Solaris 5.10 cpu hike --- _flockget - Unix

This is a discussion on Solaris 5.10 cpu hike --- _flockget - Unix ; Hi, before few times on solaris 10, one process started cosuming cpu towards 100%. I have taken pstack and gcore. And by viewing stacks of threads, its suprised to me ! 128 threads are active ------ ff140090 lwp_park (0, 0, ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: Solaris 5.10 cpu hike --- _flockget

  1. Solaris 5.10 cpu hike --- _flockget

    Hi,

    before few times on solaris 10, one process started cosuming cpu
    towards 100%.
    I have taken pstack and gcore. And by viewing stacks of threads, its
    suprised to me !

    128 threads are active


    ------
    ff140090 lwp_park (0, 0, 0)
    ff12409c _flockget (ff16c3b8, c9610, 44274, ff117594, ff16c308,
    c9610) + 94
    ff118110 fprintf (c9640, a7ae8, f7afa558, 501b0, ff118460, ff168284)
    + 50
    00059644 logstr

    many different threads (approx 100) executing/stcuk in above.

    And below two threads, one in fopen and one in write, its get locked.

    ----
    ff140090 lwp_park (0, 0, 0)
    ff12409c _flockget (ff16c3b8, c9610, 44274, ff12bb8c, ff16c308,
    c9610) + 94
    ff121478 _endopen (c98b0, a7b00, c9640, 0, 109, 0) + 12c
    ff124fac fopen (c98b0, a7b00, a7800, 3c788, 0, c9640) + 1c
    000598c8 open_logfile (c98b0, c9800, c9c00, ff052400, ff16cbc0, 1) +
    1e4
    0003da30 sig_thr (0, feefc000, 0, 0, 3d98c, 0) + a4
    ff13fff0 _lwp_start (0, 0, 0, 0, 0, 0)
    -------
    ff1218dc _flsbuf (32, c9640, ee, ff16c27c, fffffffb, ffffffff) + dc
    ff1263d4 _fwrite_unlocked (4, 1, 23, c9640, 23, febfa55d) + 414
    ff114478 _dowrite (febfa558, 23, c9640, febfa458, fe, 4) + 54
    ff116f0c _ndoprnt (a7aea, febfa520, febfa57b, febfa558, febfa458,
    ff150181) + 28bc
    ff1181a0 fprintf (c9640, a7ae8, febfa558, 501b0, ff118460, ff168284)
    + e0
    00059644 logstr (a3668, febfa8d0, 877b, dec88, ff322f58, d) + 1c4
    ....
    .... some function calls here.
    ff13fff0 _lwp_start (0, 0, 0, 0, 0, 0)
    --------


    I have tried much to debug, but no clue. [Only one doubt is i am
    using alarm() to get one thread periodically do some stuff, and in
    that above code is executed. Am i doing something stupid ? or any
    hidden Os/library bug is kidding me ?

    Tnx in advance,
    -Raxit

  2. Re: Solaris 5.10 cpu hike --- _flockget

    "Raxit@mykavita.com" writes:

    >I have tried much to debug, but no clue. [Only one doubt is i am
    >using alarm() to get one thread periodically do some stuff, and in
    >that above code is executed. Am i doing something stupid ? or any
    >hidden Os/library bug is kidding me ?


    If you use a signal ALARM handler and have him doing anything with
    stdio, you may have broken stdio (you can't use most stdio functions
    from signal handlers)

    Casper
    --
    Expressed in this posting are my opinions. They are in no way related
    to opinions held by my employer, Sun Microsystems.
    Statements on Sun products included here are not gospel and may
    be fiction rather than truth.

+ Reply to Thread