I don't know if this is relevant to this specific problem,
but I noted that some newer drives have  a "big endian vs. little endian"
problem in the way they default when writing.  Consequently, there are
different default switches to cdrecord that must be used in order to get
it to work.  I stumbled on this when trying to write CD's under a
different operating system with a newer HP SCSI CD-RW drive.

On Fri, 1 Feb 2002, Rui-Xiang Guo wrote:

> > > I just tried my CDRW(cd1). I found it did not work for me, too.
> > > I have a bit long time not to use it. ;)
> > > On my system:
> > > trm0: target 0, 8bits Narrow transfer
> > > trm0: target 0, Sync transfer 10.0 MB/sec, Offset 15
> > > cd0 at scsibus0 target 0 lun 0: <MATSHITA, CD-ROM CR-506, 8S05> SCSI2 5/cdrom removable
> > > trm0: target 1, 8bits Narrow transfer
> > > trm0: target 1, Sync transfer 10.0 MB/sec, Offset 15
> > > cd1 at scsibus0 target 1 lun 0: <SONY, CD-RW  CRX160S, 1.0a> SCSI4 5/cdrom removable
> > 
> > Does the drive really say SCSI4, or just a typo?
> 
> Yes, it prints SCSI4. It seems a bug of flash ROM on CD-RW.
> And I not yet update it, I am lazy. :)
> 
> > Anyway, does it work with other SCSI host adapters?
> > Or does it work on other OS?
> 
> However, the CD-RW works well with trm under win9x.
> I am very sure it also works well with the kernel before Tue, 8 Jan 2002,
> because when I posted mails with title "cdplay don't work" to current
> mailing list, I had tested it without problem(can mount data CDs).
> So it seems that something be changed in our SCSI layer driver.
> 
> > Doesn't your kernel print any log in dmesg on the failure?
> 
> No, I don't get any error when booting.
> 
> > I have both trm and Yamaha CDRW, but they are in
> > different machines so I cannot test them right now..
> 
> I don't think it is the problem of CDRW, cd0 still is OK,
> the problem only happen on cd1.
> 
> After I type 'mount -t cd9660 /dev/cd1a /cdrom', 
> I get these messages and fall into debug mode:
> Kernel: page fault trap,  code=0
> Stopped at      trm_msgout_abort+0x38:  andb    $0xfe,      0x2f(%eax)
> db>t
> trm_msgout_abort(
> trm_reselt(
> trm_init(
> Xintr11(
> --- interrupt ---
> idle(
> bpendtsleep(
> ...
> 
> Regards,
> -rxg
>