libata pata atapi errors

libata pata atapi errors

[Ed Sweetman]

Ok, been spending the day using libata's pata interface to my plextor 
dvd writer.   All is good, it's burning dvdr's at their rated speed and 
using dma and everything works perfectly.   Apparently, after some 
amount of disks I got the following error.

Assertion failed! qc->n_elem > 
0,drivers/scsi/libata-core.c,ata_fill_sg,line=2586
   (repeated many times)
ata6: command 0xa0 timeout, stat 0xd0 host_stat 0x60
ata6: translated ATA stat/err 0xd0/00 to SCSI SK/ASC/ASCQ 0xb/47/00
ATA: abnormal status 0xD0 on port 0x177
ATA: abnormal status 0xD0 on port 0x177
ATA: abnormal status 0xD0 on port 0x177
Device not ready. Make sure there is a disc in the drive.


Now, i used sg3-utils (from debian unstable amd64) and reset the device 
(using sg_reset) because it was stuck in a "writing" state.  I had to 
then use the "eject" utility to get the door to open because it was 
stuck in a locked state.  After all that, the device responds again, but 
is now making a strange vibration noise like the lens motor is trying to 
seek beyond where it's supposed to go.   and produces the errors again.  
Eventually the drive gets stuck in an active state and wont respond to 
any commands except spew errors when commands are attempted. 


this is with the amd/nvidia pata and sata drivers loaded on an smp 
athlon dual core system in 64bit mode.   Any additional info, just ask.

Re: libata pata atapi errors

[Alan Cox]

On Maw, 2006-02-07 at 21:14 -0500, Ed Sweetman wrote:
> Assertion failed! qc->n_elem > 
> 0,drivers/scsi/libata-core.c,ata_fill_sg,line=2586
>    (repeated many times)


Jeff was looking at some core bugs in this area. Its one of the ones I
can I think safely disclaim responsibility for (that and the fact that a
timeout of a command early on in 2.6.16-rc causes an oops)

Alan

Re: libata pata atapi errors

[Ed Sweetman]

Alan Cox wrote:

>On Maw, 2006-02-07 at 21:14 -0500, Ed Sweetman wrote:
>  
>
>>Assertion failed! qc->n_elem > 
>>0,drivers/scsi/libata-core.c,ata_fill_sg,line=2586
>>   (repeated many times)
>>    
>>
>
>
>Jeff was looking at some core bugs in this area. Its one of the ones I
>can I think safely disclaim responsibility for (that and the fact that a
>timeout of a command early on in 2.6.16-rc causes an oops)
>
>Alan
>
>  
>

Well, I was able to eventually recover the drive so that it responds 
without any errors and functions correctly.  Though, writing seems to be 
pretty much out of service.   Whenever i try writing anymore (haven't 
rebooted yet since the last errors) all it does is make a strange noise 
like the lens motor is trying to seek beyond it's limits and then I get 
the same series of errors that i originally got. 

If it helps any, I had to run a sg_sync on the drive to eventually get 
it back into a workable state.   Maybe it's having some sort of buffer 
problem that's not being internally flushed or cleared ?  I dont know.