Re: REQUEST_SENSE calls by backends

From: abel deuring (
Date: Wed Jan 17 2001 - 06:59:10 PST

  • Next message: Mark Stewart: "Re: Umax 600s/Adaptec 2906 trouble"

    Oliver Rauch wrote:
    > abel deuring wrote:
    > > I don't know, how the other OS supported by Sane will
    > > respond to a REQUEST SENSE command issued from usert space.
    > Hi Abel,
    > I think this is something that should be discussed and solved in general
    > for sane and we should make all backends work the same - if possible.
    > Is it possible to get more than 16 or 18 sense bytes without calling
    > a request sense command?


    As Douglas already wrote, we get only 16 bytes for older Linux kernels.
    I don't know, how other operating systems supported by Sane respond, but
    I think that we should in no case assume to get more than the standard
    18 bytes for an "autosense" buffer. As the example of the SCSI system in
    Linux up to 2.2.x shows, it is possible that a kernel has hard limits.

    So, _if_ a backend needs more than 18 "autosense" bytes (or 16 bytes, if
    we include older Linux kernels) from a scanner, there will be no
    guaranty that this backend will work every OS supported by Sane.

    A second question is, how to deal with REQUEST SENSE commands explicitly
    sent from a Sane backend. While my little patch to the Snapscan backend
    showed, that at least thet 1236 does not need this command, Henning has
    already explained, why the Mustek backend indeed needs this comand for
    some scanners.

    Therefore I think that we can't do more that to state, that a backend
    should avoid to issue a REQUEST SENSE in order to maintain maximum
    compatibility (as the Mustek backend seems to: REQUEST SENSE is only
    sent, " if (s->hw->flags & MUSTEK_FLAG_PRO)" -- I guess that
    MUSTEK_FLAG_PRO is only set for the Paragon Pro) -- and to add a note in
    the README file, if there will be serious incompatibilities with any OS.
    For Linux, I remember at present only the aha152x driver in lk 2.4, and
    Jürgen Fischer has already fixed the problem.


    Source code, list archive, and docs:
    To unsubscribe: echo unsubscribe sane-devel | mail

    This archive was generated by hypermail 2b29 : Wed Jan 17 2001 - 05:50:51 PST