From owner-freebsd-smp Thu Apr 2 15:15:54 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id PAA27408 for freebsd-smp-outgoing; Thu, 2 Apr 1998 15:15:54 -0800 (PST) (envelope-from owner-freebsd-smp@FreeBSD.ORG) Received: from yalta.NL.net (yalta.NL.net [193.78.240.21]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id PAA27399 for ; Thu, 2 Apr 1998 15:15:47 -0800 (PST) (envelope-from benst@terminus.stuyts.nl) Received: from alushta.NL.net ([193.78.240.22]:29961 "EHLO alushta.NL.net" ident: "IDENT-NOT-QUERIED") by yalta.NL.net with ESMTP id <24668-17222>; Fri, 3 Apr 1998 01:15:32 +0200 Received: from stuyts by alushta.NL.net with UUCP id <25549-30619>; Fri, 3 Apr 1998 01:14:01 +0200 Received: from daneel.stuyts.nl (daneel.stuyts.nl [193.78.231.7]) by terminus.stuyts.nl (8.8.8/8.8.8) with ESMTP id BAA13015; Fri, 3 Apr 1998 01:02:42 +0200 (MET DST) (envelope-from benst) Received: (from benst@localhost) by daneel.stuyts.nl (8.8.5/8.8.5) id BAA07217; Fri, 3 Apr 1998 01:02:38 +0200 (MET DST) Message-Id: <199804022302.BAA07217@daneel.stuyts.nl> Content-Type: text/plain MIME-Version: 1.0 (NeXT Mail 3.3 v118.2) In-Reply-To: <199803302332.BAA12765@pat.idi.ntnu.no> X-Nextstep-Mailer: Mail 3.3 (Enhance 1.2) Received: by NeXT.Mailer (1.118.2) From: Ben Stuyts Date: Fri, 3 Apr 98 01:02:36 +0200 To: Tor Egge Subject: Re: Log messages from latest SMP kernel cc: smp@FreeBSD.ORG Reply-To: ben@stuyts.nl References: <199803302042.WAA05032@daneel.stuyts.nl> <199803302332.BAA12765@pat.idi.ntnu.no> Sender: owner-freebsd-smp@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Hi Tor, After a few days of testing, your patch seems to have solved the problem with forward_hardclock here. Looks good. Will it be committed? (However, the -unrelated- interrupt-level buffer overflows are still there. Anybody know what causes this?) Thanks, Ben On Tue, 31 Mar 1998, Tor Egge wrote: > > Running a kernel from about March 27, I get a lot of log messages like > this: > > > > sio2: 3 more interrupt-level buffer overflows (total 69) > > > forward_hardclock: checkstate 0 > > > forward_hardclock: checkstate 0 > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > > There are two explanations for this error message > > 1. Failed IPI delivery. > > - Interrupts are disabled on the other CPU > > - The local apic TPR register value on the other CPU is too high, > blocking the IPI > > 2. Successful IPI delivery not detected > > - Incorrect declaration of some variables (missing volatile decl.). > > > If the second explanation is correct for your case, the following patch > might help you. > > Index: sys/i386/include/smp.h > =================================================================== > RCS file: /home/ncvs/src/sys/i386/include/smp.h,v > retrieving revision 1.40 > diff -u -r1.40 smp.h > --- smp.h 1998/03/07 21:34:59 1.40 > +++ smp.h 1998/03/30 22:22:23 > @@ -78,8 +78,8 @@ > extern volatile u_int stopped_cpus; > extern volatile u_int started_cpus; > > -extern unsigned int checkstate_probed_cpus; > -extern unsigned int checkstate_need_ast; > +extern volatile u_int checkstate_probed_cpus; > +extern volatile u_int checkstate_need_ast; > > /* functions in apic_ipl.s */ > void apic_eoi __P((void)); > > > - Tor Egge > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-smp" in the body of the message To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message