Messages in this thread | | | From | Paul Turner <> | Date | Thu, 14 Oct 2010 02:07:44 -0700 | Subject | Re: [PATCH v3 2/7] sched: accumulate per-cfs_rq cpu usage |
| |
On Thu, Oct 14, 2010 at 1:57 AM, Peter Zijlstra <peterz@infradead.org> wrote: > On Tue, 2010-10-12 at 13:21 +0530, Bharata B Rao wrote: >> +static u64 tg_request_cfs_quota(struct task_group *tg) >> +{ >> + struct cfs_bandwidth *cfs_b = tg_cfs_bandwidth(tg); >> + u64 delta = 0; >> + >> + if (cfs_b->runtime > 0 || cfs_b->quota == RUNTIME_INF) { >> + raw_spin_lock(&cfs_b->lock); >> + /* >> + * it's possible a bandwidth update has changed the global >> + * pool. >> + */ >> + if (cfs_b->quota == RUNTIME_INF) >> + delta = sched_cfs_bandwidth_slice(); >> + else { >> + delta = min(cfs_b->runtime, >> + sched_cfs_bandwidth_slice()); >> + cfs_b->runtime -= delta; >> + } >> + raw_spin_unlock(&cfs_b->lock); >> + } >> + return delta; >> +} > > Since you check cfs_b->quote outside of cfs_b->lock anyway, you might as > well avoid taking the lock in that case and directly return slice. >
Do you mean in the RUNTIME_INF case?
I suppose we could just avoid taking the lock there (theoretically it would be possible to slightly over-commit on a RUNTIME_INF => constrained transition, but this should be the supremely uncommon case and hey, it's transitioning from unlimited bandwidth anyway).
> Also, you possibly evaluate sched_cfs_bandwidth_slice() twice. >
Separate branch no? Should only be one evaluation. (fwiw the min macro looks to cache the evaluation) -- 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/
| |