Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264136AbTH2CPj (ORCPT ); Thu, 28 Aug 2003 22:15:39 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S264297AbTH2CPj (ORCPT ); Thu, 28 Aug 2003 22:15:39 -0400 Received: from tmr-02.dsl.thebiz.net ([216.238.38.204]:30985 "EHLO gatekeeper.tmr.com") by vger.kernel.org with ESMTP id S264136AbTH2CPa (ORCPT ); Thu, 28 Aug 2003 22:15:30 -0400 Date: Thu, 28 Aug 2003 22:06:42 -0400 (EDT) From: Bill Davidsen To: Andrew Morton cc: Christopher Swingley , linux-kernel@vger.kernel.org Subject: Re: 2.6.0-test4: Unable to handle kernel NULL pointer dereference In-Reply-To: <20030828131019.69a9f3b9.akpm@osdl.org> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1979 Lines: 55 On Thu, 28 Aug 2003, Andrew Morton wrote: > Christopher Swingley wrote: > > > > kernel: Unable to handle kernel NULL pointer dereference at virtual address 00000000 > > ... > > kernel: EIP is at find_inode_fast+0x1a/0x60 > > ... > > model name : AMD Athlon(tm) XP 1800+ > > ... > > You've been bitten by the athlon-prefetch(0)-goes-oops problem. > > Nobody seems to be working this, so I'll be sending the below in to Linus. Let's not. If this goes in the Athlon users will get bad performance, reliable operation, and no one will blame anything but the Athlon... If it keeps oopsing hopefully people will complain and it will get fixed. Taking out features instead of fixing them is a bad president, this is not a driver for an obsolete ISA card, this is a performance boost for a very current CPU. I'm surprised AMD hasn't looked at it themselves. > > --- 25/include/asm-i386/processor.h~disable-athlon-prefetch 2003-08-23 13:48:16.000000000 -0700 > +++ 25-akpm/include/asm-i386/processor.h 2003-08-23 13:48:16.000000000 -0700 > @@ -578,6 +578,8 @@ static inline void rep_nop(void) > #define ARCH_HAS_PREFETCH > extern inline void prefetch(const void *x) > { > + if (cpu_data[0].x86_vendor == X86_VENDOR_AMD) > + return; > alternative_input(ASM_NOP4, > "prefetchnta (%1)", > X86_FEATURE_XMM, > > _ > > - > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > Please read the FAQ at http://www.tux.org/lkml/ > -- bill davidsen CTO, TMR Associates, Inc Doing interesting things with little computers since 1979. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/