From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964949AbVHaW1l (ORCPT ); Wed, 31 Aug 2005 18:27:41 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S964962AbVHaW1l (ORCPT ); Wed, 31 Aug 2005 18:27:41 -0400 Received: from dsl027-180-168.sfo1.dsl.speakeasy.net ([216.27.180.168]:31137 "EHLO sunset.davemloft.net") by vger.kernel.org with ESMTP id S964949AbVHaW1l (ORCPT ); Wed, 31 Aug 2005 18:27:41 -0400 Date: Wed, 31 Aug 2005 15:27:37 -0700 (PDT) Message-Id: <20050831.152737.107306011.davem@davemloft.net> To: jkenisto@us.ibm.com Cc: linux-kernel@vger.kernel.org, akpm@osdl.org, ppancham@in.ibm.com, systemtap@sources.redhat.com Subject: Re: [PATCH] Fix kprobes handling of simultaneous probe hit/unregister From: "David S. Miller" In-Reply-To: <1125525217.2855.35.camel@localhost.localdomain> References: <1125525217.2855.35.camel@localhost.localdomain> X-Mailer: Mew version 4.2 on Emacs 21.4 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org From: Jim Keniston Date: 31 Aug 2005 14:53:37 -0700 > This bug doesn't exist on ppc64 and ia64, where a breakpoint > instruction leaves the IP pointing to the beginning of the instruction. > I don't know about sparc64. (Dave, could you please advise?) On sparc64 instructions are all 32-bit, 4-byte aligned, and a breakpoint instruction leaves the PC pointing at the beginning of that breakpoint instruction. So I think sparc64 should be OK.