Re: Infinite retries

[Boaz Harrosh <bharrosh@panasas.com>]

Alan Stern wrote:
> We do have a problem with infinite retry loops.  I'm not sure which 
> kernels are affected, but there's a good chance 2.6.27 is and an 
> excellent chance that 2.6.28-rc1 will be.
> 
> The problem is caused by buggy USB devices that return the number of 
> sectors in response to READ CAPACITY, rather than the highest sector 
> number.  Thus they appear to have one more sector than they really do.  
> We can create blacklist entries for these devices, but they constantly 
> appear and the list is never up-to-date.  So we need to handle them 
> robustly even when they aren't on the blacklist.
> 
> When the system tries to read the non-existent last sector, the devices
> send back no data.  In addition, the stupid devices return no sense
> information (even though they often do return Check Condition status).
> 
> The question is: What should usb-storage report to the midlayer when
> this happens?  Setting scmd->result to DID_ERROR doesn't help, because
> sd_done() sets it back to 0 in the NO_SENSE case.  We end up calling
> scsi_end_request() with error = 0 and bytes = 0, thereby requeuing the
> request.  There's no counter to limit the number of times the same
> request can be requeued.
> 
> I actually saw this happen to somebody in the log accompanying a bug
> report a few days ago (Bugzilla #11768).
> 
> What's the proper solution?  Should usb-storage make up some bogus 
> sense data (HARDWARE_ERROR sense key maybe)?
> 
> Alan Stern
> 
> P.S.: What about my proposed changes to scsi_io_completion()?  There 
> hasn't been any response.
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
Alan Hi

Do you have the scsi_io_completion patchset on a public git somewhere?
I would like to re-test them and review them again.

Did you try them with above problem and do they solve the issue?
Also have you looked farther into the retries/timeout issues from
block layer?

Thanks
Boaz