lkml.org 
[lkml]   [2008]   [Oct]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [PATCH] Give kjournald a IOPRIO_CLASS_RT io priority
    On Thu, 2 Oct 2008 21:40:00 -0700
    Andrew Morton <akpm@linux-foundation.org> wrote:

    > On Thu, 2 Oct 2008 21:23:55 -0700 Arjan van de Ven
    > <arjan@infradead.org> wrote:
    >
    > > On Thu, 2 Oct 2008 21:01:17 -0700
    > > Arjan van de Ven <arjan@infradead.org> wrote:
    > >
    > > > > _
    > > > >
    > > > > perhaps for varying values of "1".
    > > >
    > >
    > > caught a few, a few were over 3 1/2 seconds in stall time
    > > (specifically I know this for the last one)
    > >
    > > (I've stripped out the ? entries to keep them reasonable)
    > >
    > > [ 410.168277] WARNING: at kernel/sched.c:5652
    > > io_schedule+0x77/0xb0() [ 410.168347] Pid: 699, comm: kjournald
    > > Not tainted 2.6.27-rc8-tip #50 [ 410.168366] [<c042ee64>]
    > > warn_on_slowpath+0x41/0x65 [ 410.168414] [<c070ec83>]
    > > io_schedule+0x77/0xb0 [ 410.168421] [<c04abc72>]
    > > sync_buffer+0x33/0x37 [ 410.168429] [<c070f123>]
    > > __wait_on_bit+0x36/0x5d [ 410.168445] [<c070f1f5>]
    > > out_of_line_wait_on_bit+0xab/0xb3 [ 410.168471] [<c04abbd1>]
    > > __wait_on_buffer+0x19/0x1c [ 410.168478] [<c04de796>]
    > > journal_commit_transaction+0x484/0xcb2 [ 410.168519] [<c04e14ae>]
    > > kjournald+0xc7/0x1ea [ 410.168544] [<c043fb87>] kthread+0x3b/0x61
    > > [ 410.168559] [<c040499f>] kernel_thread_helper+0x7/0x10
    > > [ 410.168567] =======================
    > > [ 410.168572] ---[ end trace de523043f88bd9a7 ]---
    > > [ 451.605034] ------------[ cut here ]------------
    > > [ 451.605041] WARNING: at kernel/sched.c:5652
    > > io_schedule+0x77/0xb0() [ 451.605114] Pid: 699, comm: kjournald
    > > Tainted: G W 2.6.27-rc8-tip #50
    > >
    > > ...
    > >
    >
    > hm, they're all kjournald getting stuck on lock_buffer(). That _may_
    > be related to the one we care about, but it doesn't seem likely.

    one of them is (based on timestamps of the printk) is less than 3
    seconds before the "real" one, while the real one's delay was 4 seconds.
    To me that implies they're at least somewhat correlated...


    --
    Arjan van de Ven Intel Open Source Technology Centre
    For development, discussion and tips for power savings,
    visit http://www.lesswatts.org


    \
     
     \ /
      Last update: 2008-10-03 06:47    [W:0.024 / U:0.880 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site