[lkml]   [1998]   [Sep]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    SubjectRe: Today Linus redesigns the networking driver interface (was Re: tulip driver in ...)
    Alan Cox writes:
    > > First people wanted bottom half handling as a response to serial
    > > interrupts, because it made a huge difference to PPP latency. That already
    > > meant that some of the advantage of a bottom-half interrupt was no longer
    > > there. Then, with the new SMP code, it just became clear that it made no
    > > sense at all to have the difference, because the interrupt entry code
    > > became fairly involved (I definitely didn't want to have two separate
    > > copies).
    > and the people who first wanted it submitted patches that made irq handlers
    > return 1/0 according to if they needed a bh run. The instrastructure for that
    > is half in place in 2.1.1xx, so there's a valid 2.3.x project for anyone
    > who wants very fast irq handling there.

    But if your interrupt handler doesn't register/mark a BH, then the
    costs of "BH processing" (i.e. (bh_active & bh_mask)) are
    trivial. Where is the gain in that extra flag?

    > In truth for most of those cases running RTLinux is far simpler, far
    > faster and far more reliable since in the RTLinux case you _will_
    > hit your irq latency deadlines not "most of the time".

    Unfortunately RTLinux is further removed from the "real world" of
    Linux. Launching RTLinux processes is quite different (you have to
    load a module), and you don't get message queues, semaphores and the
    things people expect. I've had just this criticism from someone
    considering to use Linux/RTLinux instead of pSOS+.

    From looking at the current Linux capabilities, it looks to me like we
    can indeed give hard-RT performance. Sure, it may mean not using
    broken 8390 drivers which globally disable interrupts while spending
    1.6 ms reading a packet, but hey, we can live with that ;-)
    Being able to do all this with normal Linux without having to resort
    to RTLinux is a goal worthy of pursuit.



    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    Please read the FAQ at

     \ /
      Last update: 2005-03-22 13:44    [W:0.021 / U:1.116 seconds]
    ©2003-2017 Jasper Spaans. hosted at Digital OceanAdvertise on this site