Re: [PARPORT] ECP mode transfers in 2.4.x kernels

From: Tim Waugh (twaugh@redhat.com)
Date: Fri Aug 17 2001 - 10:20:38 EDT

  • Next message: Dave Strauss: "Re: [PARPORT] ECP mode transfers in 2.4.x kernels"

    On Fri, Aug 17, 2001 at 10:10:07AM -0400, Dave Strauss wrote:

    > Ah, much much better. I'm now getting more that 350Kbytes/sec
    > (depending on the printer and the test file, of course). Even better,
    > the garbage data is no more. I have to think that this means that the
    > ECP emulation code still has problems.

    So it seems. :-/

    There is another problem here. If software ECP was being used, WHY?
    The port->modes is meant to indicate what the hardware is capable of.
    Currently it's doing just that, technically, but it ought to give an
    indication of whether parport_pc will actually _use_ that
    capability. :-)

    I think that the functions that check whether various ECP modes are
    available should be inside a CONFIG_PARPORT_PC_FIFO ifdef.

    > Other clues that dma is being used:
    >
    > dmesg
    > [ ... ]
    > lp0: using parport0 (interrupt-driven).

    THE clue is what parport_pc says when it loads. If it says ', dma
    %d', then it knows that it can do DMA with this chipset, and will do
    so. (Earlier I said to look for 'using FIFO', but that's wrong, and
    means that parport_pc knows that the chipset can do PIO, and will use
    it. At least in current kernels; I don't remember if this was
    actually the case by 2.4.2-acX.)

    So, the next step is: are the software ECP emulation bugs fixed in
    newer kernels? There were definitely a few that were squashed since
    then, so I think probably so.

    Tim.
    */



    -- To unsubscribe, send mail to: linux-parport-request@torque.net --
    -- with the single word "unsubscribe" in the body of the message. --



    This archive was generated by hypermail 2b29 : Fri Aug 17 2001 - 10:27:27 EDT