[sane-devel] Problem with empty sense buffer in sanei_scsi (linux+hpusbscsi+vuescan)?

Jose Paulo Moitinho de Almeida moitinho@civil.ist.utl.pt
Tue, 16 Oct 2001 10:10:57 +0100


I will try to summarize my last impressions regarding this:

- vuescan is not using linked commands. But both Abel Deuring and Douglas 
Gilbert had the impression that it did when they looked at the logs. Could it 
be that there is a problem with a flag that is not properly 

- the sense buffer sent by the hpusbscsi driver is clean. In fact it doesn't 
matter wheter it is clean or not, because it is not copied by the sg driver.

- the problem can be avoided when vuescan "ignores the sense data if the low 
seven bits of the first byte isn't 0x70 and if the second byte isn't zero". 
This is not a true solution to the problem.

My doubt regards the definition of the best strategy to handle this. Assuming 
that there is a problem in sanei_scsi and Abel's patches are included I will 
eventually be able to test it with the new sg driver, but the hpusbscsi will 
not work (I think!) with kernels in the 2.2 series.


Ze Paulo