From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964837AbWGMHvX (ORCPT ); Thu, 13 Jul 2006 03:51:23 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S964816AbWGMHvX (ORCPT ); Thu, 13 Jul 2006 03:51:23 -0400 Received: from miranda.se.axis.com ([193.13.178.8]:53951 "EHLO miranda.se.axis.com") by vger.kernel.org with ESMTP id S964837AbWGMHvW convert rfc822-to-8bit (ORCPT ); Thu, 13 Jul 2006 03:51:22 -0400 From: "Mikael Starvik" To: "'Arjan van de Ven'" , "Mikael Starvik" Cc: "'Linux Kernel Mailing List'" Subject: RE: Very long startup time for a new thread Date: Thu, 13 Jul 2006 09:51:09 +0200 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 8BIT X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook, Build 10.0.6626 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1506 In-Reply-To: Importance: Normal Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Nope, no NPTL. Actually this system uses the pthread implementation in uC-libc. -----Original Message----- From: Arjan van de Ven [mailto:arjan@infradead.org] Sent: Thursday, July 13, 2006 9:15 AM To: Mikael Starvik Cc: 'Linux Kernel Mailing List' Subject: Re: Very long startup time for a new thread On Thu, 2006-07-13 at 08:07 +0200, Mikael Starvik wrote: > (This is on a 200 MIPS embedded architecture). > > On a heavily loaded system (loadavg ~4) I create a new pthread. In this > situation it takes ~4 seconds (!) before the thread is first scheduled in > (yes, I have debug outputs in the scheduler to check that). In a 2.4 based > system I don't see the same thing. I don't have any RT or FIFO tasks. Any > ideas why it takes so long time and what I can do about it? I assume you're using a new enough glibc that supports NPTL ?