From owner-svn-src-all@FreeBSD.ORG Thu Mar 31 13:59:48 2011 Return-Path: Delivered-To: svn-src-all@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 4A5D2106566B; Thu, 31 Mar 2011 13:59:48 +0000 (UTC) (envelope-from fabient@FreeBSD.org) Received: from svn.freebsd.org (svn.freebsd.org [IPv6:2001:4f8:fff6::2c]) by mx1.freebsd.org (Postfix) with ESMTP id 3AC1D8FC16; Thu, 31 Mar 2011 13:59:48 +0000 (UTC) Received: from svn.freebsd.org (localhost [127.0.0.1]) by svn.freebsd.org (8.14.3/8.14.3) with ESMTP id p2VDxm7P008486; Thu, 31 Mar 2011 13:59:48 GMT (envelope-from fabient@svn.freebsd.org) Received: (from fabient@localhost) by svn.freebsd.org (8.14.3/8.14.3/Submit) id p2VDxmOI008483; Thu, 31 Mar 2011 13:59:48 GMT (envelope-from fabient@svn.freebsd.org) Message-Id: <201103311359.p2VDxmOI008483@svn.freebsd.org> From: Fabien Thomas Date: Thu, 31 Mar 2011 13:59:48 +0000 (UTC) To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org X-SVN-Group: head MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Cc: Subject: svn commit: r220198 - head/sys/kern X-BeenThere: svn-src-all@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: "SVN commit messages for the entire src tree \(except for " user" and " projects" \)" List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 31 Mar 2011 13:59:48 -0000 Author: fabient Date: Thu Mar 31 13:59:47 2011 New Revision: 220198 URL: http://svn.freebsd.org/changeset/base/220198 Log: Clearing the flag when preempting will let the preempted thread run too much time. This can finish in a scheduler deadlock with ping-pong between two threads. One sample of this is: - device lapic (to have a preemption point on critical_exit()) - options DEVICE_POLLING with HZ>1499 (to have lapic freq = hardclock freq) - running a cpu intensive task (that does not enter the kernel) - only one CPU on SMP or no SMP. As requested by jhb@ 4BSD have received the same type of fix instead of propagating the flag to the new thread. Reviewed by: jhb, jeff MFC after: 1 month Modified: head/sys/kern/sched_4bsd.c head/sys/kern/sched_ule.c Modified: head/sys/kern/sched_4bsd.c ============================================================================== --- head/sys/kern/sched_4bsd.c Thu Mar 31 13:36:31 2011 (r220197) +++ head/sys/kern/sched_4bsd.c Thu Mar 31 13:59:47 2011 (r220198) @@ -940,13 +940,9 @@ sched_switch(struct thread *td, struct t if ((td->td_flags & TDF_NOLOAD) == 0) sched_load_rem(); - if (newtd) { - MPASS(newtd->td_lock == &sched_lock); - newtd->td_flags |= (td->td_flags & TDF_NEEDRESCHED); - } - td->td_lastcpu = td->td_oncpu; - td->td_flags &= ~TDF_NEEDRESCHED; + if (!(flags & SW_PREEMPT)) + td->td_flags &= ~TDF_NEEDRESCHED; td->td_owepreempt = 0; td->td_oncpu = NOCPU; Modified: head/sys/kern/sched_ule.c ============================================================================== --- head/sys/kern/sched_ule.c Thu Mar 31 13:36:31 2011 (r220197) +++ head/sys/kern/sched_ule.c Thu Mar 31 13:59:47 2011 (r220198) @@ -1783,7 +1783,8 @@ sched_switch(struct thread *td, struct t ts->ts_rltick = ticks; td->td_lastcpu = td->td_oncpu; td->td_oncpu = NOCPU; - td->td_flags &= ~TDF_NEEDRESCHED; + if (!(flags & SW_PREEMPT)) + td->td_flags &= ~TDF_NEEDRESCHED; td->td_owepreempt = 0; tdq->tdq_switchcnt++; /*