[Dvdrtools-users] Re: dvdrecord hangs machine -- My theory: ATA in kernel 2.4.20+ sux

[David Berry]

On Tue, 8 Jul 2003, Bryan J. Smith wrote:

> On Tue, 2003-07-08 at 05:00, David Berry wrote:
> > Problem: After dvdrecord produces the "Fixating..." message, the
> > following message is written to /var/log/messages:
> >
> > scsi : aborting command due to timeout : pid 13473, scsi0, channel 0, id 0,
> >                                       lun 0 Test Unit Ready 00 00 00 00 00
> >
> > and the whole system then hangs and needs to be rebooted. The resulting
> > disk is, however, readable when mounted - the only thing I notice is that
> > /var/log/message contains "UDF-fs: No VRS found" after mounting the disk.
> > Under Windows XP, Nero has no problems writing to the same media on the
> > same device.
> > Any help clearing up this problem would be *much* appreciated! Details
> > are:
> > O/S: Redhat 9
> > dvdrtools: v0.1.4
> > Kernel: 2.4.21
> > Media: Verbatim DataLifePlus DVD-RW
>
> I've spent about 20 hours on this with both a 3G and 4G DVD-RAM drive.
> I finally came to a common denominator, the kernel.
>
> Kernel 2.4.20+ is causing this _exact_ issue with ATA reliability.  And
> even when I _do_ get a successful record, I typically get errors.  The
> "proof is in the pudding" when I use a DVD-RAM drive and packet write
> (rewrite) DVD-RAM discs that does "verify after write" -- I get errors
> and rewrites all over the place.
>
> I've only followed the ATA development on the kernel from an "outside"
> perspective, but I believe Hendrick's ATA work is no longer in the
> 2.4.20+.  Or they did something else.  And it shows.  I've played with a
> number of stock and Red Hat kernels and this is the only thing I can
> find that is common in all situations/systems/drives.
>
> I've gone back to a Red Hat 2.4.19 kernel on some of my systems, but I
> also use NFS.  The Linux NFS client is far better in 2.4.20+ kernels.
> So I'm considering going way back to a Red Hat 2.4.9 kernel instead
> (before all the VM screwing).

Tried it with 2.4.19 kernel. It certainly made a difference - instead of
getting a single scsi timeout message followed by a system hang, I now get
a continuous stream of the following messages:

SCSI host 0 abort (pid 11858) timed out - resetting
SCSI bus is being reset for host 0 channel 0.
scsi : aborting command due to timeout : pid 11858, scsi0, channel 0, id 0,
       lun 0 Test Unit Ready 00 00 00 00 00

repeating every 0.5 second, but the system doesn't hang.


David