Bug reading audio cd without -g parameter

Hi,

cdparanoia-III-10pre0 (self compiled, downloaded from xiph.org) does not work
as expected when the -g parameter is not given.

It hangs at this stage:
# cdparanoia -B -d/dev/scd0
cdparanoia III release 10pre0 (August 29, 2006)
[..]
 (== PROGRESS == [> | ...... 32 ] == :-P o ==)

The strace output can be found here:
http://launchpadlibrarian.net/8971713/cdparanoia.log

The following lines get written to syslog when using cdparanoia:

Aug 24 14:17:25 localhost kernel: [ 6114.874753] sg_write: data in/out 56/56
bytes for SCSI command 0x12--guessing data in;
Aug 24 14:17:25 localhost kernel: [ 6114.874757] program cdparanoia not
setting count
and/or reply_len properly
Aug 24 14:17:25 localhost kernel: [ 6114.875746] sg_write: data in/out 22/22
bytes for SCSI
command 0x1a--guessing data in;
Aug 24 14:17:25 localhost kernel: [ 6114.875748] program cdparanoia not
setting count
and/or reply_len properly
Aug 24 14:17:25 localhost kernel: [ 6114.874753] sg_write: data in/out 56/56
bytes for SCSI
command 0x12--guessing data in;
Aug 24 14:17:25 localhost kernel: [ 6114.874757] program cdparanoia not
setting count
and/or reply_len properly
Aug 24 14:17:25 localhost kernel: [ 6114.875746] sg_write: data in/out 22/22
bytes for SCSI
command 0x1a--guessing data in;
Aug 24 14:17:25 localhost kernel: [ 6114.875748] program cdparanoia not
setting count
and/or reply_len properly
Aug 24 14:17:30 localhost kernel: [ 6119.960910] sg_write: data in/out
30576/30576 bytes for
SCSI command 0xd8--guessing data in;
Aug 24 14:17:30 localhost kernel: [ 6119.960913] program cdparanoia not
setting count
and/or reply_len properly

And many more of these...

Ripping with "-Z" does not hang and is super fast but the resulting wav files
could damage ones speakers.
There is nothing which sounds like the music to be ripped.

Using cdparanoia-III-alpha9.8 the music is ripped like it should be.
Even with "-Z" (which is not nearly as fast as with 3.10) everything sounds
ok.



Ripping the cd works fine if I add the "-g" parameter
(the problem is that programs which use cdparanoia do not set this parameter):

# cdparanoia -d /dev/scd0 -v -g /dev/sg0 -B -- 1
cdparanoia III release 10pre0 (August 29, 2006)
[..]
        Testing /dev/sg0 for SCSI/MMC interface
                generic device forced; not testing for SG_IO interface
                generic device: /dev/sg0
                ioctl device: /dev/scd0

Found an accessible SCSI CDROM drive.
Looking at revision of the SG interface in use...
        SG interface version 3.5.34; OK.

CDROM model sensed sensed: PLEXTOR CD-ROM PX-12TS 1.02

Checking for SCSI emulation...
        Drive is SCSI

Checking for MMC style command set...
        Drive does not have MMC CDDA support
        DMA scatter/gather table entries: 128
        table entry size: 524288 bytes
        maximum theoretical transfer: 28532 sectors
        Setting default read size to 13 sectors (30576 bytes).

Verifying CDDA command set...
        Expected command set reads OK.
This command set may use a Force Unit Access bit.
Checking drive for FUA bit support...
        Drive rejected FUA bit.


If I remove the "-g" parameter cdparanoia thinks I have an ATAPI drive
but not a SCSI drive:

cdparanoia -d /dev/scd0 -v -B -- 1
cdparanoia III release 10pre0 (August 29, 2006)
(C) 2006 Monty <monty@...> and Xiph.Org
[..]
Checking /dev/scd0 for cdrom...
        Testing /dev/scd0 for SCSI/MMC interface
                SG_IO device: /dev/scd0

CDROM model sensed sensed: PLEXTOR CD-ROM PX-12TS 1.02

Checking for SCSI emulation...
        Drive is ATAPI (using SG_IO host adaptor emulation)

Checking for MMC style command set...
        Drive does not have MMC CDDA support
        DMA scatter/gather table entries: 1
        table entry size: 524288 bytes
        maximum theoretical transfer: 222 sectors
        Setting default read size to 13 sectors (30576 bytes).

Verifying CDDA command set...
        Expected command set reads OK.

So cdparanoia's bug can be worked around by setting the "-g" parameter for
SCSI drives.


The whole bug report can also be found at
https://bugs.launchpad.net/ubuntu/+source/cdparanoia/+bug/106000


I would be happy to provide further information or test patches.

Thanks a lot in advance,
Stefan Fleiter
_______________________________________________
Paranoia mailing list
Paranoia@...
http://lists.xiph.org/mailman/listinfo/paranoia

[xiphmont]

On 8/24/07, Stefan Fleiter <stefan.fleiter@...> wrote:
> Hi,
>
> cdparanoia-III-10pre0 (self compiled, downloaded from xiph.org) does not work
> as expected when the -g parameter is not given.

It is not a cdparanoia bug; your kernel has a direction setting bug
that was fixed a few months ago.  On a recent kernel, actually, the
opposite is true; not using -g works and using -g causes the flood of
bogus syslog entries because Jens wasn't paying attention while
updating the block/SG drivers.

Which kernel is this?

Monty
_______________________________________________
Paranoia mailing list
Paranoia@...
http://lists.xiph.org/mailman/listinfo/paranoia

Hi,

I must revise what I have said.
I'm quite sure it was like I said, I could rip audio with
cdparanoia-III-alpha9.8 without the -g parameter.

But after your mail I tested the scenario again several times
and the result is:

I can only rip audio if I add the -g parameter, no matter what
cdparanoia version I use.
I am very sorry for my could have led to the wrong direction.

> It is not a cdparanoia bug; your kernel has a direction setting bug
> that was fixed a few months ago.  On a recent kernel, actually, the
> opposite is true; not using -g works and using -g causes the flood of
> bogus syslog entries because Jens wasn't paying attention while
> updating the block/SG drivers.

I have a recent kernel:
# cat /proc/version
Linux version 2.6.22-10-generic (buildd@palmer) (gcc version 4.1.3 20070812
(prerelease) (Ubuntu 4.1.2-15ubuntu2)) #1 SMP Wed Aug 22 08:11:52 GMT 2007

And syslog is flooded if I add the -g parameter.
The amazing fact is it only functions if I add -g which then produces these
error messages.
And yes: I tested this a lot of times now. :-)

What can we do now to further track the problem down?


Greetings and thanks a lot for your response,
Stefan Fleiter
_______________________________________________
Paranoia mailing list
Paranoia@...
http://lists.xiph.org/mailman/listinfo/paranoia

[xiphmont]

On 8/29/07, Stefan Fleiter <stefan.fleiter@...> wrote:
> Hi,
>
> I must revise what I have said.
> I'm quite sure it was like I said, I could rip audio with
> cdparanoia-III-alpha9.8 without the -g parameter.

cdparanoia 9.8 and 10pre prefer different default interfaces,
reflecting shifts in the kernel over the past N years.

> I can only rip audio if I add the -g parameter, no matter what
> cdparanoia version I use.

really?  that is more than very strange.  In a modern kernel (2.6.22
is modern), cdparanoia *should* be much happier without -g, unless
there are distro-specifc patches to the kernel causing trouble.


> And syslog is flooded if I add the -g parameter.

Yes, expected.  Feel free to report the kernel bug to Ubuntu.
Cdparanoiais behaving properly and to kernel spec.

> The amazing fact is it only functions if I add -g which then produces these
> error messages.
> And yes: I tested this a lot of times now. :-)
>
> What can we do now to further track the problem down?

Yes, that it does not work without -g is odd.  Could you, please, send
the output of cdparanoia -vQ?  Also, try it as root?

> Greetings and thanks a lot for your response,

No worries and thank you for actually reporting trouble (which most
people don't).

Monty
_______________________________________________
Paranoia mailing list
Paranoia@...
http://lists.xiph.org/mailman/listinfo/paranoia

Hi,

>> And syslog is flooded if I add the -g parameter.
>
> Yes, expected.  Feel free to report the kernel bug to Ubuntu.
> Cdparanoiais behaving properly and to kernel spec.

Ok, will do.
Do you have any pointers I can add to the report?
That would be very much appreciated.

> Yes, that it does not work without -g is odd.  Could you, please, send
> the output of cdparanoia -vQ?  Also, try it as root?

The output does not differ between root and non root users.

I've attached the output of this command with and without -g.

One interesting thing is, that with my other scsi cd drive
(Teac instead of Plextor) can read audio with and without -g.

The differences according to cdparanoia -vQ are that the Plextor drive
 * has no MMC CDDA support
 * rejects FUA bit


Thanks a lot again,
Stefan Fleiter

cdparanoia III release 9.8 (March 23, 2001)
(C) 2001 Monty <monty@...> and Xiphophorus

Report bugs to paranoia@...
http://www.xiph.org/paranoia/

Checking /dev/scd0 for cdrom...
        Testing /dev/scd0 for SCSI/MMC interface
                SG_IO device: /dev/scd0

CDROM model sensed sensed: PLEXTOR CD-ROM PX-12TS 1.02

Checking for SCSI emulation...
        Drive is ATAPI (using SG_IO host adaptor emulation)

Checking for MMC style command set...
        Drive does not have MMC CDDA support
        DMA scatter/gather table entries: 1
        table entry size: 524288 bytes
        maximum theoretical transfer: 222 sectors
        Setting default read size to 13 sectors (30576 bytes).

Verifying CDDA command set...
        Expected command set reads OK.

Table of contents (audio tracks only):
track        length               begin        copy pre ch
===========================================================
  1.    17540 [03:53.65]        0 [00:00.00]    no   no  2
  2.    26760 [05:56.60]    17540 [03:53.65]    no   no  2
  3.    21560 [04:47.35]    44300 [09:50.50]    no   no  2
  4.    20380 [04:31.55]    65860 [14:38.10]    no   no  2
  5.    21855 [04:51.30]    86240 [19:09.65]    no   no  2
  6.     4590 [01:01.15]   108095 [24:01.20]    no   no  2
  7.    19730 [04:23.05]   112685 [25:02.35]    no   no  2
  8.    20143 [04:28.43]   132415 [29:25.40]    no   no  2
  9.    20852 [04:38.02]   152558 [33:54.08]    no   no  2
 10.    19020 [04:13.45]   173410 [38:32.10]    no   no  2
TOTAL  192430 [42:45.55]    (audio only)


cdparanoia III release 9.8 (March 23, 2001)
(C) 2001 Monty <monty@...> and Xiphophorus

Report bugs to paranoia@...
http://www.xiph.org/paranoia/

        Testing /dev/sg0 for SCSI/MMC interface
                generic device forced; not testing for SG_IO interface
                generic device: /dev/sg0
                ioctl device: /dev/scd0

Found an accessible SCSI CDROM drive.
Looking at revision of the SG interface in use...
        SG interface version 3.5.34; OK.

CDROM model sensed sensed: PLEXTOR CD-ROM PX-12TS 1.02

Checking for SCSI emulation...
        Drive is SCSI

Checking for MMC style command set...
        Drive does not have MMC CDDA support
        DMA scatter/gather table entries: 128
        table entry size: 524288 bytes
        maximum theoretical transfer: 28532 sectors
        Setting default read size to 13 sectors (30576 bytes).

Verifying CDDA command set...
        Expected command set reads OK.
This command set may use a Force Unit Access bit.
Checking drive for FUA bit support...
        Drive rejected FUA bit.

Table of contents (audio tracks only):
track        length               begin        copy pre ch
===========================================================
  1.    17540 [03:53.65]        0 [00:00.00]    no   no  2
  2.    26760 [05:56.60]    17540 [03:53.65]    no   no  2
  3.    21560 [04:47.35]    44300 [09:50.50]    no   no  2
  4.    20380 [04:31.55]    65860 [14:38.10]    no   no  2
  5.    21855 [04:51.30]    86240 [19:09.65]    no   no  2
  6.     4590 [01:01.15]   108095 [24:01.20]    no   no  2
  7.    19730 [04:23.05]   112685 [25:02.35]    no   no  2
  8.    20143 [04:28.43]   132415 [29:25.40]    no   no  2
  9.    20852 [04:38.02]   152558 [33:54.08]    no   no  2
 10.    19020 [04:13.45]   173410 [38:32.10]    no   no  2
TOTAL  192430 [42:45.55]    (audio only)


_______________________________________________
Paranoia mailing list
Paranoia@...
http://lists.xiph.org/mailman/listinfo/paranoia