[libata] atapi_enabled problem

[libata] atapi_enabled problem

[Emmanuel Fleury]

Hi,

I'm a bit puzzled, I have a Fujistu-Siemens P7120 (see: 
http://www.labri.fr/perso/fleury/index.php?page=p7120) and the DVD-ROM 
wasn't detected at all at boot time.

After googling a bit I found this page: 
http://www.thinkwiki.org/wiki/Problems_with_SATA_and_Linux

So, I tried to compile the kernel with built-in ATAPI and S-ATA support 
(not as modules but embedded in the kernel) and I tried to pass 
libata.atapi_enabled=1 at boot time through kernel options. The 
DVD-drive was recognized at boot but generate tons of errors:

ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
ata1: no sense translation for error 0x20
ata1: no sense translation for status: 0x51
ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
sr0: CDROM (ioctl) error, command: <6>Test Unit Ready 00 00 00 00 00 00
sr: Current [descriptor]: sense key: Medium Error
     Additional sense: Unrecovered read error - auto reallocate failed
ata1: no sense translation for error 0x20
ata1: no sense translation for status: 0x51
ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
ata1: no sense translation for error 0x20
ata1: no sense translation for status: 0x51
ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
ata1: no sense translation for error 0x20
ata1: no sense translation for status: 0x51
ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
ata1: no sense translation for error 0x20
ata1: no sense translation for status: 0x51
ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
ata1: no sense translation for error 0x20
ata1: no sense translation for status: 0x51
ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
ata1: no sense translation for error 0x20
ata1: no sense translation for status: 0x51
ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
ata1: no sense translation for error 0x20
ata1: no sense translation for status: 0x51

Then I decided to just modify drivers/scsi/libata-core.c changing the 
line "atapi_enabled = 0" into "atapi_enabled = 1". Then everything 
worked as a charm.

I have no explanation yet for this behaviour (passing kernel options 
should behave the same than changing the code the way I did I suppose).

Does anyone has an idea ?

Regards
--
Emmanuel Fleury

Re: [libata] atapi_enabled problem

[Emmanuel Fleury]

Additional comments:

Emmanuel Fleury wrote:
> 
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> sr0: CDROM (ioctl) error, command: <6>Test Unit Ready 00 00 00 00 00 00
> sr: Current [descriptor]: sense key: Medium Error
>     Additional sense: Unrecovered read error - auto reallocate failed
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51

Benoit Garnier did point to me this thread: http://kerneltrap.org/node/5836

The solution proposed can't be applied to me because I have a dual
system where my HD is IDE/ATAPI and my DVD reader is S-ATA.

Indeed, I tried out several possibilities and it appeared that the way
to generate such errors was to set the Intel PIIX/ICH SATA support as
built-in (y). When set as module (m), everything is fine.

Device Drivers --->
  SCSI device support --->
    SCSI low-level drivers --->
      Intel PIIX/ICH SATA support

I might have changed this when hard-coding atapi_enabled=1 in the source.

Does someone see the light now ? (I'm still a bit in the dark)

Regards
-- 
Emmanuel Fleury              | Office: 211
Associate Professor,         | Phone: +33 (0)5 40 00 35 24
LaBRI, Domaine Universitaire | Fax:   +33 (0)5 40 00 66 69
351, Cours de la Libération  | email: emmanuel.fleury@labri.fr
33405 Talence Cedex, France  | URL: http://www.labri.fr/~fleury

Re: [libata] atapi_enabled problem

[Emmanuel Fleury]

Emmanuel Fleury wrote:
> 
> Indeed, I tried out several possibilities and it appeared that the way
> to generate such errors was to set the Intel PIIX/ICH SATA support as
> built-in (y). When set as module (m), everything is fine.
> 
> Device Drivers --->
>   SCSI device support --->
>     SCSI low-level drivers --->
>       Intel PIIX/ICH SATA support

Problem does also appear when AHCI SATA support is built-in (and Intel
PIIX/ICH SATA support is left as module).

-- 
Emmanuel Fleury

Re: [libata] atapi_enabled problem

[Randy.Dunlap]

On Thu, 20 Apr 2006 14:24:51 +0200 Emmanuel Fleury wrote:

> Hi,
> 
> I'm a bit puzzled, I have a Fujistu-Siemens P7120 (see: 
> http://www.labri.fr/perso/fleury/index.php?page=p7120) and the DVD-ROM 
> wasn't detected at all at boot time.
> 
> After googling a bit I found this page: 
> http://www.thinkwiki.org/wiki/Problems_with_SATA_and_Linux
> 
> So, I tried to compile the kernel with built-in ATAPI and S-ATA support 
> (not as modules but embedded in the kernel) and I tried to pass 
> libata.atapi_enabled=1 at boot time through kernel options. The 
> DVD-drive was recognized at boot but generate tons of errors:
> 
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> sr0: CDROM (ioctl) error, command: <6>Test Unit Ready 00 00 00 00 00 00
> sr: Current [descriptor]: sense key: Medium Error
>      Additional sense: Unrecovered read error - auto reallocate failed
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> ata1: translated ATA stat/err 0x51/20 to SCSI SK/ASC/ASCQ 0x3/11/04
> ata1: no sense translation for error 0x20
> ata1: no sense translation for status: 0x51
> 
> Then I decided to just modify drivers/scsi/libata-core.c changing the 
> line "atapi_enabled = 0" into "atapi_enabled = 1". Then everything 
> worked as a charm.
> 
> I have no explanation yet for this behaviour (passing kernel options 
> should behave the same than changing the code the way I did I suppose).

Yes, it should and it has worked for quite a few people in the past.
I suspect something more like a typo.  Anyway, recent kernels
(after 2.6.16, so 2.6.17-rc*) already have atapi_enabled set to 1.

> Does anyone has an idea ?

---
~Randy

Re: [libata] atapi_enabled problem

[Emmanuel Fleury]

Hi Randy.

Randy.Dunlap wrote:
> 
> Yes, it should and it has worked for quite a few people in the past.
> I suspect something more like a typo.  Anyway, recent kernels
> (after 2.6.16, so 2.6.17-rc*) already have atapi_enabled set to 1.

My mistake, I forgot to say I was using 2.6.16.2. Maybe this
atapi_enabled=1 does appear only in 2.6.17-rc* ?

Anyway, setting atapi_enabled to '1' in the future releases should solve
my problem. So, I guess that this isn't a bug after all (just a
difficult transition :).

Thanks to all
-- 
Emmanuel Fleury              | Office: 211
Associate Professor,         | Phone: +33 (0)5 40 00 35 24
LaBRI, Domaine Universitaire | Fax:   +33 (0)5 40 00 66 69
351, Cours de la Libération  | email: emmanuel.fleury@labri.fr
33405 Talence Cedex, France  | URL: http://www.labri.fr/~fleury

Re: [libata] atapi_enabled problem

[Randy.Dunlap]

On Thu, 20 Apr 2006 17:48:34 +0200 Emmanuel Fleury wrote:

> Hi Randy.
> 
> Randy.Dunlap wrote:
> > 
> > Yes, it should and it has worked for quite a few people in the past.
> > I suspect something more like a typo.  Anyway, recent kernels
> > (after 2.6.16, so 2.6.17-rc*) already have atapi_enabled set to 1.
> 
> My mistake, I forgot to say I was using 2.6.16.2. Maybe this
> atapi_enabled=1 does appear only in 2.6.17-rc* ?

Yes, it's not in the stable series, just in 2.6.17-rc*.

> Anyway, setting atapi_enabled to '1' in the future releases should solve
> my problem. So, I guess that this isn't a bug after all (just a
> difficult transition :).

Hopefully it's behind us then...

---
~Randy

Re: [libata] atapi_enabled problem

[Emmanuel Fleury]

Randy.Dunlap wrote:
> 
> Hopefully it's behind us then...

Not really in fact... Even with the atapi_enabled=1 set in the code, I
get the errors when either "AHCI SATA support" or "Intel PIIX/ICH SATA
support" are compiled as built-in.

Putting them as modules (both) solve the problem (don't ask me why!).

Regards
-- 
Emmanuel Fleury              | Office: 211
Associate Professor,         | Phone: +33 (0)5 40 00 35 24
LaBRI, Domaine Universitaire | Fax:   +33 (0)5 40 00 66 69
351, Cours de la Libération  | email: emmanuel.fleury@labri.fr
33405 Talence Cedex, France  | URL: http://www.labri.fr/~fleury