From owner-freebsd-alpha Sun Nov 10 3:10:51 2002 Delivered-To: freebsd-alpha@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id CF75237B401 for ; Sun, 10 Nov 2002 03:10:48 -0800 (PST) Received: from herring.nlsystems.com (mailgate.nlsystems.com [62.49.251.130]) by mx1.FreeBSD.org (Postfix) with ESMTP id BAA3143E42 for ; Sun, 10 Nov 2002 03:10:46 -0800 (PST) (envelope-from dfr@nlsystems.com) Received: from herring.nlsystems.com (herring [10.0.0.2]) by herring.nlsystems.com (8.12.6/8.12.6) with ESMTP id gAABAYVu040399; Sun, 10 Nov 2002 11:10:34 GMT (envelope-from dfr@nlsystems.com) Content-Type: text/plain; charset="iso-8859-1" From: Doug Rabson To: "Alan L. Cox" , Jeff Roberson Subject: Re: alpha: top of tree kernel blooie Date: Sun, 10 Nov 2002 11:10:34 +0000 User-Agent: KMail/1.4.3 Cc: Matthew Jacob , Andrew Gallatin , alpha@FreeBSD.ORG, alc@FreeBSD.ORG, John Baldwin References: <20021109150359.B97372-100000@mail.chesapeake.net> <3DCD7244.DEEA7387@imimic.com> In-Reply-To: <3DCD7244.DEEA7387@imimic.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Message-Id: <200211101110.34153.dfr@nlsystems.com> X-Spam-Status: No, hits=-7.7 required=5.0 tests=IN_REP_TO,QUOTED_EMAIL_TEXT,REFERENCES,SPAM_PHRASE_00_01, USER_AGENT,USER_AGENT_KMAIL version=2.41 Sender: owner-freebsd-alpha@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org On Saturday 09 November 2002 8:38 pm, Alan L. Cox wrote: > Jeff Roberson wrote: > > On Fri, 8 Nov 2002, Matthew Jacob wrote: > > > > Can you 'ps aux -o wchan' ? > > > > > > I tried a buildworld -j4 again- this time it was too far hung to > > > let anyone log in... > > > > I ran into the vm_map_delete() duplicate free again. This has been > > the same code path too many times for it to be a non logic bug. > > > > Jeff > > I'm still wondering if this isn't a false positive: > > 1. My reading of the UMA debug code is that two or more processors > can simultaneously read and modify us_freelist[]. The only lock held > during an access is the CPU private lock. > > 2. Don't we compile by default for the older Alphas that lack byte > manipulation instructions? Thus, a byte store is implemented by a > read-modify-write sequence of instructions. Thus, two simultaneous > uma_dbg_alloc()s on adjacent locations in us_freelist could cause > corruption. It was exactly this kind of breakage that prompted me to write the=20 atomic_* functions in the first place. Note that it is often possible=20 to get corruption even on a UP machine if an interrupt happens mid=20 sequence. --=20 Doug Rabson=09=09=09=09Mail: dfr@nlsystems.com =09=09=09=09=09Phone: +44 20 8348 6160 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-alpha" in the body of the message