CD burn buffer underruns on 2.6

CD burn buffer underruns on 2.6

[craig duncan]

My first attempts at burning on 2.6 resulted in a couple of coasters
(i'm running the "stock" 2.6 Debian kernel: kernel-image-2.6.0-test9-1-386).
Burn speed was 8x (max speed for my PleXwrite 8/4/32A).  With the 2.4
kernel (2.4.22-1-k7 on a 1.3 gig Athlon) i never had a problem burning
at 8x (in fact, when burning completes, it always tells me:
Buffer fill min 93%/max 100%.

At 4x, burns on 2.6 work fine, though.  At 8x, the burn always craps out
shortly after starting.  This is completely consistent... 8x fails, 4x
works (see cdrdao output below).  Nothing else is going on on my system.
Running as root or regular user doesn't matter.

After a few of these failed burns ("cdrdao simulate...", actually) i saw
this in /var/log/kern.log:

Dec 24 08:24:44 cdw kernel: cdrom_newpc_intr: 110 residual after xfer
Dec 24 08:24:57 cdw kernel: cdrom_newpc_intr: 104 residual after xfer
Dec 24 08:25:43 cdw kernel: cdrom_newpc_intr: 110 residual after xfer
Dec 24 08:25:55 cdw kernel: cdrom_newpc_intr: 104 residual after xfer
Dec 24 08:26:52 cdw kernel: cdrom_newpc_intr: 110 residual after xfer
Dec 24 08:27:03 cdw kernel: cdrom_newpc_intr: 104 residual after xfer

As a side note, under 2.6, i see these "resid: 110", "resid: 104", etc.
messages in the cdrdao output (below) which i never saw under 2.4.

Here's my cdrdao output:
-----
Starting write simulation resid: 110
at speed 8...
Pausing 10 seconds - hit CTRL-C to abort.
Process can be aborted with QUIT signal (usually CTRL-\).
resid: 104
Writing CD-TEXT lead-in...
Writing track 01 (mode AUDIO/AUDIO )...
Writing track 02 (mode AUDIO/AUDIO )...
?: Success.  : scsi sendcmd: no error
CDB:  2A 00 00 00 26 80 00 00 1A 00
status: 0x2 (CHECK CONDITION)
Sense Bytes: F0 00 03 00 00 27 18 0A 00 00 00 00 0C 09 00 00
Sense Key: 0x3 Medium Error, Segment 0
Sense Code: 0x0C Qual 0x09 (write error - loss of streaming) Fru 0x0
Sense flags: Blk 10008 (valid) 
resid: 61152
cmd finished after 0.014s timeout 180s
ERROR: Write data failed.
ERROR: Writing failed - buffer under run?
ERROR: Simulation failed.

I looked through the kernel archives and didn't see anyone else
reporting this problem but it seems kind of serious to me.

Re: CD burn buffer underruns on 2.6

[Dr. David Alan Gilbert]

* craig duncan (duncan@nycap.rr.com) wrote:
> 
> Dec 24 08:24:44 cdw kernel: cdrom_newpc_intr: 110 residual after xfer

Hmm - I'm seeing those just playing an audio CD on 2.6.0:

cdrom_newpc_intr: 3 residual after xfer

(That is a Memorex 24MAXX 1040 CD R/W on an AMD766 controller on a dual
Athlon MP board).

Dave
 -----Open up your eyes, open up your mind, open up your code -------   
/ Dr. David Alan Gilbert    | Running GNU/Linux on Alpha,68K| Happy  \ 
\ gro.gilbert @ treblig.org | MIPS,x86,ARM,SPARC,PPC & HPPA | In Hex /
 \ _________________________|_____ http://www.treblig.org   |_______/

Re: CD burn buffer underruns on 2.6

[Jens Axboe]

On Sun, Dec 28 2003, Dr. David Alan Gilbert wrote:
> * craig duncan (duncan@nycap.rr.com) wrote:
> > 
> > Dec 24 08:24:44 cdw kernel: cdrom_newpc_intr: 110 residual after xfer
> 
> Hmm - I'm seeing those just playing an audio CD on 2.6.0:
> 
> cdrom_newpc_intr: 3 residual after xfer

It's a silly debug printk, I'll make sure it gets deleted. This can
happen quite legitimately.

-- 
Jens Axboe

Re: CD burn buffer underruns on 2.6

[Claas Langbehn]

Hello,


ide-scsi is deprecated and should not be used
any longer. Instead you should use the ide 
device directly. F.ex.

cdrecord dev=dev/hdc ...

But as far as i know, cdrdao does not support
this yet :(


When I was trying to burn with ide-scsi earlier
in 2.6.0-testX, I also had lockups. I think if
you need to use cdrdao, then you might better 
stick to 2.4.x kernels until cdrdao learns it.


regards,
claas

Re: CD burn buffer underruns on 2.6

[Martin Josefsson]

[-- Attachment #1: Type: text/plain, Size: 908 bytes --]

On Mon, 2003-12-29 at 18:22, Claas Langbehn wrote:
> Hello,
> 
> 
> ide-scsi is deprecated and should not be used
> any longer. Instead you should use the ide 
> device directly. F.ex.
> 
> cdrecord dev=dev/hdc ...
> 
> But as far as i know, cdrdao does not support
> this yet :(

Some time ago I compiled cdrdao 1.1.7 with support for this (iirc I
copied libscg from cdrecord)
I use it all the time with 2.6, works like a charm at 32x.

If you are brave enough you can use it (never trust binaries offered by
strange people :) especially don't run them as root. (I never burn cd's
as root, works just fine as a normal user as long as you have
write-permission for the blockdevice, but you loose the
realtime-priority of the process when running as non-root which might
matter if you have a slow or highly loaded system while burning)

wget gandalf.hjorten.nu/cdrdao

-- 
/Martin

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: CD burn buffer underruns on 2.6

[Jens Axboe]

On Sun, Dec 28 2003, craig duncan wrote:
> My first attempts at burning on 2.6 resulted in a couple of coasters
> (i'm running the "stock" 2.6 Debian kernel: kernel-image-2.6.0-test9-1-386).
> Burn speed was 8x (max speed for my PleXwrite 8/4/32A).  With the 2.4
> kernel (2.4.22-1-k7 on a 1.3 gig Athlon) i never had a problem burning
> at 8x (in fact, when burning completes, it always tells me:
> Buffer fill min 93%/max 100%.
> 
> At 4x, burns on 2.6 work fine, though.  At 8x, the burn always craps out
> shortly after starting.  This is completely consistent... 8x fails, 4x
> works (see cdrdao output below).  Nothing else is going on on my system.
> Running as root or regular user doesn't matter.
> 
> After a few of these failed burns ("cdrdao simulate...", actually) i saw
> this in /var/log/kern.log:
> 
> Dec 24 08:24:44 cdw kernel: cdrom_newpc_intr: 110 residual after xfer
> Dec 24 08:24:57 cdw kernel: cdrom_newpc_intr: 104 residual after xfer
> Dec 24 08:25:43 cdw kernel: cdrom_newpc_intr: 110 residual after xfer
> Dec 24 08:25:55 cdw kernel: cdrom_newpc_intr: 104 residual after xfer
> Dec 24 08:26:52 cdw kernel: cdrom_newpc_intr: 110 residual after xfer
> Dec 24 08:27:03 cdw kernel: cdrom_newpc_intr: 104 residual after xfer
> 
> As a side note, under 2.6, i see these "resid: 110", "resid: 104", etc.
> messages in the cdrdao output (below) which i never saw under 2.4.
> 
> Here's my cdrdao output:
> -----
> Starting write simulation resid: 110
> at speed 8...
> Pausing 10 seconds - hit CTRL-C to abort.
> Process can be aborted with QUIT signal (usually CTRL-\).
> resid: 104
> Writing CD-TEXT lead-in...
> Writing track 01 (mode AUDIO/AUDIO )...
> Writing track 02 (mode AUDIO/AUDIO )...
> ?: Success.  : scsi sendcmd: no error
> CDB:  2A 00 00 00 26 80 00 00 1A 00
> status: 0x2 (CHECK CONDITION)
> Sense Bytes: F0 00 03 00 00 27 18 0A 00 00 00 00 0C 09 00 00
> Sense Key: 0x3 Medium Error, Segment 0
> Sense Code: 0x0C Qual 0x09 (write error - loss of streaming) Fru 0x0
> Sense flags: Blk 10008 (valid) 
> resid: 61152
> cmd finished after 0.014s timeout 180s
> ERROR: Write data failed.
> ERROR: Writing failed - buffer under run?
> ERROR: Simulation failed.
> 
> I looked through the kernel archives and didn't see anyone else
> reporting this problem but it seems kind of serious to me.

Maybe cdrdao doesn't align its buffers correctly? Can you send me a
vmstat 1 output while attempting a 8x burn on 2.6?

-- 
Jens Axboe