WSL2-Linux-Kernel/kernel/cpu.c

837 строки
20 KiB
C
Исходник Обычный вид История

/* CPU control.
* (C) 2001, 2002, 2003, 2004 Rusty Russell
*
* This code is licenced under the GPL.
*/
#include <linux/proc_fs.h>
#include <linux/smp.h>
#include <linux/init.h>
#include <linux/notifier.h>
#include <linux/sched.h>
#include <linux/unistd.h>
#include <linux/cpu.h>
cpu: introduce clear_tasks_mm_cpumask() helper Many architectures clear tasks' mm_cpumask like this: read_lock(&tasklist_lock); for_each_process(p) { if (p->mm) cpumask_clear_cpu(cpu, mm_cpumask(p->mm)); } read_unlock(&tasklist_lock); Depending on the context, the code above may have several problems, such as: 1. Working with task->mm w/o getting mm or grabing the task lock is dangerous as ->mm might disappear (exit_mm() assigns NULL under task_lock(), so tasklist lock is not enough). 2. Checking for process->mm is not enough because process' main thread may exit or detach its mm via use_mm(), but other threads may still have a valid mm. This patch implements a small helper function that does things correctly, i.e.: 1. We take the task's lock while whe handle its mm (we can't use get_task_mm()/mmput() pair as mmput() might sleep); 2. To catch exited main thread case, we use find_lock_task_mm(), which walks up all threads and returns an appropriate task (with task lock held). Also, Per Peter Zijlstra's idea, now we don't grab tasklist_lock in the new helper, instead we take the rcu read lock. We can do this because the function is called after the cpu is taken down and marked offline, so no new tasks will get this cpu set in their mm mask. Signed-off-by: Anton Vorontsov <anton.vorontsov@linaro.org> Cc: Richard Weinberger <richard@nod.at> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Mike Frysinger <vapier@gentoo.org> Cc: Paul Mundt <lethal@linux-sh.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-06-01 03:26:22 +04:00
#include <linux/oom.h>
#include <linux/rcupdate.h>
#include <linux/export.h>
#include <linux/bug.h>
#include <linux/kthread.h>
#include <linux/stop_machine.h>
#include <linux/mutex.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 11:04:11 +03:00
#include <linux/gfp.h>
#include <linux/suspend.h>
#include <linux/lockdep.h>
clockevents: Fix cpu_down() race for hrtimer based broadcasting It was found when doing a hotplug stress test on POWER, that the machine either hit softlockups or rcu_sched stall warnings. The issue was traced to commit: 7cba160ad789 ("powernv/cpuidle: Redesign idle states management") which exposed the cpu_down() race with hrtimer based broadcast mode: 5d1638acb9f6 ("tick: Introduce hrtimer based broadcast") The race is the following: Assume CPU1 is the CPU which holds the hrtimer broadcasting duty before it is taken down. CPU0 CPU1 cpu_down() take_cpu_down() disable_interrupts() cpu_die() while (CPU1 != CPU_DEAD) { msleep(100); switch_to_idle(); stop_cpu_timer(); schedule_broadcast(); } tick_cleanup_cpu_dead() take_over_broadcast() So after CPU1 disabled interrupts it cannot handle the broadcast hrtimer anymore, so CPU0 will be stuck forever. Fix this by explicitly taking over broadcast duty before cpu_die(). This is a temporary workaround. What we really want is a callback in the clockevent device which allows us to do that from the dying CPU by pushing the hrtimer onto a different cpu. That might involve an IPI and is definitely more complex than this immediate fix. Changelog was picked up from: https://lkml.org/lkml/2015/2/16/213 Suggested-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Nicolas Pitre <nico@linaro.org> Signed-off-by: Preeti U. Murthy <preeti@linux.vnet.ibm.com> Cc: linuxppc-dev@lists.ozlabs.org Cc: mpe@ellerman.id.au Cc: nicolas.pitre@linaro.org Cc: peterz@infradead.org Cc: rjw@rjwysocki.net Fixes: http://linuxppc.10917.n7.nabble.com/offlining-cpus-breakage-td88619.html Link: http://lkml.kernel.org/r/20150330092410.24979.59887.stgit@preeti.in.ibm.com [ Merged it to the latest timer tree, renamed the callback, tidied up the changelog. ] Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-03-30 12:29:19 +03:00
#include <linux/tick.h>
hotplug: Prevent alloc/free of irq descriptors during cpu up/down When a cpu goes up some architectures (e.g. x86) have to walk the irq space to set up the vector space for the cpu. While this needs extra protection at the architecture level we can avoid a few race conditions by preventing the concurrent allocation/free of irq descriptors and the associated data. When a cpu goes down it moves the interrupts which are targeted to this cpu away by reassigning the affinities. While this happens interrupts can be allocated and freed, which opens a can of race conditions in the code which reassignes the affinities because interrupt descriptors might be freed underneath. Example: CPU1 CPU2 cpu_up/down irq_desc = irq_to_desc(irq); remove_from_radix_tree(desc); raw_spin_lock(&desc->lock); free(desc); We could protect the irq descriptors with RCU, but that would require a full tree change of all accesses to interrupt descriptors. But fortunately these kind of race conditions are rather limited to a few things like cpu hotplug. The normal setup/teardown is very well serialized. So the simpler and obvious solution is: Prevent allocation and freeing of interrupt descriptors accross cpu hotplug. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: xiao jin <jin.xiao@intel.com> Cc: Joerg Roedel <jroedel@suse.de> Cc: Borislav Petkov <bp@suse.de> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20150705171102.063519515@linutronix.de
2015-07-05 20:12:30 +03:00
#include <linux/irq.h>
#include <trace/events/power.h>
#include "smpboot.h"
#ifdef CONFIG_SMP
/* Serializes the updates to cpu_online_mask, cpu_present_mask */
static DEFINE_MUTEX(cpu_add_remove_lock);
/*
CPU hotplug: Provide lockless versions of callback registration functions The following method of CPU hotplug callback registration is not safe due to the possibility of an ABBA deadlock involving the cpu_add_remove_lock and the cpu_hotplug.lock. get_online_cpus(); for_each_online_cpu(cpu) init_cpu(cpu); register_cpu_notifier(&foobar_cpu_notifier); put_online_cpus(); The deadlock is shown below: CPU 0 CPU 1 ----- ----- Acquire cpu_hotplug.lock [via get_online_cpus()] CPU online/offline operation takes cpu_add_remove_lock [via cpu_maps_update_begin()] Try to acquire cpu_add_remove_lock [via register_cpu_notifier()] CPU online/offline operation tries to acquire cpu_hotplug.lock [via cpu_hotplug_begin()] *** DEADLOCK! *** The problem here is that callback registration takes the locks in one order whereas the CPU hotplug operations take the same locks in the opposite order. To avoid this issue and to provide a race-free method to register CPU hotplug callbacks (along with initialization of already online CPUs), introduce new variants of the callback registration APIs that simply register the callbacks without holding the cpu_add_remove_lock during the registration. That way, we can avoid the ABBA scenario. However, we will need to hold the cpu_add_remove_lock throughout the entire critical section, to protect updates to the callback/notifier chain. This can be achieved by writing the callback registration code as follows: cpu_maps_update_begin(); [ or cpu_notifier_register_begin(); see below ] for_each_online_cpu(cpu) init_cpu(cpu); /* This doesn't take the cpu_add_remove_lock */ __register_cpu_notifier(&foobar_cpu_notifier); cpu_maps_update_done(); [ or cpu_notifier_register_done(); see below ] Note that we can't use get_online_cpus() here instead of cpu_maps_update_begin() because the cpu_hotplug.lock is dropped during the invocation of CPU_POST_DEAD notifiers, and hence get_online_cpus() cannot provide the necessary synchronization to protect the callback/notifier chains against concurrent reads and writes. On the other hand, since the cpu_add_remove_lock protects the entire hotplug operation (including CPU_POST_DEAD), we can use cpu_maps_update_begin/done() to guarantee proper synchronization. Also, since cpu_maps_update_begin/done() is like a super-set of get/put_online_cpus(), the former naturally protects the critical sections from concurrent hotplug operations. Since the names cpu_maps_update_begin/done() don't make much sense in CPU hotplug callback registration scenarios, we'll introduce new APIs named cpu_notifier_register_begin/done() and map them to cpu_maps_update_begin/done(). In summary, introduce the lockless variants of un/register_cpu_notifier() and also export the cpu_notifier_register_begin/done() APIs for use by modules. This way, we provide a race-free way to register hotplug callbacks as well as perform initialization for the CPUs that are already online. Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Acked-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Toshi Kani <toshi.kani@hp.com> Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com> Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-11 00:34:14 +04:00
* The following two APIs (cpu_maps_update_begin/done) must be used when
* attempting to serialize the updates to cpu_online_mask & cpu_present_mask.
* The APIs cpu_notifier_register_begin/done() must be used to protect CPU
* hotplug callback (un)registration performed using __register_cpu_notifier()
* or __unregister_cpu_notifier().
*/
void cpu_maps_update_begin(void)
{
mutex_lock(&cpu_add_remove_lock);
}
CPU hotplug: Provide lockless versions of callback registration functions The following method of CPU hotplug callback registration is not safe due to the possibility of an ABBA deadlock involving the cpu_add_remove_lock and the cpu_hotplug.lock. get_online_cpus(); for_each_online_cpu(cpu) init_cpu(cpu); register_cpu_notifier(&foobar_cpu_notifier); put_online_cpus(); The deadlock is shown below: CPU 0 CPU 1 ----- ----- Acquire cpu_hotplug.lock [via get_online_cpus()] CPU online/offline operation takes cpu_add_remove_lock [via cpu_maps_update_begin()] Try to acquire cpu_add_remove_lock [via register_cpu_notifier()] CPU online/offline operation tries to acquire cpu_hotplug.lock [via cpu_hotplug_begin()] *** DEADLOCK! *** The problem here is that callback registration takes the locks in one order whereas the CPU hotplug operations take the same locks in the opposite order. To avoid this issue and to provide a race-free method to register CPU hotplug callbacks (along with initialization of already online CPUs), introduce new variants of the callback registration APIs that simply register the callbacks without holding the cpu_add_remove_lock during the registration. That way, we can avoid the ABBA scenario. However, we will need to hold the cpu_add_remove_lock throughout the entire critical section, to protect updates to the callback/notifier chain. This can be achieved by writing the callback registration code as follows: cpu_maps_update_begin(); [ or cpu_notifier_register_begin(); see below ] for_each_online_cpu(cpu) init_cpu(cpu); /* This doesn't take the cpu_add_remove_lock */ __register_cpu_notifier(&foobar_cpu_notifier); cpu_maps_update_done(); [ or cpu_notifier_register_done(); see below ] Note that we can't use get_online_cpus() here instead of cpu_maps_update_begin() because the cpu_hotplug.lock is dropped during the invocation of CPU_POST_DEAD notifiers, and hence get_online_cpus() cannot provide the necessary synchronization to protect the callback/notifier chains against concurrent reads and writes. On the other hand, since the cpu_add_remove_lock protects the entire hotplug operation (including CPU_POST_DEAD), we can use cpu_maps_update_begin/done() to guarantee proper synchronization. Also, since cpu_maps_update_begin/done() is like a super-set of get/put_online_cpus(), the former naturally protects the critical sections from concurrent hotplug operations. Since the names cpu_maps_update_begin/done() don't make much sense in CPU hotplug callback registration scenarios, we'll introduce new APIs named cpu_notifier_register_begin/done() and map them to cpu_maps_update_begin/done(). In summary, introduce the lockless variants of un/register_cpu_notifier() and also export the cpu_notifier_register_begin/done() APIs for use by modules. This way, we provide a race-free way to register hotplug callbacks as well as perform initialization for the CPUs that are already online. Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Acked-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Toshi Kani <toshi.kani@hp.com> Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com> Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-11 00:34:14 +04:00
EXPORT_SYMBOL(cpu_notifier_register_begin);
void cpu_maps_update_done(void)
{
mutex_unlock(&cpu_add_remove_lock);
}
CPU hotplug: Provide lockless versions of callback registration functions The following method of CPU hotplug callback registration is not safe due to the possibility of an ABBA deadlock involving the cpu_add_remove_lock and the cpu_hotplug.lock. get_online_cpus(); for_each_online_cpu(cpu) init_cpu(cpu); register_cpu_notifier(&foobar_cpu_notifier); put_online_cpus(); The deadlock is shown below: CPU 0 CPU 1 ----- ----- Acquire cpu_hotplug.lock [via get_online_cpus()] CPU online/offline operation takes cpu_add_remove_lock [via cpu_maps_update_begin()] Try to acquire cpu_add_remove_lock [via register_cpu_notifier()] CPU online/offline operation tries to acquire cpu_hotplug.lock [via cpu_hotplug_begin()] *** DEADLOCK! *** The problem here is that callback registration takes the locks in one order whereas the CPU hotplug operations take the same locks in the opposite order. To avoid this issue and to provide a race-free method to register CPU hotplug callbacks (along with initialization of already online CPUs), introduce new variants of the callback registration APIs that simply register the callbacks without holding the cpu_add_remove_lock during the registration. That way, we can avoid the ABBA scenario. However, we will need to hold the cpu_add_remove_lock throughout the entire critical section, to protect updates to the callback/notifier chain. This can be achieved by writing the callback registration code as follows: cpu_maps_update_begin(); [ or cpu_notifier_register_begin(); see below ] for_each_online_cpu(cpu) init_cpu(cpu); /* This doesn't take the cpu_add_remove_lock */ __register_cpu_notifier(&foobar_cpu_notifier); cpu_maps_update_done(); [ or cpu_notifier_register_done(); see below ] Note that we can't use get_online_cpus() here instead of cpu_maps_update_begin() because the cpu_hotplug.lock is dropped during the invocation of CPU_POST_DEAD notifiers, and hence get_online_cpus() cannot provide the necessary synchronization to protect the callback/notifier chains against concurrent reads and writes. On the other hand, since the cpu_add_remove_lock protects the entire hotplug operation (including CPU_POST_DEAD), we can use cpu_maps_update_begin/done() to guarantee proper synchronization. Also, since cpu_maps_update_begin/done() is like a super-set of get/put_online_cpus(), the former naturally protects the critical sections from concurrent hotplug operations. Since the names cpu_maps_update_begin/done() don't make much sense in CPU hotplug callback registration scenarios, we'll introduce new APIs named cpu_notifier_register_begin/done() and map them to cpu_maps_update_begin/done(). In summary, introduce the lockless variants of un/register_cpu_notifier() and also export the cpu_notifier_register_begin/done() APIs for use by modules. This way, we provide a race-free way to register hotplug callbacks as well as perform initialization for the CPUs that are already online. Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Acked-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Toshi Kani <toshi.kani@hp.com> Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com> Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-11 00:34:14 +04:00
EXPORT_SYMBOL(cpu_notifier_register_done);
static RAW_NOTIFIER_HEAD(cpu_chain);
/* If set, cpu_up and cpu_down will return -EBUSY and do nothing.
* Should always be manipulated under cpu_add_remove_lock
*/
static int cpu_hotplug_disabled;
#ifdef CONFIG_HOTPLUG_CPU
static struct {
struct task_struct *active_writer;
/* wait queue to wake up the active_writer */
wait_queue_head_t wq;
/* verifies that no writer will get active while readers are active */
struct mutex lock;
/*
* Also blocks the new readers during
* an ongoing cpu hotplug operation.
*/
atomic_t refcount;
#ifdef CONFIG_DEBUG_LOCK_ALLOC
struct lockdep_map dep_map;
#endif
} cpu_hotplug = {
.active_writer = NULL,
.wq = __WAIT_QUEUE_HEAD_INITIALIZER(cpu_hotplug.wq),
.lock = __MUTEX_INITIALIZER(cpu_hotplug.lock),
#ifdef CONFIG_DEBUG_LOCK_ALLOC
.dep_map = {.name = "cpu_hotplug.lock" },
#endif
};
/* Lockdep annotations for get/put_online_cpus() and cpu_hotplug_begin/end() */
#define cpuhp_lock_acquire_read() lock_map_acquire_read(&cpu_hotplug.dep_map)
#define cpuhp_lock_acquire_tryread() \
lock_map_acquire_tryread(&cpu_hotplug.dep_map)
#define cpuhp_lock_acquire() lock_map_acquire(&cpu_hotplug.dep_map)
#define cpuhp_lock_release() lock_map_release(&cpu_hotplug.dep_map)
void get_online_cpus(void)
{
might_sleep();
if (cpu_hotplug.active_writer == current)
return;
cpuhp_lock_acquire_read();
mutex_lock(&cpu_hotplug.lock);
atomic_inc(&cpu_hotplug.refcount);
mutex_unlock(&cpu_hotplug.lock);
}
EXPORT_SYMBOL_GPL(get_online_cpus);
bool try_get_online_cpus(void)
{
if (cpu_hotplug.active_writer == current)
return true;
if (!mutex_trylock(&cpu_hotplug.lock))
return false;
cpuhp_lock_acquire_tryread();
atomic_inc(&cpu_hotplug.refcount);
mutex_unlock(&cpu_hotplug.lock);
return true;
}
EXPORT_SYMBOL_GPL(try_get_online_cpus);
void put_online_cpus(void)
{
int refcount;
if (cpu_hotplug.active_writer == current)
return;
refcount = atomic_dec_return(&cpu_hotplug.refcount);
if (WARN_ON(refcount < 0)) /* try to fix things up */
atomic_inc(&cpu_hotplug.refcount);
if (refcount <= 0 && waitqueue_active(&cpu_hotplug.wq))
wake_up(&cpu_hotplug.wq);
cpuhp_lock_release();
}
EXPORT_SYMBOL_GPL(put_online_cpus);
/*
* This ensures that the hotplug operation can begin only when the
* refcount goes to zero.
*
* Note that during a cpu-hotplug operation, the new readers, if any,
* will be blocked by the cpu_hotplug.lock
*
* Since cpu_hotplug_begin() is always called after invoking
* cpu_maps_update_begin(), we can be sure that only one writer is active.
*
* Note that theoretically, there is a possibility of a livelock:
* - Refcount goes to zero, last reader wakes up the sleeping
* writer.
* - Last reader unlocks the cpu_hotplug.lock.
* - A new reader arrives at this moment, bumps up the refcount.
* - The writer acquires the cpu_hotplug.lock finds the refcount
* non zero and goes to sleep again.
*
* However, this is very difficult to achieve in practice since
* get_online_cpus() not an api which is called all that often.
*
*/
ACPI / processor: Acquire writer lock to update CPU maps CPU system maps are protected with reader/writer locks. The reader lock, get_online_cpus(), assures that the maps are not updated while holding the lock. The writer lock, cpu_hotplug_begin(), is used to udpate the cpu maps along with cpu_maps_update_begin(). However, the ACPI processor handler updates the cpu maps without holding the the writer lock. acpi_map_lsapic() is called from acpi_processor_hotadd_init() to update cpu_possible_mask and cpu_present_mask. acpi_unmap_lsapic() is called from acpi_processor_remove() to update cpu_possible_mask. Currently, they are either unprotected or protected with the reader lock, which is not correct. For example, the get_online_cpus() below is supposed to assure that cpu_possible_mask is not changed while the code is iterating with for_each_possible_cpu(). get_online_cpus(); for_each_possible_cpu(cpu) { : } put_online_cpus(); However, this lock has no protection with CPU hotplug since the ACPI processor handler does not use the writer lock when it updates cpu_possible_mask. The reader lock does not serialize within the readers. This patch protects them with the writer lock with cpu_hotplug_begin() along with cpu_maps_update_begin(), which must be held before calling cpu_hotplug_begin(). It also protects arch_register_cpu() / arch_unregister_cpu(), which creates / deletes a sysfs cpu device interface. For this purpose it changes cpu_hotplug_begin() and cpu_hotplug_done() to global and exports them in cpu.h. Signed-off-by: Toshi Kani <toshi.kani@hp.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2013-08-12 19:45:53 +04:00
void cpu_hotplug_begin(void)
{
DEFINE_WAIT(wait);
cpu_hotplug.active_writer = current;
cpuhp_lock_acquire();
for (;;) {
mutex_lock(&cpu_hotplug.lock);
prepare_to_wait(&cpu_hotplug.wq, &wait, TASK_UNINTERRUPTIBLE);
if (likely(!atomic_read(&cpu_hotplug.refcount)))
break;
mutex_unlock(&cpu_hotplug.lock);
schedule();
}
finish_wait(&cpu_hotplug.wq, &wait);
}
ACPI / processor: Acquire writer lock to update CPU maps CPU system maps are protected with reader/writer locks. The reader lock, get_online_cpus(), assures that the maps are not updated while holding the lock. The writer lock, cpu_hotplug_begin(), is used to udpate the cpu maps along with cpu_maps_update_begin(). However, the ACPI processor handler updates the cpu maps without holding the the writer lock. acpi_map_lsapic() is called from acpi_processor_hotadd_init() to update cpu_possible_mask and cpu_present_mask. acpi_unmap_lsapic() is called from acpi_processor_remove() to update cpu_possible_mask. Currently, they are either unprotected or protected with the reader lock, which is not correct. For example, the get_online_cpus() below is supposed to assure that cpu_possible_mask is not changed while the code is iterating with for_each_possible_cpu(). get_online_cpus(); for_each_possible_cpu(cpu) { : } put_online_cpus(); However, this lock has no protection with CPU hotplug since the ACPI processor handler does not use the writer lock when it updates cpu_possible_mask. The reader lock does not serialize within the readers. This patch protects them with the writer lock with cpu_hotplug_begin() along with cpu_maps_update_begin(), which must be held before calling cpu_hotplug_begin(). It also protects arch_register_cpu() / arch_unregister_cpu(), which creates / deletes a sysfs cpu device interface. For this purpose it changes cpu_hotplug_begin() and cpu_hotplug_done() to global and exports them in cpu.h. Signed-off-by: Toshi Kani <toshi.kani@hp.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2013-08-12 19:45:53 +04:00
void cpu_hotplug_done(void)
{
cpu_hotplug.active_writer = NULL;
mutex_unlock(&cpu_hotplug.lock);
cpuhp_lock_release();
}
/*
* Wait for currently running CPU hotplug operations to complete (if any) and
* disable future CPU hotplug (from sysfs). The 'cpu_add_remove_lock' protects
* the 'cpu_hotplug_disabled' flag. The same lock is also acquired by the
* hotplug path before performing hotplug operations. So acquiring that lock
* guarantees mutual exclusion from any currently running hotplug operations.
*/
void cpu_hotplug_disable(void)
{
cpu_maps_update_begin();
cpu_hotplug_disabled = 1;
cpu_maps_update_done();
}
void cpu_hotplug_enable(void)
{
cpu_maps_update_begin();
cpu_hotplug_disabled = 0;
cpu_maps_update_done();
}
ACPI / processor: Acquire writer lock to update CPU maps CPU system maps are protected with reader/writer locks. The reader lock, get_online_cpus(), assures that the maps are not updated while holding the lock. The writer lock, cpu_hotplug_begin(), is used to udpate the cpu maps along with cpu_maps_update_begin(). However, the ACPI processor handler updates the cpu maps without holding the the writer lock. acpi_map_lsapic() is called from acpi_processor_hotadd_init() to update cpu_possible_mask and cpu_present_mask. acpi_unmap_lsapic() is called from acpi_processor_remove() to update cpu_possible_mask. Currently, they are either unprotected or protected with the reader lock, which is not correct. For example, the get_online_cpus() below is supposed to assure that cpu_possible_mask is not changed while the code is iterating with for_each_possible_cpu(). get_online_cpus(); for_each_possible_cpu(cpu) { : } put_online_cpus(); However, this lock has no protection with CPU hotplug since the ACPI processor handler does not use the writer lock when it updates cpu_possible_mask. The reader lock does not serialize within the readers. This patch protects them with the writer lock with cpu_hotplug_begin() along with cpu_maps_update_begin(), which must be held before calling cpu_hotplug_begin(). It also protects arch_register_cpu() / arch_unregister_cpu(), which creates / deletes a sysfs cpu device interface. For this purpose it changes cpu_hotplug_begin() and cpu_hotplug_done() to global and exports them in cpu.h. Signed-off-by: Toshi Kani <toshi.kani@hp.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2013-08-12 19:45:53 +04:00
#endif /* CONFIG_HOTPLUG_CPU */
/* Need to know about CPUs going up/down? */
int __ref register_cpu_notifier(struct notifier_block *nb)
{
int ret;
cpu_maps_update_begin();
ret = raw_notifier_chain_register(&cpu_chain, nb);
cpu_maps_update_done();
return ret;
}
CPU hotplug: Provide lockless versions of callback registration functions The following method of CPU hotplug callback registration is not safe due to the possibility of an ABBA deadlock involving the cpu_add_remove_lock and the cpu_hotplug.lock. get_online_cpus(); for_each_online_cpu(cpu) init_cpu(cpu); register_cpu_notifier(&foobar_cpu_notifier); put_online_cpus(); The deadlock is shown below: CPU 0 CPU 1 ----- ----- Acquire cpu_hotplug.lock [via get_online_cpus()] CPU online/offline operation takes cpu_add_remove_lock [via cpu_maps_update_begin()] Try to acquire cpu_add_remove_lock [via register_cpu_notifier()] CPU online/offline operation tries to acquire cpu_hotplug.lock [via cpu_hotplug_begin()] *** DEADLOCK! *** The problem here is that callback registration takes the locks in one order whereas the CPU hotplug operations take the same locks in the opposite order. To avoid this issue and to provide a race-free method to register CPU hotplug callbacks (along with initialization of already online CPUs), introduce new variants of the callback registration APIs that simply register the callbacks without holding the cpu_add_remove_lock during the registration. That way, we can avoid the ABBA scenario. However, we will need to hold the cpu_add_remove_lock throughout the entire critical section, to protect updates to the callback/notifier chain. This can be achieved by writing the callback registration code as follows: cpu_maps_update_begin(); [ or cpu_notifier_register_begin(); see below ] for_each_online_cpu(cpu) init_cpu(cpu); /* This doesn't take the cpu_add_remove_lock */ __register_cpu_notifier(&foobar_cpu_notifier); cpu_maps_update_done(); [ or cpu_notifier_register_done(); see below ] Note that we can't use get_online_cpus() here instead of cpu_maps_update_begin() because the cpu_hotplug.lock is dropped during the invocation of CPU_POST_DEAD notifiers, and hence get_online_cpus() cannot provide the necessary synchronization to protect the callback/notifier chains against concurrent reads and writes. On the other hand, since the cpu_add_remove_lock protects the entire hotplug operation (including CPU_POST_DEAD), we can use cpu_maps_update_begin/done() to guarantee proper synchronization. Also, since cpu_maps_update_begin/done() is like a super-set of get/put_online_cpus(), the former naturally protects the critical sections from concurrent hotplug operations. Since the names cpu_maps_update_begin/done() don't make much sense in CPU hotplug callback registration scenarios, we'll introduce new APIs named cpu_notifier_register_begin/done() and map them to cpu_maps_update_begin/done(). In summary, introduce the lockless variants of un/register_cpu_notifier() and also export the cpu_notifier_register_begin/done() APIs for use by modules. This way, we provide a race-free way to register hotplug callbacks as well as perform initialization for the CPUs that are already online. Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Acked-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Toshi Kani <toshi.kani@hp.com> Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com> Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-11 00:34:14 +04:00
int __ref __register_cpu_notifier(struct notifier_block *nb)
{
return raw_notifier_chain_register(&cpu_chain, nb);
}
static int __cpu_notify(unsigned long val, void *v, int nr_to_call,
int *nr_calls)
{
int ret;
ret = __raw_notifier_call_chain(&cpu_chain, val, v, nr_to_call,
nr_calls);
return notifier_to_errno(ret);
}
static int cpu_notify(unsigned long val, void *v)
{
return __cpu_notify(val, v, -1, NULL);
}
#ifdef CONFIG_HOTPLUG_CPU
static void cpu_notify_nofail(unsigned long val, void *v)
{
BUG_ON(cpu_notify(val, v));
}
EXPORT_SYMBOL(register_cpu_notifier);
CPU hotplug: Provide lockless versions of callback registration functions The following method of CPU hotplug callback registration is not safe due to the possibility of an ABBA deadlock involving the cpu_add_remove_lock and the cpu_hotplug.lock. get_online_cpus(); for_each_online_cpu(cpu) init_cpu(cpu); register_cpu_notifier(&foobar_cpu_notifier); put_online_cpus(); The deadlock is shown below: CPU 0 CPU 1 ----- ----- Acquire cpu_hotplug.lock [via get_online_cpus()] CPU online/offline operation takes cpu_add_remove_lock [via cpu_maps_update_begin()] Try to acquire cpu_add_remove_lock [via register_cpu_notifier()] CPU online/offline operation tries to acquire cpu_hotplug.lock [via cpu_hotplug_begin()] *** DEADLOCK! *** The problem here is that callback registration takes the locks in one order whereas the CPU hotplug operations take the same locks in the opposite order. To avoid this issue and to provide a race-free method to register CPU hotplug callbacks (along with initialization of already online CPUs), introduce new variants of the callback registration APIs that simply register the callbacks without holding the cpu_add_remove_lock during the registration. That way, we can avoid the ABBA scenario. However, we will need to hold the cpu_add_remove_lock throughout the entire critical section, to protect updates to the callback/notifier chain. This can be achieved by writing the callback registration code as follows: cpu_maps_update_begin(); [ or cpu_notifier_register_begin(); see below ] for_each_online_cpu(cpu) init_cpu(cpu); /* This doesn't take the cpu_add_remove_lock */ __register_cpu_notifier(&foobar_cpu_notifier); cpu_maps_update_done(); [ or cpu_notifier_register_done(); see below ] Note that we can't use get_online_cpus() here instead of cpu_maps_update_begin() because the cpu_hotplug.lock is dropped during the invocation of CPU_POST_DEAD notifiers, and hence get_online_cpus() cannot provide the necessary synchronization to protect the callback/notifier chains against concurrent reads and writes. On the other hand, since the cpu_add_remove_lock protects the entire hotplug operation (including CPU_POST_DEAD), we can use cpu_maps_update_begin/done() to guarantee proper synchronization. Also, since cpu_maps_update_begin/done() is like a super-set of get/put_online_cpus(), the former naturally protects the critical sections from concurrent hotplug operations. Since the names cpu_maps_update_begin/done() don't make much sense in CPU hotplug callback registration scenarios, we'll introduce new APIs named cpu_notifier_register_begin/done() and map them to cpu_maps_update_begin/done(). In summary, introduce the lockless variants of un/register_cpu_notifier() and also export the cpu_notifier_register_begin/done() APIs for use by modules. This way, we provide a race-free way to register hotplug callbacks as well as perform initialization for the CPUs that are already online. Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Acked-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Toshi Kani <toshi.kani@hp.com> Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com> Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-11 00:34:14 +04:00
EXPORT_SYMBOL(__register_cpu_notifier);
void __ref unregister_cpu_notifier(struct notifier_block *nb)
{
cpu_maps_update_begin();
raw_notifier_chain_unregister(&cpu_chain, nb);
cpu_maps_update_done();
}
EXPORT_SYMBOL(unregister_cpu_notifier);
CPU hotplug: Provide lockless versions of callback registration functions The following method of CPU hotplug callback registration is not safe due to the possibility of an ABBA deadlock involving the cpu_add_remove_lock and the cpu_hotplug.lock. get_online_cpus(); for_each_online_cpu(cpu) init_cpu(cpu); register_cpu_notifier(&foobar_cpu_notifier); put_online_cpus(); The deadlock is shown below: CPU 0 CPU 1 ----- ----- Acquire cpu_hotplug.lock [via get_online_cpus()] CPU online/offline operation takes cpu_add_remove_lock [via cpu_maps_update_begin()] Try to acquire cpu_add_remove_lock [via register_cpu_notifier()] CPU online/offline operation tries to acquire cpu_hotplug.lock [via cpu_hotplug_begin()] *** DEADLOCK! *** The problem here is that callback registration takes the locks in one order whereas the CPU hotplug operations take the same locks in the opposite order. To avoid this issue and to provide a race-free method to register CPU hotplug callbacks (along with initialization of already online CPUs), introduce new variants of the callback registration APIs that simply register the callbacks without holding the cpu_add_remove_lock during the registration. That way, we can avoid the ABBA scenario. However, we will need to hold the cpu_add_remove_lock throughout the entire critical section, to protect updates to the callback/notifier chain. This can be achieved by writing the callback registration code as follows: cpu_maps_update_begin(); [ or cpu_notifier_register_begin(); see below ] for_each_online_cpu(cpu) init_cpu(cpu); /* This doesn't take the cpu_add_remove_lock */ __register_cpu_notifier(&foobar_cpu_notifier); cpu_maps_update_done(); [ or cpu_notifier_register_done(); see below ] Note that we can't use get_online_cpus() here instead of cpu_maps_update_begin() because the cpu_hotplug.lock is dropped during the invocation of CPU_POST_DEAD notifiers, and hence get_online_cpus() cannot provide the necessary synchronization to protect the callback/notifier chains against concurrent reads and writes. On the other hand, since the cpu_add_remove_lock protects the entire hotplug operation (including CPU_POST_DEAD), we can use cpu_maps_update_begin/done() to guarantee proper synchronization. Also, since cpu_maps_update_begin/done() is like a super-set of get/put_online_cpus(), the former naturally protects the critical sections from concurrent hotplug operations. Since the names cpu_maps_update_begin/done() don't make much sense in CPU hotplug callback registration scenarios, we'll introduce new APIs named cpu_notifier_register_begin/done() and map them to cpu_maps_update_begin/done(). In summary, introduce the lockless variants of un/register_cpu_notifier() and also export the cpu_notifier_register_begin/done() APIs for use by modules. This way, we provide a race-free way to register hotplug callbacks as well as perform initialization for the CPUs that are already online. Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Acked-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Toshi Kani <toshi.kani@hp.com> Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com> Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-03-11 00:34:14 +04:00
void __ref __unregister_cpu_notifier(struct notifier_block *nb)
{
raw_notifier_chain_unregister(&cpu_chain, nb);
}
EXPORT_SYMBOL(__unregister_cpu_notifier);
/**
* clear_tasks_mm_cpumask - Safely clear tasks' mm_cpumask for a CPU
* @cpu: a CPU id
*
* This function walks all processes, finds a valid mm struct for each one and
* then clears a corresponding bit in mm's cpumask. While this all sounds
* trivial, there are various non-obvious corner cases, which this function
* tries to solve in a safe manner.
*
* Also note that the function uses a somewhat relaxed locking scheme, so it may
* be called only for an already offlined CPU.
*/
cpu: introduce clear_tasks_mm_cpumask() helper Many architectures clear tasks' mm_cpumask like this: read_lock(&tasklist_lock); for_each_process(p) { if (p->mm) cpumask_clear_cpu(cpu, mm_cpumask(p->mm)); } read_unlock(&tasklist_lock); Depending on the context, the code above may have several problems, such as: 1. Working with task->mm w/o getting mm or grabing the task lock is dangerous as ->mm might disappear (exit_mm() assigns NULL under task_lock(), so tasklist lock is not enough). 2. Checking for process->mm is not enough because process' main thread may exit or detach its mm via use_mm(), but other threads may still have a valid mm. This patch implements a small helper function that does things correctly, i.e.: 1. We take the task's lock while whe handle its mm (we can't use get_task_mm()/mmput() pair as mmput() might sleep); 2. To catch exited main thread case, we use find_lock_task_mm(), which walks up all threads and returns an appropriate task (with task lock held). Also, Per Peter Zijlstra's idea, now we don't grab tasklist_lock in the new helper, instead we take the rcu read lock. We can do this because the function is called after the cpu is taken down and marked offline, so no new tasks will get this cpu set in their mm mask. Signed-off-by: Anton Vorontsov <anton.vorontsov@linaro.org> Cc: Richard Weinberger <richard@nod.at> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Mike Frysinger <vapier@gentoo.org> Cc: Paul Mundt <lethal@linux-sh.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-06-01 03:26:22 +04:00
void clear_tasks_mm_cpumask(int cpu)
{
struct task_struct *p;
/*
* This function is called after the cpu is taken down and marked
* offline, so its not like new tasks will ever get this cpu set in
* their mm mask. -- Peter Zijlstra
* Thus, we may use rcu_read_lock() here, instead of grabbing
* full-fledged tasklist_lock.
*/
WARN_ON(cpu_online(cpu));
cpu: introduce clear_tasks_mm_cpumask() helper Many architectures clear tasks' mm_cpumask like this: read_lock(&tasklist_lock); for_each_process(p) { if (p->mm) cpumask_clear_cpu(cpu, mm_cpumask(p->mm)); } read_unlock(&tasklist_lock); Depending on the context, the code above may have several problems, such as: 1. Working with task->mm w/o getting mm or grabing the task lock is dangerous as ->mm might disappear (exit_mm() assigns NULL under task_lock(), so tasklist lock is not enough). 2. Checking for process->mm is not enough because process' main thread may exit or detach its mm via use_mm(), but other threads may still have a valid mm. This patch implements a small helper function that does things correctly, i.e.: 1. We take the task's lock while whe handle its mm (we can't use get_task_mm()/mmput() pair as mmput() might sleep); 2. To catch exited main thread case, we use find_lock_task_mm(), which walks up all threads and returns an appropriate task (with task lock held). Also, Per Peter Zijlstra's idea, now we don't grab tasklist_lock in the new helper, instead we take the rcu read lock. We can do this because the function is called after the cpu is taken down and marked offline, so no new tasks will get this cpu set in their mm mask. Signed-off-by: Anton Vorontsov <anton.vorontsov@linaro.org> Cc: Richard Weinberger <richard@nod.at> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Mike Frysinger <vapier@gentoo.org> Cc: Paul Mundt <lethal@linux-sh.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-06-01 03:26:22 +04:00
rcu_read_lock();
for_each_process(p) {
struct task_struct *t;
/*
* Main thread might exit, but other threads may still have
* a valid mm. Find one.
*/
cpu: introduce clear_tasks_mm_cpumask() helper Many architectures clear tasks' mm_cpumask like this: read_lock(&tasklist_lock); for_each_process(p) { if (p->mm) cpumask_clear_cpu(cpu, mm_cpumask(p->mm)); } read_unlock(&tasklist_lock); Depending on the context, the code above may have several problems, such as: 1. Working with task->mm w/o getting mm or grabing the task lock is dangerous as ->mm might disappear (exit_mm() assigns NULL under task_lock(), so tasklist lock is not enough). 2. Checking for process->mm is not enough because process' main thread may exit or detach its mm via use_mm(), but other threads may still have a valid mm. This patch implements a small helper function that does things correctly, i.e.: 1. We take the task's lock while whe handle its mm (we can't use get_task_mm()/mmput() pair as mmput() might sleep); 2. To catch exited main thread case, we use find_lock_task_mm(), which walks up all threads and returns an appropriate task (with task lock held). Also, Per Peter Zijlstra's idea, now we don't grab tasklist_lock in the new helper, instead we take the rcu read lock. We can do this because the function is called after the cpu is taken down and marked offline, so no new tasks will get this cpu set in their mm mask. Signed-off-by: Anton Vorontsov <anton.vorontsov@linaro.org> Cc: Richard Weinberger <richard@nod.at> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Mike Frysinger <vapier@gentoo.org> Cc: Paul Mundt <lethal@linux-sh.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-06-01 03:26:22 +04:00
t = find_lock_task_mm(p);
if (!t)
continue;
cpumask_clear_cpu(cpu, mm_cpumask(t->mm));
task_unlock(t);
}
rcu_read_unlock();
}
static inline void check_for_tasks(int dead_cpu)
{
struct task_struct *g, *p;
read_lock_irq(&tasklist_lock);
do_each_thread(g, p) {
if (!p->on_rq)
continue;
/*
* We do the check with unlocked task_rq(p)->lock.
* Order the reading to do not warn about a task,
* which was running on this cpu in the past, and
* it's just been woken on another cpu.
*/
rmb();
if (task_cpu(p) != dead_cpu)
continue;
pr_warn("Task %s (pid=%d) is on cpu %d (state=%ld, flags=%x)\n",
p->comm, task_pid_nr(p), dead_cpu, p->state, p->flags);
} while_each_thread(g, p);
read_unlock_irq(&tasklist_lock);
}
struct take_cpu_down_param {
unsigned long mod;
void *hcpu;
};
/* Take this CPU down. */
static int __ref take_cpu_down(void *_param)
{
struct take_cpu_down_param *param = _param;
int err;
/* Ensure this CPU doesn't handle any more interrupts. */
err = __cpu_disable();
if (err < 0)
[PATCH] i386 CPU hotplug (The i386 CPU hotplug patch provides infrastructure for some work which Pavel is doing as well as for ACPI S3 (suspend-to-RAM) work which Li Shaohua <shaohua.li@intel.com> is doing) The following provides i386 architecture support for safely unregistering and registering processors during runtime, updated for the current -mm tree. In order to avoid dumping cpu hotplug code into kernel/irq/* i dropped the cpu_online check in do_IRQ() by modifying fixup_irqs(). The difference being that on cpu offline, fixup_irqs() is called before we clear the cpu from cpu_online_map and a long delay in order to ensure that we never have any queued external interrupts on the APICs. There are additional changes to s390 and ppc64 to account for this change. 1) Add CONFIG_HOTPLUG_CPU 2) disable local APIC timer on dead cpus. 3) Disable preempt around irq balancing to prevent CPUs going down. 4) Print irq stats for all possible cpus. 5) Debugging check for interrupts on offline cpus. 6) Hacky fixup_irqs() to redirect irqs when cpus go off/online. 7) play_dead() for offline cpus to spin inside. 8) Handle offline cpus set in flush_tlb_others(). 9) Grab lock earlier in smp_call_function() to prevent CPUs going down. 10) Implement __cpu_disable() and __cpu_die(). 11) Enable local interrupts in cpu_enable() after fixup_irqs() 12) Don't fiddle with NMI on dead cpu, but leave intact on other cpus. 13) Program IRQ affinity whilst cpu is still in cpu_online_map on offline. Signed-off-by: Zwane Mwaikambo <zwane@linuxpower.ca> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-26 01:54:50 +04:00
return err;
cpu_notify(CPU_DYING | param->mod, param->hcpu);
/* Give up timekeeping duties */
tick_handover_do_timer();
/* Park the stopper thread */
kthread_park(current);
[PATCH] i386 CPU hotplug (The i386 CPU hotplug patch provides infrastructure for some work which Pavel is doing as well as for ACPI S3 (suspend-to-RAM) work which Li Shaohua <shaohua.li@intel.com> is doing) The following provides i386 architecture support for safely unregistering and registering processors during runtime, updated for the current -mm tree. In order to avoid dumping cpu hotplug code into kernel/irq/* i dropped the cpu_online check in do_IRQ() by modifying fixup_irqs(). The difference being that on cpu offline, fixup_irqs() is called before we clear the cpu from cpu_online_map and a long delay in order to ensure that we never have any queued external interrupts on the APICs. There are additional changes to s390 and ppc64 to account for this change. 1) Add CONFIG_HOTPLUG_CPU 2) disable local APIC timer on dead cpus. 3) Disable preempt around irq balancing to prevent CPUs going down. 4) Print irq stats for all possible cpus. 5) Debugging check for interrupts on offline cpus. 6) Hacky fixup_irqs() to redirect irqs when cpus go off/online. 7) play_dead() for offline cpus to spin inside. 8) Handle offline cpus set in flush_tlb_others(). 9) Grab lock earlier in smp_call_function() to prevent CPUs going down. 10) Implement __cpu_disable() and __cpu_die(). 11) Enable local interrupts in cpu_enable() after fixup_irqs() 12) Don't fiddle with NMI on dead cpu, but leave intact on other cpus. 13) Program IRQ affinity whilst cpu is still in cpu_online_map on offline. Signed-off-by: Zwane Mwaikambo <zwane@linuxpower.ca> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-26 01:54:50 +04:00
return 0;
}
/* Requires cpu_add_remove_lock to be held */
static int __ref _cpu_down(unsigned int cpu, int tasks_frozen)
{
int err, nr_calls = 0;
void *hcpu = (void *)(long)cpu;
unsigned long mod = tasks_frozen ? CPU_TASKS_FROZEN : 0;
struct take_cpu_down_param tcd_param = {
.mod = mod,
.hcpu = hcpu,
};
if (num_online_cpus() == 1)
return -EBUSY;
if (!cpu_online(cpu))
return -EINVAL;
cpu_hotplug_begin();
err = __cpu_notify(CPU_DOWN_PREPARE | mod, hcpu, -1, &nr_calls);
if (err) {
cpu hotplug: cpu: deliver CPU_UP_CANCELED only to NOTIFY_OKed callbacks with CPU_UP_PREPARE The functions in a CPU notifier chain is called with CPU_UP_PREPARE event before making the CPU online. If one of the callback returns NOTIFY_BAD, it stops to deliver CPU_UP_PREPARE event, and CPU online operation is canceled. Then CPU_UP_CANCELED event is delivered to the functions in a CPU notifier chain again. This CPU_UP_CANCELED event is delivered to the functions which have been called with CPU_UP_PREPARE, not delivered to the functions which haven't been called with CPU_UP_PREPARE. The problem that makes existing cpu hotplug error handlings complex is that the CPU_UP_CANCELED event is delivered to the function that has returned NOTIFY_BAD, too. Usually we don't expect to call destructor function against the object that has failed to initialize. It is like: err = register_something(); if (err) { unregister_something(); return err; } So it is natural to deliver CPU_UP_CANCELED event only to the functions that have returned NOTIFY_OK with CPU_UP_PREPARE event and not to call the function that have returned NOTIFY_BAD. This is what this patch is doing. Otherwise, every cpu hotplug notifiler has to track whether notifiler event is failed or not for each cpu. (drivers/base/topology.c is doing this with topology_dev_map) Similary this patch makes same thing with CPU_DOWN_PREPARE and CPU_DOWN_FAILED evnets. Acked-by: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: Gautham R Shenoy <ego@in.ibm.com> Cc: Oleg Nesterov <oleg@tv-sign.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 14:05:12 +04:00
nr_calls--;
__cpu_notify(CPU_DOWN_FAILED | mod, hcpu, nr_calls, NULL);
pr_warn("%s: attempt to take down CPU %u failed\n",
__func__, cpu);
Define and use new events,CPU_LOCK_ACQUIRE and CPU_LOCK_RELEASE This is an attempt to provide an alternate mechanism for postponing a hotplug event instead of using a global mechanism like lock_cpu_hotplug. The proposal is to add two new events namely CPU_LOCK_ACQUIRE and CPU_LOCK_RELEASE. The notification for these two events would be sent out before and after a cpu_hotplug event respectively. During the CPU_LOCK_ACQUIRE event, a cpu-hotplug-aware subsystem is supposed to acquire any per-subsystem hotcpu mutex ( Eg. workqueue_mutex in kernel/workqueue.c ). During the CPU_LOCK_RELEASE release event the cpu-hotplug-aware subsystem is supposed to release the per-subsystem hotcpu mutex. The reasons for defining new events as opposed to reusing the existing events like CPU_UP_PREPARE/CPU_UP_FAILED/CPU_ONLINE for locking/unlocking of per-subsystem hotcpu mutexes are as follow: - CPU_LOCK_ACQUIRE: All hotcpu mutexes are taken before subsystems start handling pre-hotplug events like CPU_UP_PREPARE/CPU_DOWN_PREPARE etc, thus ensuring a clean handling of these events. - CPU_LOCK_RELEASE: The hotcpu mutexes will be released only after all subsystems have handled post-hotplug events like CPU_DOWN_FAILED, CPU_DEAD,CPU_ONLINE etc thereby ensuring that there are no subsequent clashes amongst the interdependent subsystems after a cpu hotplugs. This patch also uses __raw_notifier_call chain in _cpu_up to take care of the dependency between the two consequetive calls to raw_notifier_call_chain. [akpm@linux-foundation.org: fix a bug] Signed-off-by: Gautham R Shenoy <ego@in.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-05-09 13:34:03 +04:00
goto out_release;
}
sched: Remove get_online_cpus() usage Remove get_online_cpus() usage from the scheduler; there's 4 sites that use it: - sched_init_smp(); where its completely superfluous since we're in 'early' boot and there simply cannot be any hotplugging. - sched_getaffinity(); we already take a raw spinlock to protect the task cpus_allowed mask, this disables preemption and therefore also stabilizes cpu_online_mask as that's modified using stop_machine. However switch to active mask for symmetry with sched_setaffinity()/set_cpus_allowed_ptr(). We guarantee active mask stability by inserting sync_rcu/sched() into _cpu_down. - sched_setaffinity(); we don't appear to need get_online_cpus() either, there's two sites where hotplug appears relevant: * cpuset_cpus_allowed(); for the !cpuset case we use possible_mask, for the cpuset case we hold task_lock, which is a spinlock and thus for mainline disables preemption (might cause pain on RT). * set_cpus_allowed_ptr(); Holds all scheduler locks and thus has preemption properly disabled; also it already deals with hotplug races explicitly where it releases them. - migrate_swap(); we can make stop_two_cpus() do the heavy lifting for us with a little trickery. By adding a sync_sched/rcu() after the CPU_DOWN_PREPARE notifier we can provide preempt/rcu guarantees for cpu_active_mask. Use these to validate that both our cpus are active when queueing the stop work before we queue the stop_machine works for take_cpu_down(). Signed-off-by: Peter Zijlstra <peterz@infradead.org> Cc: "Srivatsa S. Bhat" <srivatsa.bhat@linux.vnet.ibm.com> Cc: Paul McKenney <paulmck@linux.vnet.ibm.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Oleg Nesterov <oleg@redhat.com> Link: http://lkml.kernel.org/r/20131011123820.GV3081@twins.programming.kicks-ass.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-10-11 16:38:20 +04:00
/*
* By now we've cleared cpu_active_mask, wait for all preempt-disabled
* and RCU users of this state to go away such that all new such users
* will observe it.
*
* For CONFIG_PREEMPT we have preemptible RCU and its sync_rcu() might
* not imply sync_sched(), so explicitly call both.
*
* Do sync before park smpboot threads to take care the rcu boost case.
sched: Remove get_online_cpus() usage Remove get_online_cpus() usage from the scheduler; there's 4 sites that use it: - sched_init_smp(); where its completely superfluous since we're in 'early' boot and there simply cannot be any hotplugging. - sched_getaffinity(); we already take a raw spinlock to protect the task cpus_allowed mask, this disables preemption and therefore also stabilizes cpu_online_mask as that's modified using stop_machine. However switch to active mask for symmetry with sched_setaffinity()/set_cpus_allowed_ptr(). We guarantee active mask stability by inserting sync_rcu/sched() into _cpu_down. - sched_setaffinity(); we don't appear to need get_online_cpus() either, there's two sites where hotplug appears relevant: * cpuset_cpus_allowed(); for the !cpuset case we use possible_mask, for the cpuset case we hold task_lock, which is a spinlock and thus for mainline disables preemption (might cause pain on RT). * set_cpus_allowed_ptr(); Holds all scheduler locks and thus has preemption properly disabled; also it already deals with hotplug races explicitly where it releases them. - migrate_swap(); we can make stop_two_cpus() do the heavy lifting for us with a little trickery. By adding a sync_sched/rcu() after the CPU_DOWN_PREPARE notifier we can provide preempt/rcu guarantees for cpu_active_mask. Use these to validate that both our cpus are active when queueing the stop work before we queue the stop_machine works for take_cpu_down(). Signed-off-by: Peter Zijlstra <peterz@infradead.org> Cc: "Srivatsa S. Bhat" <srivatsa.bhat@linux.vnet.ibm.com> Cc: Paul McKenney <paulmck@linux.vnet.ibm.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Oleg Nesterov <oleg@redhat.com> Link: http://lkml.kernel.org/r/20131011123820.GV3081@twins.programming.kicks-ass.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-10-11 16:38:20 +04:00
*/
#ifdef CONFIG_PREEMPT
synchronize_sched();
#endif
synchronize_rcu();
smpboot_park_threads(cpu);
sched: Remove get_online_cpus() usage Remove get_online_cpus() usage from the scheduler; there's 4 sites that use it: - sched_init_smp(); where its completely superfluous since we're in 'early' boot and there simply cannot be any hotplugging. - sched_getaffinity(); we already take a raw spinlock to protect the task cpus_allowed mask, this disables preemption and therefore also stabilizes cpu_online_mask as that's modified using stop_machine. However switch to active mask for symmetry with sched_setaffinity()/set_cpus_allowed_ptr(). We guarantee active mask stability by inserting sync_rcu/sched() into _cpu_down. - sched_setaffinity(); we don't appear to need get_online_cpus() either, there's two sites where hotplug appears relevant: * cpuset_cpus_allowed(); for the !cpuset case we use possible_mask, for the cpuset case we hold task_lock, which is a spinlock and thus for mainline disables preemption (might cause pain on RT). * set_cpus_allowed_ptr(); Holds all scheduler locks and thus has preemption properly disabled; also it already deals with hotplug races explicitly where it releases them. - migrate_swap(); we can make stop_two_cpus() do the heavy lifting for us with a little trickery. By adding a sync_sched/rcu() after the CPU_DOWN_PREPARE notifier we can provide preempt/rcu guarantees for cpu_active_mask. Use these to validate that both our cpus are active when queueing the stop work before we queue the stop_machine works for take_cpu_down(). Signed-off-by: Peter Zijlstra <peterz@infradead.org> Cc: "Srivatsa S. Bhat" <srivatsa.bhat@linux.vnet.ibm.com> Cc: Paul McKenney <paulmck@linux.vnet.ibm.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Oleg Nesterov <oleg@redhat.com> Link: http://lkml.kernel.org/r/20131011123820.GV3081@twins.programming.kicks-ass.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-10-11 16:38:20 +04:00
/*
hotplug: Prevent alloc/free of irq descriptors during cpu up/down When a cpu goes up some architectures (e.g. x86) have to walk the irq space to set up the vector space for the cpu. While this needs extra protection at the architecture level we can avoid a few race conditions by preventing the concurrent allocation/free of irq descriptors and the associated data. When a cpu goes down it moves the interrupts which are targeted to this cpu away by reassigning the affinities. While this happens interrupts can be allocated and freed, which opens a can of race conditions in the code which reassignes the affinities because interrupt descriptors might be freed underneath. Example: CPU1 CPU2 cpu_up/down irq_desc = irq_to_desc(irq); remove_from_radix_tree(desc); raw_spin_lock(&desc->lock); free(desc); We could protect the irq descriptors with RCU, but that would require a full tree change of all accesses to interrupt descriptors. But fortunately these kind of race conditions are rather limited to a few things like cpu hotplug. The normal setup/teardown is very well serialized. So the simpler and obvious solution is: Prevent allocation and freeing of interrupt descriptors accross cpu hotplug. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: xiao jin <jin.xiao@intel.com> Cc: Joerg Roedel <jroedel@suse.de> Cc: Borislav Petkov <bp@suse.de> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20150705171102.063519515@linutronix.de
2015-07-05 20:12:30 +03:00
* Prevent irq alloc/free while the dying cpu reorganizes the
* interrupt affinities.
sched: Remove get_online_cpus() usage Remove get_online_cpus() usage from the scheduler; there's 4 sites that use it: - sched_init_smp(); where its completely superfluous since we're in 'early' boot and there simply cannot be any hotplugging. - sched_getaffinity(); we already take a raw spinlock to protect the task cpus_allowed mask, this disables preemption and therefore also stabilizes cpu_online_mask as that's modified using stop_machine. However switch to active mask for symmetry with sched_setaffinity()/set_cpus_allowed_ptr(). We guarantee active mask stability by inserting sync_rcu/sched() into _cpu_down. - sched_setaffinity(); we don't appear to need get_online_cpus() either, there's two sites where hotplug appears relevant: * cpuset_cpus_allowed(); for the !cpuset case we use possible_mask, for the cpuset case we hold task_lock, which is a spinlock and thus for mainline disables preemption (might cause pain on RT). * set_cpus_allowed_ptr(); Holds all scheduler locks and thus has preemption properly disabled; also it already deals with hotplug races explicitly where it releases them. - migrate_swap(); we can make stop_two_cpus() do the heavy lifting for us with a little trickery. By adding a sync_sched/rcu() after the CPU_DOWN_PREPARE notifier we can provide preempt/rcu guarantees for cpu_active_mask. Use these to validate that both our cpus are active when queueing the stop work before we queue the stop_machine works for take_cpu_down(). Signed-off-by: Peter Zijlstra <peterz@infradead.org> Cc: "Srivatsa S. Bhat" <srivatsa.bhat@linux.vnet.ibm.com> Cc: Paul McKenney <paulmck@linux.vnet.ibm.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Oleg Nesterov <oleg@redhat.com> Link: http://lkml.kernel.org/r/20131011123820.GV3081@twins.programming.kicks-ass.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-10-11 16:38:20 +04:00
*/
hotplug: Prevent alloc/free of irq descriptors during cpu up/down When a cpu goes up some architectures (e.g. x86) have to walk the irq space to set up the vector space for the cpu. While this needs extra protection at the architecture level we can avoid a few race conditions by preventing the concurrent allocation/free of irq descriptors and the associated data. When a cpu goes down it moves the interrupts which are targeted to this cpu away by reassigning the affinities. While this happens interrupts can be allocated and freed, which opens a can of race conditions in the code which reassignes the affinities because interrupt descriptors might be freed underneath. Example: CPU1 CPU2 cpu_up/down irq_desc = irq_to_desc(irq); remove_from_radix_tree(desc); raw_spin_lock(&desc->lock); free(desc); We could protect the irq descriptors with RCU, but that would require a full tree change of all accesses to interrupt descriptors. But fortunately these kind of race conditions are rather limited to a few things like cpu hotplug. The normal setup/teardown is very well serialized. So the simpler and obvious solution is: Prevent allocation and freeing of interrupt descriptors accross cpu hotplug. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: xiao jin <jin.xiao@intel.com> Cc: Joerg Roedel <jroedel@suse.de> Cc: Borislav Petkov <bp@suse.de> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20150705171102.063519515@linutronix.de
2015-07-05 20:12:30 +03:00
irq_lock_sparse();
sched: Remove get_online_cpus() usage Remove get_online_cpus() usage from the scheduler; there's 4 sites that use it: - sched_init_smp(); where its completely superfluous since we're in 'early' boot and there simply cannot be any hotplugging. - sched_getaffinity(); we already take a raw spinlock to protect the task cpus_allowed mask, this disables preemption and therefore also stabilizes cpu_online_mask as that's modified using stop_machine. However switch to active mask for symmetry with sched_setaffinity()/set_cpus_allowed_ptr(). We guarantee active mask stability by inserting sync_rcu/sched() into _cpu_down. - sched_setaffinity(); we don't appear to need get_online_cpus() either, there's two sites where hotplug appears relevant: * cpuset_cpus_allowed(); for the !cpuset case we use possible_mask, for the cpuset case we hold task_lock, which is a spinlock and thus for mainline disables preemption (might cause pain on RT). * set_cpus_allowed_ptr(); Holds all scheduler locks and thus has preemption properly disabled; also it already deals with hotplug races explicitly where it releases them. - migrate_swap(); we can make stop_two_cpus() do the heavy lifting for us with a little trickery. By adding a sync_sched/rcu() after the CPU_DOWN_PREPARE notifier we can provide preempt/rcu guarantees for cpu_active_mask. Use these to validate that both our cpus are active when queueing the stop work before we queue the stop_machine works for take_cpu_down(). Signed-off-by: Peter Zijlstra <peterz@infradead.org> Cc: "Srivatsa S. Bhat" <srivatsa.bhat@linux.vnet.ibm.com> Cc: Paul McKenney <paulmck@linux.vnet.ibm.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Oleg Nesterov <oleg@redhat.com> Link: http://lkml.kernel.org/r/20131011123820.GV3081@twins.programming.kicks-ass.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-10-11 16:38:20 +04:00
hotplug: Prevent alloc/free of irq descriptors during cpu up/down When a cpu goes up some architectures (e.g. x86) have to walk the irq space to set up the vector space for the cpu. While this needs extra protection at the architecture level we can avoid a few race conditions by preventing the concurrent allocation/free of irq descriptors and the associated data. When a cpu goes down it moves the interrupts which are targeted to this cpu away by reassigning the affinities. While this happens interrupts can be allocated and freed, which opens a can of race conditions in the code which reassignes the affinities because interrupt descriptors might be freed underneath. Example: CPU1 CPU2 cpu_up/down irq_desc = irq_to_desc(irq); remove_from_radix_tree(desc); raw_spin_lock(&desc->lock); free(desc); We could protect the irq descriptors with RCU, but that would require a full tree change of all accesses to interrupt descriptors. But fortunately these kind of race conditions are rather limited to a few things like cpu hotplug. The normal setup/teardown is very well serialized. So the simpler and obvious solution is: Prevent allocation and freeing of interrupt descriptors accross cpu hotplug. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: xiao jin <jin.xiao@intel.com> Cc: Joerg Roedel <jroedel@suse.de> Cc: Borislav Petkov <bp@suse.de> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20150705171102.063519515@linutronix.de
2015-07-05 20:12:30 +03:00
/*
* So now all preempt/rcu users must observe !cpu_active().
*/
err = __stop_machine(take_cpu_down, &tcd_param, cpumask_of(cpu));
if (err) {
/* CPU didn't die: tell everyone. Can't complain. */
cpu_notify_nofail(CPU_DOWN_FAILED | mod, hcpu);
hotplug: Prevent alloc/free of irq descriptors during cpu up/down When a cpu goes up some architectures (e.g. x86) have to walk the irq space to set up the vector space for the cpu. While this needs extra protection at the architecture level we can avoid a few race conditions by preventing the concurrent allocation/free of irq descriptors and the associated data. When a cpu goes down it moves the interrupts which are targeted to this cpu away by reassigning the affinities. While this happens interrupts can be allocated and freed, which opens a can of race conditions in the code which reassignes the affinities because interrupt descriptors might be freed underneath. Example: CPU1 CPU2 cpu_up/down irq_desc = irq_to_desc(irq); remove_from_radix_tree(desc); raw_spin_lock(&desc->lock); free(desc); We could protect the irq descriptors with RCU, but that would require a full tree change of all accesses to interrupt descriptors. But fortunately these kind of race conditions are rather limited to a few things like cpu hotplug. The normal setup/teardown is very well serialized. So the simpler and obvious solution is: Prevent allocation and freeing of interrupt descriptors accross cpu hotplug. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: xiao jin <jin.xiao@intel.com> Cc: Joerg Roedel <jroedel@suse.de> Cc: Borislav Petkov <bp@suse.de> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20150705171102.063519515@linutronix.de
2015-07-05 20:12:30 +03:00
irq_unlock_sparse();
goto out_release;
}
BUG_ON(cpu_online(cpu));
/*
* The migration_call() CPU_DYING callback will have removed all
* runnable tasks from the cpu, there's only the idle task left now
* that the migration thread is done doing the stop_machine thing.
*
* Wait for the stop thread to go away.
*/
while (!per_cpu(cpu_dead_idle, cpu))
cpu_relax();
smp_mb(); /* Read from cpu_dead_idle before __cpu_die(). */
per_cpu(cpu_dead_idle, cpu) = false;
hotplug: Prevent alloc/free of irq descriptors during cpu up/down When a cpu goes up some architectures (e.g. x86) have to walk the irq space to set up the vector space for the cpu. While this needs extra protection at the architecture level we can avoid a few race conditions by preventing the concurrent allocation/free of irq descriptors and the associated data. When a cpu goes down it moves the interrupts which are targeted to this cpu away by reassigning the affinities. While this happens interrupts can be allocated and freed, which opens a can of race conditions in the code which reassignes the affinities because interrupt descriptors might be freed underneath. Example: CPU1 CPU2 cpu_up/down irq_desc = irq_to_desc(irq); remove_from_radix_tree(desc); raw_spin_lock(&desc->lock); free(desc); We could protect the irq descriptors with RCU, but that would require a full tree change of all accesses to interrupt descriptors. But fortunately these kind of race conditions are rather limited to a few things like cpu hotplug. The normal setup/teardown is very well serialized. So the simpler and obvious solution is: Prevent allocation and freeing of interrupt descriptors accross cpu hotplug. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: xiao jin <jin.xiao@intel.com> Cc: Joerg Roedel <jroedel@suse.de> Cc: Borislav Petkov <bp@suse.de> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20150705171102.063519515@linutronix.de
2015-07-05 20:12:30 +03:00
/* Interrupts are moved away from the dying cpu, reenable alloc/free */
irq_unlock_sparse();
clockevents: Fix cpu_down() race for hrtimer based broadcasting It was found when doing a hotplug stress test on POWER, that the machine either hit softlockups or rcu_sched stall warnings. The issue was traced to commit: 7cba160ad789 ("powernv/cpuidle: Redesign idle states management") which exposed the cpu_down() race with hrtimer based broadcast mode: 5d1638acb9f6 ("tick: Introduce hrtimer based broadcast") The race is the following: Assume CPU1 is the CPU which holds the hrtimer broadcasting duty before it is taken down. CPU0 CPU1 cpu_down() take_cpu_down() disable_interrupts() cpu_die() while (CPU1 != CPU_DEAD) { msleep(100); switch_to_idle(); stop_cpu_timer(); schedule_broadcast(); } tick_cleanup_cpu_dead() take_over_broadcast() So after CPU1 disabled interrupts it cannot handle the broadcast hrtimer anymore, so CPU0 will be stuck forever. Fix this by explicitly taking over broadcast duty before cpu_die(). This is a temporary workaround. What we really want is a callback in the clockevent device which allows us to do that from the dying CPU by pushing the hrtimer onto a different cpu. That might involve an IPI and is definitely more complex than this immediate fix. Changelog was picked up from: https://lkml.org/lkml/2015/2/16/213 Suggested-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Nicolas Pitre <nico@linaro.org> Signed-off-by: Preeti U. Murthy <preeti@linux.vnet.ibm.com> Cc: linuxppc-dev@lists.ozlabs.org Cc: mpe@ellerman.id.au Cc: nicolas.pitre@linaro.org Cc: peterz@infradead.org Cc: rjw@rjwysocki.net Fixes: http://linuxppc.10917.n7.nabble.com/offlining-cpus-breakage-td88619.html Link: http://lkml.kernel.org/r/20150330092410.24979.59887.stgit@preeti.in.ibm.com [ Merged it to the latest timer tree, renamed the callback, tidied up the changelog. ] Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-03-30 12:29:19 +03:00
hotplug_cpu__broadcast_tick_pull(cpu);
/* This actually kills the CPU. */
__cpu_die(cpu);
/* CPU is completely dead: tell everyone. Too late to complain. */
tick_cleanup_dead_cpu(cpu);
cpu_notify_nofail(CPU_DEAD | mod, hcpu);
check_for_tasks(cpu);
Define and use new events,CPU_LOCK_ACQUIRE and CPU_LOCK_RELEASE This is an attempt to provide an alternate mechanism for postponing a hotplug event instead of using a global mechanism like lock_cpu_hotplug. The proposal is to add two new events namely CPU_LOCK_ACQUIRE and CPU_LOCK_RELEASE. The notification for these two events would be sent out before and after a cpu_hotplug event respectively. During the CPU_LOCK_ACQUIRE event, a cpu-hotplug-aware subsystem is supposed to acquire any per-subsystem hotcpu mutex ( Eg. workqueue_mutex in kernel/workqueue.c ). During the CPU_LOCK_RELEASE release event the cpu-hotplug-aware subsystem is supposed to release the per-subsystem hotcpu mutex. The reasons for defining new events as opposed to reusing the existing events like CPU_UP_PREPARE/CPU_UP_FAILED/CPU_ONLINE for locking/unlocking of per-subsystem hotcpu mutexes are as follow: - CPU_LOCK_ACQUIRE: All hotcpu mutexes are taken before subsystems start handling pre-hotplug events like CPU_UP_PREPARE/CPU_DOWN_PREPARE etc, thus ensuring a clean handling of these events. - CPU_LOCK_RELEASE: The hotcpu mutexes will be released only after all subsystems have handled post-hotplug events like CPU_DOWN_FAILED, CPU_DEAD,CPU_ONLINE etc thereby ensuring that there are no subsequent clashes amongst the interdependent subsystems after a cpu hotplugs. This patch also uses __raw_notifier_call chain in _cpu_up to take care of the dependency between the two consequetive calls to raw_notifier_call_chain. [akpm@linux-foundation.org: fix a bug] Signed-off-by: Gautham R Shenoy <ego@in.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-05-09 13:34:03 +04:00
out_release:
cpu_hotplug_done();
if (!err)
cpu_notify_nofail(CPU_POST_DEAD | mod, hcpu);
return err;
}
int __ref cpu_down(unsigned int cpu)
{
int err;
cpu_maps_update_begin();
cpu hotplug, sched: Introduce cpu_active_map and redo sched domain managment (take 2) This is based on Linus' idea of creating cpu_active_map that prevents scheduler load balancer from migrating tasks to the cpu that is going down. It allows us to simplify domain management code and avoid unecessary domain rebuilds during cpu hotplug event handling. Please ignore the cpusets part for now. It needs some more work in order to avoid crazy lock nesting. Although I did simplfy and unify domain reinitialization logic. We now simply call partition_sched_domains() in all the cases. This means that we're using exact same code paths as in cpusets case and hence the test below cover cpusets too. Cpuset changes to make rebuild_sched_domains() callable from various contexts are in the separate patch (right next after this one). This not only boots but also easily handles while true; do make clean; make -j 8; done and while true; do on-off-cpu 1; done at the same time. (on-off-cpu 1 simple does echo 0/1 > /sys/.../cpu1/online thing). Suprisingly the box (dual-core Core2) is quite usable. In fact I'm typing this on right now in gnome-terminal and things are moving just fine. Also this is running with most of the debug features enabled (lockdep, mutex, etc) no BUG_ONs or lockdep complaints so far. I believe I addressed all of the Dmitry's comments for original Linus' version. I changed both fair and rt balancer to mask out non-active cpus. And replaced cpu_is_offline() with !cpu_active() in the main scheduler code where it made sense (to me). Signed-off-by: Max Krasnyanskiy <maxk@qualcomm.com> Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Acked-by: Gregory Haskins <ghaskins@novell.com> Cc: dmitry.adamushko@gmail.com Cc: pj@sgi.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-15 15:43:49 +04:00
if (cpu_hotplug_disabled) {
err = -EBUSY;
cpu hotplug, sched: Introduce cpu_active_map and redo sched domain managment (take 2) This is based on Linus' idea of creating cpu_active_map that prevents scheduler load balancer from migrating tasks to the cpu that is going down. It allows us to simplify domain management code and avoid unecessary domain rebuilds during cpu hotplug event handling. Please ignore the cpusets part for now. It needs some more work in order to avoid crazy lock nesting. Although I did simplfy and unify domain reinitialization logic. We now simply call partition_sched_domains() in all the cases. This means that we're using exact same code paths as in cpusets case and hence the test below cover cpusets too. Cpuset changes to make rebuild_sched_domains() callable from various contexts are in the separate patch (right next after this one). This not only boots but also easily handles while true; do make clean; make -j 8; done and while true; do on-off-cpu 1; done at the same time. (on-off-cpu 1 simple does echo 0/1 > /sys/.../cpu1/online thing). Suprisingly the box (dual-core Core2) is quite usable. In fact I'm typing this on right now in gnome-terminal and things are moving just fine. Also this is running with most of the debug features enabled (lockdep, mutex, etc) no BUG_ONs or lockdep complaints so far. I believe I addressed all of the Dmitry's comments for original Linus' version. I changed both fair and rt balancer to mask out non-active cpus. And replaced cpu_is_offline() with !cpu_active() in the main scheduler code where it made sense (to me). Signed-off-by: Max Krasnyanskiy <maxk@qualcomm.com> Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Acked-by: Gregory Haskins <ghaskins@novell.com> Cc: dmitry.adamushko@gmail.com Cc: pj@sgi.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-15 15:43:49 +04:00
goto out;
}
err = _cpu_down(cpu, 0);
cpu hotplug, sched: Introduce cpu_active_map and redo sched domain managment (take 2) This is based on Linus' idea of creating cpu_active_map that prevents scheduler load balancer from migrating tasks to the cpu that is going down. It allows us to simplify domain management code and avoid unecessary domain rebuilds during cpu hotplug event handling. Please ignore the cpusets part for now. It needs some more work in order to avoid crazy lock nesting. Although I did simplfy and unify domain reinitialization logic. We now simply call partition_sched_domains() in all the cases. This means that we're using exact same code paths as in cpusets case and hence the test below cover cpusets too. Cpuset changes to make rebuild_sched_domains() callable from various contexts are in the separate patch (right next after this one). This not only boots but also easily handles while true; do make clean; make -j 8; done and while true; do on-off-cpu 1; done at the same time. (on-off-cpu 1 simple does echo 0/1 > /sys/.../cpu1/online thing). Suprisingly the box (dual-core Core2) is quite usable. In fact I'm typing this on right now in gnome-terminal and things are moving just fine. Also this is running with most of the debug features enabled (lockdep, mutex, etc) no BUG_ONs or lockdep complaints so far. I believe I addressed all of the Dmitry's comments for original Linus' version. I changed both fair and rt balancer to mask out non-active cpus. And replaced cpu_is_offline() with !cpu_active() in the main scheduler code where it made sense (to me). Signed-off-by: Max Krasnyanskiy <maxk@qualcomm.com> Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Acked-by: Gregory Haskins <ghaskins@novell.com> Cc: dmitry.adamushko@gmail.com Cc: pj@sgi.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-15 15:43:49 +04:00
out:
cpu_maps_update_done();
return err;
}
EXPORT_SYMBOL(cpu_down);
#endif /*CONFIG_HOTPLUG_CPU*/
/*
* Unpark per-CPU smpboot kthreads at CPU-online time.
*/
static int smpboot_thread_call(struct notifier_block *nfb,
unsigned long action, void *hcpu)
{
int cpu = (long)hcpu;
switch (action & ~CPU_TASKS_FROZEN) {
case CPU_DOWN_FAILED:
case CPU_ONLINE:
smpboot_unpark_threads(cpu);
break;
default:
break;
}
return NOTIFY_OK;
}
static struct notifier_block smpboot_thread_notifier = {
.notifier_call = smpboot_thread_call,
.priority = CPU_PRI_SMPBOOT,
};
void smpboot_thread_init(void)
{
register_cpu_notifier(&smpboot_thread_notifier);
}
/* Requires cpu_add_remove_lock to be held */
static int _cpu_up(unsigned int cpu, int tasks_frozen)
{
Define and use new events,CPU_LOCK_ACQUIRE and CPU_LOCK_RELEASE This is an attempt to provide an alternate mechanism for postponing a hotplug event instead of using a global mechanism like lock_cpu_hotplug. The proposal is to add two new events namely CPU_LOCK_ACQUIRE and CPU_LOCK_RELEASE. The notification for these two events would be sent out before and after a cpu_hotplug event respectively. During the CPU_LOCK_ACQUIRE event, a cpu-hotplug-aware subsystem is supposed to acquire any per-subsystem hotcpu mutex ( Eg. workqueue_mutex in kernel/workqueue.c ). During the CPU_LOCK_RELEASE release event the cpu-hotplug-aware subsystem is supposed to release the per-subsystem hotcpu mutex. The reasons for defining new events as opposed to reusing the existing events like CPU_UP_PREPARE/CPU_UP_FAILED/CPU_ONLINE for locking/unlocking of per-subsystem hotcpu mutexes are as follow: - CPU_LOCK_ACQUIRE: All hotcpu mutexes are taken before subsystems start handling pre-hotplug events like CPU_UP_PREPARE/CPU_DOWN_PREPARE etc, thus ensuring a clean handling of these events. - CPU_LOCK_RELEASE: The hotcpu mutexes will be released only after all subsystems have handled post-hotplug events like CPU_DOWN_FAILED, CPU_DEAD,CPU_ONLINE etc thereby ensuring that there are no subsequent clashes amongst the interdependent subsystems after a cpu hotplugs. This patch also uses __raw_notifier_call chain in _cpu_up to take care of the dependency between the two consequetive calls to raw_notifier_call_chain. [akpm@linux-foundation.org: fix a bug] Signed-off-by: Gautham R Shenoy <ego@in.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-05-09 13:34:03 +04:00
int ret, nr_calls = 0;
void *hcpu = (void *)(long)cpu;
unsigned long mod = tasks_frozen ? CPU_TASKS_FROZEN : 0;
struct task_struct *idle;
cpu_hotplug_begin();
ACPI / processor: prevent cpu from becoming online Even if acpi_processor_handle_eject() offlines cpu, there is a chance to online the cpu after that. So the patch closes the window by using get/put_online_cpus(). Why does the patch change _cpu_up() logic? The patch cares the race of hot-remove cpu and _cpu_up(). If the patch does not change it, there is the following race. hot-remove cpu | _cpu_up() ------------------------------------- ------------------------------------ call acpi_processor_handle_eject() | call cpu_down() | call get_online_cpus() | | call cpu_hotplug_begin() and stop here call arch_unregister_cpu() | call acpi_unmap_lsapic() | call put_online_cpus() | | start and continue _cpu_up() return acpi_processor_remove() | continue hot-remove the cpu | So _cpu_up() can continue to itself. And hot-remove cpu can also continue itself. If the patch changes _cpu_up() logic, the race disappears as below: hot-remove cpu | _cpu_up() ----------------------------------------------------------------------- call acpi_processor_handle_eject() | call cpu_down() | call get_online_cpus() | | call cpu_hotplug_begin() and stop here call arch_unregister_cpu() | call acpi_unmap_lsapic() | cpu's cpu_present is set | to false by set_cpu_present()| call put_online_cpus() | | start _cpu_up() | check cpu_present() and return -EINVAL return acpi_processor_remove() | continue hot-remove the cpu | Signed-off-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com> Reviewed-by: Toshi Kani <toshi.kani@hp.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2012-10-23 03:30:54 +04:00
if (cpu_online(cpu) || !cpu_present(cpu)) {
ret = -EINVAL;
goto out;
}
idle = idle_thread_get(cpu);
if (IS_ERR(idle)) {
ret = PTR_ERR(idle);
goto out;
}
ret = smpboot_create_threads(cpu);
if (ret)
goto out;
ret = __cpu_notify(CPU_UP_PREPARE | mod, hcpu, -1, &nr_calls);
if (ret) {
cpu hotplug: cpu: deliver CPU_UP_CANCELED only to NOTIFY_OKed callbacks with CPU_UP_PREPARE The functions in a CPU notifier chain is called with CPU_UP_PREPARE event before making the CPU online. If one of the callback returns NOTIFY_BAD, it stops to deliver CPU_UP_PREPARE event, and CPU online operation is canceled. Then CPU_UP_CANCELED event is delivered to the functions in a CPU notifier chain again. This CPU_UP_CANCELED event is delivered to the functions which have been called with CPU_UP_PREPARE, not delivered to the functions which haven't been called with CPU_UP_PREPARE. The problem that makes existing cpu hotplug error handlings complex is that the CPU_UP_CANCELED event is delivered to the function that has returned NOTIFY_BAD, too. Usually we don't expect to call destructor function against the object that has failed to initialize. It is like: err = register_something(); if (err) { unregister_something(); return err; } So it is natural to deliver CPU_UP_CANCELED event only to the functions that have returned NOTIFY_OK with CPU_UP_PREPARE event and not to call the function that have returned NOTIFY_BAD. This is what this patch is doing. Otherwise, every cpu hotplug notifiler has to track whether notifiler event is failed or not for each cpu. (drivers/base/topology.c is doing this with topology_dev_map) Similary this patch makes same thing with CPU_DOWN_PREPARE and CPU_DOWN_FAILED evnets. Acked-by: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: Gautham R Shenoy <ego@in.ibm.com> Cc: Oleg Nesterov <oleg@tv-sign.ru> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-10-18 14:05:12 +04:00
nr_calls--;
pr_warn("%s: attempt to bring up CPU %u failed\n",
__func__, cpu);
goto out_notify;
}
/* Arch-specific enabling code. */
ret = __cpu_up(cpu, idle);
hotplug: Prevent alloc/free of irq descriptors during cpu up/down When a cpu goes up some architectures (e.g. x86) have to walk the irq space to set up the vector space for the cpu. While this needs extra protection at the architecture level we can avoid a few race conditions by preventing the concurrent allocation/free of irq descriptors and the associated data. When a cpu goes down it moves the interrupts which are targeted to this cpu away by reassigning the affinities. While this happens interrupts can be allocated and freed, which opens a can of race conditions in the code which reassignes the affinities because interrupt descriptors might be freed underneath. Example: CPU1 CPU2 cpu_up/down irq_desc = irq_to_desc(irq); remove_from_radix_tree(desc); raw_spin_lock(&desc->lock); free(desc); We could protect the irq descriptors with RCU, but that would require a full tree change of all accesses to interrupt descriptors. But fortunately these kind of race conditions are rather limited to a few things like cpu hotplug. The normal setup/teardown is very well serialized. So the simpler and obvious solution is: Prevent allocation and freeing of interrupt descriptors accross cpu hotplug. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Peter Zijlstra <peterz@infradead.org> Cc: xiao jin <jin.xiao@intel.com> Cc: Joerg Roedel <jroedel@suse.de> Cc: Borislav Petkov <bp@suse.de> Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com> Link: http://lkml.kernel.org/r/20150705171102.063519515@linutronix.de
2015-07-05 20:12:30 +03:00
if (ret != 0)
goto out_notify;
BUG_ON(!cpu_online(cpu));
/* Now call notifier in preparation. */
cpu_notify(CPU_ONLINE | mod, hcpu);
out_notify:
if (ret != 0)
__cpu_notify(CPU_UP_CANCELED | mod, hcpu, nr_calls, NULL);
out:
cpu_hotplug_done();
return ret;
}
int cpu_up(unsigned int cpu)
{
int err = 0;
if (!cpu_possible(cpu)) {
pr_err("can't online cpu %d because it is not configured as may-hotadd at boot time\n",
cpu);
#if defined(CONFIG_IA64)
pr_err("please check additional_cpus= boot parameter\n");
#endif
return -EINVAL;
}
err = try_online_node(cpu_to_node(cpu));
if (err)
return err;
cpu_maps_update_begin();
cpu hotplug, sched: Introduce cpu_active_map and redo sched domain managment (take 2) This is based on Linus' idea of creating cpu_active_map that prevents scheduler load balancer from migrating tasks to the cpu that is going down. It allows us to simplify domain management code and avoid unecessary domain rebuilds during cpu hotplug event handling. Please ignore the cpusets part for now. It needs some more work in order to avoid crazy lock nesting. Although I did simplfy and unify domain reinitialization logic. We now simply call partition_sched_domains() in all the cases. This means that we're using exact same code paths as in cpusets case and hence the test below cover cpusets too. Cpuset changes to make rebuild_sched_domains() callable from various contexts are in the separate patch (right next after this one). This not only boots but also easily handles while true; do make clean; make -j 8; done and while true; do on-off-cpu 1; done at the same time. (on-off-cpu 1 simple does echo 0/1 > /sys/.../cpu1/online thing). Suprisingly the box (dual-core Core2) is quite usable. In fact I'm typing this on right now in gnome-terminal and things are moving just fine. Also this is running with most of the debug features enabled (lockdep, mutex, etc) no BUG_ONs or lockdep complaints so far. I believe I addressed all of the Dmitry's comments for original Linus' version. I changed both fair and rt balancer to mask out non-active cpus. And replaced cpu_is_offline() with !cpu_active() in the main scheduler code where it made sense (to me). Signed-off-by: Max Krasnyanskiy <maxk@qualcomm.com> Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Acked-by: Gregory Haskins <ghaskins@novell.com> Cc: dmitry.adamushko@gmail.com Cc: pj@sgi.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-15 15:43:49 +04:00
if (cpu_hotplug_disabled) {
err = -EBUSY;
cpu hotplug, sched: Introduce cpu_active_map and redo sched domain managment (take 2) This is based on Linus' idea of creating cpu_active_map that prevents scheduler load balancer from migrating tasks to the cpu that is going down. It allows us to simplify domain management code and avoid unecessary domain rebuilds during cpu hotplug event handling. Please ignore the cpusets part for now. It needs some more work in order to avoid crazy lock nesting. Although I did simplfy and unify domain reinitialization logic. We now simply call partition_sched_domains() in all the cases. This means that we're using exact same code paths as in cpusets case and hence the test below cover cpusets too. Cpuset changes to make rebuild_sched_domains() callable from various contexts are in the separate patch (right next after this one). This not only boots but also easily handles while true; do make clean; make -j 8; done and while true; do on-off-cpu 1; done at the same time. (on-off-cpu 1 simple does echo 0/1 > /sys/.../cpu1/online thing). Suprisingly the box (dual-core Core2) is quite usable. In fact I'm typing this on right now in gnome-terminal and things are moving just fine. Also this is running with most of the debug features enabled (lockdep, mutex, etc) no BUG_ONs or lockdep complaints so far. I believe I addressed all of the Dmitry's comments for original Linus' version. I changed both fair and rt balancer to mask out non-active cpus. And replaced cpu_is_offline() with !cpu_active() in the main scheduler code where it made sense (to me). Signed-off-by: Max Krasnyanskiy <maxk@qualcomm.com> Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Acked-by: Gregory Haskins <ghaskins@novell.com> Cc: dmitry.adamushko@gmail.com Cc: pj@sgi.com Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-15 15:43:49 +04:00
goto out;
}
err = _cpu_up(cpu, 0);
out:
cpu_maps_update_done();
return err;
}
EXPORT_SYMBOL_GPL(cpu_up);
#ifdef CONFIG_PM_SLEEP_SMP
static cpumask_var_t frozen_cpus;
int disable_nonboot_cpus(void)
{
int cpu, first_cpu, error = 0;
cpu_maps_update_begin();
first_cpu = cpumask_first(cpu_online_mask);
/*
* We take down all of the non-boot CPUs in one shot to avoid races
* with the userspace trying to use the CPU hotplug at the same time
*/
cpumask_clear(frozen_cpus);
pr_info("Disabling non-boot CPUs ...\n");
for_each_online_cpu(cpu) {
if (cpu == first_cpu)
continue;
trace_suspend_resume(TPS("CPU_OFF"), cpu, true);
error = _cpu_down(cpu, 1);
trace_suspend_resume(TPS("CPU_OFF"), cpu, false);
if (!error)
cpumask_set_cpu(cpu, frozen_cpus);
else {
pr_err("Error taking CPU%d down: %d\n", cpu, error);
break;
}
}
if (!error) {
BUG_ON(num_online_cpus() > 1);
/* Make sure the CPUs won't be enabled by someone else */
cpu_hotplug_disabled = 1;
} else {
pr_err("Non-boot CPUs are not disabled\n");
}
cpu_maps_update_done();
return error;
}
void __weak arch_enable_nonboot_cpus_begin(void)
{
}
void __weak arch_enable_nonboot_cpus_end(void)
{
}
void __ref enable_nonboot_cpus(void)
{
int cpu, error;
/* Allow everyone to use the CPU hotplug again */
cpu_maps_update_begin();
cpu_hotplug_disabled = 0;
if (cpumask_empty(frozen_cpus))
goto out;
pr_info("Enabling non-boot CPUs ...\n");
arch_enable_nonboot_cpus_begin();
for_each_cpu(cpu, frozen_cpus) {
trace_suspend_resume(TPS("CPU_ON"), cpu, true);
error = _cpu_up(cpu, 1);
trace_suspend_resume(TPS("CPU_ON"), cpu, false);
if (!error) {
pr_info("CPU%d is up\n", cpu);
continue;
}
pr_warn("Error taking CPU%d up: %d\n", cpu, error);
}
arch_enable_nonboot_cpus_end();
cpumask_clear(frozen_cpus);
out:
cpu_maps_update_done();
}
static int __init alloc_frozen_cpus(void)
{
if (!alloc_cpumask_var(&frozen_cpus, GFP_KERNEL|__GFP_ZERO))
return -ENOMEM;
return 0;
}
core_initcall(alloc_frozen_cpus);
/*
* When callbacks for CPU hotplug notifications are being executed, we must
* ensure that the state of the system with respect to the tasks being frozen
* or not, as reported by the notification, remains unchanged *throughout the
* duration* of the execution of the callbacks.
* Hence we need to prevent the freezer from racing with regular CPU hotplug.
*
* This synchronization is implemented by mutually excluding regular CPU
* hotplug and Suspend/Hibernate call paths by hooking onto the Suspend/
* Hibernate notifications.
*/
static int
cpu_hotplug_pm_callback(struct notifier_block *nb,
unsigned long action, void *ptr)
{
switch (action) {
case PM_SUSPEND_PREPARE:
case PM_HIBERNATION_PREPARE:
cpu_hotplug_disable();
break;
case PM_POST_SUSPEND:
case PM_POST_HIBERNATION:
cpu_hotplug_enable();
break;
default:
return NOTIFY_DONE;
}
return NOTIFY_OK;
}
static int __init cpu_hotplug_pm_sync_init(void)
{
/*
* cpu_hotplug_pm_callback has higher priority than x86
* bsp_pm_callback which depends on cpu_hotplug_pm_callback
* to disable cpu hotplug to avoid cpu hotplug race.
*/
pm_notifier(cpu_hotplug_pm_callback, 0);
return 0;
}
core_initcall(cpu_hotplug_pm_sync_init);
#endif /* CONFIG_PM_SLEEP_SMP */
/**
* notify_cpu_starting(cpu) - call the CPU_STARTING notifiers
* @cpu: cpu that just started
*
* This function calls the cpu_chain notifiers with CPU_STARTING.
* It must be called by the arch code on the new cpu, before the new cpu
* enables interrupts and before the "boot" cpu returns from __cpu_up().
*/
void notify_cpu_starting(unsigned int cpu)
{
unsigned long val = CPU_STARTING;
#ifdef CONFIG_PM_SLEEP_SMP
if (frozen_cpus != NULL && cpumask_test_cpu(cpu, frozen_cpus))
val = CPU_STARTING_FROZEN;
#endif /* CONFIG_PM_SLEEP_SMP */
cpu_notify(val, (void *)(long)cpu);
}
#endif /* CONFIG_SMP */
cpu masks: optimize and clean up cpumask_of_cpu() Clean up and optimize cpumask_of_cpu(), by sharing all the zero words. Instead of stupidly generating all possible i=0...NR_CPUS 2^i patterns creating a huge array of constant bitmasks, realize that the zero words can be shared. In other words, on a 64-bit architecture, we only ever need 64 of these arrays - with a different bit set in one single world (with enough zero words around it so that we can create any bitmask by just offsetting in that big array). And then we just put enough zeroes around it that we can point every single cpumask to be one of those things. So when we have 4k CPU's, instead of having 4k arrays (of 4k bits each, with one bit set in each array - 2MB memory total), we have exactly 64 arrays instead, each 8k bits in size (64kB total). And then we just point cpumask(n) to the right position (which we can calculate dynamically). Once we have the right arrays, getting "cpumask(n)" ends up being: static inline const cpumask_t *get_cpu_mask(unsigned int cpu) { const unsigned long *p = cpu_bit_bitmap[1 + cpu % BITS_PER_LONG]; p -= cpu / BITS_PER_LONG; return (const cpumask_t *)p; } This brings other advantages and simplifications as well: - we are not wasting memory that is just filled with a single bit in various different places - we don't need all those games to re-create the arrays in some dense format, because they're already going to be dense enough. if we compile a kernel for up to 4k CPU's, "wasting" that 64kB of memory is a non-issue (especially since by doing this "overlapping" trick we probably get better cache behaviour anyway). [ mingo@elte.hu: Converted Linus's mails into a commit. See: http://lkml.org/lkml/2008/7/27/156 http://lkml.org/lkml/2008/7/28/320 Also applied a family filter - which also has the side-effect of leaving out the bits where Linus calls me an idio... Oh, never mind ;-) ] Signed-off-by: Ingo Molnar <mingo@elte.hu> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Al Viro <viro@ZenIV.linux.org.uk> Cc: Mike Travis <travis@sgi.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-28 22:32:33 +04:00
/*
* cpu_bit_bitmap[] is a special, "compressed" data structure that
* represents all NR_CPUS bits binary values of 1<<nr.
*
* It is used by cpumask_of() to get a constant address to a CPU
cpu masks: optimize and clean up cpumask_of_cpu() Clean up and optimize cpumask_of_cpu(), by sharing all the zero words. Instead of stupidly generating all possible i=0...NR_CPUS 2^i patterns creating a huge array of constant bitmasks, realize that the zero words can be shared. In other words, on a 64-bit architecture, we only ever need 64 of these arrays - with a different bit set in one single world (with enough zero words around it so that we can create any bitmask by just offsetting in that big array). And then we just put enough zeroes around it that we can point every single cpumask to be one of those things. So when we have 4k CPU's, instead of having 4k arrays (of 4k bits each, with one bit set in each array - 2MB memory total), we have exactly 64 arrays instead, each 8k bits in size (64kB total). And then we just point cpumask(n) to the right position (which we can calculate dynamically). Once we have the right arrays, getting "cpumask(n)" ends up being: static inline const cpumask_t *get_cpu_mask(unsigned int cpu) { const unsigned long *p = cpu_bit_bitmap[1 + cpu % BITS_PER_LONG]; p -= cpu / BITS_PER_LONG; return (const cpumask_t *)p; } This brings other advantages and simplifications as well: - we are not wasting memory that is just filled with a single bit in various different places - we don't need all those games to re-create the arrays in some dense format, because they're already going to be dense enough. if we compile a kernel for up to 4k CPU's, "wasting" that 64kB of memory is a non-issue (especially since by doing this "overlapping" trick we probably get better cache behaviour anyway). [ mingo@elte.hu: Converted Linus's mails into a commit. See: http://lkml.org/lkml/2008/7/27/156 http://lkml.org/lkml/2008/7/28/320 Also applied a family filter - which also has the side-effect of leaving out the bits where Linus calls me an idio... Oh, never mind ;-) ] Signed-off-by: Ingo Molnar <mingo@elte.hu> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Al Viro <viro@ZenIV.linux.org.uk> Cc: Mike Travis <travis@sgi.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-28 22:32:33 +04:00
* mask value that has a single bit set only.
*/
cpu masks: optimize and clean up cpumask_of_cpu() Clean up and optimize cpumask_of_cpu(), by sharing all the zero words. Instead of stupidly generating all possible i=0...NR_CPUS 2^i patterns creating a huge array of constant bitmasks, realize that the zero words can be shared. In other words, on a 64-bit architecture, we only ever need 64 of these arrays - with a different bit set in one single world (with enough zero words around it so that we can create any bitmask by just offsetting in that big array). And then we just put enough zeroes around it that we can point every single cpumask to be one of those things. So when we have 4k CPU's, instead of having 4k arrays (of 4k bits each, with one bit set in each array - 2MB memory total), we have exactly 64 arrays instead, each 8k bits in size (64kB total). And then we just point cpumask(n) to the right position (which we can calculate dynamically). Once we have the right arrays, getting "cpumask(n)" ends up being: static inline const cpumask_t *get_cpu_mask(unsigned int cpu) { const unsigned long *p = cpu_bit_bitmap[1 + cpu % BITS_PER_LONG]; p -= cpu / BITS_PER_LONG; return (const cpumask_t *)p; } This brings other advantages and simplifications as well: - we are not wasting memory that is just filled with a single bit in various different places - we don't need all those games to re-create the arrays in some dense format, because they're already going to be dense enough. if we compile a kernel for up to 4k CPU's, "wasting" that 64kB of memory is a non-issue (especially since by doing this "overlapping" trick we probably get better cache behaviour anyway). [ mingo@elte.hu: Converted Linus's mails into a commit. See: http://lkml.org/lkml/2008/7/27/156 http://lkml.org/lkml/2008/7/28/320 Also applied a family filter - which also has the side-effect of leaving out the bits where Linus calls me an idio... Oh, never mind ;-) ] Signed-off-by: Ingo Molnar <mingo@elte.hu> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Al Viro <viro@ZenIV.linux.org.uk> Cc: Mike Travis <travis@sgi.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-28 22:32:33 +04:00
/* cpu_bit_bitmap[0] is empty - so we can back into it */
#define MASK_DECLARE_1(x) [x+1][0] = (1UL << (x))
cpu masks: optimize and clean up cpumask_of_cpu() Clean up and optimize cpumask_of_cpu(), by sharing all the zero words. Instead of stupidly generating all possible i=0...NR_CPUS 2^i patterns creating a huge array of constant bitmasks, realize that the zero words can be shared. In other words, on a 64-bit architecture, we only ever need 64 of these arrays - with a different bit set in one single world (with enough zero words around it so that we can create any bitmask by just offsetting in that big array). And then we just put enough zeroes around it that we can point every single cpumask to be one of those things. So when we have 4k CPU's, instead of having 4k arrays (of 4k bits each, with one bit set in each array - 2MB memory total), we have exactly 64 arrays instead, each 8k bits in size (64kB total). And then we just point cpumask(n) to the right position (which we can calculate dynamically). Once we have the right arrays, getting "cpumask(n)" ends up being: static inline const cpumask_t *get_cpu_mask(unsigned int cpu) { const unsigned long *p = cpu_bit_bitmap[1 + cpu % BITS_PER_LONG]; p -= cpu / BITS_PER_LONG; return (const cpumask_t *)p; } This brings other advantages and simplifications as well: - we are not wasting memory that is just filled with a single bit in various different places - we don't need all those games to re-create the arrays in some dense format, because they're already going to be dense enough. if we compile a kernel for up to 4k CPU's, "wasting" that 64kB of memory is a non-issue (especially since by doing this "overlapping" trick we probably get better cache behaviour anyway). [ mingo@elte.hu: Converted Linus's mails into a commit. See: http://lkml.org/lkml/2008/7/27/156 http://lkml.org/lkml/2008/7/28/320 Also applied a family filter - which also has the side-effect of leaving out the bits where Linus calls me an idio... Oh, never mind ;-) ] Signed-off-by: Ingo Molnar <mingo@elte.hu> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Al Viro <viro@ZenIV.linux.org.uk> Cc: Mike Travis <travis@sgi.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-28 22:32:33 +04:00
#define MASK_DECLARE_2(x) MASK_DECLARE_1(x), MASK_DECLARE_1(x+1)
#define MASK_DECLARE_4(x) MASK_DECLARE_2(x), MASK_DECLARE_2(x+2)
#define MASK_DECLARE_8(x) MASK_DECLARE_4(x), MASK_DECLARE_4(x+4)
cpu masks: optimize and clean up cpumask_of_cpu() Clean up and optimize cpumask_of_cpu(), by sharing all the zero words. Instead of stupidly generating all possible i=0...NR_CPUS 2^i patterns creating a huge array of constant bitmasks, realize that the zero words can be shared. In other words, on a 64-bit architecture, we only ever need 64 of these arrays - with a different bit set in one single world (with enough zero words around it so that we can create any bitmask by just offsetting in that big array). And then we just put enough zeroes around it that we can point every single cpumask to be one of those things. So when we have 4k CPU's, instead of having 4k arrays (of 4k bits each, with one bit set in each array - 2MB memory total), we have exactly 64 arrays instead, each 8k bits in size (64kB total). And then we just point cpumask(n) to the right position (which we can calculate dynamically). Once we have the right arrays, getting "cpumask(n)" ends up being: static inline const cpumask_t *get_cpu_mask(unsigned int cpu) { const unsigned long *p = cpu_bit_bitmap[1 + cpu % BITS_PER_LONG]; p -= cpu / BITS_PER_LONG; return (const cpumask_t *)p; } This brings other advantages and simplifications as well: - we are not wasting memory that is just filled with a single bit in various different places - we don't need all those games to re-create the arrays in some dense format, because they're already going to be dense enough. if we compile a kernel for up to 4k CPU's, "wasting" that 64kB of memory is a non-issue (especially since by doing this "overlapping" trick we probably get better cache behaviour anyway). [ mingo@elte.hu: Converted Linus's mails into a commit. See: http://lkml.org/lkml/2008/7/27/156 http://lkml.org/lkml/2008/7/28/320 Also applied a family filter - which also has the side-effect of leaving out the bits where Linus calls me an idio... Oh, never mind ;-) ] Signed-off-by: Ingo Molnar <mingo@elte.hu> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Al Viro <viro@ZenIV.linux.org.uk> Cc: Mike Travis <travis@sgi.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-28 22:32:33 +04:00
const unsigned long cpu_bit_bitmap[BITS_PER_LONG+1][BITS_TO_LONGS(NR_CPUS)] = {
MASK_DECLARE_8(0), MASK_DECLARE_8(8),
MASK_DECLARE_8(16), MASK_DECLARE_8(24),
#if BITS_PER_LONG > 32
MASK_DECLARE_8(32), MASK_DECLARE_8(40),
MASK_DECLARE_8(48), MASK_DECLARE_8(56),
#endif
};
cpu masks: optimize and clean up cpumask_of_cpu() Clean up and optimize cpumask_of_cpu(), by sharing all the zero words. Instead of stupidly generating all possible i=0...NR_CPUS 2^i patterns creating a huge array of constant bitmasks, realize that the zero words can be shared. In other words, on a 64-bit architecture, we only ever need 64 of these arrays - with a different bit set in one single world (with enough zero words around it so that we can create any bitmask by just offsetting in that big array). And then we just put enough zeroes around it that we can point every single cpumask to be one of those things. So when we have 4k CPU's, instead of having 4k arrays (of 4k bits each, with one bit set in each array - 2MB memory total), we have exactly 64 arrays instead, each 8k bits in size (64kB total). And then we just point cpumask(n) to the right position (which we can calculate dynamically). Once we have the right arrays, getting "cpumask(n)" ends up being: static inline const cpumask_t *get_cpu_mask(unsigned int cpu) { const unsigned long *p = cpu_bit_bitmap[1 + cpu % BITS_PER_LONG]; p -= cpu / BITS_PER_LONG; return (const cpumask_t *)p; } This brings other advantages and simplifications as well: - we are not wasting memory that is just filled with a single bit in various different places - we don't need all those games to re-create the arrays in some dense format, because they're already going to be dense enough. if we compile a kernel for up to 4k CPU's, "wasting" that 64kB of memory is a non-issue (especially since by doing this "overlapping" trick we probably get better cache behaviour anyway). [ mingo@elte.hu: Converted Linus's mails into a commit. See: http://lkml.org/lkml/2008/7/27/156 http://lkml.org/lkml/2008/7/28/320 Also applied a family filter - which also has the side-effect of leaving out the bits where Linus calls me an idio... Oh, never mind ;-) ] Signed-off-by: Ingo Molnar <mingo@elte.hu> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Al Viro <viro@ZenIV.linux.org.uk> Cc: Mike Travis <travis@sgi.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-28 22:32:33 +04:00
EXPORT_SYMBOL_GPL(cpu_bit_bitmap);
const DECLARE_BITMAP(cpu_all_bits, NR_CPUS) = CPU_BITS_ALL;
EXPORT_SYMBOL(cpu_all_bits);
#ifdef CONFIG_INIT_ALL_POSSIBLE
static DECLARE_BITMAP(cpu_possible_bits, CONFIG_NR_CPUS) __read_mostly
= CPU_BITS_ALL;
#else
static DECLARE_BITMAP(cpu_possible_bits, CONFIG_NR_CPUS) __read_mostly;
#endif
const struct cpumask *const cpu_possible_mask = to_cpumask(cpu_possible_bits);
EXPORT_SYMBOL(cpu_possible_mask);
static DECLARE_BITMAP(cpu_online_bits, CONFIG_NR_CPUS) __read_mostly;
const struct cpumask *const cpu_online_mask = to_cpumask(cpu_online_bits);
EXPORT_SYMBOL(cpu_online_mask);
static DECLARE_BITMAP(cpu_present_bits, CONFIG_NR_CPUS) __read_mostly;
const struct cpumask *const cpu_present_mask = to_cpumask(cpu_present_bits);
EXPORT_SYMBOL(cpu_present_mask);
static DECLARE_BITMAP(cpu_active_bits, CONFIG_NR_CPUS) __read_mostly;
const struct cpumask *const cpu_active_mask = to_cpumask(cpu_active_bits);
EXPORT_SYMBOL(cpu_active_mask);
void set_cpu_possible(unsigned int cpu, bool possible)
{
if (possible)
cpumask_set_cpu(cpu, to_cpumask(cpu_possible_bits));
else
cpumask_clear_cpu(cpu, to_cpumask(cpu_possible_bits));
}
void set_cpu_present(unsigned int cpu, bool present)
{
if (present)
cpumask_set_cpu(cpu, to_cpumask(cpu_present_bits));
else
cpumask_clear_cpu(cpu, to_cpumask(cpu_present_bits));
}
void set_cpu_online(unsigned int cpu, bool online)
{
if (online) {
cpumask_set_cpu(cpu, to_cpumask(cpu_online_bits));
cpumask_set_cpu(cpu, to_cpumask(cpu_active_bits));
} else {
cpumask_clear_cpu(cpu, to_cpumask(cpu_online_bits));
}
}
void set_cpu_active(unsigned int cpu, bool active)
{
if (active)
cpumask_set_cpu(cpu, to_cpumask(cpu_active_bits));
else
cpumask_clear_cpu(cpu, to_cpumask(cpu_active_bits));
}
void init_cpu_present(const struct cpumask *src)
{
cpumask_copy(to_cpumask(cpu_present_bits), src);
}
void init_cpu_possible(const struct cpumask *src)
{
cpumask_copy(to_cpumask(cpu_possible_bits), src);
}
void init_cpu_online(const struct cpumask *src)
{
cpumask_copy(to_cpumask(cpu_online_bits), src);
}