#include <linux/mm.h>
#include <linux/cpumask.h>
#include <linux/cpu.h>
-#include <linux/profile.h>
#include <linux/highmem.h>
#include <linux/mutex.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
#include <asm/sections.h>
-#include <asm/semaphore.h>
#include <asm/irq_regs.h>
+#include <asm/ptrace.h>
struct profile_hit {
u32 pc, hits;
#define NR_PROFILE_GRP (NR_PROFILE_HIT/PROFILE_GRPSZ)
/* Oprofile timer tick hook */
-int (*timer_hook)(struct pt_regs *) __read_mostly;
+static int (*timer_hook)(struct pt_regs *) __read_mostly;
static atomic_t *prof_buffer;
static unsigned long prof_len, prof_shift;
-static int prof_on __read_mostly;
-static cpumask_t prof_cpu_mask = CPU_MASK_ALL;
+
+int prof_on __read_mostly;
+EXPORT_SYMBOL_GPL(prof_on);
+
+static cpumask_var_t prof_cpu_mask;
#ifdef CONFIG_SMP
static DEFINE_PER_CPU(struct profile_hit *[2], cpu_profile_hits);
static DEFINE_PER_CPU(int, cpu_profile_flip);
static DEFINE_MUTEX(profile_flip_mutex);
#endif /* CONFIG_SMP */
-static int __init profile_setup(char * str)
+int profile_setup(char *str)
{
- static char __initdata schedstr[] = "schedule";
+ static char schedstr[] = "schedule";
+ static char sleepstr[] = "sleep";
+ static char kvmstr[] = "kvm";
int par;
- if (!strncmp(str, schedstr, strlen(schedstr))) {
+ if (!strncmp(str, sleepstr, strlen(sleepstr))) {
+#ifdef CONFIG_SCHEDSTATS
+ prof_on = SLEEP_PROFILING;
+ if (str[strlen(sleepstr)] == ',')
+ str += strlen(sleepstr) + 1;
+ if (get_option(&str, &par))
+ prof_shift = par;
+ printk(KERN_INFO
+ "kernel sleep profiling enabled (shift: %ld)\n",
+ prof_shift);
+#else
+ printk(KERN_WARNING
+ "kernel sleep profiling requires CONFIG_SCHEDSTATS\n");
+#endif /* CONFIG_SCHEDSTATS */
+ } else if (!strncmp(str, schedstr, strlen(schedstr))) {
prof_on = SCHED_PROFILING;
if (str[strlen(schedstr)] == ',')
str += strlen(schedstr) + 1;
printk(KERN_INFO
"kernel schedule profiling enabled (shift: %ld)\n",
prof_shift);
+ } else if (!strncmp(str, kvmstr, strlen(kvmstr))) {
+ prof_on = KVM_PROFILING;
+ if (str[strlen(kvmstr)] == ',')
+ str += strlen(kvmstr) + 1;
+ if (get_option(&str, &par))
+ prof_shift = par;
+ printk(KERN_INFO
+ "kernel KVM profiling enabled (shift: %ld)\n",
+ prof_shift);
} else if (get_option(&str, &par)) {
prof_shift = par;
prof_on = CPU_PROFILING;
__setup("profile=", profile_setup);
-void __init profile_init(void)
+int __ref profile_init(void)
{
- if (!prof_on)
- return;
-
+ int buffer_bytes;
+ if (!prof_on)
+ return 0;
+
/* only text is profiled */
prof_len = (_etext - _stext) >> prof_shift;
- prof_buffer = alloc_bootmem(prof_len*sizeof(atomic_t));
+ buffer_bytes = prof_len*sizeof(atomic_t);
+
+ if (!alloc_cpumask_var(&prof_cpu_mask, GFP_KERNEL))
+ return -ENOMEM;
+
+ cpumask_copy(prof_cpu_mask, cpu_possible_mask);
+
+ prof_buffer = kzalloc(buffer_bytes, GFP_KERNEL|__GFP_NOWARN);
+ if (prof_buffer)
+ return 0;
+
+ prof_buffer = alloc_pages_exact(buffer_bytes,
+ GFP_KERNEL|__GFP_ZERO|__GFP_NOWARN);
+ if (prof_buffer)
+ return 0;
+
+ prof_buffer = vmalloc(buffer_bytes);
+ if (prof_buffer)
+ return 0;
+
+ free_cpumask_var(prof_cpu_mask);
+ return -ENOMEM;
}
/* Profile event notifications */
-
-#ifdef CONFIG_PROFILING
-
+
static BLOCKING_NOTIFIER_HEAD(task_exit_notifier);
static ATOMIC_NOTIFIER_HEAD(task_free_notifier);
static BLOCKING_NOTIFIER_HEAD(munmap_notifier);
-
-void profile_task_exit(struct task_struct * task)
+
+void profile_task_exit(struct task_struct *task)
{
blocking_notifier_call_chain(&task_exit_notifier, 0, task);
}
-
-int profile_handoff_task(struct task_struct * task)
+
+int profile_handoff_task(struct task_struct *task)
{
int ret;
ret = atomic_notifier_call_chain(&task_free_notifier, 0, task);
blocking_notifier_call_chain(&munmap_notifier, 0, (void *)addr);
}
-int task_handoff_register(struct notifier_block * n)
+int task_handoff_register(struct notifier_block *n)
{
return atomic_notifier_chain_register(&task_free_notifier, n);
}
+EXPORT_SYMBOL_GPL(task_handoff_register);
-int task_handoff_unregister(struct notifier_block * n)
+int task_handoff_unregister(struct notifier_block *n)
{
return atomic_notifier_chain_unregister(&task_free_notifier, n);
}
+EXPORT_SYMBOL_GPL(task_handoff_unregister);
-int profile_event_register(enum profile_type type, struct notifier_block * n)
+int profile_event_register(enum profile_type type, struct notifier_block *n)
{
int err = -EINVAL;
-
+
switch (type) {
- case PROFILE_TASK_EXIT:
- err = blocking_notifier_chain_register(
- &task_exit_notifier, n);
- break;
- case PROFILE_MUNMAP:
- err = blocking_notifier_chain_register(
- &munmap_notifier, n);
- break;
+ case PROFILE_TASK_EXIT:
+ err = blocking_notifier_chain_register(
+ &task_exit_notifier, n);
+ break;
+ case PROFILE_MUNMAP:
+ err = blocking_notifier_chain_register(
+ &munmap_notifier, n);
+ break;
}
-
+
return err;
}
+EXPORT_SYMBOL_GPL(profile_event_register);
-
-int profile_event_unregister(enum profile_type type, struct notifier_block * n)
+int profile_event_unregister(enum profile_type type, struct notifier_block *n)
{
int err = -EINVAL;
-
+
switch (type) {
- case PROFILE_TASK_EXIT:
- err = blocking_notifier_chain_unregister(
- &task_exit_notifier, n);
- break;
- case PROFILE_MUNMAP:
- err = blocking_notifier_chain_unregister(
- &munmap_notifier, n);
- break;
+ case PROFILE_TASK_EXIT:
+ err = blocking_notifier_chain_unregister(
+ &task_exit_notifier, n);
+ break;
+ case PROFILE_MUNMAP:
+ err = blocking_notifier_chain_unregister(
+ &munmap_notifier, n);
+ break;
}
return err;
}
+EXPORT_SYMBOL_GPL(profile_event_unregister);
int register_timer_hook(int (*hook)(struct pt_regs *))
{
timer_hook = hook;
return 0;
}
+EXPORT_SYMBOL_GPL(register_timer_hook);
void unregister_timer_hook(int (*hook)(struct pt_regs *))
{
/* make sure all CPUs see the NULL hook */
synchronize_sched(); /* Allow ongoing interrupts to complete. */
}
-
-EXPORT_SYMBOL_GPL(register_timer_hook);
EXPORT_SYMBOL_GPL(unregister_timer_hook);
-EXPORT_SYMBOL_GPL(task_handoff_register);
-EXPORT_SYMBOL_GPL(task_handoff_unregister);
-
-#endif /* CONFIG_PROFILING */
-EXPORT_SYMBOL_GPL(profile_event_register);
-EXPORT_SYMBOL_GPL(profile_event_unregister);
#ifdef CONFIG_SMP
/*
* positions to which hits are accounted during short intervals (e.g.
* several seconds) is usually very small. Exclusion from buffer
* flipping is provided by interrupt disablement (note that for
- * SCHED_PROFILING profile_hit() may be called from process context).
+ * SCHED_PROFILING or SLEEP_PROFILING profile_hit() may be called from
+ * process context).
* The hash function is meant to be lightweight as opposed to strong,
* and was vaguely inspired by ppc64 firmware-supported inverted
* pagetable hash functions, but uses a full hashtable full of finite
mutex_lock(&profile_flip_mutex);
j = per_cpu(cpu_profile_flip, get_cpu());
put_cpu();
- on_each_cpu(__profile_flip_buffers, NULL, 0, 1);
+ on_each_cpu(__profile_flip_buffers, NULL, 1);
for_each_online_cpu(cpu) {
struct profile_hit *hits = per_cpu(cpu_profile_hits, cpu)[j];
for (i = 0; i < NR_PROFILE_HIT; ++i) {
mutex_lock(&profile_flip_mutex);
i = per_cpu(cpu_profile_flip, get_cpu());
put_cpu();
- on_each_cpu(__profile_flip_buffers, NULL, 0, 1);
+ on_each_cpu(__profile_flip_buffers, NULL, 1);
for_each_online_cpu(cpu) {
struct profile_hit *hits = per_cpu(cpu_profile_hits, cpu)[i];
memset(hits, 0, NR_PROFILE_HIT*sizeof(struct profile_hit));
mutex_unlock(&profile_flip_mutex);
}
-void profile_hit(int type, void *__pc)
+void profile_hits(int type, void *__pc, unsigned int nr_hits)
{
unsigned long primary, secondary, flags, pc = (unsigned long)__pc;
int i, j, cpu;
put_cpu();
return;
}
+ /*
+ * We buffer the global profiler buffer into a per-CPU
+ * queue and thus reduce the number of global (and possibly
+ * NUMA-alien) accesses. The write-queue is self-coalescing:
+ */
local_irq_save(flags);
do {
for (j = 0; j < PROFILE_GRPSZ; ++j) {
if (hits[i + j].pc == pc) {
- hits[i + j].hits++;
+ hits[i + j].hits += nr_hits;
goto out;
} else if (!hits[i + j].hits) {
hits[i + j].pc = pc;
- hits[i + j].hits = 1;
+ hits[i + j].hits = nr_hits;
goto out;
}
}
i = (i + secondary) & (NR_PROFILE_HIT - 1);
} while (i != primary);
- atomic_inc(&prof_buffer[pc]);
+
+ /*
+ * Add the current hit(s) and flush the write-queue out
+ * to the global buffer:
+ */
+ atomic_add(nr_hits, &prof_buffer[pc]);
for (i = 0; i < NR_PROFILE_HIT; ++i) {
atomic_add(hits[i].hits, &prof_buffer[hits[i].pc]);
hits[i].pc = hits[i].hits = 0;
put_cpu();
}
-#ifdef CONFIG_HOTPLUG_CPU
-static int __devinit profile_cpu_callback(struct notifier_block *info,
+static int __cpuinit profile_cpu_callback(struct notifier_block *info,
unsigned long action, void *__cpu)
{
int node, cpu = (unsigned long)__cpu;
switch (action) {
case CPU_UP_PREPARE:
+ case CPU_UP_PREPARE_FROZEN:
node = cpu_to_node(cpu);
per_cpu(cpu_profile_flip, cpu) = 0;
if (!per_cpu(cpu_profile_hits, cpu)[1]) {
- page = alloc_pages_node(node,
- GFP_KERNEL | __GFP_ZERO | GFP_THISNODE,
+ page = alloc_pages_exact_node(node,
+ GFP_KERNEL | __GFP_ZERO,
0);
if (!page)
return NOTIFY_BAD;
per_cpu(cpu_profile_hits, cpu)[1] = page_address(page);
}
if (!per_cpu(cpu_profile_hits, cpu)[0]) {
- page = alloc_pages_node(node,
- GFP_KERNEL | __GFP_ZERO | GFP_THISNODE,
+ page = alloc_pages_exact_node(node,
+ GFP_KERNEL | __GFP_ZERO,
0);
if (!page)
goto out_free;
per_cpu(cpu_profile_hits, cpu)[0] = page_address(page);
}
break;
- out_free:
+out_free:
page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[1]);
per_cpu(cpu_profile_hits, cpu)[1] = NULL;
__free_page(page);
return NOTIFY_BAD;
case CPU_ONLINE:
- cpu_set(cpu, prof_cpu_mask);
+ case CPU_ONLINE_FROZEN:
+ if (prof_cpu_mask != NULL)
+ cpumask_set_cpu(cpu, prof_cpu_mask);
break;
case CPU_UP_CANCELED:
+ case CPU_UP_CANCELED_FROZEN:
case CPU_DEAD:
- cpu_clear(cpu, prof_cpu_mask);
+ case CPU_DEAD_FROZEN:
+ if (prof_cpu_mask != NULL)
+ cpumask_clear_cpu(cpu, prof_cpu_mask);
if (per_cpu(cpu_profile_hits, cpu)[0]) {
page = virt_to_page(per_cpu(cpu_profile_hits, cpu)[0]);
per_cpu(cpu_profile_hits, cpu)[0] = NULL;
}
return NOTIFY_OK;
}
-#endif /* CONFIG_HOTPLUG_CPU */
#else /* !CONFIG_SMP */
#define profile_flip_buffers() do { } while (0)
#define profile_discard_flip_buffers() do { } while (0)
+#define profile_cpu_callback NULL
-void profile_hit(int type, void *__pc)
+void profile_hits(int type, void *__pc, unsigned int nr_hits)
{
unsigned long pc;
if (prof_on != type || !prof_buffer)
return;
pc = ((unsigned long)__pc - (unsigned long)_stext) >> prof_shift;
- atomic_inc(&prof_buffer[min(pc, prof_len - 1)]);
+ atomic_add(nr_hits, &prof_buffer[min(pc, prof_len - 1)]);
}
#endif /* !CONFIG_SMP */
+EXPORT_SYMBOL_GPL(profile_hits);
void profile_tick(int type)
{
if (type == CPU_PROFILING && timer_hook)
timer_hook(regs);
- if (!user_mode(regs) && cpu_isset(smp_processor_id(), prof_cpu_mask))
+ if (!user_mode(regs) && prof_cpu_mask != NULL &&
+ cpumask_test_cpu(smp_processor_id(), prof_cpu_mask))
profile_hit(type, (void *)profile_pc(regs));
}
#ifdef CONFIG_PROC_FS
#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
#include <asm/uaccess.h>
-#include <asm/ptrace.h>
-static int prof_cpu_mask_read_proc (char *page, char **start, off_t off,
- int count, int *eof, void *data)
+static int prof_cpu_mask_proc_show(struct seq_file *m, void *v)
+{
+ seq_cpumask(m, prof_cpu_mask);
+ seq_putc(m, '\n');
+ return 0;
+}
+
+static int prof_cpu_mask_proc_open(struct inode *inode, struct file *file)
{
- int len = cpumask_scnprintf(page, count, *(cpumask_t *)data);
- if (count - len < 2)
- return -EINVAL;
- len += sprintf(page + len, "\n");
- return len;
+ return single_open(file, prof_cpu_mask_proc_show, NULL);
}
-static int prof_cpu_mask_write_proc (struct file *file, const char __user *buffer,
- unsigned long count, void *data)
+static ssize_t prof_cpu_mask_proc_write(struct file *file,
+ const char __user *buffer, size_t count, loff_t *pos)
{
- cpumask_t *mask = (cpumask_t *)data;
- unsigned long full_count = count, err;
- cpumask_t new_value;
+ cpumask_var_t new_value;
+ int err;
- err = cpumask_parse_user(buffer, count, new_value);
- if (err)
- return err;
+ if (!alloc_cpumask_var(&new_value, GFP_KERNEL))
+ return -ENOMEM;
- *mask = new_value;
- return full_count;
+ err = cpumask_parse_user(buffer, count, new_value);
+ if (!err) {
+ cpumask_copy(prof_cpu_mask, new_value);
+ err = count;
+ }
+ free_cpumask_var(new_value);
+ return err;
}
+static const struct file_operations prof_cpu_mask_proc_fops = {
+ .open = prof_cpu_mask_proc_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = single_release,
+ .write = prof_cpu_mask_proc_write,
+};
+
void create_prof_cpu_mask(struct proc_dir_entry *root_irq_dir)
{
- struct proc_dir_entry *entry;
-
/* create /proc/irq/prof_cpu_mask */
- if (!(entry = create_proc_entry("prof_cpu_mask", 0600, root_irq_dir)))
- return;
- entry->nlink = 1;
- entry->data = (void *)&prof_cpu_mask;
- entry->read_proc = prof_cpu_mask_read_proc;
- entry->write_proc = prof_cpu_mask_write_proc;
+ proc_create("prof_cpu_mask", 0600, root_irq_dir, &prof_cpu_mask_proc_fops);
}
/*
{
unsigned long p = *ppos;
ssize_t read;
- char * pnt;
+ char *pnt;
unsigned int sample_step = 1 << prof_shift;
profile_flip_buffers();
read = 0;
while (p < sizeof(unsigned int) && count > 0) {
- put_user(*((char *)(&sample_step)+p),buf);
+ if (put_user(*((char *)(&sample_step)+p), buf))
+ return -EFAULT;
buf++; p++; count--; read++;
}
pnt = (char *)prof_buffer + p - sizeof(atomic_t);
- if (copy_to_user(buf,(void *)pnt,count))
+ if (copy_to_user(buf, (void *)pnt, count))
return -EFAULT;
read += count;
*ppos += read;
size_t count, loff_t *ppos)
{
#ifdef CONFIG_SMP
- extern int setup_profiling_timer (unsigned int multiplier);
+ extern int setup_profiling_timer(unsigned int multiplier);
if (count == sizeof(int)) {
unsigned int multiplier;
return count;
}
-static struct file_operations proc_profile_operations = {
+static const struct file_operations proc_profile_operations = {
.read = read_profile,
.write = write_profile,
};
#ifdef CONFIG_SMP
-static void __init profile_nop(void *unused)
+static void profile_nop(void *unused)
{
}
-static int __init create_hash_tables(void)
+static int create_hash_tables(void)
{
int cpu;
int node = cpu_to_node(cpu);
struct page *page;
- page = alloc_pages_node(node,
+ page = alloc_pages_exact_node(node,
GFP_KERNEL | __GFP_ZERO | GFP_THISNODE,
0);
if (!page)
goto out_cleanup;
per_cpu(cpu_profile_hits, cpu)[1]
= (struct profile_hit *)page_address(page);
- page = alloc_pages_node(node,
+ page = alloc_pages_exact_node(node,
GFP_KERNEL | __GFP_ZERO | GFP_THISNODE,
0);
if (!page)
out_cleanup:
prof_on = 0;
smp_mb();
- on_each_cpu(profile_nop, NULL, 0, 1);
+ on_each_cpu(profile_nop, NULL, 1);
for_each_online_cpu(cpu) {
struct page *page;
#define create_hash_tables() ({ 0; })
#endif
-static int __init create_proc_profile(void)
+int __ref create_proc_profile(void) /* false positive from hotcpu_notifier */
{
struct proc_dir_entry *entry;
if (!prof_on)
return 0;
if (create_hash_tables())
- return -1;
- if (!(entry = create_proc_entry("profile", S_IWUSR | S_IRUGO, NULL)))
+ return -ENOMEM;
+ entry = proc_create("profile", S_IWUSR | S_IRUGO,
+ NULL, &proc_profile_operations);
+ if (!entry)
return 0;
- entry->proc_fops = &proc_profile_operations;
entry->size = (1+prof_len) * sizeof(atomic_t);
hotcpu_notifier(profile_cpu_callback, 0);
return 0;