From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264472AbTLVLZG (ORCPT ); Mon, 22 Dec 2003 06:25:06 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S264493AbTLVLZG (ORCPT ); Mon, 22 Dec 2003 06:25:06 -0500 Received: from mx2.elte.hu ([157.181.151.9]:21958 "EHLO mx2.elte.hu") by vger.kernel.org with ESMTP id S264472AbTLVLZC (ORCPT ); Mon, 22 Dec 2003 06:25:02 -0500 Date: Mon, 22 Dec 2003 12:25:58 +0100 From: Ingo Molnar To: Gabor MICSKO Cc: LKML Subject: Re: [test] exec-shield vs. paxtest 0.9.5 horrible results Message-ID: <20031222112558.GA10337@elte.hu> References: <1072090466.1471.4.camel@gmicsko03> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1072090466.1471.4.camel@gmicsko03> User-Agent: Mutt/1.4.1i X-ELTE-SpamVersion: SpamAssassin ELTE 1.0 X-ELTE-VirusStatus: clean X-ELTE-SpamCheck: no X-ELTE-SpamCheck-Details: score=-4.9, required 5.9, BAYES_00 -4.90 X-ELTE-SpamLevel: X-ELTE-SpamScore: -4 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org * Gabor MICSKO wrote: > Any idea? yes. Undo the patch below. The paxtest author decided to add this pointless mprotect(stackptr, PROT_EXEC) to make sure the test lists exec-shield as 'vulnerable' while listing PaX as non-vulnerable. I sent the fix but (not surprisingly) it was not added. Marketing via testsuite eh? Ingo --- paxtest-0.9.4/body.c +++ paxtest-0.9.5/body.c @@ -29,6 +29,7 @@ fflush( stdout ); if( fork() == 0 ) { + do_mprotect((unsigned long)argv & ~4095U, 4096, PROT_READ|PROT_WRITE|PROT_EXEC); doit(); } else { wait( &status );