[net-next-2.6.git] / mm / backing-dev.c


#include <linux/wait.h>
#include <linux/backing-dev.h>
#include <linux/kthread.h>
#include <linux/freezer.h>
#include <linux/fs.h>
#include <linux/pagemap.h>
#include <linux/mm.h>
#include <linux/sched.h>
#include <linux/module.h>
#include <linux/writeback.h>
#include <linux/device.h>
#include <trace/events/writeback.h>

static atomic_long_t bdi_seq = ATOMIC_LONG_INIT(0);

void default_unplug_io_fn(struct backing_dev_info *bdi, struct page *page)
{
}
EXPORT_SYMBOL(default_unplug_io_fn);

struct backing_dev_info default_backing_dev_info = {
	.name		= "default",
	.ra_pages	= VM_MAX_READAHEAD * 1024 / PAGE_CACHE_SIZE,
	.state		= 0,
	.capabilities	= BDI_CAP_MAP_COPY,
	.unplug_io_fn	= default_unplug_io_fn,
};
EXPORT_SYMBOL_GPL(default_backing_dev_info);

struct backing_dev_info noop_backing_dev_info = {
	.name		= "noop",
};
EXPORT_SYMBOL_GPL(noop_backing_dev_info);

static struct class *bdi_class;

/*
 * bdi_lock protects updates to bdi_list and bdi_pending_list, as well as
 * reader side protection for bdi_pending_list. bdi_list has RCU reader side
 * locking.
 */
DEFINE_SPINLOCK(bdi_lock);
LIST_HEAD(bdi_list);
LIST_HEAD(bdi_pending_list);

static struct task_struct *sync_supers_tsk;
static struct timer_list sync_supers_timer;

static int bdi_sync_supers(void *);
static void sync_supers_timer_fn(unsigned long);

#ifdef CONFIG_DEBUG_FS
#include <linux/debugfs.h>
#include <linux/seq_file.h>

static struct dentry *bdi_debug_root;

static void bdi_debug_init(void)
{
	bdi_debug_root = debugfs_create_dir("bdi", NULL);
}

static int bdi_debug_stats_show(struct seq_file *m, void *v)
{
	struct backing_dev_info *bdi = m->private;
	struct bdi_writeback *wb = &bdi->wb;
	unsigned long background_thresh;
	unsigned long dirty_thresh;
	unsigned long bdi_thresh;
	unsigned long nr_dirty, nr_io, nr_more_io, nr_wb;
	struct inode *inode;

	nr_wb = nr_dirty = nr_io = nr_more_io = 0;
	spin_lock(&inode_lock);
	list_for_each_entry(inode, &wb->b_dirty, i_list)
		nr_dirty++;
	list_for_each_entry(inode, &wb->b_io, i_list)
		nr_io++;
	list_for_each_entry(inode, &wb->b_more_io, i_list)
		nr_more_io++;
	spin_unlock(&inode_lock);

	get_dirty_limits(&background_thresh, &dirty_thresh, &bdi_thresh, bdi);

#define K(x) ((x) << (PAGE_SHIFT - 10))
	seq_printf(m,
		   "BdiWriteback:     %8lu kB\n"
		   "BdiReclaimable:   %8lu kB\n"
		   "BdiDirtyThresh:   %8lu kB\n"
		   "DirtyThresh:      %8lu kB\n"
		   "BackgroundThresh: %8lu kB\n"
		   "b_dirty:          %8lu\n"
		   "b_io:             %8lu\n"
		   "b_more_io:        %8lu\n"
		   "bdi_list:         %8u\n"
		   "state:            %8lx\n",
		   (unsigned long) K(bdi_stat(bdi, BDI_WRITEBACK)),
		   (unsigned long) K(bdi_stat(bdi, BDI_RECLAIMABLE)),
		   K(bdi_thresh), K(dirty_thresh),
		   K(background_thresh), nr_dirty, nr_io, nr_more_io,
		   !list_empty(&bdi->bdi_list), bdi->state);
#undef K

	return 0;
}

static int bdi_debug_stats_open(struct inode *inode, struct file *file)
{
	return single_open(file, bdi_debug_stats_show, inode->i_private);
}

static const struct file_operations bdi_debug_stats_fops = {
	.open		= bdi_debug_stats_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= single_release,
};

static void bdi_debug_register(struct backing_dev_info *bdi, const char *name)
{
	bdi->debug_dir = debugfs_create_dir(name, bdi_debug_root);
	bdi->debug_stats = debugfs_create_file("stats", 0444, bdi->debug_dir,
					       bdi, &bdi_debug_stats_fops);
}

static void bdi_debug_unregister(struct backing_dev_info *bdi)
{
	debugfs_remove(bdi->debug_stats);
	debugfs_remove(bdi->debug_dir);
}
#else
static inline void bdi_debug_init(void)
{
}
static inline void bdi_debug_register(struct backing_dev_info *bdi,
				      const char *name)
{
}
static inline void bdi_debug_unregister(struct backing_dev_info *bdi)
{
}
#endif

static ssize_t read_ahead_kb_store(struct device *dev,
				  struct device_attribute *attr,
				  const char *buf, size_t count)
{
	struct backing_dev_info *bdi = dev_get_drvdata(dev);
	char *end;
	unsigned long read_ahead_kb;
	ssize_t ret = -EINVAL;

	read_ahead_kb = simple_strtoul(buf, &end, 10);
	if (*buf && (end[0] == '\0' || (end[0] == '\n' && end[1] == '\0'))) {
		bdi->ra_pages = read_ahead_kb >> (PAGE_SHIFT - 10);
		ret = count;
	}
	return ret;
}

#define K(pages) ((pages) << (PAGE_SHIFT - 10))

#define BDI_SHOW(name, expr)						\
static ssize_t name##_show(struct device *dev,				\
			   struct device_attribute *attr, char *page)	\
{									\
	struct backing_dev_info *bdi = dev_get_drvdata(dev);		\
									\
	return snprintf(page, PAGE_SIZE-1, "%lld\n", (long long)expr);	\
}

BDI_SHOW(read_ahead_kb, K(bdi->ra_pages))

static ssize_t min_ratio_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t count)
{
	struct backing_dev_info *bdi = dev_get_drvdata(dev);
	char *end;
	unsigned int ratio;
	ssize_t ret = -EINVAL;

	ratio = simple_strtoul(buf, &end, 10);
	if (*buf && (end[0] == '\0' || (end[0] == '\n' && end[1] == '\0'))) {
		ret = bdi_set_min_ratio(bdi, ratio);
		if (!ret)
			ret = count;
	}
	return ret;
}
BDI_SHOW(min_ratio, bdi->min_ratio)

static ssize_t max_ratio_store(struct device *dev,
		struct device_attribute *attr, const char *buf, size_t count)
{
	struct backing_dev_info *bdi = dev_get_drvdata(dev);
	char *end;
	unsigned int ratio;
	ssize_t ret = -EINVAL;

	ratio = simple_strtoul(buf, &end, 10);
	if (*buf && (end[0] == '\0' || (end[0] == '\n' && end[1] == '\0'))) {
		ret = bdi_set_max_ratio(bdi, ratio);
		if (!ret)
			ret = count;
	}
	return ret;
}
BDI_SHOW(max_ratio, bdi->max_ratio)

#define __ATTR_RW(attr) __ATTR(attr, 0644, attr##_show, attr##_store)

static struct device_attribute bdi_dev_attrs[] = {
	__ATTR_RW(read_ahead_kb),
	__ATTR_RW(min_ratio),
	__ATTR_RW(max_ratio),
	__ATTR_NULL,
};

static __init int bdi_class_init(void)
{
	bdi_class = class_create(THIS_MODULE, "bdi");
	if (IS_ERR(bdi_class))
		return PTR_ERR(bdi_class);

	bdi_class->dev_attrs = bdi_dev_attrs;
	bdi_debug_init();
	return 0;
}
postcore_initcall(bdi_class_init);

static int __init default_bdi_init(void)
{
	int err;

	sync_supers_tsk = kthread_run(bdi_sync_supers, NULL, "sync_supers");
	BUG_ON(IS_ERR(sync_supers_tsk));

	init_timer(&sync_supers_timer);
	setup_timer(&sync_supers_timer, sync_supers_timer_fn, 0);
	bdi_arm_supers_timer();

	err = bdi_init(&default_backing_dev_info);
	if (!err)
		bdi_register(&default_backing_dev_info, NULL, "default");

	return err;
}
subsys_initcall(default_bdi_init);

static void bdi_wb_init(struct bdi_writeback *wb, struct backing_dev_info *bdi)
{
	memset(wb, 0, sizeof(*wb));

	wb->bdi = bdi;
	wb->last_old_flush = jiffies;
	INIT_LIST_HEAD(&wb->b_dirty);
	INIT_LIST_HEAD(&wb->b_io);
	INIT_LIST_HEAD(&wb->b_more_io);
}

int bdi_has_dirty_io(struct backing_dev_info *bdi)
{
	return wb_has_dirty_io(&bdi->wb);
}

static void bdi_flush_io(struct backing_dev_info *bdi)
{
	struct writeback_control wbc = {
		.sync_mode		= WB_SYNC_NONE,
		.older_than_this	= NULL,
		.range_cyclic		= 1,
		.nr_to_write		= 1024,
	};

	writeback_inodes_wb(&bdi->wb, &wbc);
}

/*
 * kupdated() used to do this. We cannot do it from the bdi_forker_thread()
 * or we risk deadlocking on ->s_umount. The longer term solution would be
 * to implement sync_supers_bdi() or similar and simply do it from the
 * bdi writeback thread individually.
 */
static int bdi_sync_supers(void *unused)
{
	set_user_nice(current, 0);

	while (!kthread_should_stop()) {
		set_current_state(TASK_INTERRUPTIBLE);
		schedule();

		/*
		 * Do this periodically, like kupdated() did before.
		 */
		sync_supers();
	}

	return 0;
}

void bdi_arm_supers_timer(void)
{
	unsigned long next;

	if (!dirty_writeback_interval)
		return;

	next = msecs_to_jiffies(dirty_writeback_interval * 10) + jiffies;
	mod_timer(&sync_supers_timer, round_jiffies_up(next));
}

static void sync_supers_timer_fn(unsigned long unused)
{
	wake_up_process(sync_supers_tsk);
	bdi_arm_supers_timer();
}

static int bdi_forker_thread(void *ptr)
{
	struct bdi_writeback *me = ptr;

	current->flags |= PF_FLUSHER | PF_SWAPWRITE;
	set_freezable();

	/*
	 * Our parent may run at a different priority, just set us to normal
	 */
	set_user_nice(current, 0);

	for (;;) {
		bool fork = false;
		struct task_struct *task;
		struct backing_dev_info *bdi;

		/*
		 * Temporary measure, we want to make sure we don't see
		 * dirty data on the default backing_dev_info
		 */
		if (wb_has_dirty_io(me) || !list_empty(&me->bdi->work_list))
			wb_do_writeback(me, 0);

		spin_lock_bh(&bdi_lock);
		set_current_state(TASK_INTERRUPTIBLE);

		/*
		 * Check if any existing bdi's have dirty data without
		 * a thread registered. If so, set that up.
		 */
		list_for_each_entry(bdi, &bdi_list, bdi_list) {
			if (!bdi_cap_writeback_dirty(bdi))
				continue;
			if (bdi->wb.task)
				continue;
			if (list_empty(&bdi->work_list) &&
			    !bdi_has_dirty_io(bdi))
				continue;

			WARN(!test_bit(BDI_registered, &bdi->state),
			     "bdi %p/%s is not registered!\n", bdi, bdi->name);

			fork = true;

			/*
			 * Set the pending bit - if someone will try to
			 * unregister this bdi - it'll wait on this bit.
			 */
			set_bit(BDI_pending, &bdi->state);
			break;
		}
		spin_unlock_bh(&bdi_lock);

		/* Keep working if default bdi still has things to do */
		if (!list_empty(&me->bdi->work_list))
			__set_current_state(TASK_RUNNING);

		if (!fork) {
			unsigned long wait;

			wait = msecs_to_jiffies(dirty_writeback_interval * 10);
			if (wait)
				schedule_timeout(wait);
			else
				schedule();
			try_to_freeze();
			continue;
		}

		__set_current_state(TASK_RUNNING);

		task = kthread_run(bdi_writeback_thread, &bdi->wb, "flush-%s",
				   dev_name(bdi->dev));
		if (IS_ERR(task)) {
			/*
			 * If thread creation fails, force writeout of the bdi
			 * from the thread.
			 */
			bdi_flush_io(bdi);
		} else
			bdi->wb.task = task;
	}

	return 0;
}

/*
 * Remove bdi from bdi_list, and ensure that it is no longer visible
 */
static void bdi_remove_from_list(struct backing_dev_info *bdi)
{
	spin_lock_bh(&bdi_lock);
	list_del_rcu(&bdi->bdi_list);
	spin_unlock_bh(&bdi_lock);

	synchronize_rcu();
}

int bdi_register(struct backing_dev_info *bdi, struct device *parent,
		const char *fmt, ...)
{
	va_list args;
	int ret = 0;
	struct device *dev;

	if (bdi->dev)	/* The driver needs to use separate queues per device */
		goto exit;

	va_start(args, fmt);
	dev = device_create_vargs(bdi_class, parent, MKDEV(0, 0), bdi, fmt, args);
	va_end(args);
	if (IS_ERR(dev)) {
		ret = PTR_ERR(dev);
		goto exit;
	}

	spin_lock_bh(&bdi_lock);
	list_add_tail_rcu(&bdi->bdi_list, &bdi_list);
	spin_unlock_bh(&bdi_lock);

	bdi->dev = dev;

	/*
	 * Just start the forker thread for our default backing_dev_info,
	 * and add other bdi's to the list. They will get a thread created
	 * on-demand when they need it.
	 */
	if (bdi_cap_flush_forker(bdi)) {
		struct bdi_writeback *wb = &bdi->wb;

		wb->task = kthread_run(bdi_forker_thread, wb, "bdi-%s",
						dev_name(dev));
		if (IS_ERR(wb->task)) {
			wb->task = NULL;
			ret = -ENOMEM;

			bdi_remove_from_list(bdi);
			goto exit;
		}
	}

	bdi_debug_register(bdi, dev_name(dev));
	set_bit(BDI_registered, &bdi->state);
	trace_writeback_bdi_register(bdi);
exit:
	return ret;
}
EXPORT_SYMBOL(bdi_register);

int bdi_register_dev(struct backing_dev_info *bdi, dev_t dev)
{
	return bdi_register(bdi, NULL, "%u:%u", MAJOR(dev), MINOR(dev));
}
EXPORT_SYMBOL(bdi_register_dev);

/*
 * Remove bdi from the global list and shutdown any threads we have running
 */
static void bdi_wb_shutdown(struct backing_dev_info *bdi)
{
	if (!bdi_cap_writeback_dirty(bdi))
		return;

	/*
	 * If setup is pending, wait for that to complete first
	 */
	wait_on_bit(&bdi->state, BDI_pending, bdi_sched_wait,
			TASK_UNINTERRUPTIBLE);

	/*
	 * Make sure nobody finds us on the bdi_list anymore
	 */
	bdi_remove_from_list(bdi);

	/*
	 * Finally, kill the kernel thread. We don't need to be RCU
	 * safe anymore, since the bdi is gone from visibility. Force
	 * unfreeze of the thread before calling kthread_stop(), otherwise
	 * it would never exet if it is currently stuck in the refrigerator.
	 */
	if (bdi->wb.task) {
		thaw_process(bdi->wb.task);
		kthread_stop(bdi->wb.task);
	}
}

/*
 * This bdi is going away now, make sure that no super_blocks point to it
 */
static void bdi_prune_sb(struct backing_dev_info *bdi)
{
	struct super_block *sb;

	spin_lock(&sb_lock);
	list_for_each_entry(sb, &super_blocks, s_list) {
		if (sb->s_bdi == bdi)
			sb->s_bdi = NULL;
	}
	spin_unlock(&sb_lock);
}

void bdi_unregister(struct backing_dev_info *bdi)
{
	if (bdi->dev) {
		trace_writeback_bdi_unregister(bdi);
		bdi_prune_sb(bdi);

		if (!bdi_cap_flush_forker(bdi))
			bdi_wb_shutdown(bdi);
		bdi_debug_unregister(bdi);
		device_unregister(bdi->dev);
		bdi->dev = NULL;
	}
}
EXPORT_SYMBOL(bdi_unregister);

int bdi_init(struct backing_dev_info *bdi)
{
	int i, err;

	bdi->dev = NULL;

	bdi->min_ratio = 0;
	bdi->max_ratio = 100;
	bdi->max_prop_frac = PROP_FRAC_BASE;
	spin_lock_init(&bdi->wb_lock);
	INIT_LIST_HEAD(&bdi->bdi_list);
	INIT_LIST_HEAD(&bdi->work_list);

	bdi_wb_init(&bdi->wb, bdi);

	for (i = 0; i < NR_BDI_STAT_ITEMS; i++) {
		err = percpu_counter_init(&bdi->bdi_stat[i], 0);
		if (err)
			goto err;
	}

	bdi->dirty_exceeded = 0;
	err = prop_local_init_percpu(&bdi->completions);

	if (err) {
err:
		while (i--)
			percpu_counter_destroy(&bdi->bdi_stat[i]);
	}

	return err;
}
EXPORT_SYMBOL(bdi_init);

void bdi_destroy(struct backing_dev_info *bdi)
{
	int i;

	/*
	 * Splice our entries to the default_backing_dev_info, if this
	 * bdi disappears
	 */
	if (bdi_has_dirty_io(bdi)) {
		struct bdi_writeback *dst = &default_backing_dev_info.wb;

		spin_lock(&inode_lock);
		list_splice(&bdi->wb.b_dirty, &dst->b_dirty);
		list_splice(&bdi->wb.b_io, &dst->b_io);
		list_splice(&bdi->wb.b_more_io, &dst->b_more_io);
		spin_unlock(&inode_lock);
	}

	bdi_unregister(bdi);

	for (i = 0; i < NR_BDI_STAT_ITEMS; i++)
		percpu_counter_destroy(&bdi->bdi_stat[i]);

	prop_local_destroy_percpu(&bdi->completions);
}
EXPORT_SYMBOL(bdi_destroy);

/*
 * For use from filesystems to quickly init and register a bdi associated
 * with dirty writeback
 */
int bdi_setup_and_register(struct backing_dev_info *bdi, char *name,
			   unsigned int cap)
{
	char tmp[32];
	int err;

	bdi->name = name;
	bdi->capabilities = cap;
	err = bdi_init(bdi);
	if (err)
		return err;

	sprintf(tmp, "%.28s%s", name, "-%d");
	err = bdi_register(bdi, NULL, tmp, atomic_long_inc_return(&bdi_seq));
	if (err) {
		bdi_destroy(bdi);
		return err;
	}

	return 0;
}
EXPORT_SYMBOL(bdi_setup_and_register);

static wait_queue_head_t congestion_wqh[2] = {
		__WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[0]),
		__WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[1])
	};

void clear_bdi_congested(struct backing_dev_info *bdi, int sync)
{
	enum bdi_state bit;
	wait_queue_head_t *wqh = &congestion_wqh[sync];

	bit = sync ? BDI_sync_congested : BDI_async_congested;
	clear_bit(bit, &bdi->state);
	smp_mb__after_clear_bit();
	if (waitqueue_active(wqh))
		wake_up(wqh);
}
EXPORT_SYMBOL(clear_bdi_congested);

void set_bdi_congested(struct backing_dev_info *bdi, int sync)
{
	enum bdi_state bit;

	bit = sync ? BDI_sync_congested : BDI_async_congested;
	set_bit(bit, &bdi->state);
}
EXPORT_SYMBOL(set_bdi_congested);

/**
 * congestion_wait - wait for a backing_dev to become uncongested
 * @sync: SYNC or ASYNC IO
 * @timeout: timeout in jiffies
 *
 * Waits for up to @timeout jiffies for a backing_dev (any backing_dev) to exit
 * write congestion.  If no backing_devs are congested then just wait for the
 * next write to be completed.
 */
long congestion_wait(int sync, long timeout)
{
	long ret;
	DEFINE_WAIT(wait);
	wait_queue_head_t *wqh = &congestion_wqh[sync];

	prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
	ret = io_schedule_timeout(timeout);
	finish_wait(wqh, &wait);
	return ret;
}
EXPORT_SYMBOL(congestion_wait);
Commit	Line	Data
3fcfab16 AM	1
	2	#include <linux/wait.h>
	3	#include <linux/backing-dev.h>
03ba3782 JA	4	#include <linux/kthread.h>
03ba3782 JA	5	#include <linux/freezer.h>
3fcfab16	6	#include <linux/fs.h>
26160158	7	#include <linux/pagemap.h>
03ba3782	8	#include <linux/mm.h>
3fcfab16 AM	9	#include <linux/sched.h>
3fcfab16 AM	10	#include <linux/module.h>
cf0ca9fe PZ	11	#include <linux/writeback.h>
cf0ca9fe PZ	12	#include <linux/device.h>
455b2864	13	#include <trace/events/writeback.h>
cf0ca9fe	14
c3c53206 JA	15	static atomic_long_t bdi_seq = ATOMIC_LONG_INIT(0);
c3c53206 JA	16
26160158 JA	17	void default_unplug_io_fn(struct backing_dev_info bdi, struct page page)
	18	{
	19	}
	20	EXPORT_SYMBOL(default_unplug_io_fn);
	21
	22	struct backing_dev_info default_backing_dev_info = {
d993831f	23	.name = "default",
26160158 JA	24	.ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_CACHE_SIZE,
	25	.state = 0,
	26	.capabilities = BDI_CAP_MAP_COPY,
	27	.unplug_io_fn = default_unplug_io_fn,
	28	};
	29	EXPORT_SYMBOL_GPL(default_backing_dev_info);
cf0ca9fe	30
5129a469 JE	31	struct backing_dev_info noop_backing_dev_info = {
	32	.name = "noop",
	33	};
	34	EXPORT_SYMBOL_GPL(noop_backing_dev_info);
	35
cf0ca9fe	36	static struct class *bdi_class;
cfc4ba53 JA	37
	38	/*
	39	* bdi_lock protects updates to bdi_list and bdi_pending_list, as well as
	40	* reader side protection for bdi_pending_list. bdi_list has RCU reader side
	41	* locking.
	42	*/
03ba3782	43	DEFINE_SPINLOCK(bdi_lock);
66f3b8e2	44	LIST_HEAD(bdi_list);
03ba3782 JA	45	LIST_HEAD(bdi_pending_list);
	46
	47	static struct task_struct *sync_supers_tsk;
	48	static struct timer_list sync_supers_timer;
	49
	50	static int bdi_sync_supers(void *);
	51	static void sync_supers_timer_fn(unsigned long);
03ba3782	52
76f1418b MS	53	#ifdef CONFIG_DEBUG_FS
	54	#include <linux/debugfs.h>
	55	#include <linux/seq_file.h>
	56
	57	static struct dentry *bdi_debug_root;
	58
	59	static void bdi_debug_init(void)
	60	{
	61	bdi_debug_root = debugfs_create_dir("bdi", NULL);
	62	}
	63
	64	static int bdi_debug_stats_show(struct seq_file m, void v)
	65	{
	66	struct backing_dev_info *bdi = m->private;
c1955ce3	67	struct bdi_writeback *wb = &bdi->wb;
364aeb28 DR	68	unsigned long background_thresh;
	69	unsigned long dirty_thresh;
	70	unsigned long bdi_thresh;
f09b00d3 JA	71	unsigned long nr_dirty, nr_io, nr_more_io, nr_wb;
	72	struct inode *inode;
	73
f09b00d3 JA	74	nr_wb = nr_dirty = nr_io = nr_more_io = 0;
f09b00d3 JA	75	spin_lock(&inode_lock);
c1955ce3 CH	76	list_for_each_entry(inode, &wb->b_dirty, i_list)
	77	nr_dirty++;
	78	list_for_each_entry(inode, &wb->b_io, i_list)
	79	nr_io++;
	80	list_for_each_entry(inode, &wb->b_more_io, i_list)
	81	nr_more_io++;
f09b00d3	82	spin_unlock(&inode_lock);
76f1418b MS	83
	84	get_dirty_limits(&background_thresh, &dirty_thresh, &bdi_thresh, bdi);
	85
	86	#define K(x) ((x) << (PAGE_SHIFT - 10))
	87	seq_printf(m,
	88	"BdiWriteback: %8lu kB\n"
	89	"BdiReclaimable: %8lu kB\n"
	90	"BdiDirtyThresh: %8lu kB\n"
	91	"DirtyThresh: %8lu kB\n"
f09b00d3	92	"BackgroundThresh: %8lu kB\n"
f09b00d3 JA	93	"b_dirty: %8lu\n"
	94	"b_io: %8lu\n"
	95	"b_more_io: %8lu\n"
	96	"bdi_list: %8u\n"
c1955ce3	97	"state: %8lx\n",
76f1418b MS	98	(unsigned long) K(bdi_stat(bdi, BDI_WRITEBACK)),
76f1418b MS	99	(unsigned long) K(bdi_stat(bdi, BDI_RECLAIMABLE)),
f09b00d3	100	K(bdi_thresh), K(dirty_thresh),
c1955ce3 CH	101	K(background_thresh), nr_dirty, nr_io, nr_more_io,
c1955ce3 CH	102	!list_empty(&bdi->bdi_list), bdi->state);
76f1418b MS	103	#undef K
	104
	105	return 0;
	106	}
	107
	108	static int bdi_debug_stats_open(struct inode inode, struct file file)
	109	{
	110	return single_open(file, bdi_debug_stats_show, inode->i_private);
	111	}
	112
	113	static const struct file_operations bdi_debug_stats_fops = {
	114	.open = bdi_debug_stats_open,
	115	.read = seq_read,
	116	.llseek = seq_lseek,
	117	.release = single_release,
	118	};
	119
	120	static void bdi_debug_register(struct backing_dev_info bdi, const char name)
	121	{
	122	bdi->debug_dir = debugfs_create_dir(name, bdi_debug_root);
	123	bdi->debug_stats = debugfs_create_file("stats", 0444, bdi->debug_dir,
	124	bdi, &bdi_debug_stats_fops);
	125	}
	126
	127	static void bdi_debug_unregister(struct backing_dev_info *bdi)
	128	{
	129	debugfs_remove(bdi->debug_stats);
	130	debugfs_remove(bdi->debug_dir);
	131	}
	132	#else
	133	static inline void bdi_debug_init(void)
	134	{
	135	}
	136	static inline void bdi_debug_register(struct backing_dev_info *bdi,
	137	const char *name)
	138	{
	139	}
	140	static inline void bdi_debug_unregister(struct backing_dev_info *bdi)
	141	{
	142	}
	143	#endif
	144
cf0ca9fe PZ	145	static ssize_t read_ahead_kb_store(struct device *dev,
	146	struct device_attribute *attr,
	147	const char *buf, size_t count)
	148	{
	149	struct backing_dev_info *bdi = dev_get_drvdata(dev);
	150	char *end;
	151	unsigned long read_ahead_kb;
	152	ssize_t ret = -EINVAL;
	153
	154	read_ahead_kb = simple_strtoul(buf, &end, 10);
	155	if (*buf && (end[0] == '\0' \|\| (end[0] == '\n' && end[1] == '\0'))) {
	156	bdi->ra_pages = read_ahead_kb >> (PAGE_SHIFT - 10);
	157	ret = count;
	158	}
	159	return ret;
	160	}
	161
	162	#define K(pages) ((pages) << (PAGE_SHIFT - 10))
	163
	164	#define BDI_SHOW(name, expr) \
	165	static ssize_t name##_show(struct device *dev, \
	166	struct device_attribute attr, char page) \
	167	{ \
	168	struct backing_dev_info *bdi = dev_get_drvdata(dev); \
	169	\
	170	return snprintf(page, PAGE_SIZE-1, "%lld\n", (long long)expr); \
	171	}
	172
	173	BDI_SHOW(read_ahead_kb, K(bdi->ra_pages))
	174
189d3c4a PZ	175	static ssize_t min_ratio_store(struct device *dev,
	176	struct device_attribute attr, const char buf, size_t count)
	177	{
	178	struct backing_dev_info *bdi = dev_get_drvdata(dev);
	179	char *end;
	180	unsigned int ratio;
	181	ssize_t ret = -EINVAL;
	182
	183	ratio = simple_strtoul(buf, &end, 10);
	184	if (*buf && (end[0] == '\0' \|\| (end[0] == '\n' && end[1] == '\0'))) {
	185	ret = bdi_set_min_ratio(bdi, ratio);
	186	if (!ret)
	187	ret = count;
	188	}
	189	return ret;
	190	}
	191	BDI_SHOW(min_ratio, bdi->min_ratio)
	192
a42dde04 PZ	193	static ssize_t max_ratio_store(struct device *dev,
	194	struct device_attribute attr, const char buf, size_t count)
	195	{
	196	struct backing_dev_info *bdi = dev_get_drvdata(dev);
	197	char *end;
	198	unsigned int ratio;
	199	ssize_t ret = -EINVAL;
	200
	201	ratio = simple_strtoul(buf, &end, 10);
	202	if (*buf && (end[0] == '\0' \|\| (end[0] == '\n' && end[1] == '\0'))) {
	203	ret = bdi_set_max_ratio(bdi, ratio);
	204	if (!ret)
	205	ret = count;
	206	}
	207	return ret;
	208	}
	209	BDI_SHOW(max_ratio, bdi->max_ratio)
	210
cf0ca9fe PZ	211	#define __ATTR_RW(attr) __ATTR(attr, 0644, attr##_show, attr##_store)
	212
	213	static struct device_attribute bdi_dev_attrs[] = {
	214	__ATTR_RW(read_ahead_kb),
189d3c4a	215	__ATTR_RW(min_ratio),
a42dde04	216	__ATTR_RW(max_ratio),
cf0ca9fe PZ	217	__ATTR_NULL,
	218	};
	219
	220	static __init int bdi_class_init(void)
	221	{
	222	bdi_class = class_create(THIS_MODULE, "bdi");
14421453 AB	223	if (IS_ERR(bdi_class))
	224	return PTR_ERR(bdi_class);
	225
cf0ca9fe	226	bdi_class->dev_attrs = bdi_dev_attrs;
76f1418b	227	bdi_debug_init();
cf0ca9fe PZ	228	return 0;
cf0ca9fe PZ	229	}
76f1418b	230	postcore_initcall(bdi_class_init);
cf0ca9fe	231
26160158 JA	232	static int __init default_bdi_init(void)
	233	{
	234	int err;
	235
03ba3782 JA	236	sync_supers_tsk = kthread_run(bdi_sync_supers, NULL, "sync_supers");
	237	BUG_ON(IS_ERR(sync_supers_tsk));
	238
	239	init_timer(&sync_supers_timer);
	240	setup_timer(&sync_supers_timer, sync_supers_timer_fn, 0);
6423104b	241	bdi_arm_supers_timer();
03ba3782	242
26160158 JA	243	err = bdi_init(&default_backing_dev_info);
	244	if (!err)
	245	bdi_register(&default_backing_dev_info, NULL, "default");
	246
	247	return err;
	248	}
	249	subsys_initcall(default_bdi_init);
	250
03ba3782 JA	251	static void bdi_wb_init(struct bdi_writeback wb, struct backing_dev_info bdi)
	252	{
	253	memset(wb, 0, sizeof(*wb));
	254
	255	wb->bdi = bdi;
	256	wb->last_old_flush = jiffies;
	257	INIT_LIST_HEAD(&wb->b_dirty);
	258	INIT_LIST_HEAD(&wb->b_io);
	259	INIT_LIST_HEAD(&wb->b_more_io);
	260	}
	261
03ba3782 JA	262	int bdi_has_dirty_io(struct backing_dev_info *bdi)
	263	{
	264	return wb_has_dirty_io(&bdi->wb);
	265	}
	266
	267	static void bdi_flush_io(struct backing_dev_info *bdi)
	268	{
	269	struct writeback_control wbc = {
03ba3782 JA	270	.sync_mode = WB_SYNC_NONE,
	271	.older_than_this = NULL,
	272	.range_cyclic = 1,
	273	.nr_to_write = 1024,
	274	};
	275
9c3a8ee8	276	writeback_inodes_wb(&bdi->wb, &wbc);
03ba3782 JA	277	}
	278
	279	/*
6f904ff0	280	* kupdated() used to do this. We cannot do it from the bdi_forker_thread()
03ba3782 JA	281	* or we risk deadlocking on ->s_umount. The longer term solution would be
03ba3782 JA	282	* to implement sync_supers_bdi() or similar and simply do it from the
6f904ff0	283	* bdi writeback thread individually.
03ba3782 JA	284	*/
	285	static int bdi_sync_supers(void *unused)
	286	{
	287	set_user_nice(current, 0);
	288
	289	while (!kthread_should_stop()) {
	290	set_current_state(TASK_INTERRUPTIBLE);
	291	schedule();
	292
	293	/*
	294	* Do this periodically, like kupdated() did before.
	295	*/
	296	sync_supers();
	297	}
	298
	299	return 0;
	300	}
	301
6423104b	302	void bdi_arm_supers_timer(void)
03ba3782 JA	303	{
	304	unsigned long next;
	305
6423104b JA	306	if (!dirty_writeback_interval)
	307	return;
	308
03ba3782 JA	309	next = msecs_to_jiffies(dirty_writeback_interval * 10) + jiffies;
	310	mod_timer(&sync_supers_timer, round_jiffies_up(next));
	311	}
	312
	313	static void sync_supers_timer_fn(unsigned long unused)
	314	{
	315	wake_up_process(sync_supers_tsk);
6423104b	316	bdi_arm_supers_timer();
03ba3782 JA	317	}
03ba3782 JA	318
6f904ff0	319	static int bdi_forker_thread(void *ptr)
03ba3782 JA	320	{
	321	struct bdi_writeback *me = ptr;
	322
c1955ce3 CH	323	current->flags \|= PF_FLUSHER \| PF_SWAPWRITE;
	324	set_freezable();
	325
	326	/*
	327	* Our parent may run at a different priority, just set us to normal
	328	*/
	329	set_user_nice(current, 0);
03ba3782 JA	330
03ba3782 JA	331	for (;;) {
080dcec4	332	bool fork = false;
94eac5e6	333	struct task_struct *task;
78c40cb6	334	struct backing_dev_info *bdi;
03ba3782 JA	335
	336	/*
	337	* Temporary measure, we want to make sure we don't see
	338	* dirty data on the default backing_dev_info
	339	*/
	340	if (wb_has_dirty_io(me) \|\| !list_empty(&me->bdi->work_list))
	341	wb_do_writeback(me, 0);
	342
cfc4ba53	343	spin_lock_bh(&bdi_lock);
c5f7ad23	344	set_current_state(TASK_INTERRUPTIBLE);
03ba3782 JA	345
	346	/*
	347	* Check if any existing bdi's have dirty data without
	348	* a thread registered. If so, set that up.
	349	*/
78c40cb6	350	list_for_each_entry(bdi, &bdi_list, bdi_list) {
080dcec4 AB	351	if (!bdi_cap_writeback_dirty(bdi))
080dcec4 AB	352	continue;
03ba3782 JA	353	if (bdi->wb.task)
	354	continue;
	355	if (list_empty(&bdi->work_list) &&
	356	!bdi_has_dirty_io(bdi))
	357	continue;
	358
080dcec4 AB	359	WARN(!test_bit(BDI_registered, &bdi->state),
	360	"bdi %p/%s is not registered!\n", bdi, bdi->name);
	361
080dcec4	362	fork = true;
78c40cb6 AB	363
	364	/*
	365	* Set the pending bit - if someone will try to
	366	* unregister this bdi - it'll wait on this bit.
	367	*/
	368	set_bit(BDI_pending, &bdi->state);
080dcec4	369	break;
03ba3782	370	}
080dcec4	371	spin_unlock_bh(&bdi_lock);
03ba3782	372
c4ec7908 AB	373	/* Keep working if default bdi still has things to do */
	374	if (!list_empty(&me->bdi->work_list))
	375	__set_current_state(TASK_RUNNING);
	376
080dcec4	377	if (!fork) {
03ba3782 JA	378	unsigned long wait;
03ba3782 JA	379
03ba3782	380	wait = msecs_to_jiffies(dirty_writeback_interval * 10);
6423104b JA	381	if (wait)
	382	schedule_timeout(wait);
	383	else
	384	schedule();
03ba3782 JA	385	try_to_freeze();
	386	continue;
	387	}
	388
	389	__set_current_state(TASK_RUNNING);
	390
94eac5e6 AB	391	task = kthread_run(bdi_writeback_thread, &bdi->wb, "flush-%s",
	392	dev_name(bdi->dev));
	393	if (IS_ERR(task)) {
03ba3782	394	/*
78c40cb6 AB	395	* If thread creation fails, force writeout of the bdi
78c40cb6 AB	396	* from the thread.
03ba3782	397	*/
03ba3782	398	bdi_flush_io(bdi);
94eac5e6 AB	399	} else
94eac5e6 AB	400	bdi->wb.task = task;
03ba3782 JA	401	}
	402
	403	return 0;
	404	}
	405
cfc4ba53 JA	406	/*
	407	* Remove bdi from bdi_list, and ensure that it is no longer visible
	408	*/
	409	static void bdi_remove_from_list(struct backing_dev_info *bdi)
	410	{
	411	spin_lock_bh(&bdi_lock);
	412	list_del_rcu(&bdi->bdi_list);
	413	spin_unlock_bh(&bdi_lock);
	414
	415	synchronize_rcu();
	416	}
	417
cf0ca9fe PZ	418	int bdi_register(struct backing_dev_info bdi, struct device parent,
	419	const char *fmt, ...)
	420	{
cf0ca9fe PZ	421	va_list args;
	422	int ret = 0;
	423	struct device *dev;
	424
69fc208b	425	if (bdi->dev) /* The driver needs to use separate queues per device */
f1d0b063 KS	426	goto exit;
f1d0b063 KS	427
cf0ca9fe	428	va_start(args, fmt);
19051c50	429	dev = device_create_vargs(bdi_class, parent, MKDEV(0, 0), bdi, fmt, args);
cf0ca9fe	430	va_end(args);
cf0ca9fe PZ	431	if (IS_ERR(dev)) {
	432	ret = PTR_ERR(dev);
	433	goto exit;
	434	}
	435
cfc4ba53 JA	436	spin_lock_bh(&bdi_lock);
	437	list_add_tail_rcu(&bdi->bdi_list, &bdi_list);
	438	spin_unlock_bh(&bdi_lock);
66f3b8e2	439
cf0ca9fe	440	bdi->dev = dev;
cf0ca9fe	441
03ba3782 JA	442	/*
	443	* Just start the forker thread for our default backing_dev_info,
	444	* and add other bdi's to the list. They will get a thread created
	445	* on-demand when they need it.
	446	*/
	447	if (bdi_cap_flush_forker(bdi)) {
	448	struct bdi_writeback *wb = &bdi->wb;
	449
6f904ff0	450	wb->task = kthread_run(bdi_forker_thread, wb, "bdi-%s",
03ba3782 JA	451	dev_name(dev));
	452	if (IS_ERR(wb->task)) {
	453	wb->task = NULL;
	454	ret = -ENOMEM;
	455
cfc4ba53	456	bdi_remove_from_list(bdi);
03ba3782 JA	457	goto exit;
	458	}
	459	}
	460
	461	bdi_debug_register(bdi, dev_name(dev));
500b067c	462	set_bit(BDI_registered, &bdi->state);
455b2864	463	trace_writeback_bdi_register(bdi);
cf0ca9fe	464	exit:
cf0ca9fe PZ	465	return ret;
	466	}
	467	EXPORT_SYMBOL(bdi_register);
	468
	469	int bdi_register_dev(struct backing_dev_info *bdi, dev_t dev)
	470	{
	471	return bdi_register(bdi, NULL, "%u:%u", MAJOR(dev), MINOR(dev));
	472	}
	473	EXPORT_SYMBOL(bdi_register_dev);
	474
03ba3782 JA	475	/*
	476	* Remove bdi from the global list and shutdown any threads we have running
	477	*/
	478	static void bdi_wb_shutdown(struct backing_dev_info *bdi)
66f3b8e2	479	{
03ba3782 JA	480	if (!bdi_cap_writeback_dirty(bdi))
	481	return;
	482
	483	/*
	484	* If setup is pending, wait for that to complete first
	485	*/
	486	wait_on_bit(&bdi->state, BDI_pending, bdi_sched_wait,
	487	TASK_UNINTERRUPTIBLE);
	488
	489	/*
	490	* Make sure nobody finds us on the bdi_list anymore
	491	*/
cfc4ba53	492	bdi_remove_from_list(bdi);
03ba3782 JA	493
03ba3782 JA	494	/*
c1955ce3	495	* Finally, kill the kernel thread. We don't need to be RCU
c62b17a5 RD	496	* safe anymore, since the bdi is gone from visibility. Force
	497	* unfreeze of the thread before calling kthread_stop(), otherwise
	498	* it would never exet if it is currently stuck in the refrigerator.
03ba3782	499	*/
c1955ce3 CH	500	if (bdi->wb.task) {
	501	thaw_process(bdi->wb.task);
	502	kthread_stop(bdi->wb.task);
c62b17a5	503	}
66f3b8e2 JA	504	}
66f3b8e2 JA	505
592b09a4 JA	506	/*
	507	* This bdi is going away now, make sure that no super_blocks point to it
	508	*/
	509	static void bdi_prune_sb(struct backing_dev_info *bdi)
	510	{
	511	struct super_block *sb;
	512
	513	spin_lock(&sb_lock);
	514	list_for_each_entry(sb, &super_blocks, s_list) {
	515	if (sb->s_bdi == bdi)
	516	sb->s_bdi = NULL;
	517	}
	518	spin_unlock(&sb_lock);
	519	}
	520
cf0ca9fe PZ	521	void bdi_unregister(struct backing_dev_info *bdi)
	522	{
	523	if (bdi->dev) {
455b2864	524	trace_writeback_bdi_unregister(bdi);
8c4db335 JA	525	bdi_prune_sb(bdi);
8c4db335 JA	526
03ba3782 JA	527	if (!bdi_cap_flush_forker(bdi))
03ba3782 JA	528	bdi_wb_shutdown(bdi);
76f1418b	529	bdi_debug_unregister(bdi);
cf0ca9fe PZ	530	device_unregister(bdi->dev);
	531	bdi->dev = NULL;
	532	}
	533	}
	534	EXPORT_SYMBOL(bdi_unregister);
3fcfab16	535
b2e8fb6e PZ	536	int bdi_init(struct backing_dev_info *bdi)
b2e8fb6e PZ	537	{
03ba3782	538	int i, err;
b2e8fb6e	539
cf0ca9fe PZ	540	bdi->dev = NULL;
cf0ca9fe PZ	541
189d3c4a	542	bdi->min_ratio = 0;
a42dde04 PZ	543	bdi->max_ratio = 100;
a42dde04 PZ	544	bdi->max_prop_frac = PROP_FRAC_BASE;
03ba3782	545	spin_lock_init(&bdi->wb_lock);
66f3b8e2	546	INIT_LIST_HEAD(&bdi->bdi_list);
03ba3782 JA	547	INIT_LIST_HEAD(&bdi->work_list);
	548
	549	bdi_wb_init(&bdi->wb, bdi);
	550
b2e8fb6e	551	for (i = 0; i < NR_BDI_STAT_ITEMS; i++) {
ea319518	552	err = percpu_counter_init(&bdi->bdi_stat[i], 0);
04fbfdc1 PZ	553	if (err)
	554	goto err;
	555	}
	556
	557	bdi->dirty_exceeded = 0;
	558	err = prop_local_init_percpu(&bdi->completions);
	559
	560	if (err) {
	561	err:
4b01a0b1	562	while (i--)
04fbfdc1	563	percpu_counter_destroy(&bdi->bdi_stat[i]);
b2e8fb6e PZ	564	}
	565
	566	return err;
	567	}
	568	EXPORT_SYMBOL(bdi_init);
	569
	570	void bdi_destroy(struct backing_dev_info *bdi)
	571	{
	572	int i;
	573
ce5f8e77 JA	574	/*
	575	* Splice our entries to the default_backing_dev_info, if this
	576	* bdi disappears
	577	*/
	578	if (bdi_has_dirty_io(bdi)) {
	579	struct bdi_writeback *dst = &default_backing_dev_info.wb;
	580
	581	spin_lock(&inode_lock);
	582	list_splice(&bdi->wb.b_dirty, &dst->b_dirty);
	583	list_splice(&bdi->wb.b_io, &dst->b_io);
	584	list_splice(&bdi->wb.b_more_io, &dst->b_more_io);
	585	spin_unlock(&inode_lock);
	586	}
66f3b8e2	587
cf0ca9fe PZ	588	bdi_unregister(bdi);
cf0ca9fe PZ	589
b2e8fb6e PZ	590	for (i = 0; i < NR_BDI_STAT_ITEMS; i++)
b2e8fb6e PZ	591	percpu_counter_destroy(&bdi->bdi_stat[i]);
04fbfdc1 PZ	592
04fbfdc1 PZ	593	prop_local_destroy_percpu(&bdi->completions);
b2e8fb6e PZ	594	}
	595	EXPORT_SYMBOL(bdi_destroy);
	596
c3c53206 JA	597	/*
	598	* For use from filesystems to quickly init and register a bdi associated
	599	* with dirty writeback
	600	*/
	601	int bdi_setup_and_register(struct backing_dev_info bdi, char name,
	602	unsigned int cap)
	603	{
	604	char tmp[32];
	605	int err;
	606
	607	bdi->name = name;
	608	bdi->capabilities = cap;
	609	err = bdi_init(bdi);
	610	if (err)
	611	return err;
	612
	613	sprintf(tmp, "%.28s%s", name, "-%d");
	614	err = bdi_register(bdi, NULL, tmp, atomic_long_inc_return(&bdi_seq));
	615	if (err) {
	616	bdi_destroy(bdi);
	617	return err;
	618	}
	619
	620	return 0;
	621	}
	622	EXPORT_SYMBOL(bdi_setup_and_register);
	623
3fcfab16 AM	624	static wait_queue_head_t congestion_wqh[2] = {
	625	__WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[0]),
	626	__WAIT_QUEUE_HEAD_INITIALIZER(congestion_wqh[1])
	627	};
	628
1faa16d2	629	void clear_bdi_congested(struct backing_dev_info *bdi, int sync)
3fcfab16 AM	630	{
3fcfab16 AM	631	enum bdi_state bit;
1faa16d2	632	wait_queue_head_t *wqh = &congestion_wqh[sync];
3fcfab16	633
1faa16d2	634	bit = sync ? BDI_sync_congested : BDI_async_congested;
3fcfab16 AM	635	clear_bit(bit, &bdi->state);
	636	smp_mb__after_clear_bit();
	637	if (waitqueue_active(wqh))
	638	wake_up(wqh);
	639	}
	640	EXPORT_SYMBOL(clear_bdi_congested);
	641
1faa16d2	642	void set_bdi_congested(struct backing_dev_info *bdi, int sync)
3fcfab16 AM	643	{
	644	enum bdi_state bit;
	645
1faa16d2	646	bit = sync ? BDI_sync_congested : BDI_async_congested;
3fcfab16 AM	647	set_bit(bit, &bdi->state);
	648	}
	649	EXPORT_SYMBOL(set_bdi_congested);
	650
	651	/**
	652	* congestion_wait - wait for a backing_dev to become uncongested
8aa7e847	653	* @sync: SYNC or ASYNC IO
3fcfab16 AM	654	* @timeout: timeout in jiffies
	655	*
	656	* Waits for up to @timeout jiffies for a backing_dev (any backing_dev) to exit
	657	* write congestion. If no backing_devs are congested then just wait for the
	658	* next write to be completed.
	659	*/
8aa7e847	660	long congestion_wait(int sync, long timeout)
3fcfab16 AM	661	{
	662	long ret;
	663	DEFINE_WAIT(wait);
8aa7e847	664	wait_queue_head_t *wqh = &congestion_wqh[sync];
3fcfab16 AM	665
	666	prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
	667	ret = io_schedule_timeout(timeout);
	668	finish_wait(wqh, &wait);
	669	return ret;
	670	}
	671	EXPORT_SYMBOL(congestion_wait);
04fbfdc1	672