diff options
author | 2010-02-26 16:38:56 -0800 | |
---|---|---|
committer | 2010-02-27 09:53:52 +0100 | |
commit | a47cd880b50e14b0b6f5e9d426ae9a2676c9c474 (patch) | |
tree | c134dcec52450ec92ea853f4aeeef8e3967a36cb /kernel | |
parent | rcu: Export rcu_scheduler_active (diff) | |
download | linux-a47cd880b50e14b0b6f5e9d426ae9a2676c9c474.tar.gz linux-a47cd880b50e14b0b6f5e9d426ae9a2676c9c474.tar.bz2 linux-a47cd880b50e14b0b6f5e9d426ae9a2676c9c474.zip |
rcu: Fix accelerated grace periods for last non-dynticked CPU
It is invalid to invoke __rcu_process_callbacks() with irqs
disabled, so do it indirectly via raise_softirq(). This
requires a state-machine implementation to cycle through the
grace-period machinery the required number of times.
Located-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: laijs@cn.fujitsu.com
Cc: dipankar@in.ibm.com
Cc: mathieu.desnoyers@polymtl.ca
Cc: josh@joshtriplett.org
Cc: dvhltc@us.ibm.com
Cc: niv@us.ibm.com
Cc: peterz@infradead.org
Cc: rostedt@goodmis.org
Cc: Valdis.Kletnieks@vt.edu
Cc: dhowells@redhat.com
LKML-Reference: <1267231138-27856-1-git-send-email-paulmck@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Diffstat (limited to 'kernel')
-rw-r--r-- | kernel/rcutree.c | 3 | ||||
-rw-r--r-- | kernel/rcutree.h | 1 | ||||
-rw-r--r-- | kernel/rcutree_plugin.h | 73 |
3 files changed, 57 insertions, 20 deletions
diff --git a/kernel/rcutree.c b/kernel/rcutree.c index 335bfe4f0076..3ec8160fc75f 100644 --- a/kernel/rcutree.c +++ b/kernel/rcutree.c @@ -1341,6 +1341,9 @@ static void rcu_process_callbacks(struct softirq_action *unused) * grace-period manipulations above. */ smp_mb(); /* See above block comment. */ + + /* If we are last CPU on way to dyntick-idle mode, accelerate it. */ + rcu_needs_cpu_flush(); } static void diff --git a/kernel/rcutree.h b/kernel/rcutree.h index 2ceb08388582..1439eb504c22 100644 --- a/kernel/rcutree.h +++ b/kernel/rcutree.h @@ -373,5 +373,6 @@ static int rcu_preempt_needs_cpu(int cpu); static void __cpuinit rcu_preempt_init_percpu_data(int cpu); static void rcu_preempt_send_cbs_to_orphanage(void); static void __init __rcu_init_preempt(void); +static void rcu_needs_cpu_flush(void); #endif /* #ifndef RCU_TREE_NONCORE */ diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h index 3516de7091a1..ed241fc478f0 100644 --- a/kernel/rcutree_plugin.h +++ b/kernel/rcutree_plugin.h @@ -973,9 +973,19 @@ int rcu_needs_cpu(int cpu) return rcu_needs_cpu_quick_check(cpu); } +/* + * Check to see if we need to continue a callback-flush operations to + * allow the last CPU to enter dyntick-idle mode. But fast dyntick-idle + * entry is not configured, so we never do need to. + */ +static void rcu_needs_cpu_flush(void) +{ +} + #else /* #if !defined(CONFIG_RCU_FAST_NO_HZ) */ #define RCU_NEEDS_CPU_FLUSHES 5 +static DEFINE_PER_CPU(int, rcu_dyntick_drain); /* * Check to see if any future RCU-related work will need to be done @@ -988,39 +998,62 @@ int rcu_needs_cpu(int cpu) * only if all other CPUs are already in dynticks-idle mode. This will * allow the CPU cores to be powered down immediately, as opposed to after * waiting many milliseconds for grace periods to elapse. + * + * Because it is not legal to invoke rcu_process_callbacks() with irqs + * disabled, we do one pass of force_quiescent_state(), then do a + * raise_softirq() to cause rcu_process_callbacks() to be invoked later. + * The per-cpu rcu_dyntick_drain variable controls the sequencing. */ int rcu_needs_cpu(int cpu) { - int c = 1; - int i; + int c = 0; int thatcpu; /* Don't bother unless we are the last non-dyntick-idle CPU. */ for_each_cpu_not(thatcpu, nohz_cpu_mask) - if (thatcpu != cpu) + if (thatcpu != cpu) { + per_cpu(rcu_dyntick_drain, cpu) = 0; return rcu_needs_cpu_quick_check(cpu); - - /* Try to push remaining RCU-sched and RCU-bh callbacks through. */ - for (i = 0; i < RCU_NEEDS_CPU_FLUSHES && c; i++) { - c = 0; - if (per_cpu(rcu_sched_data, cpu).nxtlist) { - rcu_sched_qs(cpu); - force_quiescent_state(&rcu_sched_state, 0); - __rcu_process_callbacks(&rcu_sched_state, - &per_cpu(rcu_sched_data, cpu)); - c = !!per_cpu(rcu_sched_data, cpu).nxtlist; - } - if (per_cpu(rcu_bh_data, cpu).nxtlist) { - rcu_bh_qs(cpu); - force_quiescent_state(&rcu_bh_state, 0); - __rcu_process_callbacks(&rcu_bh_state, - &per_cpu(rcu_bh_data, cpu)); - c = !!per_cpu(rcu_bh_data, cpu).nxtlist; } + + /* Check and update the rcu_dyntick_drain sequencing. */ + if (per_cpu(rcu_dyntick_drain, cpu) <= 0) { + /* First time through, initialize the counter. */ + per_cpu(rcu_dyntick_drain, cpu) = RCU_NEEDS_CPU_FLUSHES; + } else if (--per_cpu(rcu_dyntick_drain, cpu) <= 0) { + /* We have hit the limit, so time to give up. */ + return rcu_needs_cpu_quick_check(cpu); + } + + /* Do one step pushing remaining RCU callbacks through. */ + if (per_cpu(rcu_sched_data, cpu).nxtlist) { + rcu_sched_qs(cpu); + force_quiescent_state(&rcu_sched_state, 0); + c = c || per_cpu(rcu_sched_data, cpu).nxtlist; + } + if (per_cpu(rcu_bh_data, cpu).nxtlist) { + rcu_bh_qs(cpu); + force_quiescent_state(&rcu_bh_state, 0); + c = c || per_cpu(rcu_bh_data, cpu).nxtlist; } /* If RCU callbacks are still pending, RCU still needs this CPU. */ + if (c) + raise_softirq(RCU_SOFTIRQ); return c; } +/* + * Check to see if we need to continue a callback-flush operations to + * allow the last CPU to enter dyntick-idle mode. + */ +static void rcu_needs_cpu_flush(void) +{ + int cpu = smp_processor_id(); + + if (per_cpu(rcu_dyntick_drain, cpu) <= 0) + return; + (void)rcu_needs_cpu(cpu); +} + #endif /* #else #if !defined(CONFIG_RCU_FAST_NO_HZ) */ |