On 29/06/2016 19:36, Jarkko Lavinen wrote: > On Tue, Jun 28, 2016 at 07:14:55PM +0200, Paolo Bonzini wrote: >> This is wrong, because INQUIRY's byte 3 is defined to be part of the >> length in modern SCSI standards. > >> This is wrong, because INQUIRY's byte 3 is defined to be part of the >> length in modern SCSI standards. > Ok. I was using outdated ANSI spec too pedantically. The first patch is > not needed at all. > >> SCAN conflicts with START_STOP. Add a comment please saying that >> START_STOP has cmd->xfer set to 0 in scsi_req_xfer for non-scanner >> devices. > Done. > > I also removed this: > > case GET_DATA_BUFFER_STATUS: > cmd->xfer = buf[8] | (buf[7] << 8); > > Since it is again unneedes and handled by scsi_cdb_xfer. > > I added longer sense buffer in Request Sense command for scanners. > Multi Pro has at least one sense bit in byte 18 and Minolta's own SW > uses 20 bytes as an allocation size in Request Sense command. > > The extra sense bit seems to be a kind of internal busy/film holder > moving status. After having this sense bit available scanning programs > have been very stable and every feature seems to be working.
Thanks! Out of curiosity what HBA are you using on the host? Paolo