lkml.org 
[lkml]   [2004]   [Sep]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    SubjectRe: [patch] voluntary-preempt-2.6.9-rc1-bk12-R5
    From
    Date
    On Sun, 2004-09-05 at 15:12, Ingo Molnar wrote:
    > * Lee Revell <rlrevell@joe-job.com> wrote:
    >
    > > Ok, first new one in a while. This was with -R0, but I haven't seen
    > > anyone else report it. Let me know if you need the complete trace.
    > >
    > > preemption latency trace v1.0.2
    > > -------------------------------
    > > latency: 511 us, entries: 951 (951)
    > > -----------------
    > > | task: dbench/4810, uid:1000 nice:0 policy:0 rt_prio:0
    > > -----------------
    > > => started at: kill_pg_info+0x10/0x50
    > > => ended at: kill_pg_info+0x2e/0x50
    > > =======>
    > > 00000001 0.000ms (+0.000ms): kill_pg_info (sys_kill)
    > > 00000001 0.000ms (+0.000ms): __kill_pg_info (kill_pg_info)
    > > 00000001 0.000ms (+0.000ms): find_pid (__kill_pg_info)
    >
    > this is quite hard to fix - lots of processes were SIGKILL-ed (or
    > SIGTERM-ed) and the signal semantics require us to deliver signals
    > atomically. The only fix would be to turn the signal locks into
    > semaphores but that's quite hard. (it's also a bit problematic for
    > interrupt-delivered signals.)
    >
    > Ingo
    >

    Here is a histogram I generated using realfeel2. This should provide
    better data than my jackd histograms because the latter are dependent on
    the ALSA driver, jackd's design, etc.

    I had to modify the amlat utilities to use usecs instead of msecs, this
    is a very good sign. ;-)

    http://krustophenia.net/testresults.php?dataset=2.6.9-rc1-R0#/var/www/2.6.9-rc1-R0/foo.hist

    I find the two smaller spikes to either side of the central spike really
    odd. These showed up in my jackd tests too, I had attributed them to
    some measurement artifact, but they seem real. Maybe a rounding bug, or
    some kind of weird cache effect?

    Lee


    -
    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/

    \
     
     \ /
      Last update: 2005-03-22 14:05    [W:0.021 / U:30.024 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site