From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christian Borntraeger Subject: Re: [patch] mm: fix XIP file writes Date: Mon, 10 Dec 2007 15:38:20 +0100 Message-ID: <200712101538.20619.borntraeger@de.ibm.com> References: <20071204042628.GA26636@wotan.suse.de> <20071204113549.GA6751@wotan.suse.de> <20071204130050.GB6751@wotan.suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Cc: Andrew Morton , Linux Kernel Mailing List , linux-fsdevel@vger.kernel.org, "Eric W. Biederman" , rob@landley.net, Jens Axboe , cotte@de.ibm.com To: Nick Piggin Return-path: Received: from mtagate2.uk.ibm.com ([195.212.29.135]:64502 "EHLO mtagate2.uk.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751407AbXLJOjA (ORCPT ); Mon, 10 Dec 2007 09:39:00 -0500 In-Reply-To: <20071204130050.GB6751@wotan.suse.de> Content-Disposition: inline Sender: linux-fsdevel-owner@vger.kernel.org List-ID: Hi Nick, > Here we go. See, brd already found a bug ;) > Can you apply the ext2 XIP patch too? And I'll resend the brd XIP patch. [...] > Writing to XIP files at a non-page-aligned offset results in data corruption > because the writes were always sent to the start of the page. [...] > @@ -314,7 +314,7 @@ __xip_file_write(struct file *filp, cons > fault_in_pages_readable(buf, bytes); > kaddr = kmap_atomic(page, KM_USER0); > copied = bytes - > - __copy_from_user_inatomic_nocache(kaddr, buf, bytes); > + __copy_from_user_inatomic_nocache(kaddr + offset, buf, bytes); > kunmap_atomic(kaddr, KM_USER0); > flush_dcache_page(page); I asked myself why this problem never happened before. So I asked our testers to reproduce this problem on 2.6.23 and service levels. As the testcase did not trigger, I looked into the 2.6.23 code. This problem was introduced by commit 4a9e5ef1f4f15205e477817a5cefc34bd3f65f55 (mm: write iovec cleanup from Nick Piggin) during 2.6.24-rc: --------snip------- - copied = filemap_copy_from_user(page, offset, buf, bytes); [...] + copied = bytes - + __copy_from_user_inatomic_nocache(kaddr, buf, bytes); ------------------- So yes, its good to have xip on brd. It even tests your changes ;-) Good news is, that we dont need anything for stable. Christian