From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: libata/sata_sil24 cache alignment problem? Date: Tue, 12 Feb 2008 20:32:55 -0600 Message-ID: <1202869975.3137.186.camel@localhost.localdomain> References: <20080212180254.GA28237@postdiluvian.org> <20080212221257.12eeab13@core> <1202859388.3137.164.camel@localhost.localdomain> <20080213021338.6946650e@core> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from accolon.hansenpartnership.com ([76.243.235.52]:34757 "EHLO accolon.hansenpartnership.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754268AbYBMCdA (ORCPT ); Tue, 12 Feb 2008 21:33:00 -0500 In-Reply-To: <20080213021338.6946650e@core> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Alan Cox Cc: Mark Mason , linux-ide@vger.kernel.org, jgarzik@pobox.com On Wed, 2008-02-13 at 02:13 +0000, Alan Cox wrote: > > I tell you what ... find me a parisc box that actually has IDE and we > > might have told you ... > > The NS87415 variant IDE has been tested on parisc and didn't blow up - > must just be lucky. Actually, it's only a specific class of machines: the PCX (that's most of the 715,725 series) that are fully incoherent. PCXL hack around this by making the page uncacheable and the 64 bit machines have coherence index handling which fixes this up (mostly). > > (actually, the pa8800's have IDE CD's on a cmd640 chip, but that oopses > > on boot for no reason we've tracked yet). > > With libata, old IDE or both ? and what does the oops look like ? Unfortunately, there's no oops; we get a high priority machine check which means something has gone wrong somewhere and it's not pretty. It's on my todo list to track down ... I just don't have one of these boxes and CDs are hard to play with remotely. James