Re: Bug#395390: wodim: cdrecord -checkdrive freezes with CyberDrv CW088D

[Eduard Bloch <edi@gmx.de>]

reassign 395390 kernel
retitle 395390 IDE driver (via?) freezes ignoring ATAPI timeout
severity 395390 important
thanks

Unfortunately I don't know a good way to continue with debugging it. I
am reassigning your issue to the kernel team.

Eduard.

#include <hallo.h>
* Francois Gouget [Fri, Oct 27 2006, 04:02:57AM]:
> On Fri, 27 Oct 2006, Eduard Bloch wrote:
> [...]
> > > ---
> > > Oct 26 18:39:03 amboise kernel: hdd: status error: status=0x58 {
> > > DriveReady SeekComplete DataRequest }
> > > Oct 26 18:39:03 amboise kernel: ide: failed opcode was: unknown
> > > Oct 26 18:39:03 amboise kernel: hdd: drive not ready for command
> > > Oct 26 18:39:03 amboise kernel: hdd: status timeout: status=0xd0 { Busy }
> > > Oct 26 18:39:03 amboise kernel: ide: failed opcode was: unknown
> > > Oct 26 18:39:03 amboise kernel: hdd: DMA disabled
> > > Oct 26 18:39:03 amboise kernel: hdd: drive not ready for command
> > > ---
> > 
> > This looks like a kernel bug. I assume that your drive freezes, and the
> > driver does not abort, ignoring the requested timeout.
> 
> Ok. Seems like a longstanding bug then since I pretty sure it was 
> already present in 2.6.13. Would you happen to know of kernel debug 
> options to turn on to get more data? Or hints as to where I should dig 
> (this is a standard IDE CD writer and a VIA 'VT8375 [KM266/KL266]' 
> chipset).
> 
>  
> > > A couple more notes: I used to be able to write CDs with this burner. I 
> > > am sure it worked with a 2.4 kernel. It may have stopped working when I 
> > > switched to a 2.6 kernel, a long time ago.
> > 
> > Could you make the wodim binary suid-root and retry the same thing?
> 
> Actually I ran the above command as root (to match xcdroast which I runs 
> it as root).
> 
> 
> > > Executing 'read buffer' command on Bus 1 Target 1, Lun 0 timeout 40s
> > > CDB:  3C 00 00 00 00 00 00 FC 00 00
> > > ---
> > > This is where cdrecord freezes and thus where the traces end.
> > 
> > It waits for the kernel to return. A program cannot abort a syscal just
> > so, so it's kernel's turn, no matter what the outcome is.
> 
> Yes, that does seem to point to the kernel.
> 
> -- 
> Francois Gouget <fgouget@free.fr>              http://fgouget.free.fr/
>             "Lotto: A tax on people who are bad at math." -- unknown
>           "Windows: Microsoft's tax on computer illiterates." -- WE7U
> 

-- 
* Tolimar nimmt Madkiss und wirft ihn .. autsch, mein Rücken!
<Madkiss> towo: haaaaaaaaaaaar har har
* Wulf schenkt Madkiss ein neues nick complete
<Madkiss> hm.
<Madkiss> waldi: danke.