Re: Problem with SCSI Scanner on Irix 6.2

From: David Paschal (paschal@ns.net)
Date: Mon Jan 17 2000 - 01:28:09 PST

  • Next message: Carlos Puchol: "re-aligning? re-calibrating? an hp scanjet IIc"

    > has someone running a HP Scanjet IIp (or another HP scanner) on Irix 6.2
    > ?
    > There is a problem with such a combination that the scanner is
    > recognized
    > during the SCSI INQUIRE command (vendor, model and firmware is read
    > back),
    > but the scanner does not give any response to other commands.
    > The TEST UNIT READY seems also to work.
    > Has someone encountered a similar problem on Irix and how was it solved
    > ?

    Hi, Peter. I don't know if this is related, but I had trouble with partial
    SCL responses when I was developing my experimental support for HP OfficeJets
    connected to HP JetDirect print servers (see http://www.ns.net/~paschal/sane).
    There were situations where the device apparently split up the SCL response
    into different packets, which caused read() on the client to return only the
    first part. This caused a great deal of confusion with the upper levels of
    the HP SCL backend, due to both the short initial response, and the remainder
    of the response, which was interpreted as a malformed response to a subsequent
    request. I solved these problems by adding a select-and-read loop to my I/O
    code, with support for different timeouts for the first and subsequent
    select()s. See hp_jdex_read() and hp_jdex_select() in hp-jdex.c.

    Perhaps the ScanJetIIp/Irix6.2 problem is caused by a similar sensitivity
    in the HP backend to unusual timeout behaviour in the underlying OS, although
    it's strange that read() would return immediately with no data. I wonder if
    the fd is somehow defaulting to non-blocking mode? Just speculating aloud
    here...

    David

    --
    Source code, list archive, and docs: http://www.mostang.com/sane/
    To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com
    



    This archive was generated by hypermail 2b29 : Mon Jan 17 2000 - 01:28:11 PST