5 files changed, 158 insertions, 189 deletions

diff --git a/kernel/cpu.c b/kernel/cpu.c
index 10ba5f1004a5..29510d68338a 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -216,7 +216,6 @@ static int __ref take_cpu_down(void *_param)
 static int __ref _cpu_down(unsigned int cpu, int tasks_frozen)
 {
         int err, nr_calls = 0;
-        struct task_struct *p;
         cpumask_t old_allowed, tmp;
         void *hcpu = (void *)(long)cpu;
         unsigned long mod = tasks_frozen ? CPU_TASKS_FROZEN : 0;
@@ -249,21 +248,18 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen)
         cpus_setall(tmp);
         cpu_clear(cpu, tmp);
         set_cpus_allowed_ptr(current, &tmp);
+        tmp = cpumask_of_cpu(cpu);
 
-        p = __stop_machine_run(take_cpu_down, &tcd_param, cpu);
-
-        if (IS_ERR(p) || cpu_online(cpu)) {
+        err = __stop_machine(take_cpu_down, &tcd_param, &tmp);
+        if (err) {
                 /* CPU didn't die: tell everyone.  Can't complain. */
                 if (raw_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED | mod,
                                             hcpu) == NOTIFY_BAD)
                         BUG();
 
-                if (IS_ERR(p)) {
-                        err = PTR_ERR(p);
-                        goto out_allowed;
-                }
-                goto out_thread;
+                goto out_allowed;
         }
+        BUG_ON(cpu_online(cpu));
 
         /* Wait for it to sleep (leaving idle task). */
         while (!idle_cpu(cpu))
@@ -279,8 +275,6 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen)
 
         check_for_tasks(cpu);
 
-out_thread:
-        err = kthread_stop(p);
 out_allowed:
         set_cpus_allowed_ptr(current, &old_allowed);
 out_release:
diff --git a/kernel/module.c b/kernel/module.c
index d8b5605132a0..61d212120df4 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -325,18 +325,6 @@ static unsigned long find_symbol(const char *name,
         return -ENOENT;
 }
 
-/* lookup symbol in given range of kernel_symbols */
-static const struct kernel_symbol *lookup_symbol(const char *name,
-        const struct kernel_symbol *start,
-        const struct kernel_symbol *stop)
-{
-        const struct kernel_symbol *ks = start;
-        for (; ks < stop; ks++)
-                if (strcmp(ks->name, name) == 0)
-                        return ks;
-        return NULL;
-}
-
 /* Search for module by name: must hold module_mutex. */
 static struct module *find_module(const char *name)
 {
@@ -690,7 +678,7 @@ static int try_stop_module(struct module *mod, int flags, int *forced)
         if (flags & O_NONBLOCK) {
                 struct stopref sref = { mod, flags, forced };
 
-                return stop_machine_run(__try_stop_module, &sref, NR_CPUS);
+                return stop_machine(__try_stop_module, &sref, NULL);
         } else {
                 /* We don't need to stop the machine for this. */
                 mod->state = MODULE_STATE_GOING;
@@ -1428,7 +1416,7 @@ static int __unlink_module(void *_mod)
 static void free_module(struct module *mod)
 {
         /* Delete from various lists */
-        stop_machine_run(__unlink_module, mod, NR_CPUS);
+        stop_machine(__unlink_module, mod, NULL);
         remove_notes_attrs(mod);
         remove_sect_attrs(mod);
         mod_kobject_remove(mod);
@@ -1703,6 +1691,19 @@ static void setup_modinfo(struct module *mod, Elf_Shdr *sechdrs,
 }
 
 #ifdef CONFIG_KALLSYMS
+
+/* lookup symbol in given range of kernel_symbols */
+static const struct kernel_symbol *lookup_symbol(const char *name,
+        const struct kernel_symbol *start,
+        const struct kernel_symbol *stop)
+{
+        const struct kernel_symbol *ks = start;
+        for (; ks < stop; ks++)
+                if (strcmp(ks->name, name) == 0)
+                        return ks;
+        return NULL;
+}
+
 static int is_exported(const char *name, const struct module *mod)
 {
         if (!mod && lookup_symbol(name, __start___ksymtab, __stop___ksymtab))
@@ -2196,7 +2197,7 @@ static struct module *load_module(void __user *umod,
         /* Now sew it into the lists so we can get lockdep and oops
          * info during argument parsing.  Noone should access us, since
          * strong_try_module_get() will fail. */
-        stop_machine_run(__link_module, mod, NR_CPUS);
+        stop_machine(__link_module, mod, NULL);
 
         /* Size of section 0 is 0, so this works well if no params */
         err = parse_args(mod->name, mod->args,
@@ -2230,7 +2231,7 @@ static struct module *load_module(void __user *umod,
         return mod;
 
  unlink:
-        stop_machine_run(__unlink_module, mod, NR_CPUS);
+        stop_machine(__unlink_module, mod, NULL);
         module_arch_cleanup(mod);
  cleanup:
         kobject_del(&mod->mkobj.kobj);
diff --git a/kernel/rcuclassic.c b/kernel/rcuclassic.c
index 6f8696c502f4..aad93cdc9f68 100644
--- a/kernel/rcuclassic.c
+++ b/kernel/rcuclassic.c
@@ -91,8 +91,8 @@ static void force_quiescent_state(struct rcu_data *rdp,
                  * rdp->cpu is the current cpu.
                  *
                  * cpu_online_map is updated by the _cpu_down()
-                 * using stop_machine_run(). Since we're in irqs disabled
-                 * section, stop_machine_run() is not exectuting, hence
+                 * using __stop_machine(). Since we're in irqs disabled
+                 * section, __stop_machine() is not exectuting, hence
                  * the cpu_online_map is stable.
                  *
                  * However,  a cpu might have been offlined _just_ before
diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c
index 738b411ff2d3..e446c7c7d6a9 100644
--- a/kernel/stop_machine.c
+++ b/kernel/stop_machine.c
@@ -1,4 +1,4 @@
-/* Copyright 2005 Rusty Russell rusty@rustcorp.com.au IBM Corporation.
+/* Copyright 2008, 2005 Rusty Russell rusty@rustcorp.com.au IBM Corporation.
  * GPL v2 and any later version.
  */
 #include <linux/cpu.h>
@@ -13,204 +13,178 @@
 #include <asm/atomic.h>
 #include <asm/uaccess.h>
 
-/* Since we effect priority and affinity (both of which are visible
- * to, and settable by outside processes) we do indirection via a
- * kthread. */
-
-/* Thread to stop each CPU in user context. */
+/* This controls the threads on each CPU. */
 enum stopmachine_state {
-        STOPMACHINE_WAIT,
+        /* Dummy starting state for thread. */
+        STOPMACHINE_NONE,
+        /* Awaiting everyone to be scheduled. */
         STOPMACHINE_PREPARE,
+        /* Disable interrupts. */
         STOPMACHINE_DISABLE_IRQ,
+        /* Run the function */
+        STOPMACHINE_RUN,
+        /* Exit */
         STOPMACHINE_EXIT,
 };
+static enum stopmachine_state state;
 
-static enum stopmachine_state stopmachine_state;
-static unsigned int stopmachine_num_threads;
-static atomic_t stopmachine_thread_ack;
-
-static int stopmachine(void *cpu)
-{
-        int irqs_disabled = 0;
-        int prepared = 0;
-        cpumask_of_cpu_ptr(cpumask, (int)(long)cpu);
-
-        set_cpus_allowed_ptr(current, cpumask);
-
-        /* Ack: we are alive */
-        smp_mb(); /* Theoretically the ack = 0 might not be on this CPU yet. */
-        atomic_inc(&stopmachine_thread_ack);
-
-        /* Simple state machine */
-        while (stopmachine_state != STOPMACHINE_EXIT) {
-                if (stopmachine_state == STOPMACHINE_DISABLE_IRQ 
-                    && !irqs_disabled) {
-                        local_irq_disable();
-                        hard_irq_disable();
-                        irqs_disabled = 1;
-                        /* Ack: irqs disabled. */
-                        smp_mb(); /* Must read state first. */
-                        atomic_inc(&stopmachine_thread_ack);
-                } else if (stopmachine_state == STOPMACHINE_PREPARE
-                           && !prepared) {
-                        /* Everyone is in place, hold CPU. */
-                        preempt_disable();
-                        prepared = 1;
-                        smp_mb(); /* Must read state first. */
-                        atomic_inc(&stopmachine_thread_ack);
-                }
-                /* Yield in first stage: migration threads need to
-                 * help our sisters onto their CPUs. */
-                if (!prepared && !irqs_disabled)
-                        yield();
-                cpu_relax();
-        }
-
-        /* Ack: we are exiting. */
-        smp_mb(); /* Must read state first. */
-        atomic_inc(&stopmachine_thread_ack);
-
-        if (irqs_disabled)
-                local_irq_enable();
-        if (prepared)
-                preempt_enable();
+struct stop_machine_data {
+        int (*fn)(void *);
+        void *data;
+        int fnret;
+};
 
-        return 0;
-}
+/* Like num_online_cpus(), but hotplug cpu uses us, so we need this. */
+static unsigned int num_threads;
+static atomic_t thread_ack;
+static struct completion finished;
+static DEFINE_MUTEX(lock);
 
-/* Change the thread state */
-static void stopmachine_set_state(enum stopmachine_state state)
+static void set_state(enum stopmachine_state newstate)
 {
-        atomic_set(&stopmachine_thread_ack, 0);
+        /* Reset ack counter. */
+        atomic_set(&thread_ack, num_threads);
         smp_wmb();
-        stopmachine_state = state;
-        while (atomic_read(&stopmachine_thread_ack) != stopmachine_num_threads)
-                cpu_relax();
+        state = newstate;
 }
 
-static int stop_machine(void)
+/* Last one to ack a state moves to the next state. */
+static void ack_state(void)
 {
-        int i, ret = 0;
-
-        atomic_set(&stopmachine_thread_ack, 0);
-        stopmachine_num_threads = 0;
-        stopmachine_state = STOPMACHINE_WAIT;
-
-        for_each_online_cpu(i) {
-                if (i == raw_smp_processor_id())
-                        continue;
-                ret = kernel_thread(stopmachine, (void *)(long)i,CLONE_KERNEL);
-                if (ret < 0)
-                        break;
-                stopmachine_num_threads++;
-        }
-
-        /* Wait for them all to come to life. */
-        while (atomic_read(&stopmachine_thread_ack) != stopmachine_num_threads) {
-                yield();
-                cpu_relax();
+        if (atomic_dec_and_test(&thread_ack)) {
+                /* If we're the last one to ack the EXIT, we're finished. */
+                if (state == STOPMACHINE_EXIT)
+                        complete(&finished);
+                else
+                        set_state(state + 1);
         }
+}
 
-        /* If some failed, kill them all. */
-        if (ret < 0) {
-                stopmachine_set_state(STOPMACHINE_EXIT);
-                return ret;
-        }
+/* This is the actual thread which stops the CPU.  It exits by itself rather
+ * than waiting for kthread_stop(), because it's easier for hotplug CPU. */
+static int stop_cpu(struct stop_machine_data *smdata)
+{
+        enum stopmachine_state curstate = STOPMACHINE_NONE;
+        int uninitialized_var(ret);
 
-        /* Now they are all started, make them hold the CPUs, ready. */
-        preempt_disable();
-        stopmachine_set_state(STOPMACHINE_PREPARE);
+        /* Simple state machine */
+        do {
+                /* Chill out and ensure we re-read stopmachine_state. */
+                cpu_relax();
+                if (state != curstate) {
+                        curstate = state;
+                        switch (curstate) {
+                        case STOPMACHINE_DISABLE_IRQ:
+                                local_irq_disable();
+                                hard_irq_disable();
+                                break;
+                        case STOPMACHINE_RUN:
+                                /* |= allows error detection if functions on
+                                 * multiple CPUs. */
+                                smdata->fnret |= smdata->fn(smdata->data);
+                                break;
+                        default:
+                                break;
+                        }
+                        ack_state();
+                }
+        } while (curstate != STOPMACHINE_EXIT);
 
-        /* Make them disable irqs. */
-        local_irq_disable();
-        hard_irq_disable();
-        stopmachine_set_state(STOPMACHINE_DISABLE_IRQ);
+        local_irq_enable();
+        do_exit(0);
+}
 
+/* Callback for CPUs which aren't supposed to do anything. */
+static int chill(void *unused)
+{
         return 0;
 }
 
-static void restart_machine(void)
+int __stop_machine(int (*fn)(void *), void *data, const cpumask_t *cpus)
 {
-        stopmachine_set_state(STOPMACHINE_EXIT);
-        local_irq_enable();
-        preempt_enable_no_resched();
-}
+        int i, err;
+        struct stop_machine_data active, idle;
+        struct task_struct **threads;
+
+        active.fn = fn;
+        active.data = data;
+        active.fnret = 0;
+        idle.fn = chill;
+        idle.data = NULL;
+
+        /* This could be too big for stack on large machines. */
+        threads = kcalloc(NR_CPUS, sizeof(threads[0]), GFP_KERNEL);
+        if (!threads)
+                return -ENOMEM;
+
+        /* Set up initial state. */
+        mutex_lock(&lock);
+        init_completion(&finished);
+        num_threads = num_online_cpus();
+        set_state(STOPMACHINE_PREPARE);
 
-struct stop_machine_data {
-        int (*fn)(void *);
-        void *data;
-        struct completion done;
-};
+        for_each_online_cpu(i) {
+                struct stop_machine_data *smdata = &idle;
+                struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
 
-static int do_stop(void *_smdata)
-{
-        struct stop_machine_data *smdata = _smdata;
-        int ret;
+                if (!cpus) {
+                        if (i == first_cpu(cpu_online_map))
+                                smdata = &active;
+                } else {
+                        if (cpu_isset(i, *cpus))
+                                smdata = &active;
+                }
 
-        ret = stop_machine();
-        if (ret == 0) {
-                ret = smdata->fn(smdata->data);
-                restart_machine();
-        }
+                threads[i] = kthread_create((void *)stop_cpu, smdata, "kstop%u",
+                                            i);
+                if (IS_ERR(threads[i])) {
+                        err = PTR_ERR(threads[i]);
+                        threads[i] = NULL;
+                        goto kill_threads;
+                }
 
-        /* We're done: you can kthread_stop us now */
-        complete(&smdata->done);
+                /* Place it onto correct cpu. */
+                kthread_bind(threads[i], i);
 
-        /* Wait for kthread_stop */
-        set_current_state(TASK_INTERRUPTIBLE);
-        while (!kthread_should_stop()) {
-                schedule();
-                set_current_state(TASK_INTERRUPTIBLE);
+                /* Make it highest prio. */
+                if (sched_setscheduler_nocheck(threads[i], SCHED_FIFO, &param))
+                        BUG();
         }
-        __set_current_state(TASK_RUNNING);
-        return ret;
-}
 
-struct task_struct *__stop_machine_run(int (*fn)(void *), void *data,
-                                       unsigned int cpu)
-{
-        static DEFINE_MUTEX(stopmachine_mutex);
-        struct stop_machine_data smdata;
-        struct task_struct *p;
+        /* We've created all the threads.  Wake them all: hold this CPU so one
+         * doesn't hit this CPU until we're ready. */
+        get_cpu();
+        for_each_online_cpu(i)
+                wake_up_process(threads[i]);
 
-        smdata.fn = fn;
-        smdata.data = data;
-        init_completion(&smdata.done);
+        /* This will release the thread on our CPU. */
+        put_cpu();
+        wait_for_completion(&finished);
+        mutex_unlock(&lock);
 
-        mutex_lock(&stopmachine_mutex);
+        kfree(threads);
 
-        /* If they don't care which CPU fn runs on, bind to any online one. */
-        if (cpu == NR_CPUS)
-                cpu = raw_smp_processor_id();
+        return active.fnret;
 
-        p = kthread_create(do_stop, &smdata, "kstopmachine");
-        if (!IS_ERR(p)) {
-                struct sched_param param = { .sched_priority = MAX_RT_PRIO-1 };
+kill_threads:
+        for_each_online_cpu(i)
+                if (threads[i])
+                        kthread_stop(threads[i]);
+        mutex_unlock(&lock);
 
-                /* One high-prio thread per cpu.  We'll do this one. */
-                sched_setscheduler_nocheck(p, SCHED_FIFO, &param);
-                kthread_bind(p, cpu);
-                wake_up_process(p);
-                wait_for_completion(&smdata.done);
-        }
-        mutex_unlock(&stopmachine_mutex);
-        return p;
+        kfree(threads);
+        return err;
 }
 
-int stop_machine_run(int (*fn)(void *), void *data, unsigned int cpu)
+int stop_machine(int (*fn)(void *), void *data, const cpumask_t *cpus)
 {
-        struct task_struct *p;
         int ret;
 
         /* No CPUs can come up or down during this. */
         get_online_cpus();
-        p = __stop_machine_run(fn, data, cpu);
-        if (!IS_ERR(p))
-                ret = kthread_stop(p);
-        else
-                ret = PTR_ERR(p);
+        ret = __stop_machine(fn, data, cpus);
         put_online_cpus();
 
         return ret;
 }
-EXPORT_SYMBOL_GPL(stop_machine_run);
+EXPORT_SYMBOL_GPL(stop_machine);
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 4231a3dc224a..f6e3af31b403 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -587,7 +587,7 @@ static int __ftrace_modify_code(void *data)
 
 static void ftrace_run_update_code(int command)
 {
-        stop_machine_run(__ftrace_modify_code, &command, NR_CPUS);
+        stop_machine(__ftrace_modify_code, &command, NULL);
 }
 
 void ftrace_disable_daemon(void)
@@ -787,7 +787,7 @@ static int ftrace_update_code(void)
             !ftrace_enabled || !ftraced_trigger)
                 return 0;
 
-        stop_machine_run(__ftrace_update_code, NULL, NR_CPUS);
+        stop_machine(__ftrace_update_code, NULL, NULL);
 
         return 1;
 }
@@ -1564,7 +1564,7 @@ static int __init ftrace_dynamic_init(void)
 
         addr = (unsigned long)ftrace_record_ip;
 
-        stop_machine_run(ftrace_dyn_arch_init, &addr, NR_CPUS);
+        stop_machine(ftrace_dyn_arch_init, &addr, NULL);
 
         /* ftrace_dyn_arch_init places the return code in addr */
         if (addr) {