From mboxrd@z Thu Jan 1 00:00:00 1970 From: Zoltan Boszormenyi Subject: Re: Bad CD disk disables IDE DMA Date: Thu, 16 Aug 2007 14:39:25 +0200 Message-ID: <46C4457D.5030703@dunaweb.hu> References: <8Squ5-2cU-19@gated-at.bofh.it> <8SIKl-5IK-13@gated-at.bofh.it> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from linux.dunaweb.hu ([62.77.196.1]:40563 "EHLO linux.dunaweb.hu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753112AbXHPMjr convert rfc822-to-8bit (ORCPT ); Thu, 16 Aug 2007 08:39:47 -0400 In-Reply-To: Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: 7eggert@gmx.de Cc: Michal Piotrowski , linux-kernel , IDE/ATA development list Bodo Eggert =EDrta: > Michal Piotrowski wrote: > =20 >> On 15/08/07, Zoltan Boszormenyi wrote: >> =20 > > =20 >>> I noticed that a bad CD of mine makes DMA disabled: >>> =20 > > [...] > =20 >>> hda: cdrom_decode_status: error=3D0x40 { LastFailedSense=3D0x04 } >>> ide: failed opcode was: unknown >>> hda: DMA disabled >>> hda: ide_intr: huh? expected NULL handler on exit >>> hda: ATAPI reset complete >>> >>> Every time I put the said CD into to drive and DMA is on, I get the >>> above messages. >>> =20 >> This might be intended. >> =20 > > Maybe, and maybe only a certain effect might be intended. And maybe I= can > help by asking these questions: > > 1) Does disabling DMA fix the seek errors, or does it hide them by th= e PIO > interfce not printing them? > =20 With DMA off no seek error occurs. When the transfer rate is limited to UDMA3 I get this: hda: drive_cmd: status=3D0x51 { DriveReady SeekComplete Error } hda: drive_cmd: error=3D0x04 { AbortedCommand } ide: failed opcode was: 0xec Note the difference: the error code is different and it knows the failed opcode as well. But DMA remains on. Can it be a bad 80w cable? I don't think so, considering that it doesn't occur for DVDs when using UDMA4... > 2) If it does hide them, would filtering for seek errors be a sane th= ing to > do, or does the DMA engine (or HDD) behave badly on these errors, = or > do bad-DMA devices report seek errors, too? > > 3) If it does not hide them, would re-enabling DMA on disk change be = a > feasable workaround? > (proposed interface: use hdparm -d $num where $num =3D > 0, 1: as before > 3: DMA on, if switched off automatically, will be set to 2 > 2: DMA off, will be turned on (set to 3) on disk change) > =20 The automatic DMA re-enabling would be a good solution. > 4) Does libata work well enough for making all effort put into that o= ld > IDE layer be a waste of time? > =20 Unfortunately libata (or pata_amd) doesn't work very well on this machi= ne, I tried to switch to libata for the CD drive but it lost the DVD-playin= g=20 capability. It seems the error was that libdvdcss couldn't get the keys via the scd= =20 device. This is the main obstacle that stops me switching to Fedora 7 from FC6. I reported it some time ago but got no answer. Look for subject "DVD-playing, ide-cd vs scsi-cd drivers". Best regards, Zolt=E1n B=F6sz=F6rm=E9nyi