From baf25731e54d06eb13dc4eda78c6dc7da4ce84e8 Mon Sep 17 00:00:00 2001 From: "Zhang, Yanmin" Date: Tue, 9 Sep 2008 11:26:33 +0800 Subject: sched: fix 2.6.27-rc5 couldn't boot on tulsa machine randomly On my tulsa x86-64 machine, kernel 2.6.25-rc5 couldn't boot randomly. Basically, function __enable_runtime forgets to reset rt_rq->rt_throttled to 0. When every cpu is up, per-cpu migration_thread is created and it runs very fast, sometimes to mark the corresponding rt_rq->rt_throttled to 1 very quickly. After all cpus are up, with below calling chain: sched_init_smp => arch_init_sched_domains => build_sched_domains => ... => cpu_attach_domain => rq_attach_root => set_rq_online => ... => _enable_runtime _enable_runtime is called against every rt_rq again, so rt_rq->rt_time is reset to 0, but rt_rq->rt_throttled might be still 1. Later on function do_sched_rt_period_timer couldn't reset it, and all RT tasks couldn't be scheduled to run on that cpu. here is RT task migration_thread which is woken up when a task is migrated to another cpu. Below patch fixes it against 2.6.27-rc5. Signed-off-by: Zhang Yanmin Signed-off-by: Ingo Molnar --- kernel/sched_rt.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 552310798da..1113157b205 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -350,6 +350,7 @@ static void __enable_runtime(struct rq *rq) spin_lock(&rt_rq->rt_runtime_lock); rt_rq->rt_runtime = rt_b->rt_runtime; rt_rq->rt_time = 0; + rt_rq->rt_throttled = 0; spin_unlock(&rt_rq->rt_runtime_lock); spin_unlock(&rt_b->rt_runtime_lock); } -- cgit v1.2.3-70-g09d2 From ec5d498991e87c74730509508b25c3959192b7e7 Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Wed, 10 Sep 2008 17:00:19 -0700 Subject: sched: fix deadlock in setting scheduler parameter to zero Andrei Gusev wrote: > I played witch scheduler settings. After doing something like: > echo -n 1000000 >sched_rt_period_us > > command is locked. I found in kernel.log: > > Sep 11 00:39:34 zaratustra > Sep 11 00:39:34 zaratustra Pid: 4495, comm: bash Tainted: G W > (2.6.26.3 #12) > Sep 11 00:39:34 zaratustra EIP: 0060:[] EFLAGS: 00210246 CPU: 0 > Sep 11 00:39:34 zaratustra EIP is at div64_u64+0x57/0x80 > Sep 11 00:39:34 zaratustra EAX: 0000389f EBX: 00000000 ECX: 00000000 > EDX: 00000000 > Sep 11 00:39:34 zaratustra ESI: d9800000 EDI: d9800000 EBP: 0000389f > ESP: ea7a6edc > Sep 11 00:39:34 zaratustra DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068 > Sep 11 00:39:34 zaratustra Process bash (pid: 4495, ti=ea7a6000 > task=ea744000 task.ti=ea7a6000) > Sep 11 00:39:34 zaratustra Stack: 00000000 000003e8 d9800000 0000389f > c0119042 00000000 00000000 00000001 > Sep 11 00:39:34 zaratustra 00000000 00000000 ea7a6f54 00010000 00000000 > c04d2e80 00000001 000e7ef0 > Sep 11 00:39:34 zaratustra c01191a3 00000000 00000000 ea7a6fa0 00000001 > ffffffff c04d2e80 ea5b2480 > Sep 11 00:39:34 zaratustra Call Trace: > Sep 11 00:39:34 zaratustra [] __rt_schedulable+0x52/0x130 > Sep 11 00:39:34 zaratustra [] sched_rt_handler+0x83/0x120 > Sep 11 00:39:34 zaratustra [] proc_sys_call_handler+0xb6/0xd0 > Sep 11 00:39:34 zaratustra [] proc_sys_write+0x0/0x20 > Sep 11 00:39:34 zaratustra [] proc_sys_write+0x19/0x20 > Sep 11 00:39:34 zaratustra [] vfs_write+0xa8/0x140 > Sep 11 00:39:34 zaratustra [] sys_write+0x41/0x80 > Sep 11 00:39:34 zaratustra [] sysenter_past_esp+0x6a/0x91 > Sep 11 00:39:34 zaratustra ======================= > Sep 11 00:39:34 zaratustra Code: c8 41 0f ad f3 d3 ee f6 c1 20 0f 45 de > 31 f6 0f ad ef d3 ed f6 c1 20 0f 45 fd 0f 45 ee 31 c9 39 eb 89 fe 89 ea > 77 08 89 e8 31 d2 f3 89 c1 89 f0 8b 7c 24 08 f7 f3 8b 74 24 04 89 > ca 8b 1c 24 > Sep 11 00:39:34 zaratustra EIP: [] div64_u64+0x57/0x80 SS:ESP > 0068:ea7a6edc > Sep 11 00:39:34 zaratustra ---[ end trace 4eaa2a86a8e2da22 ]--- fix the boundary condition. sysctl_sched_rt_period=0 makes exception at to_ratio(). Signed-off-by: Hiroshi Shimamoto Signed-off-by: Ingo Molnar --- kernel/sched.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index cc1f81b50b8..98890807375 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -8909,6 +8909,9 @@ static int sched_rt_global_constraints(void) u64 rt_runtime, rt_period; int ret = 0; + if (sysctl_sched_rt_period <= 0) + return -EINVAL; + rt_period = ktime_to_ns(tg->rt_bandwidth.rt_period); rt_runtime = tg->rt_bandwidth.rt_runtime; @@ -8925,6 +8928,9 @@ static int sched_rt_global_constraints(void) unsigned long flags; int i; + if (sysctl_sched_rt_period <= 0) + return -EINVAL; + spin_lock_irqsave(&def_rt_bandwidth.rt_runtime_lock, flags); for_each_possible_cpu(i) { struct rt_rq *rt_rq = &cpu_rq(i)->rt; -- cgit v1.2.3-70-g09d2 From 4e74339af6a59c061cf02f1ac497766bca1de19a Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Sat, 13 Sep 2008 02:33:08 -0700 Subject: cpuset: avoid changing cpuset's cpus when -errno returned After the patch: commit 0b2f630a28d53b5a2082a5275bc3334b10373508 Author: Miao Xie Date: Fri Jul 25 01:47:21 2008 -0700 cpusets: restructure the function update_cpumask() and update_nodemask() It might happen that 'echo 0 > /cpuset/sub/cpus' returned failure but 'cpus' has been changed, because cpus was changed before calling heap_init() which may return -ENOMEM. This patch restores the orginal behavior. Signed-off-by: Li Zefan Acked-by: Paul Menage Cc: Paul Jackson Cc: Miao Xie Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- kernel/cpuset.c | 37 +++++++++++++++---------------------- 1 file changed, 15 insertions(+), 22 deletions(-) (limited to 'kernel') diff --git a/kernel/cpuset.c b/kernel/cpuset.c index f227bc17269..827cd9adccb 100644 --- a/kernel/cpuset.c +++ b/kernel/cpuset.c @@ -843,37 +843,25 @@ static void cpuset_change_cpumask(struct task_struct *tsk, /** * update_tasks_cpumask - Update the cpumasks of tasks in the cpuset. * @cs: the cpuset in which each task's cpus_allowed mask needs to be changed + * @heap: if NULL, defer allocating heap memory to cgroup_scan_tasks() * * Called with cgroup_mutex held * * The cgroup_scan_tasks() function will scan all the tasks in a cgroup, * calling callback functions for each. * - * Return 0 if successful, -errno if not. + * No return value. It's guaranteed that cgroup_scan_tasks() always returns 0 + * if @heap != NULL. */ -static int update_tasks_cpumask(struct cpuset *cs) +static void update_tasks_cpumask(struct cpuset *cs, struct ptr_heap *heap) { struct cgroup_scanner scan; - struct ptr_heap heap; - int retval; - - /* - * cgroup_scan_tasks() will initialize heap->gt for us. - * heap_init() is still needed here for we should not change - * cs->cpus_allowed when heap_init() fails. - */ - retval = heap_init(&heap, PAGE_SIZE, GFP_KERNEL, NULL); - if (retval) - return retval; scan.cg = cs->css.cgroup; scan.test_task = cpuset_test_cpumask; scan.process_task = cpuset_change_cpumask; - scan.heap = &heap; - retval = cgroup_scan_tasks(&scan); - - heap_free(&heap); - return retval; + scan.heap = heap; + cgroup_scan_tasks(&scan); } /** @@ -883,6 +871,7 @@ static int update_tasks_cpumask(struct cpuset *cs) */ static int update_cpumask(struct cpuset *cs, const char *buf) { + struct ptr_heap heap; struct cpuset trialcs; int retval; int is_load_balanced; @@ -917,6 +906,10 @@ static int update_cpumask(struct cpuset *cs, const char *buf) if (cpus_equal(cs->cpus_allowed, trialcs.cpus_allowed)) return 0; + retval = heap_init(&heap, PAGE_SIZE, GFP_KERNEL, NULL); + if (retval) + return retval; + is_load_balanced = is_sched_load_balance(&trialcs); mutex_lock(&callback_mutex); @@ -927,9 +920,9 @@ static int update_cpumask(struct cpuset *cs, const char *buf) * Scan tasks in the cpuset, and update the cpumasks of any * that need an update. */ - retval = update_tasks_cpumask(cs); - if (retval < 0) - return retval; + update_tasks_cpumask(cs, &heap); + + heap_free(&heap); if (is_load_balanced) async_rebuild_sched_domains(); @@ -1965,7 +1958,7 @@ static void scan_for_empty_cpusets(const struct cpuset *root) nodes_empty(cp->mems_allowed)) remove_tasks_in_empty_cpuset(cp); else { - update_tasks_cpumask(cp); + update_tasks_cpumask(cp, NULL); update_tasks_nodemask(cp, &oldmems); } } -- cgit v1.2.3-70-g09d2 From 2344abbcbdb82140050e8be29d3d55e4f6fe860b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 16 Sep 2008 11:32:50 -0700 Subject: clockevents: make device shutdown robust The device shut down does not cleanup the next_event variable of the clock event device. So when the device is reactivated the possible stale next_event value can prevent the device to be reprogrammed as it claims to wait on a event already. This is the root cause of the resurfacing suspend/resume problem, where systems need key press to come back to life. Fix this by setting next_event to KTIME_MAX when the device is shut down. Use a separate function for shutdown which takes care of that and only keep the direct set mode call in the broadcast code, where we can not touch the next_event value. Signed-off-by: Thomas Gleixner --- kernel/time/clockevents.c | 12 +++++++++++- kernel/time/tick-broadcast.c | 9 ++++----- kernel/time/tick-common.c | 4 ++-- kernel/time/tick-internal.h | 2 ++ 4 files changed, 19 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/time/clockevents.c b/kernel/time/clockevents.c index 1876b526c77..f8d968063ce 100644 --- a/kernel/time/clockevents.c +++ b/kernel/time/clockevents.c @@ -71,6 +71,16 @@ void clockevents_set_mode(struct clock_event_device *dev, } } +/** + * clockevents_shutdown - shutdown the device and clear next_event + * @dev: device to shutdown + */ +void clockevents_shutdown(struct clock_event_device *dev) +{ + clockevents_set_mode(dev, CLOCK_EVT_MODE_SHUTDOWN); + dev->next_event.tv64 = KTIME_MAX; +} + /** * clockevents_program_event - Reprogram the clock event device. * @expires: absolute expiry time (monotonic clock) @@ -206,7 +216,7 @@ void clockevents_exchange_device(struct clock_event_device *old, if (new) { BUG_ON(new->mode != CLOCK_EVT_MODE_UNUSED); - clockevents_set_mode(new, CLOCK_EVT_MODE_SHUTDOWN); + clockevents_shutdown(new); } local_irq_restore(flags); } diff --git a/kernel/time/tick-broadcast.c b/kernel/time/tick-broadcast.c index 2f5a38294bf..f1f3eee2811 100644 --- a/kernel/time/tick-broadcast.c +++ b/kernel/time/tick-broadcast.c @@ -236,8 +236,7 @@ static void tick_do_broadcast_on_off(void *why) if (!cpu_isset(cpu, tick_broadcast_mask)) { cpu_set(cpu, tick_broadcast_mask); if (td->mode == TICKDEV_MODE_PERIODIC) - clockevents_set_mode(dev, - CLOCK_EVT_MODE_SHUTDOWN); + clockevents_shutdown(dev); } if (*reason == CLOCK_EVT_NOTIFY_BROADCAST_FORCE) tick_broadcast_force = 1; @@ -254,7 +253,7 @@ static void tick_do_broadcast_on_off(void *why) if (cpus_empty(tick_broadcast_mask)) { if (!bc_stopped) - clockevents_set_mode(bc, CLOCK_EVT_MODE_SHUTDOWN); + clockevents_shutdown(bc); } else if (bc_stopped) { if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC) tick_broadcast_start_periodic(bc); @@ -306,7 +305,7 @@ void tick_shutdown_broadcast(unsigned int *cpup) if (tick_broadcast_device.mode == TICKDEV_MODE_PERIODIC) { if (bc && cpus_empty(tick_broadcast_mask)) - clockevents_set_mode(bc, CLOCK_EVT_MODE_SHUTDOWN); + clockevents_shutdown(bc); } spin_unlock_irqrestore(&tick_broadcast_lock, flags); @@ -321,7 +320,7 @@ void tick_suspend_broadcast(void) bc = tick_broadcast_device.evtdev; if (bc) - clockevents_set_mode(bc, CLOCK_EVT_MODE_SHUTDOWN); + clockevents_shutdown(bc); spin_unlock_irqrestore(&tick_broadcast_lock, flags); } diff --git a/kernel/time/tick-common.c b/kernel/time/tick-common.c index c4777193d56..019315ebf9d 100644 --- a/kernel/time/tick-common.c +++ b/kernel/time/tick-common.c @@ -249,7 +249,7 @@ static int tick_check_new_device(struct clock_event_device *newdev) * not give it back to the clockevents layer ! */ if (tick_is_broadcast_device(curdev)) { - clockevents_set_mode(curdev, CLOCK_EVT_MODE_SHUTDOWN); + clockevents_shutdown(curdev); curdev = NULL; } clockevents_exchange_device(curdev, newdev); @@ -311,7 +311,7 @@ static void tick_suspend(void) unsigned long flags; spin_lock_irqsave(&tick_device_lock, flags); - clockevents_set_mode(td->evtdev, CLOCK_EVT_MODE_SHUTDOWN); + clockevents_shutdown(td->evtdev); spin_unlock_irqrestore(&tick_device_lock, flags); } diff --git a/kernel/time/tick-internal.h b/kernel/time/tick-internal.h index 0ffc2918ea6..6e9db9734aa 100644 --- a/kernel/time/tick-internal.h +++ b/kernel/time/tick-internal.h @@ -10,6 +10,8 @@ extern int tick_do_timer_cpu __read_mostly; extern void tick_setup_periodic(struct clock_event_device *dev, int broadcast); extern void tick_handle_periodic(struct clock_event_device *dev); +extern void clockevents_shutdown(struct clock_event_device *dev); + /* * NO_HZ / high resolution timer shared code */ -- cgit v1.2.3-70-g09d2