From owner-freebsd-sparc64@FreeBSD.ORG Wed Jan 12 23:41:09 2005 Return-Path: Delivered-To: freebsd-sparc64@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id AF3B616A4CE for ; Wed, 12 Jan 2005 23:41:09 +0000 (GMT) Received: from obsecurity.dyndns.org (CPE0050040655c8-CM00111ae02aac.cpe.net.cable.rogers.com [69.199.47.57]) by mx1.FreeBSD.org (Postfix) with ESMTP id 5902443D45 for ; Wed, 12 Jan 2005 23:41:09 +0000 (GMT) (envelope-from kris@obsecurity.org) Received: by obsecurity.dyndns.org (Postfix, from userid 1000) id 4808D51274; Wed, 12 Jan 2005 15:41:08 -0800 (PST) Date: Wed, 12 Jan 2005 15:41:08 -0800 From: Kris Kennaway To: Kris Kennaway Message-ID: <20050112234108.GA17895@xor.obsecurity.org> References: <20050103104025.A6665@carver.gumbysoft.com> <20050104014112.23160.qmail@web17309.mail.tpe.yahoo.com> <6.2.0.14.2.20050104102555.049faa18@white.dogwood.com> <20050104222258.GB79661@xor.obsecurity.org> <6.2.0.14.2.20050104155036.04b3bd68@white.dogwood.com> <20050105055845.GA92720@xor.obsecurity.org> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="qDbXVdCdHGoSgWSk" Content-Disposition: inline In-Reply-To: <20050105055845.GA92720@xor.obsecurity.org> User-Agent: Mutt/1.4.2.1i cc: freebsd-sparc@freebsd.org Subject: Re: Fast Data Access MMU Miss problem X-BeenThere: freebsd-sparc64@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Porting FreeBSD to the Sparc List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 12 Jan 2005 23:41:09 -0000 --qDbXVdCdHGoSgWSk Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Tue, Jan 04, 2005 at 09:58:45PM -0800, Kris Kennaway wrote: > On Tue, Jan 04, 2005 at 04:00:20PM -1000, David Cornejo wrote: > > You mean I have to do something besides gripe? :-) > >=20 > > I have pasted in below the results of three crashes - the first two are= =20 > > pretty interesting in that they happen in _mtx_lock_sleep() when it=20 > > recurses. The third blows away the theory that it's that routine. The= =20 > > common thing seems to be locking, but that could just be coincidence. > >=20 > > I suppose I could try gdb on the kernel. I only have one sparc machine= , I=20 > > presume an x86 gdb won't work on this, is there anyway to get a sparc g= db=20 > > built under x86? I know gdb 6 seems to be messed up when looking at th= e=20 > > crash dumps, should I try 5.3 on the live kernel also? >=20 > I've not had luck using anything other than the gdb53 port which > you're using, but this is fine. Actually, looking at your traces I've > seen the first two myself - there seems to be a problem with file > descriptor reference counting in RELENG_5 and HEAD, and some other > memory use-after-free bugs that I've not had any luck getting tracked > down so far. FYI, bmilekic has a memory debugging patch that should help us track these down (in fact it's already caught one problem condition, which I'm waiting to hear back from him for analysis). Stay tuned. Kris --qDbXVdCdHGoSgWSk Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (FreeBSD) iD8DBQFB5bWTWry0BWjoQKURAgV6AKCVrWn+p3Qz1eE/Kz9ADBRCPFaZcACgmrpO 2LhbhJQOBA/IYVxNmAXPoYQ= =E4Tb -----END PGP SIGNATURE----- --qDbXVdCdHGoSgWSk--