From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S266442AbUGBEH6 (ORCPT ); Fri, 2 Jul 2004 00:07:58 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S266454AbUGBEH6 (ORCPT ); Fri, 2 Jul 2004 00:07:58 -0400 Received: from mail.shareable.org ([81.29.64.88]:32942 "EHLO mail.shareable.org") by vger.kernel.org with ESMTP id S266442AbUGBEH4 (ORCPT ); Fri, 2 Jul 2004 00:07:56 -0400 Date: Fri, 2 Jul 2004 05:07:52 +0100 From: Jamie Lokier To: John David Anglin Cc: linux-kernel@vger.kernel.org, parisc-linux@parisc-linux.org, Matthew Wilcox , Grant Grundler Subject: Re: [parisc-linux] Comparing PROT_EXEC-only pages on different CPUs Message-ID: <20040702040752.GB10366@mail.shareable.org> References: <20040701224016.GB7928@mail.shareable.org> <200407020324.i623OIZ9011855@hiauly1.hia.nrc.ca> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200407020324.i623OIZ9011855@hiauly1.hia.nrc.ca> User-Agent: Mutt/1.4.1i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org John David Anglin wrote: > > Richard raises an interesting point: exec-only pages are useless if > > the code needs to read jump tables and constant pools. It seems very > > likely Alpha and IA64 have these. > > HPPA GCC code also needs to be able to read jump tables in .text. As Richard Henderson points out, the toolchain should be requesting PROT_READ|PROT_EXEC if it generates code which needs data read access. ELF is particularly good for tracking generating those flags, so do HPPA code segments have the read flag set at the moment? I.e. if it does, then it would be fine to change the kernel to honour PROT_EXEC-only. (There's also the question of whether that code needs data read access: do jump table instructions do data reads or do they count as code reads? I don't know HPPA well enough to know). -- Jamie