[patch] Real-Time Preemption, -RT-2.6.10-rc1-mm2-V0.7.1

From: Ingo Molnar (mingo_at_elte.hu)
Date: 11/03/04

  • Next message: Andi Kleen: "Re: 2.6.8 and 2.6.9 Dual Opteron glitches" 
    Date:	Wed, 3 Nov 2004 11:58:40 +0100
To: linux-kernel@vger.kernel.org

    i have released the -V0.7.1 Real-Time Preemption patch, which can be
    downloaded from:

        http://redhat.com/~mingo/realtime-preempt/

    this release is mainly a merge of -V0.6.9 to 2.6.10-rc2-mm2.

    I havent done a proper changelog for a couple of days so here is a list
    of bigger changes since -V0.4:

     - implemented a first version of the priority inheritance handling and
       priority inversion avoidance logic. This feature, after some initial
       stability problems, solved the jackd and rtc_wakeup latencies that
       were introduced by the ultra-finegrained locking in the -V series.

       (the -T/U series had a coarser locking scheme triggered much lower
       levels of priority inversion scenarios. The locking in the -V series
       was clearly the tipping point.)

       The new PI code covers all synchronization objects in Linux (on
       PREEMPT_REALTIME): spinlocks, rwlocks, semaphores and rwsems.
       Feedback on the design of this code would be welcome, and patches as
       well, if you have a better scheme. The code is pretty modular so feel
       free to experiment with alternative schemes.

     - completely reworked the debugging framework. All lock types
       (spinlocks, rwlocks, semaphores and rwsems) are now tracked, both
       their symbolic name and their place of acquire are traced and printed
       out upon detection of a deadlock. More and better information is
       printed upon a deadlock. Got rid of the 'semaphore owners array' in
       debugging mode, this reduces the footprint of semaphores quite
       significantly and speeds up deadlock detection.

     - got rid of the separate 'counted semaphores' implementation, it was
       too intrusive. Made the core 'generic semaphores' implementation
       compatible with vanilla Linux counted semaphore semantics. This also
       enabled the unrolling of the completion-handling cleanups which,
       while being very nice, were getting intrusive as well.

     - countless build and driver related reports/fixes from lots of people

     - more latency breaks in the remaining critical sections. A
       particularly important one was the irqs-off latency bugfix from
       Thomas Gleixner.

     - sped up the i8259 PIC and the PIT timer hardirq handling routines -
       these are now in the path of the longest latency.

     - cleaned up IRQ and signal preemption - there were missed
       check-rescheds and possibilities for IRQ recursion.

     - made ALSA's ioctl()s not use the BKL - this fixes more jackd
       latencies.

    to create a -V0.7.1 tree from scratch, the patching order is:

       http://kernel.org/pub/linux/kernel/v2.6/linux-2.6.9.tar.bz2
       http://kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.10-rc1.bz2
       http://kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.10-rc1/2.6.10-rc1-mm2/2.6.10-rc1-mm2.bz2
       http://redhat.com/~mingo/realtime-preempt/realtime-preempt-2.6.10-rc1-mm2-V0.7.1

            Ingo
    -
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/

  • Next message: Andi Kleen: "Re: 2.6.8 and 2.6.9 Dual Opteron glitches"

    Relevant Pages

    • Re: [patch] Real-Time Preemption, -RT-2.6.10-rc1-mm2-V0.7.1
      ... > levels of priority inversion scenarios. ... > (spinlocks, rwlocks, semaphores and rwsems) are now tracked, both ... > out upon detection of a deadlock. ... > - more latency breaks in the remaining critical sections. ...
      (Linux-Kernel)
    • Re: Copy Constructors
      ... >> shallow copy will result in deadlock. ... > Deadlocks, semaphores, synch access, critical sections, are all OT here. ...
      (comp.lang.cpp)
    Loading