* [PATCH] Hot unplug CPU 4/4
@ 2002-09-13 1:56 Rusty Russell
0 siblings, 0 replies; only message in thread
From: Rusty Russell @ 2002-09-13 1:56 UTC (permalink / raw)
To: linux-kernel; +Cc: Patrick Mochel, mingo, Kimio Suganuma, torvalds
[ Patrick, I put the cpus as root/sys/cpus/0000/xxx, is that OK? ]
[ Ingo, scheduler changes in two places (online checks before pulling
tasks), plus the extraction of common code from migration thread ]
[ Kimio, I added the CPU_DEAD notifier ]
Name: Hotplug CPU Remove Generic Code
Author: Rusty Russell
Status: Experimental
Depends: Hotcpu/set-cpus-can-fail.patch.gz
Depends: Misc/daemonize-reparent.patch.2.5.33.gz
D: This adds the generic infrastructure to allow removing of CPUs
D: in a running kernel.
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/arch/i386/kernel/smpboot.c .7234-linux-2.5.34.updated/arch/i386/kernel/smpboot.c
--- .7234-linux-2.5.34/arch/i386/kernel/smpboot.c 2002-09-13 11:46:29.000000000 +1000
+++ .7234-linux-2.5.34.updated/arch/i386/kernel/smpboot.c 2002-09-13 11:47:04.000000000 +1000
@@ -1220,6 +1220,17 @@ int __devinit __cpu_up(unsigned int cpu)
return 0;
}
+int __cpu_disable(void)
+{
+ return -ENOSYS;
+}
+
+/* Since we fail __cpu_disable, this is never called. */
+void __cpu_die(unsigned int cpu)
+{
+ BUG();
+}
+
void __init smp_cpus_done(unsigned int max_cpus)
{
zap_low_mappings();
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/arch/ppc/kernel/smp.c .7234-linux-2.5.34.updated/arch/ppc/kernel/smp.c
--- .7234-linux-2.5.34/arch/ppc/kernel/smp.c 2002-09-13 11:46:29.000000000 +1000
+++ .7234-linux-2.5.34.updated/arch/ppc/kernel/smp.c 2002-09-13 11:47:04.000000000 +1000
@@ -438,6 +438,17 @@ int __cpu_up(unsigned int cpu)
return 0;
}
+int __cpu_disable(unsigned int cpu)
+{
+ return -ENOSYS;
+}
+
+/* Since we fail __cpu_disable, this is never called. */
+void __cpu_die(unsigned int cpu)
+{
+ BUG();
+}
+
void smp_cpus_done(unsigned int max_cpus)
{
smp_ops->setup_cpu(0);
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/include/asm-i386/smp.h .7234-linux-2.5.34.updated/include/asm-i386/smp.h
--- .7234-linux-2.5.34/include/asm-i386/smp.h 2002-09-13 11:46:28.000000000 +1000
+++ .7234-linux-2.5.34.updated/include/asm-i386/smp.h 2002-09-13 11:47:04.000000000 +1000
@@ -123,6 +123,9 @@ static inline int num_booting_cpus(void)
return hweight32(cpu_callout_map);
}
+extern int __cpu_disable(void);
+extern void __cpu_die(unsigned int cpu);
+
#endif /* !__ASSEMBLY__ */
#define NO_PROC_ID 0xFF /* No processor magic marker */
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/include/asm-ppc/smp.h .7234-linux-2.5.34.updated/include/asm-ppc/smp.h
--- .7234-linux-2.5.34/include/asm-ppc/smp.h 2002-09-13 11:46:28.000000000 +1000
+++ .7234-linux-2.5.34.updated/include/asm-ppc/smp.h 2002-09-13 11:47:04.000000000 +1000
@@ -68,6 +68,8 @@ static inline int any_online_cpu(const u
}
extern int __cpu_up(unsigned int cpu);
+extern int __cpu_disable(void);
+extern void __cpu_die(unsigned int cpu);
extern int smp_hw_index[];
#define hard_smp_processor_id() (smp_hw_index[smp_processor_id()])
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/include/linux/notifier.h .7234-linux-2.5.34.updated/include/linux/notifier.h
--- .7234-linux-2.5.34/include/linux/notifier.h 2002-07-27 15:24:39.000000000 +1000
+++ .7234-linux-2.5.34.updated/include/linux/notifier.h 2002-09-13 11:47:04.000000000 +1000
@@ -60,7 +60,9 @@ extern int notifier_call_chain(struct no
#define NETLINK_URELEASE 0x0001 /* Unicast netlink socket released */
+#define CPU_OFFLINE 0x0001 /* CPU (unsigned)v going down */
#define CPU_ONLINE 0x0002 /* CPU (unsigned)v coming up */
+#define CPU_DEAD 0x0003 /* CPU (unsigned)v dead */
#endif /* __KERNEL__ */
#endif /* _LINUX_NOTIFIER_H */
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/include/linux/sched.h .7234-linux-2.5.34.updated/include/linux/sched.h
--- .7234-linux-2.5.34/include/linux/sched.h 2002-09-13 11:46:29.000000000 +1000
+++ .7234-linux-2.5.34.updated/include/linux/sched.h 2002-09-13 11:47:04.000000000 +1000
@@ -431,6 +431,9 @@ do { if (atomic_dec_and_test(&(tsk)->usa
#if CONFIG_SMP
extern int set_cpus_allowed(task_t *p, const unsigned long new_mask[]);
+#ifdef CONFIG_HOTPLUG
+extern void migrate_all_tasks(void);
+#endif
#else
static inline int set_cpus_allowed(struct task_struct *p,
const unsigned long *new_mask)
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/include/linux/smp.h .7234-linux-2.5.34.updated/include/linux/smp.h
--- .7234-linux-2.5.34/include/linux/smp.h 2002-09-13 11:46:28.000000000 +1000
+++ .7234-linux-2.5.34.updated/include/linux/smp.h 2002-09-13 11:47:04.000000000 +1000
@@ -78,6 +78,7 @@ extern int register_cpu_notifier(struct
extern void unregister_cpu_notifier(struct notifier_block *nb);
int cpu_up(unsigned int cpu);
+int cpu_down(unsigned int cpu);
#else /* !SMP */
/*
@@ -106,6 +107,10 @@ static inline int register_cpu_notifier(
static inline void unregister_cpu_notifier(struct notifier_block *nb)
{
}
+
+/* Bring a CPU down/up */
+static inline int cpu_down(unsigned int cpu) { return -EBUSY; }
+static inline int cpu_up(unsigned int cpu) { return -ENOSYS; }
#endif /* !SMP */
#define get_cpu() ({ preempt_disable(); smp_processor_id(); })
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/kernel/cpu.c .7234-linux-2.5.34.updated/kernel/cpu.c
--- .7234-linux-2.5.34/kernel/cpu.c 2002-08-28 09:29:53.000000000 +1000
+++ .7234-linux-2.5.34.updated/kernel/cpu.c 2002-09-13 11:47:04.000000000 +1000
@@ -8,13 +8,220 @@
#include <linux/notifier.h>
#include <linux/sched.h>
#include <linux/unistd.h>
+#include <linux/kmod.h> /* for hotplug_path */
+#include <linux/brlock.h>
+#include <linux/device.h>
#include <asm/semaphore.h>
+#include <asm/uaccess.h>
/* This protects CPUs going up and down... */
DECLARE_MUTEX(cpucontrol);
static struct notifier_block *cpu_chain = NULL;
+#ifdef CONFIG_HOTPLUG
+/* Notify userspace when a cpu event occurs, by running '/sbin/hotplug
+ * cpu' with certain environment variables set. */
+static int cpu_run_sbin_hotplug(unsigned int cpu, const char *action)
+{
+ char *argv[3], *envp[5], cpu_str[12], action_str[32];
+ int i;
+
+ sprintf(cpu_str, "CPU=%d", cpu);
+ sprintf(action_str, "ACTION=%s", action);
+
+ i = 0;
+ argv[i++] = hotplug_path;
+ argv[i++] = "cpu";
+ argv[i] = NULL;
+
+ i = 0;
+ /* minimal command environment */
+ envp [i++] = "HOME=/";
+ envp [i++] = "PATH=/sbin:/bin:/usr/sbin:/usr/bin";
+ envp [i++] = cpu_str;
+ envp [i++] = action_str;
+ envp [i] = NULL;
+
+ return call_usermodehelper(argv [0], argv, envp);
+}
+
+int cpu_down(unsigned int cpu)
+{
+ int ret;
+ DECLARE_BITMAP(mask, NR_CPUS) = CPU_MASK_ALL;
+
+ if ((ret = down_interruptible(&cpucontrol)) != 0)
+ return ret;
+
+ if (!cpu_online(cpu)) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ if (num_online_cpus() == 1) {
+ ret = -EBUSY;
+ goto out;
+ }
+
+ /* Schedule ourselves on the dying CPU. */
+ migrate_to_cpu(cpu);
+
+ preempt_disable();
+
+ /* Disable CPU. */
+ ret = __cpu_disable();
+ if (ret != 0) {
+ printk("CPU disable failed: %i\n", ret);
+ goto preempt_out;
+ }
+ BUG_ON(cpu_online(cpu));
+
+ /* Move other tasks off to other CPUs (simple since they are
+ not running now). */
+ migrate_all_tasks();
+
+ /* Move off dying CPU, which will revert to idle process. */
+ __clear_bit(cpu, mask);
+ set_cpus_allowed(current, mask);
+ preempt_enable();
+
+ /* CPU has been disabled: tell everyone */
+ notifier_call_chain(&cpu_chain, CPU_OFFLINE, (void *)(long)cpu);
+
+ /* Die, CPU, die!. */
+ __cpu_die(cpu);
+
+ /* CPU has is completely dead: tell everyone */
+ notifier_call_chain(&cpu_chain, CPU_DEAD, (void *)(long)cpu);
+
+ printk("Running CPU hotplug remove for cpu %u\n", cpu);
+ cpu_run_sbin_hotplug(cpu, "remove");
+ printk("Done sbin hotplug\n");
+ up(&cpucontrol);
+ printk("Done with cpucontrol\n");
+ {
+ struct task_struct *p;
+
+ write_lock_irq(&tasklist_lock);
+ for_each_task(p) {
+ if (p->thread_info->cpu == cpu)
+ printk("Left %p: %s\n", p, p->comm);
+ }
+ write_unlock_irq(&tasklist_lock);
+ }
+ return ret;
+
+ preempt_out:
+ preempt_enable();
+ out:
+ up(&cpucontrol);
+ return ret;
+}
+
+struct cpu_device
+{
+ struct device dev;
+ unsigned int cpu;
+};
+
+static ssize_t show_online(struct device *dev,
+ char *buf,
+ size_t count,
+ loff_t off)
+{
+ char out[3];
+ struct cpu_device *cpudev = container_of(dev, struct cpu_device, dev);
+
+ sprintf(out, "%i\n", !!cpu_online(cpudev->cpu));
+ if (off >= strlen(out)) return 0;
+ if (off + count > strlen(out)) count = strlen(out) - off;
+ memcpy(buf, out+off, count);
+ return (ssize_t)count;
+}
+
+static ssize_t store_online(struct device *dev,
+ const char *buf,
+ size_t count,
+ loff_t off)
+{
+ struct cpu_device *cpudev = container_of(dev, struct cpu_device, dev);
+ ssize_t ret;
+
+ if (off != 0)
+ return -EINVAL;
+ switch (buf[0]) {
+ case '0':
+ ret = cpu_down(cpudev->cpu);
+ break;
+ case '1':
+ ret = cpu_up(cpudev->cpu);
+ break;
+ default:
+ ret = -EINVAL;
+ }
+
+ if (ret == 0)
+ ret = count;
+ return ret;
+}
+
+static void __init create_entries(struct device *parent)
+{
+ struct device_attribute *online;
+
+ online = kmalloc(sizeof(*online), GFP_KERNEL);
+ online->attr.name = "online";
+ online->attr.mode = 0600;
+ online->show = show_online;
+ online->store = store_online;
+ device_create_file(parent, online);
+}
+
+struct device cpu_dev = {
+ .name = "cpus",
+ .bus_id = "cpus"
+};
+
+static int __init create_per_cpu_entries(void)
+{
+ unsigned int i;
+
+ register_sys_device(&cpu_dev);
+
+ for (i = 0; i < NR_CPUS; i++) {
+ if (cpu_possible(i)) {
+ struct cpu_device *cpudir;
+ cpudir = kmalloc(sizeof(*cpudir), GFP_KERNEL);
+ memset(cpudir, 0, sizeof(*cpudir));
+
+ sprintf(cpudir->dev.name, "%i", i);
+ sprintf(cpudir->dev.bus_id, "%04x", i);
+ cpudir->dev.parent = &cpu_dev;
+ cpudir->cpu = i;
+
+ device_register(&cpudir->dev);
+ create_entries(&cpudir->dev);
+ }
+ }
+ return 0;
+}
+
+__initcall(create_per_cpu_entries);
+
+#else /* !CONFIG_HOTPLUG */
+
+int cpu_down(unsigned int cpu)
+{
+ return -ENOSYS;
+}
+
+static inline int cpu_run_sbin_hotplug(unsigned int cpu, const char *action)
+{
+ return 0;
+}
+#endif
+
/* Need to know about CPUs going up/down? */
int register_cpu_notifier(struct notifier_block *nb)
{
@@ -48,7 +255,11 @@ int __devinit cpu_up(unsigned int cpu)
printk("CPU %u IS NOW UP!\n", cpu);
notifier_call_chain(&cpu_chain, CPU_ONLINE, (void *)(long)cpu);
+#if 0 /* FIXME: Don't do this during boot. --RR */
+ cpu_run_sbin_hotplug(cpu, "add");
+#endif
out:
up(&cpucontrol);
return ret;
}
+
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/kernel/sched.c .7234-linux-2.5.34.updated/kernel/sched.c
--- .7234-linux-2.5.34/kernel/sched.c 2002-09-13 11:46:29.000000000 +1000
+++ .7234-linux-2.5.34.updated/kernel/sched.c 2002-09-13 11:47:04.000000000 +1000
@@ -412,11 +412,13 @@ repeat_lock_task:
if (!p->array) {
/*
* Fast-migrate the task if it's not running or runnable
- * currently. Do not violate hard affinity.
+ * currently. Do not violate hard affinity. Do not pull
+ * onto offline CPU.
*/
if (unlikely(sync && !task_running(rq, p) &&
(task_cpu(p) != smp_processor_id()) &&
- (test_bit(smp_processor_id(), p->cpus_allowed)))) {
+ (test_bit(smp_processor_id(), p->cpus_allowed)) &&
+ cpu_online(smp_processor_id()))) {
set_task_cpu(p, smp_processor_id());
task_rq_unlock(rq, &flags);
@@ -744,6 +746,11 @@ static void load_balance(runqueue_t *thi
struct list_head *head, *curr;
task_t *tmp;
+ /* CPU going down is a special case: we don't pull more tasks
+ onboard */
+ if (unlikely(!cpu_online(this_cpu)))
+ goto out;
+
busiest = find_busiest_queue(this_rq, this_cpu, idle, &imbalance);
if (!busiest)
goto out;
@@ -1968,6 +1975,105 @@ int set_cpus_allowed(task_t *p, const un
return ret;
}
+/* Move (not current) task off this cpu, onto dest cpu. Reference to
+ task must be held. */
+static void move_task_away(struct task_struct *p, unsigned int dest_cpu)
+{
+ runqueue_t *rq_dest;
+ unsigned long flags;
+
+ rq_dest = cpu_rq(dest_cpu);
+
+ if (task_cpu(p) != smp_processor_id())
+ return; /* Already moved */
+
+ local_irq_save(flags);
+ double_rq_lock(this_rq(), rq_dest);
+ if (task_cpu(p) != smp_processor_id())
+ goto out; /* Already moved */
+
+ set_task_cpu(p, dest_cpu);
+ if (p->array) {
+ deactivate_task(p, this_rq());
+ activate_task(p, rq_dest);
+ }
+ out:
+ double_rq_unlock(this_rq(), rq_dest);
+ local_irq_restore(flags);
+}
+
+#ifdef CONFIG_HOTPLUG
+/* Slow but sure. We don't fight against load_balance, new people
+ setting affinity, or try_to_wake_up's fast path pulling things in,
+ as cpu_online() no longer true. */
+static int move_all_tasks(unsigned int kill_it)
+{
+ unsigned int num_signalled = 0;
+ unsigned int dest_cpu;
+ struct task_struct *p;
+ DECLARE_BITMAP(cpus_allowed, NR_CPUS);
+
+ again:
+ read_lock(&tasklist_lock);
+ for_each_task(p) {
+ if (p == current)
+ continue;
+
+ /* Kernel threads which are bound to specific
+ processors need to look after themselves
+ with their own callbacks */
+ if (p->mm == NULL
+ && find_first_zero_bit(p->cpus_allowed,NR_CPUS) != NR_CPUS)
+ continue;
+
+ if (task_cpu(p) != smp_processor_id())
+ continue;
+
+ get_task_struct(p);
+ break;
+ }
+ read_unlock(&tasklist_lock);
+
+ /* Did we reach the end? */
+ if (p == &init_task)
+ return num_signalled;
+
+ memcpy(cpus_allowed, p->cpus_allowed, sizeof(cpus_allowed));
+ clear_bit(smp_processor_id(), cpus_allowed);
+ dest_cpu = any_online_cpu(cpus_allowed);
+ if (dest_cpu == NR_CPUS) {
+ num_signalled++;
+ if (!kill_it) {
+ /* FIXME: New signal needed? --RR */
+ force_sig(SIGPWR, p);
+ goto again;
+ }
+ /* Kill it (it can die on any CPU). */
+ memset(p->cpus_allowed, 0xFF, sizeof(p->cpus_allowed));
+ clear_bit(smp_processor_id(), p->cpus_allowed);
+ dest_cpu = any_online_cpu(p->cpus_allowed);
+ force_sig(SIGKILL, p);
+ }
+ move_task_away(p, dest_cpu);
+ goto again;
+}
+
+/* Move non-kernel-thread tasks off this (offline) CPU, except us. */
+void migrate_all_tasks(void)
+{
+ preempt_disable();
+ if (move_all_tasks(0)) {
+ /* Wait for processes to react to signal */
+ schedule_timeout(30*HZ);
+ move_all_tasks(1);
+ }
+}
+#endif /* CONFIG_HOTPLUG */
+
+/* This is the CPU to stop, and who to wake about it */
+static int migration_stop = -1;
+static struct completion migration_stopped;
+
/*
* migration_thread - this is a highprio system thread that performs
* thread migration by 'pulling' threads into the target runqueue.
@@ -2000,10 +2106,8 @@ static int migration_thread(void * data)
sprintf(current->comm, "migration_CPU%d", smp_processor_id());
- for (;;) {
- runqueue_t *rq_src, *rq_dest;
+ while (migration_stop != cpu) {
struct list_head *head;
- int cpu_src, cpu_dest;
migration_req_t *req;
unsigned long flags;
task_t *p;
@@ -2021,31 +2125,33 @@ static int migration_thread(void * data)
spin_unlock_irqrestore(&rq->lock, flags);
p = req->task;
- cpu_dest = any_online_cpu(p->cpus_allowed);
- rq_dest = cpu_rq(cpu_dest);
-repeat:
- cpu_src = task_cpu(p);
- rq_src = cpu_rq(cpu_src);
-
- local_irq_save(flags);
- double_rq_lock(rq_src, rq_dest);
- if (task_cpu(p) != cpu_src) {
- double_rq_unlock(rq_src, rq_dest);
- local_irq_restore(flags);
- goto repeat;
- }
- if (rq_src == rq) {
- set_task_cpu(p, cpu_dest);
- if (p->array) {
- deactivate_task(p, rq_src);
- activate_task(p, rq_dest);
- }
- }
- double_rq_unlock(rq_src, rq_dest);
- local_irq_restore(flags);
-
+ move_task_away(p, any_online_cpu(p->cpus_allowed));
complete(&req->done);
}
+ current->state = TASK_RUNNING;
+
+ printk("Migration thread for %u exiting\n", cpu);
+ rq->migration_thread = NULL;
+ complete(&migration_stopped);
+
+ return 0;
+}
+
+/* No locking required: CPU notifiers are serialized */
+static void stop_migration_thread(unsigned int cpu)
+{
+ /* We want to wake it, but it may exit first. */
+ struct task_struct *migthread = cpu_rq(cpu)->migration_thread;
+
+ get_task_struct(migthread);
+ init_completion(&migration_stopped);
+ /* They must not access completion until it's initialized. */
+ wmb();
+ migration_stop = cpu;
+ wake_up_process(cpu_rq(cpu)->migration_thread);
+ wait_for_completion(&migration_stopped);
+ put_task_struct(migthread);
+ migration_stop = -1;
}
/*
@@ -2065,6 +2171,9 @@ static int migration_call(struct notifie
while (!cpu_rq((long)hcpu)->migration_thread)
yield();
break;
+ case CPU_OFFLINE:
+ stop_migration_thread((long)hcpu);
+ break;
}
return NOTIFY_OK;
}
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/kernel/softirq.c .7234-linux-2.5.34.updated/kernel/softirq.c
--- .7234-linux-2.5.34/kernel/softirq.c 2002-09-13 11:46:28.000000000 +1000
+++ .7234-linux-2.5.34.updated/kernel/softirq.c 2002-09-13 11:47:04.000000000 +1000
@@ -352,11 +352,20 @@ void __run_task_queue(task_queue *list)
}
}
+/* This is the CPU to stop, and who to wake about it */
+static int ksoftirq_stop = -1;
+static struct task_struct *ksoftirq_killer = NULL;
+
static int ksoftirqd(void * __bind_cpu)
{
int cpu = (int) (long) __bind_cpu;
+ if (ksoftirqd_task(cpu))
+ BUG();
+
+ sprintf(current->comm, "ksoftirqd_CPU%d", cpu);
daemonize();
+ reparent_to_init();
set_user_nice(current, 19);
current->flags |= PF_IOTHREAD;
sigfillset(¤t->blocked);
@@ -372,7 +381,8 @@ static int ksoftirqd(void * __bind_cpu)
ksoftirqd_task(cpu) = current;
- for (;;) {
+ while (ksoftirq_stop != cpu) {
+ rmb();
if (!softirq_pending(cpu))
schedule();
@@ -385,25 +395,98 @@ static int ksoftirqd(void * __bind_cpu)
__set_current_state(TASK_INTERRUPTIBLE);
}
+ set_current_state(TASK_RUNNING);
+
+ printk("ksoftirqd for %i dying\n", cpu);
+ ksoftirqd_task(cpu) = NULL;
+ wmb();
+ wake_up_process(ksoftirq_killer);
+
+ return 0;
}
static int __devinit cpu_callback(struct notifier_block *nfb,
unsigned long action,
void *hcpu)
{
- int hotcpu = (unsigned long)hcpu;
+ unsigned int i, hotcpu = (unsigned long)hcpu;
if (action == CPU_ONLINE) {
if (kernel_thread(ksoftirqd, hcpu,
CLONE_FS | CLONE_FILES | CLONE_SIGNAL) < 0) {
- printk("ksoftirqd for %i failed\n", hotcpu);
+ printk("ksoftirqd for %u failed\n", hotcpu);
return NOTIFY_BAD;
}
while (!ksoftirqd_task(hotcpu))
yield();
return NOTIFY_OK;
- }
+ }
+
+ if (action == CPU_OFFLINE) {
+ struct task_struct *kd_task;
+
+ printk("Killing ksoftirqd for %u\n", hotcpu);
+ /* Kill ksoftirqd: get ref in case it exits before we
+ wake it */
+ ksoftirq_killer = current;
+ kd_task = ksoftirqd_task(hotcpu);
+ get_task_struct(kd_task);
+ set_current_state(TASK_INTERRUPTIBLE);
+ ksoftirq_stop = hotcpu;
+ wake_up_process(kd_task);
+ while (ksoftirqd_task(hotcpu)) {
+ schedule();
+ set_current_state(TASK_INTERRUPTIBLE);
+ }
+ set_current_state(TASK_RUNNING);
+ put_task_struct(kd_task);
+ ksoftirq_stop = -1;
+ return NOTIFY_OK;
+ }
+
+ if (action == CPU_DEAD) {
+ __u32 pending;
+ struct tasklet_struct *list, *t;
+ unsigned int cpu = get_cpu();
+
+ printk("Moving ksoftirqs for %u\n", hotcpu);
+
+ /* move pending softirqs */
+ local_irq_disable();
+
+ pending = softirq_pending(hotcpu);
+ softirq_pending(hotcpu) = 0;
+
+ for (i=0; i<32; i++)
+ if (pending & (1<<i))
+ __cpu_raise_softirq(cpu, i);
+ list = per_cpu(tasklet_vec, hotcpu).list;
+ if (list != NULL) {
+ t = list;
+ while (list->next != NULL)
+ list = list->next;
+ list->next = __get_cpu_var(tasklet_vec).list;
+ __get_cpu_var(tasklet_vec).list = t;
+ per_cpu(tasklet_vec, hotcpu).list = NULL;
+ }
+
+ list = per_cpu(tasklet_hi_vec, hotcpu).list;
+ if (list != NULL) {
+ t = list;
+ while (list->next != NULL)
+ list = list->next;
+ list->next = __get_cpu_var(tasklet_hi_vec).list;
+ __get_cpu_var(tasklet_hi_vec).list = t;
+ per_cpu(tasklet_hi_vec, hotcpu).list = NULL;
+ }
+ local_irq_enable();
+ put_cpu();
+
+ return NOTIFY_OK;
+ }
+
+ /* We fear change! */
return NOTIFY_BAD;
}
diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7234-linux-2.5.34/net/core/dev.c .7234-linux-2.5.34.updated/net/core/dev.c
--- .7234-linux-2.5.34/net/core/dev.c 2002-09-01 12:23:08.000000000 +1000
+++ .7234-linux-2.5.34.updated/net/core/dev.c 2002-09-13 11:47:04.000000000 +1000
@@ -105,6 +105,7 @@
#include <linux/init.h>
#include <linux/kmod.h>
#include <linux/module.h>
+#include <linux/smp.h>
#if defined(CONFIG_NET_RADIO) || defined(CONFIG_NET_PCMCIA_RADIO)
#include <linux/wireless.h> /* Note : will define WIRELESS_EXT */
#include <net/iw_handler.h>
@@ -2864,3 +2865,67 @@ static int net_run_sbin_hotplug(struct n
return call_usermodehelper(argv [0], argv, envp);
}
#endif
+
+static int dev_cpu_callback(struct notifier_block *nfb, unsigned long action, void * ocpu)
+{
+ struct sk_buff *list_sk, *sk_head;
+ struct net_device *list_net, *net_head;
+ struct softnet_data *queue;
+ struct sk_buff *skb;
+ unsigned int cpu = smp_processor_id();
+ unsigned long oldcpu = (unsigned long) ocpu;
+ unsigned long flags;
+
+ if (action != CPU_OFFLINE)
+ return 0;
+
+ local_irq_save(flags);
+
+ /* Move completion queue */
+
+ list_sk = softnet_data[oldcpu].completion_queue;
+ if (list_sk != NULL) {
+ sk_head = list_sk;
+ while (list_sk->next != NULL)
+ list_sk = list_sk->next;
+ list_sk->next = softnet_data[cpu].completion_queue;
+ softnet_data[cpu].completion_queue = sk_head;
+ softnet_data[oldcpu].completion_queue = NULL;
+ }
+
+ /* Move output_queue */
+
+ list_net = softnet_data[oldcpu].output_queue;
+ if (list_net != NULL) {
+ net_head = list_net;
+ while (list_net->next != NULL)
+ list_net = list_net->next_sched;
+ list_net->next_sched = softnet_data[cpu].output_queue;
+ softnet_data[cpu].output_queue = net_head;
+ softnet_data[oldcpu].output_queue = NULL;
+ }
+
+ local_irq_restore(flags);
+
+ /* Move input_pkt_queue */
+
+ queue = &softnet_data[oldcpu];
+ for (;;) {
+ skb = __skb_dequeue(&queue->input_pkt_queue);
+ if (skb == NULL)
+ break;
+ netif_rx(skb);
+ }
+
+ return 0;
+}
+
+static struct notifier_block cpu_callback_nfb = {&dev_cpu_callback, NULL, 0 };
+
+static int __init dev_cpu_callback_init(void)
+{
+ register_cpu_notifier(&cpu_callback_nfb);
+ return 0;
+}
+
+__initcall(dev_cpu_callback_init);
--
Anyone who quotes me in their sig is an idiot. -- Rusty Russell.
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2002-09-13 1:56 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2002-09-13 1:56 [PATCH] Hot unplug CPU 4/4 Rusty Russell
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).