linux-ide: hda: possibly failed opcode: 0x25 on Alpha with 2.6.34-rc3

linux-ide: hda: possibly failed opcode: 0x25 on Alpha with 2.6.34-rc3

[Adrian Glaubitz]

Dear IDE maintainers,

I recently installed Debian unstable on an old digital personal
workstation 433au alpha (Miata) [1] with the default kernel 2.6.32
from Debian unstable. There were no obvious problems except for the
tulip network driver which produced a lot of noise [2]. Joe Perches was
so kind to provide a patch to reduce the noise. I applied the patch
against 2.6.34-rc3 and rebuilt the kernel with the 2.6.32 Debian
config using "make oldconfig". The kernel build took around 18 hours ;).

Anyway, after rebooting the machine into the new kernel (2.6.34-rc3), the
IDE driver obviously triggered a lot of errors [3] resulting in the kernel
not being able to mount the root partition.

The error message is always the same:

hda: possibly failed opcode: 0x25
drive not ready for command
status error: status=0x5a { DriveReady SeekComplete DataRequest Index }

But please see the screenshots in [3].

The system finally dropped into a busybox. I issued a "reboot" and tried
to reboot into the Debian stock 2.6.32 kernel. However, after the IDE problem
with 2.6.34-rc3, aboot was unable to read the bootsector. When I powercycled
the machine, I could reboot properly into 2.6.32 without any problems. I
already checked the IDE cabling, the harddisk is new, btw. The problem
reproduces every time I boot into 2.6.34-rc3.

It looks like that some changes in the IDE driver between 2.6.32 and
2.6.34-rc3 somehow messed up the code on Alpha. I didn't test the
whole thing on 2.6.33.2 yet, however the machine is busy right now
building that kernel so I can try tomorrow.

The lspci output of the hardware looks like this:

test-adrian1:~# lspci
00:03.0 Ethernet controller: Digital Equipment Corporation DECchip 21142/43 (rev 30)
00:07.0 ISA bridge: Contaq Microsystems 82c693
00:07.1 IDE interface: Contaq Microsystems 82c693
00:07.2 IDE interface: Contaq Microsystems 82c693
00:07.3 USB Controller: Contaq Microsystems 82c693
00:0b.0 VGA compatible controller: Matrox Graphics, Inc. MGA 2064W [Millennium] (rev 01)
00:14.0 PCI bridge: Digital Equipment Corporation DECchip 21152 (rev 03)
01:04.0 SCSI storage controller: QLogic Corp. ISP1020 Fast-wide SCSI (rev 05)
01:09.0 Ethernet controller: Intel Corporation 82541GI Gigabit Ethernet Controller

If you need anymore verbose or debug output, please let me know.

Regards,

Adrian

[1] http://en.wikipedia.org/wiki/Digital_Personal_Workstation
[2] http://lkml.org/lkml/2010/4/5/104
[3] http://users.physik.fu-berlin.de/~glaubitz/alpha-crash/

Re: linux-ide: hda: possibly failed opcode: 0x25 on Alpha with 2.6.34-rc3

[Bartlomiej Zolnierkiewicz]

On Wednesday 07 April 2010 01:26:59 am Adrian Glaubitz wrote:
> Dear IDE maintainers,

I'm flattered but nowadays I'm just a maintainer of a modest
out-of-tree project or two.. ;-)

> I recently installed Debian unstable on an old digital personal
> workstation 433au alpha (Miata) [1] with the default kernel 2.6.32
> from Debian unstable. There were no obvious problems except for the
> tulip network driver which produced a lot of noise [2]. Joe Perches was
> so kind to provide a patch to reduce the noise. I applied the patch
> against 2.6.34-rc3 and rebuilt the kernel with the 2.6.32 Debian
> config using "make oldconfig". The kernel build took around 18 hours ;).
> 
> Anyway, after rebooting the machine into the new kernel (2.6.34-rc3), the
> IDE driver obviously triggered a lot of errors [3] resulting in the kernel
> not being able to mount the root partition.

I think that the highest chance of bringing the right people's
attention to this kernel regression would be achieved by filling
the official bug-report at https://bugzilla.kernel.org/.

Thanks.
--
Bartlomiej Zolnierkiewicz

Re: linux-ide: hda: possibly failed opcode: 0x25 on Alpha with 2.6.34-rc3

[Adrian Glaubitz]

Hi Bartlomiej,

thank you very much for your kind reply!

On Tue, Apr 13, 2010 at 05:34:57PM +0200, Bartlomiej Zolnierkiewicz wrote:
> On Wednesday 07 April 2010 01:26:59 am Adrian Glaubitz wrote:
> > Dear IDE maintainers,
> 
> I'm flattered but nowadays I'm just a maintainer of a modest
> out-of-tree project or two.. ;-)

Hehe, just trying to keep good manners in public ;).

> 
> > I recently installed Debian unstable on an old digital personal
> > workstation 433au alpha (Miata) [1] with the default kernel 2.6.32
> > from Debian unstable. There were no obvious problems except for the
> > tulip network driver which produced a lot of noise [2]. Joe Perches was
> > so kind to provide a patch to reduce the noise. I applied the patch
> > against 2.6.34-rc3 and rebuilt the kernel with the 2.6.32 Debian
> > config using "make oldconfig". The kernel build took around 18 hours ;).
> > 
> > Anyway, after rebooting the machine into the new kernel (2.6.34-rc3), the
> > IDE driver obviously triggered a lot of errors [3] resulting in the kernel
> > not being able to mount the root partition.
> 
> I think that the highest chance of bringing the right people's
> attention to this kernel regression would be achieved by filling
> the official bug-report at https://bugzilla.kernel.org/.

Ok, I will report the problem there. Btw, the problem doesn't occur
with 2.6.33.2 either so it definetely seems to be a regression with
2.6.34. Unfortunately, I don't have the machine at hand anymore, so I
won't be able to test in near future :(. Very sorry. However, I know
someone who still has such a workstation, I will ask him to test.


Thanks,

Adrian