lkml.org 
[lkml]   [2011]   [Jan]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    From
    Date
    SubjectRe: [patch 1/2] sched: move periodic share updates to entity_tick()
    Back from the Holidays and catching up on email -- Peter, looks like
    this fix still needs merging?

    On Thu, Dec 16, 2010 at 6:26 AM, Mike Galbraith <efault@gmx.de> wrote:
    > On Thu, 2010-12-16 at 12:03 +0100, Peter Zijlstra wrote:
    >> I made that:
    >
    > Box is ill, I'll test these as soon as it's all better.
    >
    >> ---
    >> Subject: sched: Move periodic share updates to entity_tick()
    >> From: Paul Turner <pjt@google.com>
    >> Date: Wed, 15 Dec 2010 19:10:17 -0800
    >>
    >> Long running entities that do not block (dequeue) require periodic updates to
    >> maintain accurate share values.  (Note: group entities with several threads are
    >> quite likely to be non-blocking in many circumstances).
    >>
    >> By virtue of being long-running however, we will see entity ticks (otherwise
    >> the required update occurs in dequeue/put and we are done).  Thus we can move
    >> the detection (and associated work) for these updates into the periodic path.
    >>
    >> This restores the 'atomicity' of update_curr() with respect to accounting.
    >>
    >> Signed-off-by: Paul Turner <pjt@google.com>
    >> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
    >> LKML-Reference: <20101216031038.067028969@google.com>
    >> ---
    >>  kernel/sched_fair.c |   21 +++++++++++++++++----
    >>  1 file changed, 17 insertions(+), 4 deletions(-)
    >>
    >> Index: linux-2.6/kernel/sched_fair.c
    >> ===================================================================
    >> --- linux-2.6.orig/kernel/sched_fair.c
    >> +++ linux-2.6/kernel/sched_fair.c
    >> @@ -564,10 +564,6 @@ __update_curr(struct cfs_rq *cfs_rq, str
    >>
    >>  #if defined CONFIG_SMP && defined CONFIG_FAIR_GROUP_SCHED
    >>       cfs_rq->load_unacc_exec_time += delta_exec;
    >> -     if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) {
    >> -             update_cfs_load(cfs_rq, 0);
    >> -             update_cfs_shares(cfs_rq, 0);
    >> -     }
    >>  #endif
    >>  }
    >>
    >> @@ -809,6 +805,14 @@ static void update_cfs_shares(struct cfs
    >>
    >>       reweight_entity(cfs_rq_of(se), se, shares);
    >>  }
    >> +
    >> +static void update_entity_shares_tick(struct cfs_rq *cfs_rq)
    >> +{
    >> +     if (cfs_rq->load_unacc_exec_time > sysctl_sched_shares_window) {
    >> +             update_cfs_load(cfs_rq, 0);
    >> +             update_cfs_shares(cfs_rq, 0);
    >> +     }
    >> +}
    >>  #else /* CONFIG_FAIR_GROUP_SCHED */
    >>  static void update_cfs_load(struct cfs_rq *cfs_rq, int global_update)
    >>  {
    >> @@ -817,6 +821,10 @@ static void update_cfs_load(struct cfs_r
    >>  static inline void update_cfs_shares(struct cfs_rq *cfs_rq, long weight_delta)
    >>  {
    >>  }
    >> +
    >> +static inline void update_entity_shares_tick(struct cfs_rq *cfs_rq)
    >> +{
    >> +}
    >>  #endif /* CONFIG_FAIR_GROUP_SCHED */
    >>
    >>  static void enqueue_sleeper(struct cfs_rq *cfs_rq, struct sched_entity *se)
    >> @@ -1133,6 +1141,11 @@ entity_tick(struct cfs_rq *cfs_rq, struc
    >>        */
    >>       update_curr(cfs_rq);
    >>
    >> +     /*
    >> +      * Update share accounting for long-running entities.
    >> +      */
    >> +     update_entity_shares_tick(cfs_rq);
    >> +
    >>  #ifdef CONFIG_SCHED_HRTICK
    >>       /*
    >>        * queued ticks are scheduled to match the slice, so don't bother
    >>
    >
    >
    >
    --
    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: 2011-01-11 00:53    [W:0.050 / U:4.608 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site