linux/kernel/softirq.c

909 lines
22 KiB
C
Raw Normal View History

/*
* linux/kernel/softirq.c
*
* Copyright (C) 1992 Linus Torvalds
*
* Distribute under GPLv2.
*
* Rewritten. Old one was good in 2.2, but in 2.3 it was immoral. --ANK (990903)
*
* Remote softirq infrastructure is by Jens Axboe.
*/
#include <linux/module.h>
#include <linux/kernel_stat.h>
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/notifier.h>
#include <linux/percpu.h>
#include <linux/cpu.h>
#include <linux/freezer.h>
#include <linux/kthread.h>
#include <linux/rcupdate.h>
#include <linux/ftrace.h>
#include <linux/smp.h>
#include <linux/tick.h>
#define CREATE_TRACE_POINTS
#include <trace/events/irq.h>
#include <asm/irq.h>
/*
- No shared variables, all the data are CPU local.
- If a softirq needs serialization, let it serialize itself
by its own spinlocks.
- Even if softirq is serialized, only local cpu is marked for
execution. Hence, we get something sort of weak cpu binding.
Though it is still not clear, will it result in better locality
or will not.
Examples:
- NET RX softirq. It is multithreaded and does not require
any global serialization.
- NET TX softirq. It kicks software netdevice queues, hence
it is logically serialized per device, but this serialization
is invisible to common code.
- Tasklets: serialized wrt itself.
*/
#ifndef __ARCH_IRQ_STAT
irq_cpustat_t irq_stat[NR_CPUS] ____cacheline_aligned;
EXPORT_SYMBOL(irq_stat);
#endif
static struct softirq_action softirq_vec[NR_SOFTIRQS] __cacheline_aligned_in_smp;
static DEFINE_PER_CPU(struct task_struct *, ksoftirqd);
char *softirq_to_name[NR_SOFTIRQS] = {
"HI", "TIMER", "NET_TX", "NET_RX", "BLOCK", "BLOCK_IOPOLL",
"TASKLET", "SCHED", "HRTIMER", "RCU"
};
/*
* we cannot loop indefinitely here to avoid userspace starvation,
* but we also don't want to introduce a worst case 1/HZ latency
* to the pending events, so lets the scheduler to balance
* the softirq load for us.
*/
static void wakeup_softirqd(void)
{
/* Interrupts are disabled: no need to stop preemption */
struct task_struct *tsk = __this_cpu_read(ksoftirqd);
if (tsk && tsk->state != TASK_RUNNING)
wake_up_process(tsk);
}
/*
* preempt_count and SOFTIRQ_OFFSET usage:
* - preempt_count is changed by SOFTIRQ_OFFSET on entering or leaving
* softirq processing.
* - preempt_count is changed by SOFTIRQ_DISABLE_OFFSET (= 2 * SOFTIRQ_OFFSET)
* on local_bh_disable or local_bh_enable.
* This lets us distinguish between whether we are currently processing
* softirq and whether we just have bh disabled.
*/
/*
* This one is for softirq.c-internal use,
* where hardirqs are disabled legitimately:
*/
[PATCH] Reducing local_bh_enable/disable overhead in irqtrace The recent changes from irqtrace feature has added overheads to local_bh_disable and local_bh_enable that reduces UDP performance across x86_64 and IA64, even though IA64 does not support the irqtrace feature. Patch in question is [PATCH]lockdep: irqtrace subsystem, core http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=c ommit;h=de30a2b355ea85350ca2f58f3b9bf4e5bc007986 Prior to this patch, local_bh_disable was a short macro. Now it is a function which calls __local_bh_disable with added irq flags save and restore. The irq flags save and restore were also added to local_bh_enable, probably for injecting the trace irqs code. This overhead is on the generic code path across all architectures. On a IA_64 test machine (Itanium-2 1.6 GHz) running a benchmark like netperf's UDP streaming test, the added overhead results in a drop of 3% in throughput, as udp_sendmsg calls the local_bh_enable/disable several times. Other workloads that have heavy usages of local_bh_enable/disable could also be affected. The patch ideally should not have affected IA-64 performance as it does not have IRQ tracing support. A significant portion of the overhead is in the added irq flags save and restore, which I think is not needed if IRQ tracing is unused. A suggested patch is attached below that recovers the lost performance. However, the "ifdef"s in the patch are a bit ugly. Signed-off-by: Tim Chen <tim.c.chen@intel.com> Acked-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-30 12:04:02 +02:00
#ifdef CONFIG_TRACE_IRQFLAGS
static void __local_bh_disable(unsigned long ip, unsigned int cnt)
{
unsigned long flags;
WARN_ON_ONCE(in_irq());
raw_local_irq_save(flags);
/*
* The preempt tracer hooks into add_preempt_count and will break
* lockdep because it calls back into lockdep after SOFTIRQ_OFFSET
* is set and before current->softirq_enabled is cleared.
* We must manually increment preempt_count here and manually
* call the trace_preempt_off later.
*/
preempt_count() += cnt;
/*
* Were softirqs turned off above:
*/
if (softirq_count() == cnt)
trace_softirqs_off(ip);
raw_local_irq_restore(flags);
if (preempt_count() == cnt)
trace_preempt_off(CALLER_ADDR0, get_parent_ip(CALLER_ADDR1));
}
[PATCH] Reducing local_bh_enable/disable overhead in irqtrace The recent changes from irqtrace feature has added overheads to local_bh_disable and local_bh_enable that reduces UDP performance across x86_64 and IA64, even though IA64 does not support the irqtrace feature. Patch in question is [PATCH]lockdep: irqtrace subsystem, core http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=c ommit;h=de30a2b355ea85350ca2f58f3b9bf4e5bc007986 Prior to this patch, local_bh_disable was a short macro. Now it is a function which calls __local_bh_disable with added irq flags save and restore. The irq flags save and restore were also added to local_bh_enable, probably for injecting the trace irqs code. This overhead is on the generic code path across all architectures. On a IA_64 test machine (Itanium-2 1.6 GHz) running a benchmark like netperf's UDP streaming test, the added overhead results in a drop of 3% in throughput, as udp_sendmsg calls the local_bh_enable/disable several times. Other workloads that have heavy usages of local_bh_enable/disable could also be affected. The patch ideally should not have affected IA-64 performance as it does not have IRQ tracing support. A significant portion of the overhead is in the added irq flags save and restore, which I think is not needed if IRQ tracing is unused. A suggested patch is attached below that recovers the lost performance. However, the "ifdef"s in the patch are a bit ugly. Signed-off-by: Tim Chen <tim.c.chen@intel.com> Acked-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-30 12:04:02 +02:00
#else /* !CONFIG_TRACE_IRQFLAGS */
static inline void __local_bh_disable(unsigned long ip, unsigned int cnt)
[PATCH] Reducing local_bh_enable/disable overhead in irqtrace The recent changes from irqtrace feature has added overheads to local_bh_disable and local_bh_enable that reduces UDP performance across x86_64 and IA64, even though IA64 does not support the irqtrace feature. Patch in question is [PATCH]lockdep: irqtrace subsystem, core http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=c ommit;h=de30a2b355ea85350ca2f58f3b9bf4e5bc007986 Prior to this patch, local_bh_disable was a short macro. Now it is a function which calls __local_bh_disable with added irq flags save and restore. The irq flags save and restore were also added to local_bh_enable, probably for injecting the trace irqs code. This overhead is on the generic code path across all architectures. On a IA_64 test machine (Itanium-2 1.6 GHz) running a benchmark like netperf's UDP streaming test, the added overhead results in a drop of 3% in throughput, as udp_sendmsg calls the local_bh_enable/disable several times. Other workloads that have heavy usages of local_bh_enable/disable could also be affected. The patch ideally should not have affected IA-64 performance as it does not have IRQ tracing support. A significant portion of the overhead is in the added irq flags save and restore, which I think is not needed if IRQ tracing is unused. A suggested patch is attached below that recovers the lost performance. However, the "ifdef"s in the patch are a bit ugly. Signed-off-by: Tim Chen <tim.c.chen@intel.com> Acked-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-30 12:04:02 +02:00
{
add_preempt_count(cnt);
[PATCH] Reducing local_bh_enable/disable overhead in irqtrace The recent changes from irqtrace feature has added overheads to local_bh_disable and local_bh_enable that reduces UDP performance across x86_64 and IA64, even though IA64 does not support the irqtrace feature. Patch in question is [PATCH]lockdep: irqtrace subsystem, core http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=c ommit;h=de30a2b355ea85350ca2f58f3b9bf4e5bc007986 Prior to this patch, local_bh_disable was a short macro. Now it is a function which calls __local_bh_disable with added irq flags save and restore. The irq flags save and restore were also added to local_bh_enable, probably for injecting the trace irqs code. This overhead is on the generic code path across all architectures. On a IA_64 test machine (Itanium-2 1.6 GHz) running a benchmark like netperf's UDP streaming test, the added overhead results in a drop of 3% in throughput, as udp_sendmsg calls the local_bh_enable/disable several times. Other workloads that have heavy usages of local_bh_enable/disable could also be affected. The patch ideally should not have affected IA-64 performance as it does not have IRQ tracing support. A significant portion of the overhead is in the added irq flags save and restore, which I think is not needed if IRQ tracing is unused. A suggested patch is attached below that recovers the lost performance. However, the "ifdef"s in the patch are a bit ugly. Signed-off-by: Tim Chen <tim.c.chen@intel.com> Acked-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-30 12:04:02 +02:00
barrier();
}
#endif /* CONFIG_TRACE_IRQFLAGS */
void local_bh_disable(void)
{
__local_bh_disable((unsigned long)__builtin_return_address(0),
SOFTIRQ_DISABLE_OFFSET);
}
EXPORT_SYMBOL(local_bh_disable);
static void __local_bh_enable(unsigned int cnt)
{
WARN_ON_ONCE(in_irq());
WARN_ON_ONCE(!irqs_disabled());
if (softirq_count() == cnt)
trace_softirqs_on((unsigned long)__builtin_return_address(0));
sub_preempt_count(cnt);
}
/*
* Special-case - softirqs can safely be enabled in
* cond_resched_softirq(), or by __do_softirq(),
* without processing still-pending softirqs:
*/
void _local_bh_enable(void)
{
__local_bh_enable(SOFTIRQ_DISABLE_OFFSET);
}
EXPORT_SYMBOL(_local_bh_enable);
static inline void _local_bh_enable_ip(unsigned long ip)
{
WARN_ON_ONCE(in_irq() || irqs_disabled());
[PATCH] Reducing local_bh_enable/disable overhead in irqtrace The recent changes from irqtrace feature has added overheads to local_bh_disable and local_bh_enable that reduces UDP performance across x86_64 and IA64, even though IA64 does not support the irqtrace feature. Patch in question is [PATCH]lockdep: irqtrace subsystem, core http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=c ommit;h=de30a2b355ea85350ca2f58f3b9bf4e5bc007986 Prior to this patch, local_bh_disable was a short macro. Now it is a function which calls __local_bh_disable with added irq flags save and restore. The irq flags save and restore were also added to local_bh_enable, probably for injecting the trace irqs code. This overhead is on the generic code path across all architectures. On a IA_64 test machine (Itanium-2 1.6 GHz) running a benchmark like netperf's UDP streaming test, the added overhead results in a drop of 3% in throughput, as udp_sendmsg calls the local_bh_enable/disable several times. Other workloads that have heavy usages of local_bh_enable/disable could also be affected. The patch ideally should not have affected IA-64 performance as it does not have IRQ tracing support. A significant portion of the overhead is in the added irq flags save and restore, which I think is not needed if IRQ tracing is unused. A suggested patch is attached below that recovers the lost performance. However, the "ifdef"s in the patch are a bit ugly. Signed-off-by: Tim Chen <tim.c.chen@intel.com> Acked-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-30 12:04:02 +02:00
#ifdef CONFIG_TRACE_IRQFLAGS
local_irq_disable();
[PATCH] Reducing local_bh_enable/disable overhead in irqtrace The recent changes from irqtrace feature has added overheads to local_bh_disable and local_bh_enable that reduces UDP performance across x86_64 and IA64, even though IA64 does not support the irqtrace feature. Patch in question is [PATCH]lockdep: irqtrace subsystem, core http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=c ommit;h=de30a2b355ea85350ca2f58f3b9bf4e5bc007986 Prior to this patch, local_bh_disable was a short macro. Now it is a function which calls __local_bh_disable with added irq flags save and restore. The irq flags save and restore were also added to local_bh_enable, probably for injecting the trace irqs code. This overhead is on the generic code path across all architectures. On a IA_64 test machine (Itanium-2 1.6 GHz) running a benchmark like netperf's UDP streaming test, the added overhead results in a drop of 3% in throughput, as udp_sendmsg calls the local_bh_enable/disable several times. Other workloads that have heavy usages of local_bh_enable/disable could also be affected. The patch ideally should not have affected IA-64 performance as it does not have IRQ tracing support. A significant portion of the overhead is in the added irq flags save and restore, which I think is not needed if IRQ tracing is unused. A suggested patch is attached below that recovers the lost performance. However, the "ifdef"s in the patch are a bit ugly. Signed-off-by: Tim Chen <tim.c.chen@intel.com> Acked-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-30 12:04:02 +02:00
#endif
/*
* Are softirqs going to be turned on now:
*/
if (softirq_count() == SOFTIRQ_DISABLE_OFFSET)
trace_softirqs_on(ip);
/*
* Keep preemption disabled until we are done with
* softirq processing:
*/
sub_preempt_count(SOFTIRQ_DISABLE_OFFSET - 1);
if (unlikely(!in_interrupt() && local_softirq_pending()))
do_softirq();
dec_preempt_count();
[PATCH] Reducing local_bh_enable/disable overhead in irqtrace The recent changes from irqtrace feature has added overheads to local_bh_disable and local_bh_enable that reduces UDP performance across x86_64 and IA64, even though IA64 does not support the irqtrace feature. Patch in question is [PATCH]lockdep: irqtrace subsystem, core http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=c ommit;h=de30a2b355ea85350ca2f58f3b9bf4e5bc007986 Prior to this patch, local_bh_disable was a short macro. Now it is a function which calls __local_bh_disable with added irq flags save and restore. The irq flags save and restore were also added to local_bh_enable, probably for injecting the trace irqs code. This overhead is on the generic code path across all architectures. On a IA_64 test machine (Itanium-2 1.6 GHz) running a benchmark like netperf's UDP streaming test, the added overhead results in a drop of 3% in throughput, as udp_sendmsg calls the local_bh_enable/disable several times. Other workloads that have heavy usages of local_bh_enable/disable could also be affected. The patch ideally should not have affected IA-64 performance as it does not have IRQ tracing support. A significant portion of the overhead is in the added irq flags save and restore, which I think is not needed if IRQ tracing is unused. A suggested patch is attached below that recovers the lost performance. However, the "ifdef"s in the patch are a bit ugly. Signed-off-by: Tim Chen <tim.c.chen@intel.com> Acked-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-30 12:04:02 +02:00
#ifdef CONFIG_TRACE_IRQFLAGS
local_irq_enable();
[PATCH] Reducing local_bh_enable/disable overhead in irqtrace The recent changes from irqtrace feature has added overheads to local_bh_disable and local_bh_enable that reduces UDP performance across x86_64 and IA64, even though IA64 does not support the irqtrace feature. Patch in question is [PATCH]lockdep: irqtrace subsystem, core http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=c ommit;h=de30a2b355ea85350ca2f58f3b9bf4e5bc007986 Prior to this patch, local_bh_disable was a short macro. Now it is a function which calls __local_bh_disable with added irq flags save and restore. The irq flags save and restore were also added to local_bh_enable, probably for injecting the trace irqs code. This overhead is on the generic code path across all architectures. On a IA_64 test machine (Itanium-2 1.6 GHz) running a benchmark like netperf's UDP streaming test, the added overhead results in a drop of 3% in throughput, as udp_sendmsg calls the local_bh_enable/disable several times. Other workloads that have heavy usages of local_bh_enable/disable could also be affected. The patch ideally should not have affected IA-64 performance as it does not have IRQ tracing support. A significant portion of the overhead is in the added irq flags save and restore, which I think is not needed if IRQ tracing is unused. A suggested patch is attached below that recovers the lost performance. However, the "ifdef"s in the patch are a bit ugly. Signed-off-by: Tim Chen <tim.c.chen@intel.com> Acked-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-30 12:04:02 +02:00
#endif
preempt_check_resched();
}
void local_bh_enable(void)
{
_local_bh_enable_ip((unsigned long)__builtin_return_address(0));
}
EXPORT_SYMBOL(local_bh_enable);
void local_bh_enable_ip(unsigned long ip)
{
_local_bh_enable_ip(ip);
}
EXPORT_SYMBOL(local_bh_enable_ip);
/*
* We restart softirq processing MAX_SOFTIRQ_RESTART times,
* and we fall back to softirqd after that.
*
* This number has been established via experimentation.
* The two things to balance is latency against fairness -
* we want to handle softirqs as soon as possible, but they
* should not be able to lock up the box.
*/
#define MAX_SOFTIRQ_RESTART 10
asmlinkage void __do_softirq(void)
{
struct softirq_action *h;
__u32 pending;
int max_restart = MAX_SOFTIRQ_RESTART;
int cpu;
pending = local_softirq_pending();
account_system_vtime(current);
__local_bh_disable((unsigned long)__builtin_return_address(0),
SOFTIRQ_OFFSET);
lockdep_softirq_enter();
cpu = smp_processor_id();
restart:
/* Reset the pending bitmask before enabling irqs */
set_softirq_pending(0);
local_irq_enable();
h = softirq_vec;
do {
if (pending & 1) {
unsigned int vec_nr = h - softirq_vec;
int prev_count = preempt_count();
kstat_incr_softirqs_this_cpu(vec_nr);
trace_softirq_entry(vec_nr);
h->action(h);
trace_softirq_exit(vec_nr);
if (unlikely(prev_count != preempt_count())) {
printk(KERN_ERR "huh, entered softirq %u %s %p"
"with preempt_count %08x,"
" exited with %08x?\n", vec_nr,
softirq_to_name[vec_nr], h->action,
prev_count, preempt_count());
preempt_count() = prev_count;
}
rcu_bh_qs(cpu);
}
h++;
pending >>= 1;
} while (pending);
local_irq_disable();
pending = local_softirq_pending();
if (pending && --max_restart)
goto restart;
if (pending)
wakeup_softirqd();
lockdep_softirq_exit();
account_system_vtime(current);
__local_bh_enable(SOFTIRQ_OFFSET);
}
#ifndef __ARCH_HAS_DO_SOFTIRQ
asmlinkage void do_softirq(void)
{
__u32 pending;
unsigned long flags;
if (in_interrupt())
return;
local_irq_save(flags);
pending = local_softirq_pending();
if (pending)
__do_softirq();
local_irq_restore(flags);
}
#endif
/*
* Enter an interrupt context.
*/
void irq_enter(void)
{
int cpu = smp_processor_id();
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 21:55:32 +01:00
rcu_irq_enter();
if (idle_cpu(cpu) && !in_interrupt()) {
/*
* Prevent raise_softirq from needlessly waking up ksoftirqd
* here, as softirq will be serviced on return from interrupt.
*/
local_bh_disable();
tick_check_idle(cpu);
_local_bh_enable();
}
__irq_enter();
}
#ifdef __ARCH_IRQ_EXIT_IRQS_DISABLED
# define invoke_softirq() __do_softirq()
#else
# define invoke_softirq() do_softirq()
#endif
/*
* Exit an interrupt context. Process softirqs if needed and possible:
*/
void irq_exit(void)
{
account_system_vtime(current);
trace_hardirq_exit();
sub_preempt_count(IRQ_EXIT_OFFSET);
if (!in_interrupt() && local_softirq_pending())
invoke_softirq();
rcu_irq_exit();
#ifdef CONFIG_NO_HZ
/* Make sure that timer wheel updates are propagated */
"Tree RCU": scalable classic RCU implementation This patch fixes a long-standing performance bug in classic RCU that results in massive internal-to-RCU lock contention on systems with more than a few hundred CPUs. Although this patch creates a separate flavor of RCU for ease of review and patch maintenance, it is intended to replace classic RCU. This patch still handles stress better than does mainline, so I am still calling it ready for inclusion. This patch is against the -tip tree. Nevertheless, experience on an actual 1000+ CPU machine would still be most welcome. Most of the changes noted below were found while creating an rcutiny (which should permit ejecting the current rcuclassic) and while doing detailed line-by-line documentation. Updates from v9 (http://lkml.org/lkml/2008/12/2/334): o Fixes from remainder of line-by-line code walkthrough, including comment spelling, initialization, undesirable narrowing due to type conversion, removing redundant memory barriers, removing redundant local-variable initialization, and removing redundant local variables. I do not believe that any of these fixes address the CPU-hotplug issues that Andi Kleen was seeing, but please do give it a whirl in case the machine is smarter than I am. A writeup from the walkthrough may be found at the following URL, in case you are suffering from terminal insomnia or masochism: http://www.kernel.org/pub/linux/kernel/people/paulmck/tmp/rcutree-walkthrough.2008.12.16a.pdf o Made rcutree tracing use seq_file, as suggested some time ago by Lai Jiangshan. o Added a .csv variant of the rcudata debugfs trace file, to allow people having thousands of CPUs to drop the data into a spreadsheet. Tested with oocalc and gnumeric. Updated documentation to suit. Updates from v8 (http://lkml.org/lkml/2008/11/15/139): o Fix a theoretical race between grace-period initialization and force_quiescent_state() that could occur if more than three jiffies were required to carry out the grace-period initialization. Which it might, if you had enough CPUs. o Apply Ingo's printk-standardization patch. o Substitute local variables for repeated accesses to global variables. o Fix comment misspellings and redundant (but harmless) increments of ->n_rcu_pending (this latter after having explicitly added it). o Apply checkpatch fixes. Updates from v7 (http://lkml.org/lkml/2008/10/10/291): o Fixed a number of problems noted by Gautham Shenoy, including the cpu-stall-detection bug that he was having difficulty convincing me was real. ;-) o Changed cpu-stall detection to wait for ten seconds rather than three in order to reduce false positive, as suggested by Ingo Molnar. o Produced a design document (http://lwn.net/Articles/305782/). The act of writing this document uncovered a number of both theoretical and "here and now" bugs as noted below. o Fix dynticks_nesting accounting confusion, simplify WARN_ON() condition, fix kerneldoc comments, and add memory barriers in dynticks interface functions. o Add more data to tracing. o Remove unused "rcu_barrier" field from rcu_data structure. o Count calls to rcu_pending() from scheduling-clock interrupt to use as a surrogate timebase should jiffies stop counting. o Fix a theoretical race between force_quiescent_state() and grace-period initialization. Yes, initialization does have to go on for some jiffies for this race to occur, but given enough CPUs... Updates from v6 (http://lkml.org/lkml/2008/9/23/448): o Fix a number of checkpatch.pl complaints. o Apply review comments from Ingo Molnar and Lai Jiangshan on the stall-detection code. o Fix several bugs in !CONFIG_SMP builds. o Fix a misspelled config-parameter name so that RCU now announces at boot time if stall detection is configured. o Run tests on numerous combinations of configurations parameters, which after the fixes above, now build and run correctly. Updates from v5 (http://lkml.org/lkml/2008/9/15/92, bad subject line): o Fix a compiler error in the !CONFIG_FANOUT_EXACT case (blew a changeset some time ago, and finally got around to retesting this option). o Fix some tracing bugs in rcupreempt that caused incorrect totals to be printed. o I now test with a more brutal random-selection online/offline script (attached). Probably more brutal than it needs to be on the people reading it as well, but so it goes. o A number of optimizations and usability improvements: o Make rcu_pending() ignore the grace-period timeout when there is no grace period in progress. o Make force_quiescent_state() avoid going for a global lock in the case where there is no grace period in progress. o Rearrange struct fields to improve struct layout. o Make call_rcu() initiate a grace period if RCU was idle, rather than waiting for the next scheduling clock interrupt. o Invoke rcu_irq_enter() and rcu_irq_exit() only when idle, as suggested by Andi Kleen. I still don't completely trust this change, and might back it out. o Make CONFIG_RCU_TRACE be the single config variable manipulated for all forms of RCU, instead of the prior confusion. o Document tracing files and formats for both rcupreempt and rcutree. Updates from v4 for those missing v5 given its bad subject line: o Separated dynticks interface so that NMIs and irqs call separate functions, greatly simplifying it. In particular, this code no longer requires a proof of correctness. ;-) o Separated dynticks state out into its own per-CPU structure, avoiding the duplicated accounting. o The case where a dynticks-idle CPU runs an irq handler that invokes call_rcu() is now correctly handled, forcing that CPU out of dynticks-idle mode. o Review comments have been applied (thank you all!!!). For but one example, fixed the dynticks-ordering issue that Manfred pointed out, saving me much debugging. ;-) o Adjusted rcuclassic and rcupreempt to handle dynticks changes. Attached is an updated patch to Classic RCU that applies a hierarchy, greatly reducing the contention on the top-level lock for large machines. This passes 10-hour concurrent rcutorture and online-offline testing on 128-CPU ppc64 without dynticks enabled, and exposes some timekeeping bugs in presence of dynticks (exciting working on a system where "sleep 1" hangs until interrupted...), which were fixed in the 2.6.27 kernel. It is getting more reliable than mainline by some measures, so the next version will be against -tip for inclusion. See also Manfred Spraul's recent patches (or his earlier work from 2004 at http://marc.info/?l=linux-kernel&m=108546384711797&w=2). We will converge onto a common patch in the fullness of time, but are currently exploring different regions of the design space. That said, I have already gratefully stolen quite a few of Manfred's ideas. This patch provides CONFIG_RCU_FANOUT, which controls the bushiness of the RCU hierarchy. Defaults to 32 on 32-bit machines and 64 on 64-bit machines. If CONFIG_NR_CPUS is less than CONFIG_RCU_FANOUT, there is no hierarchy. By default, the RCU initialization code will adjust CONFIG_RCU_FANOUT to balance the hierarchy, so strongly NUMA architectures may choose to set CONFIG_RCU_FANOUT_EXACT to disable this balancing, allowing the hierarchy to be exactly aligned to the underlying hardware. Up to two levels of hierarchy are permitted (in addition to the root node), allowing up to 16,384 CPUs on 32-bit systems and up to 262,144 CPUs on 64-bit systems. I just know that I am going to regret saying this, but this seems more than sufficient for the foreseeable future. (Some architectures might wish to set CONFIG_RCU_FANOUT=4, which would limit such architectures to 64 CPUs. If this becomes a real problem, additional levels can be added, but I doubt that it will make a significant difference on real hardware.) In the common case, a given CPU will manipulate its private rcu_data structure and the rcu_node structure that it shares with its immediate neighbors. This can reduce both lock and memory contention by multiple orders of magnitude, which should eliminate the need for the strange manipulations that are reported to be required when running Linux on very large systems. Some shortcomings: o More bugs will probably surface as a result of an ongoing line-by-line code inspection. Patches will be provided as required. o There are probably hangs, rcutorture failures, &c. Seems quite stable on a 128-CPU machine, but that is kind of small compared to 4096 CPUs. However, seems to do better than mainline. Patches will be provided as required. o The memory footprint of this version is several KB larger than rcuclassic. A separate UP-only rcutiny patch will be provided, which will reduce the memory footprint significantly, even compared to the old rcuclassic. One such patch passes light testing, and has a memory footprint smaller even than rcuclassic. Initial reaction from various embedded guys was "it is not worth it", so am putting it aside. Credits: o Manfred Spraul for ideas, review comments, and bugs spotted, as well as some good friendly competition. ;-) o Josh Triplett, Ingo Molnar, Peter Zijlstra, Mathieu Desnoyers, Lai Jiangshan, Andi Kleen, Andy Whitcroft, and Andrew Morton for reviews and comments. o Thomas Gleixner for much-needed help with some timer issues (see patches below). o Jon M. Tollefson, Tim Pepper, Andrew Theurer, Jose R. Santos, Andy Whitcroft, Darrick Wong, Nishanth Aravamudan, Anton Blanchard, Dave Kleikamp, and Nathan Lynch for keeping machines alive despite my heavy abuse^Wtesting. Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-12-18 21:55:32 +01:00
if (idle_cpu(smp_processor_id()) && !in_interrupt() && !need_resched())
tick_nohz_stop_sched_tick(0);
#endif
preempt_enable_no_resched();
}
/*
* This function must run with irqs disabled!
*/
inline void raise_softirq_irqoff(unsigned int nr)
{
__raise_softirq_irqoff(nr);
/*
* If we're in an interrupt or softirq, we're done
* (this also catches softirq-disabled code). We will
* actually run the softirq once we return from
* the irq or softirq.
*
* Otherwise we wake up ksoftirqd to make sure we
* schedule the softirq soon.
*/
if (!in_interrupt())
wakeup_softirqd();
}
void raise_softirq(unsigned int nr)
{
unsigned long flags;
local_irq_save(flags);
raise_softirq_irqoff(nr);
local_irq_restore(flags);
}
void open_softirq(int nr, void (*action)(struct softirq_action *))
{
softirq_vec[nr].action = action;
}
/*
* Tasklets
*/
struct tasklet_head
{
struct tasklet_struct *head;
struct tasklet_struct **tail;
};
static DEFINE_PER_CPU(struct tasklet_head, tasklet_vec);
static DEFINE_PER_CPU(struct tasklet_head, tasklet_hi_vec);
void __tasklet_schedule(struct tasklet_struct *t)
{
unsigned long flags;
local_irq_save(flags);
t->next = NULL;
*__this_cpu_read(tasklet_vec.tail) = t;
__this_cpu_write(tasklet_vec.tail, &(t->next));
raise_softirq_irqoff(TASKLET_SOFTIRQ);
local_irq_restore(flags);
}
EXPORT_SYMBOL(__tasklet_schedule);
void __tasklet_hi_schedule(struct tasklet_struct *t)
{
unsigned long flags;
local_irq_save(flags);
t->next = NULL;
*__this_cpu_read(tasklet_hi_vec.tail) = t;
__this_cpu_write(tasklet_hi_vec.tail, &(t->next));
raise_softirq_irqoff(HI_SOFTIRQ);
local_irq_restore(flags);
}
EXPORT_SYMBOL(__tasklet_hi_schedule);
void __tasklet_hi_schedule_first(struct tasklet_struct *t)
{
BUG_ON(!irqs_disabled());
t->next = __this_cpu_read(tasklet_hi_vec.head);
__this_cpu_write(tasklet_hi_vec.head, t);
__raise_softirq_irqoff(HI_SOFTIRQ);
}
EXPORT_SYMBOL(__tasklet_hi_schedule_first);
static void tasklet_action(struct softirq_action *a)
{
struct tasklet_struct *list;
local_irq_disable();
list = __this_cpu_read(tasklet_vec.head);
__this_cpu_write(tasklet_vec.head, NULL);
__this_cpu_write(tasklet_vec.tail, &__get_cpu_var(tasklet_vec).head);
local_irq_enable();
while (list) {
struct tasklet_struct *t = list;
list = list->next;
if (tasklet_trylock(t)) {
if (!atomic_read(&t->count)) {
if (!test_and_clear_bit(TASKLET_STATE_SCHED, &t->state))
BUG();
t->func(t->data);
tasklet_unlock(t);
continue;
}
tasklet_unlock(t);
}
local_irq_disable();
t->next = NULL;
*__this_cpu_read(tasklet_vec.tail) = t;
__this_cpu_write(tasklet_vec.tail, &(t->next));
__raise_softirq_irqoff(TASKLET_SOFTIRQ);
local_irq_enable();
}
}
static void tasklet_hi_action(struct softirq_action *a)
{
struct tasklet_struct *list;
local_irq_disable();
list = __this_cpu_read(tasklet_hi_vec.head);
__this_cpu_write(tasklet_hi_vec.head, NULL);
__this_cpu_write(tasklet_hi_vec.tail, &__get_cpu_var(tasklet_hi_vec).head);
local_irq_enable();
while (list) {
struct tasklet_struct *t = list;
list = list->next;
if (tasklet_trylock(t)) {
if (!atomic_read(&t->count)) {
if (!test_and_clear_bit(TASKLET_STATE_SCHED, &t->state))
BUG();
t->func(t->data);
tasklet_unlock(t);
continue;
}
tasklet_unlock(t);
}
local_irq_disable();
t->next = NULL;
*__this_cpu_read(tasklet_hi_vec.tail) = t;
__this_cpu_write(tasklet_hi_vec.tail, &(t->next));
__raise_softirq_irqoff(HI_SOFTIRQ);
local_irq_enable();
}
}
void tasklet_init(struct tasklet_struct *t,
void (*func)(unsigned long), unsigned long data)
{
t->next = NULL;
t->state = 0;
atomic_set(&t->count, 0);
t->func = func;
t->data = data;
}
EXPORT_SYMBOL(tasklet_init);
void tasklet_kill(struct tasklet_struct *t)
{
if (in_interrupt())
printk("Attempt to kill tasklet from interrupt\n");
while (test_and_set_bit(TASKLET_STATE_SCHED, &t->state)) {
do {
yield();
} while (test_bit(TASKLET_STATE_SCHED, &t->state));
}
tasklet_unlock_wait(t);
clear_bit(TASKLET_STATE_SCHED, &t->state);
}
EXPORT_SYMBOL(tasklet_kill);
/*
* tasklet_hrtimer
*/
/*
* The trampoline is called when the hrtimer expires. It schedules a tasklet
* to run __tasklet_hrtimer_trampoline() which in turn will call the intended
* hrtimer callback, but from softirq context.
*/
static enum hrtimer_restart __hrtimer_tasklet_trampoline(struct hrtimer *timer)
{
struct tasklet_hrtimer *ttimer =
container_of(timer, struct tasklet_hrtimer, timer);
tasklet_hi_schedule(&ttimer->tasklet);
return HRTIMER_NORESTART;
}
/*
* Helper function which calls the hrtimer callback from
* tasklet/softirq context
*/
static void __tasklet_hrtimer_trampoline(unsigned long data)
{
struct tasklet_hrtimer *ttimer = (void *)data;
enum hrtimer_restart restart;
restart = ttimer->function(&ttimer->timer);
if (restart != HRTIMER_NORESTART)
hrtimer_restart(&ttimer->timer);
}
/**
* tasklet_hrtimer_init - Init a tasklet/hrtimer combo for softirq callbacks
* @ttimer: tasklet_hrtimer which is initialized
* @function: hrtimer callback funtion which gets called from softirq context
* @which_clock: clock id (CLOCK_MONOTONIC/CLOCK_REALTIME)
* @mode: hrtimer mode (HRTIMER_MODE_ABS/HRTIMER_MODE_REL)
*/
void tasklet_hrtimer_init(struct tasklet_hrtimer *ttimer,
enum hrtimer_restart (*function)(struct hrtimer *),
clockid_t which_clock, enum hrtimer_mode mode)
{
hrtimer_init(&ttimer->timer, which_clock, mode);
ttimer->timer.function = __hrtimer_tasklet_trampoline;
tasklet_init(&ttimer->tasklet, __tasklet_hrtimer_trampoline,
(unsigned long)ttimer);
ttimer->function = function;
}
EXPORT_SYMBOL_GPL(tasklet_hrtimer_init);
/*
* Remote softirq bits
*/
DEFINE_PER_CPU(struct list_head [NR_SOFTIRQS], softirq_work_list);
EXPORT_PER_CPU_SYMBOL(softirq_work_list);
static void __local_trigger(struct call_single_data *cp, int softirq)
{
struct list_head *head = &__get_cpu_var(softirq_work_list[softirq]);
list_add_tail(&cp->list, head);
/* Trigger the softirq only if the list was previously empty. */
if (head->next == &cp->list)
raise_softirq_irqoff(softirq);
}
#ifdef CONFIG_USE_GENERIC_SMP_HELPERS
static void remote_softirq_receive(void *data)
{
struct call_single_data *cp = data;
unsigned long flags;
int softirq;
softirq = cp->priv;
local_irq_save(flags);
__local_trigger(cp, softirq);
local_irq_restore(flags);
}
static int __try_remote_softirq(struct call_single_data *cp, int cpu, int softirq)
{
if (cpu_online(cpu)) {
cp->func = remote_softirq_receive;
cp->info = cp;
cp->flags = 0;
cp->priv = softirq;
__smp_call_function_single(cpu, cp, 0);
return 0;
}
return 1;
}
#else /* CONFIG_USE_GENERIC_SMP_HELPERS */
static int __try_remote_softirq(struct call_single_data *cp, int cpu, int softirq)
{
return 1;
}
#endif
/**
* __send_remote_softirq - try to schedule softirq work on a remote cpu
* @cp: private SMP call function data area
* @cpu: the remote cpu
* @this_cpu: the currently executing cpu
* @softirq: the softirq for the work
*
* Attempt to schedule softirq work on a remote cpu. If this cannot be
* done, the work is instead queued up on the local cpu.
*
* Interrupts must be disabled.
*/
void __send_remote_softirq(struct call_single_data *cp, int cpu, int this_cpu, int softirq)
{
if (cpu == this_cpu || __try_remote_softirq(cp, cpu, softirq))
__local_trigger(cp, softirq);
}
EXPORT_SYMBOL(__send_remote_softirq);
/**
* send_remote_softirq - try to schedule softirq work on a remote cpu
* @cp: private SMP call function data area
* @cpu: the remote cpu
* @softirq: the softirq for the work
*
* Like __send_remote_softirq except that disabling interrupts and
* computing the current cpu is done for the caller.
*/
void send_remote_softirq(struct call_single_data *cp, int cpu, int softirq)
{
unsigned long flags;
int this_cpu;
local_irq_save(flags);
this_cpu = smp_processor_id();
__send_remote_softirq(cp, cpu, this_cpu, softirq);
local_irq_restore(flags);
}
EXPORT_SYMBOL(send_remote_softirq);
static int __cpuinit remote_softirq_cpu_notify(struct notifier_block *self,
unsigned long action, void *hcpu)
{
/*
* If a CPU goes away, splice its entries to the current CPU
* and trigger a run of the softirq
*/
if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) {
int cpu = (unsigned long) hcpu;
int i;
local_irq_disable();
for (i = 0; i < NR_SOFTIRQS; i++) {
struct list_head *head = &per_cpu(softirq_work_list[i], cpu);
struct list_head *local_head;
if (list_empty(head))
continue;
local_head = &__get_cpu_var(softirq_work_list[i]);
list_splice_init(head, local_head);
raise_softirq_irqoff(i);
}
local_irq_enable();
}
return NOTIFY_OK;
}
static struct notifier_block __cpuinitdata remote_softirq_cpu_notifier = {
.notifier_call = remote_softirq_cpu_notify,
};
void __init softirq_init(void)
{
int cpu;
for_each_possible_cpu(cpu) {
int i;
per_cpu(tasklet_vec, cpu).tail =
&per_cpu(tasklet_vec, cpu).head;
per_cpu(tasklet_hi_vec, cpu).tail =
&per_cpu(tasklet_hi_vec, cpu).head;
for (i = 0; i < NR_SOFTIRQS; i++)
INIT_LIST_HEAD(&per_cpu(softirq_work_list[i], cpu));
}
register_hotcpu_notifier(&remote_softirq_cpu_notifier);
open_softirq(TASKLET_SOFTIRQ, tasklet_action);
open_softirq(HI_SOFTIRQ, tasklet_hi_action);
}
static int run_ksoftirqd(void * __bind_cpu)
{
set_current_state(TASK_INTERRUPTIBLE);
current->flags |= PF_KSOFTIRQD;
while (!kthread_should_stop()) {
preempt_disable();
if (!local_softirq_pending()) {
preempt_enable_no_resched();
schedule();
preempt_disable();
}
__set_current_state(TASK_RUNNING);
while (local_softirq_pending()) {
/* Preempt disable stops cpu going offline.
If already offline, we'll be on wrong CPU:
don't process */
if (cpu_is_offline((long)__bind_cpu))
goto wait_to_die;
do_softirq();
preempt_enable_no_resched();
cond_resched();
preempt_disable();
rcu: refactor RCU's context-switch handling The addition of preemptible RCU to treercu resulted in a bit of confusion and inefficiency surrounding the handling of context switches for RCU-sched and for RCU-preempt. For RCU-sched, a context switch is a quiescent state, pure and simple, just like it always has been. For RCU-preempt, a context switch is in no way a quiescent state, but special handling is required when a task blocks in an RCU read-side critical section. However, the callout from the scheduler and the outer loop in ksoftirqd still calls something named rcu_sched_qs(), whose name is no longer accurate. Furthermore, when rcu_check_callbacks() notes an RCU-sched quiescent state, it ends up unnecessarily (though harmlessly, aside from the performance hit) enqueuing the current task if it happens to be running in an RCU-preempt read-side critical section. This not only increases the maximum latency of scheduler_tick(), it also needlessly increases the overhead of the next outermost rcu_read_unlock() invocation. This patch addresses this situation by separating the notion of RCU's context-switch handling from that of RCU-sched's quiescent states. The context-switch handling is covered by rcu_note_context_switch() in general and by rcu_preempt_note_context_switch() for preemptible RCU. This permits rcu_sched_qs() to handle quiescent states and only quiescent states. It also reduces the maximum latency of scheduler_tick(), though probably by much less than a microsecond. Finally, it means that tasks within preemptible-RCU read-side critical sections avoid incurring the overhead of queuing unless there really is a context switch. Suggested-by: Lai Jiangshan <laijs@cn.fujitsu.com> Acked-by: Lai Jiangshan <laijs@cn.fujitsu.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <peterz@infradead.org>
2010-04-02 02:37:01 +02:00
rcu_note_context_switch((long)__bind_cpu);
}
preempt_enable();
set_current_state(TASK_INTERRUPTIBLE);
}
__set_current_state(TASK_RUNNING);
return 0;
wait_to_die:
preempt_enable();
/* Wait for kthread_stop */
set_current_state(TASK_INTERRUPTIBLE);
while (!kthread_should_stop()) {
schedule();
set_current_state(TASK_INTERRUPTIBLE);
}
__set_current_state(TASK_RUNNING);
return 0;
}
#ifdef CONFIG_HOTPLUG_CPU
/*
* tasklet_kill_immediate is called to remove a tasklet which can already be
* scheduled for execution on @cpu.
*
* Unlike tasklet_kill, this function removes the tasklet
* _immediately_, even if the tasklet is in TASKLET_STATE_SCHED state.
*
* When this function is called, @cpu must be in the CPU_DEAD state.
*/
void tasklet_kill_immediate(struct tasklet_struct *t, unsigned int cpu)
{
struct tasklet_struct **i;
BUG_ON(cpu_online(cpu));
BUG_ON(test_bit(TASKLET_STATE_RUN, &t->state));
if (!test_bit(TASKLET_STATE_SCHED, &t->state))
return;
/* CPU is dead, so no lock needed. */
for (i = &per_cpu(tasklet_vec, cpu).head; *i; i = &(*i)->next) {
if (*i == t) {
*i = t->next;
/* If this was the tail element, move the tail ptr */
if (*i == NULL)
per_cpu(tasklet_vec, cpu).tail = i;
return;
}
}
BUG();
}
static void takeover_tasklets(unsigned int cpu)
{
/* CPU is dead, so no lock needed. */
local_irq_disable();
/* Find end, append list for that CPU. */
if (&per_cpu(tasklet_vec, cpu).head != per_cpu(tasklet_vec, cpu).tail) {
*__this_cpu_read(tasklet_vec.tail) = per_cpu(tasklet_vec, cpu).head;
this_cpu_write(tasklet_vec.tail, per_cpu(tasklet_vec, cpu).tail);
per_cpu(tasklet_vec, cpu).head = NULL;
per_cpu(tasklet_vec, cpu).tail = &per_cpu(tasklet_vec, cpu).head;
}
raise_softirq_irqoff(TASKLET_SOFTIRQ);
if (&per_cpu(tasklet_hi_vec, cpu).head != per_cpu(tasklet_hi_vec, cpu).tail) {
*__this_cpu_read(tasklet_hi_vec.tail) = per_cpu(tasklet_hi_vec, cpu).head;
__this_cpu_write(tasklet_hi_vec.tail, per_cpu(tasklet_hi_vec, cpu).tail);
per_cpu(tasklet_hi_vec, cpu).head = NULL;
per_cpu(tasklet_hi_vec, cpu).tail = &per_cpu(tasklet_hi_vec, cpu).head;
}
raise_softirq_irqoff(HI_SOFTIRQ);
local_irq_enable();
}
#endif /* CONFIG_HOTPLUG_CPU */
static int __cpuinit cpu_callback(struct notifier_block *nfb,
unsigned long action,
void *hcpu)
{
int hotcpu = (unsigned long)hcpu;
struct task_struct *p;
switch (action) {
case CPU_UP_PREPARE:
case CPU_UP_PREPARE_FROZEN:
p = kthread_create(run_ksoftirqd, hcpu, "ksoftirqd/%d", hotcpu);
if (IS_ERR(p)) {
printk("ksoftirqd for %i failed\n", hotcpu);
return notifier_from_errno(PTR_ERR(p));
}
kthread_bind(p, hotcpu);
per_cpu(ksoftirqd, hotcpu) = p;
break;
case CPU_ONLINE:
case CPU_ONLINE_FROZEN:
wake_up_process(per_cpu(ksoftirqd, hotcpu));
break;
#ifdef CONFIG_HOTPLUG_CPU
case CPU_UP_CANCELED:
case CPU_UP_CANCELED_FROZEN:
if (!per_cpu(ksoftirqd, hotcpu))
break;
/* Unbind so it can run. Fall thru. */
kthread_bind(per_cpu(ksoftirqd, hotcpu),
cpumask_any(cpu_online_mask));
case CPU_DEAD:
case CPU_DEAD_FROZEN: {
static const struct sched_param param = {
.sched_priority = MAX_RT_PRIO-1
};
p = per_cpu(ksoftirqd, hotcpu);
per_cpu(ksoftirqd, hotcpu) = NULL;
sched_setscheduler_nocheck(p, SCHED_FIFO, &param);
kthread_stop(p);
takeover_tasklets(hotcpu);
break;
}
#endif /* CONFIG_HOTPLUG_CPU */
}
return NOTIFY_OK;
}
static struct notifier_block __cpuinitdata cpu_nfb = {
.notifier_call = cpu_callback
};
static __init int spawn_ksoftirqd(void)
{
void *cpu = (void *)(long)smp_processor_id();
int err = cpu_callback(&cpu_nfb, CPU_UP_PREPARE, cpu);
BUG_ON(err != NOTIFY_OK);
cpu_callback(&cpu_nfb, CPU_ONLINE, cpu);
register_cpu_notifier(&cpu_nfb);
return 0;
}
early_initcall(spawn_ksoftirqd);
/*
* [ These __weak aliases are kept in a separate compilation unit, so that
* GCC does not inline them incorrectly. ]
*/
int __init __weak early_irq_init(void)
{
return 0;
}
#ifdef CONFIG_GENERIC_HARDIRQS
int __init __weak arch_probe_nr_irqs(void)
{
return NR_IRQS_LEGACY;
}
int __init __weak arch_early_irq_init(void)
{
return 0;
}
#endif