Score: -100 
>
>Will,
>
>Thanks for your message.
>The article you mentioned was usefull.
>I believe the method described in the article is used
>in the linux kernel to create processor masks:
>cpu_sibling_mask and cpu_core_mask.
>cpu_sibling_map is a mask for logical processors
>sharing the same core, and smp_core_mask for logical
>processors sharing the same physical cpu (socket)
>
>Using cpu_sibling_mask instead of smp_num_siblings,
>the following oprofile patch correctly detects if 
>HT is enabled or not.
>My concern is that with hot plug CPU support the cpu masks
>can change and oprofile can misbehave if there is a change
>sometime between initialization and profiling
>Any suggestion on how to deal with this? Can we ignore
>CPU configuration changes? If not can oprofile deal with
>changing cpu_type?
>
>Thanks
>
>Renato
>
>==============================================
>Signed-off-by: Jose Renato Santos <jsantos@hpl.hp.com>
>
>diff -pruN linux-2.6.19.1/arch/i386/oprofile/nmi_int.c
>linux-2.6.19.1-new/arch/i386/oprofile/nmi_int.c
>--- linux-2.6.19.1/arch/i386/oprofile/nmi_int.c	2006-12-11
>11:32:53.000000000 -0800
>+++ linux-2.6.19.1-new/arch/i386/oprofile/nmi_int.c	2006-12-19
>11:31:04.000000000 -0800
>@@ -342,7 +342,8 @@ static int __init p4_init(char ** cpu_ty
> 	model = &op_p4_spec;
> 	return 1;
> #else
>-	switch (smp_num_siblings) {
>+	/* assume sibling configuration is uniform across all CPUs */
>+	switch (cpus_weight(cpu_sibling_map[0])) {
> 		case 1:
> 			*cpu_type = "i386/p4";
> 			model = &op_p4_spec;
>
>
>> -----Original Message-----
>> From: William Cohen [mailto:wcohen@nc.rr.com] 
>> Sent: Tuesday, December 19, 2006 5:26 AM
>> To: Santos, Jose Renato G
>> Cc: oprofile-list@lists.sourceforge.net
>> Subject: Re: How to make OProfile use cpu model P4 when HT is 
>> disabled by BIOS
>> 
>> Santos, Jose Renato G wrote:
>> > Oprofile assumes cpu type "p4-ht" even when hyperthreading 
>> is disabled 
>> > by the BIOS. Basically oprofile chooses ht mode if 
>> smp_num_siblings > 
>> > 1. Linux sets this variable based on info returned by cpuid that 
>> > informs the maximum number of logical processors per physical CPU. 
>> > This is always 2 on an HT capable CPU, regardeless if HT is 
>> disabled 
>> > by the BIOS or not (at least on my machine).
>> > It would be nice if we detected that hyperthreading is disabled and 
>> > used the richer set of events available for the non-ht 
>> version of the 
>> > "p4" model.
>> > Any suggestions on how we could do that?
>> > 
>> > Thanks
>> > 
>> > Renato
>> 
>> At one time OProfile detected whether hyper-threading was 
>> actually in use. I have seen it both in p4 and p4-ht on the 
>> same processor. I have not tried it lately and there could be 
>> problems with it. The code can be confused by multi-core 
>> processors. The code needs to be updated to make a 
>> distinction between a sibling that is a separate core and a 
>> sibling that is a logical processor. The following article 
>> describes methods to accomplish this:
>> 
>> http://www3.intel.com/cd/ids/developer/asmo-na/eng/275339.htm
>> 
>> 
>> -Will
>> 
>
>-------------------------------------------------------------------------
>Take Surveys. Earn Cash. Influence the Future of IT
>Join SourceForge.net's Techsay panel and you'll get the chance to share your
>opinions on IT & business topics through brief surveys - and earn cash
>http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
>_______________________________________________
>oprofile-list mailing list
>oprofile-list@lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/oprofile-list

Actions: View | Diff

Attachments on bug 217134: 144140 | 226031