buffer.c

 * ll_rw_block: low-level access to block devices (DEPRECATED)
 * @rw: whether to %READ or %WRITE or %SWRITE or maybe %READA (readahead)
 * @nr: number of &struct buffer_heads in the array
 * @bhs: array of pointers to &struct buffer_head
 *
 * ll_rw_block() takes an array of pointers to &struct buffer_heads, and
 * requests an I/O operation on them, either a %READ or a %WRITE.  The third
 * %SWRITE is like %WRITE only we make sure that the *current* data in buffers
 * are sent to disk. The fourth %READA option is described in the documentation
 * for generic_make_request() which ll_rw_block() calls.
 *
 * This function drops any buffer that it cannot get a lock on (with the
 * BH_Lock state bit) unless SWRITE is required, any buffer that appears to be
 * clean when doing a write request, and any buffer that appears to be
 * up-to-date when doing read request.  Further it marks as clean buffers that
 * are processed for writing (the buffer cache won't assume that they are
 * actually clean until the buffer gets unlocked).
 *
 * ll_rw_block sets b_end_io to simple completion handler that marks
 * the buffer up-to-date (if approriate), unlocks the buffer and wakes
 * any waiters. 
 *
 * All of the buffers must be for the same device, and must also be a
 * multiple of the current approved size for the device.
 */
void ll_rw_block(int rw, int nr, struct buffer_head *bhs[])
{
	int i;

	for (i = 0; i < nr; i++) {
		struct buffer_head *bh = bhs[i];

		if (rw == SWRITE || rw == SWRITE_SYNC || rw == SWRITE_SYNC_PLUG)
			lock_buffer(bh);
		else if (!trylock_buffer(bh))
			continue;

		if (rw == WRITE || rw == SWRITE || rw == SWRITE_SYNC ||
		    rw == SWRITE_SYNC_PLUG) {
			if (test_clear_buffer_dirty(bh)) {
				bh->b_end_io = end_buffer_write_sync;
				get_bh(bh);
				if (rw == SWRITE_SYNC)
					submit_bh(WRITE_SYNC, bh);
				else
					submit_bh(WRITE, bh);
				continue;
			}
		} else {
			if (!buffer_uptodate(bh)) {
				bh->b_end_io = end_buffer_read_sync;
				get_bh(bh);
				submit_bh(rw, bh);
				continue;
			}
		}
		unlock_buffer(bh);
	}
}
EXPORT_SYMBOL(ll_rw_block);

/*
 * For a data-integrity writeout, we need to wait upon any in-progress I/O
 * and then start new I/O and then wait upon it.  The caller must have a ref on
 * the buffer_head.
 */
int sync_dirty_buffer(struct buffer_head *bh)
{
	int ret = 0;

	WARN_ON(atomic_read(&bh->b_count) < 1);
	lock_buffer(bh);
	if (test_clear_buffer_dirty(bh)) {
		get_bh(bh);
		bh->b_end_io = end_buffer_write_sync;
		ret = submit_bh(WRITE_SYNC, bh);
		wait_on_buffer(bh);
		if (buffer_eopnotsupp(bh)) {
			clear_buffer_eopnotsupp(bh);
			ret = -EOPNOTSUPP;
		}
		if (!ret && !buffer_uptodate(bh))
			ret = -EIO;
	} else {
		unlock_buffer(bh);
	}
	return ret;
}
EXPORT_SYMBOL(sync_dirty_buffer);

/*
 * try_to_free_buffers() checks if all the buffers on this particular page
 * are unused, and releases them if so.
 *
 * Exclusion against try_to_free_buffers may be obtained by either
 * locking the page or by holding its mapping's private_lock.
 *
 * If the page is dirty but all the buffers are clean then we need to
 * be sure to mark the page clean as well.  This is because the page
 * may be against a block device, and a later reattachment of buffers
 * to a dirty page will set *all* buffers dirty.  Which would corrupt
 * filesystem data on the same device.
 *
 * The same applies to regular filesystem pages: if all the buffers are
 * clean then we set the page clean and proceed.  To do that, we require
 * total exclusion from __set_page_dirty_buffers().  That is obtained with
 * private_lock.
 *
 * try_to_free_buffers() is non-blocking.
 */
static inline int buffer_busy(struct buffer_head *bh)
{
	return atomic_read(&bh->b_count) |
		(bh->b_state & ((1 << BH_Dirty) | (1 << BH_Lock)));
}

static int
drop_buffers(struct page *page, struct buffer_head **buffers_to_free)
{
	struct buffer_head *head = page_buffers(page);
	struct buffer_head *bh;

	bh = head;
	do {
		if (buffer_write_io_error(bh) && page->mapping)
			set_bit(AS_EIO, &page->mapping->flags);
		if (buffer_busy(bh))
			goto failed;
		bh = bh->b_this_page;
	} while (bh != head);

	do {
		struct buffer_head *next = bh->b_this_page;

		if (bh->b_assoc_map)
			__remove_assoc_queue(bh);
		bh = next;
	} while (bh != head);
	*buffers_to_free = head;
	__clear_page_buffers(page);
	return 1;
failed:
	return 0;
}

int try_to_free_buffers(struct page *page)
{
	struct address_space * const mapping = page->mapping;
	struct buffer_head *buffers_to_free = NULL;
	int ret = 0;

	BUG_ON(!PageLocked(page));
	if (PageWriteback(page))
		return 0;

	if (mapping == NULL) {		/* can this still happen? */
		ret = drop_buffers(page, &buffers_to_free);
		goto out;
	}

	spin_lock(&mapping->private_lock);
	ret = drop_buffers(page, &buffers_to_free);

	/*
	 * If the filesystem writes its buffers by hand (eg ext3)
	 * then we can have clean buffers against a dirty page.  We
	 * clean the page here; otherwise the VM will never notice
	 * that the filesystem did any IO at all.
	 *
	 * Also, during truncate, discard_buffer will have marked all
	 * the page's buffers clean.  We discover that here and clean
	 * the page also.
	 *
	 * private_lock must be held over this entire operation in order
	 * to synchronise against __set_page_dirty_buffers and prevent the
	 * dirty bit from being lost.
	 */
	if (ret)
		cancel_dirty_page(page, PAGE_CACHE_SIZE);
	spin_unlock(&mapping->private_lock);
out:
	if (buffers_to_free) {
		struct buffer_head *bh = buffers_to_free;

		do {
			struct buffer_head *next = bh->b_this_page;
			free_buffer_head(bh);
			bh = next;
		} while (bh != buffers_to_free);
	}
	return ret;
}
EXPORT_SYMBOL(try_to_free_buffers);

void block_sync_page(struct page *page)
{
	struct address_space *mapping;

	smp_mb();
	mapping = page_mapping(page);
	if (mapping)
		blk_run_backing_dev(mapping->backing_dev_info, page);
}
EXPORT_SYMBOL(block_sync_page);

/*
 * There are no bdflush tunables left.  But distributions are
 * still running obsolete flush daemons, so we terminate them here.
 *
 * Use of bdflush() is deprecated and will be removed in a future kernel.
 * The `flush-X' kernel threads fully replace bdflush daemons and this call.
 */
SYSCALL_DEFINE2(bdflush, int, func, long, data)
{
	static int msg_count;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	if (msg_count < 5) {
		msg_count++;
		printk(KERN_INFO
			"warning: process `%s' used the obsolete bdflush"
			" system call\n", current->comm);
		printk(KERN_INFO "Fix your initscripts?\n");
	}

	if (func == 1)
		do_exit(0);
	return 0;
}

/*
 * Buffer-head allocation
 */
static struct kmem_cache *bh_cachep;

/*
 * Once the number of bh's in the machine exceeds this level, we start
 * stripping them in writeback.
 */
static int max_buffer_heads;

int buffer_heads_over_limit;

struct bh_accounting {
	int nr;			/* Number of live bh's */
	int ratelimit;		/* Limit cacheline bouncing */
};

static DEFINE_PER_CPU(struct bh_accounting, bh_accounting) = {0, 0};

static void recalc_bh_state(void)
{
	int i;
	int tot = 0;

	if (__get_cpu_var(bh_accounting).ratelimit++ < 4096)
		return;
	__get_cpu_var(bh_accounting).ratelimit = 0;
	for_each_online_cpu(i)
		tot += per_cpu(bh_accounting, i).nr;
	buffer_heads_over_limit = (tot > max_buffer_heads);
}
	
struct buffer_head *alloc_buffer_head(gfp_t gfp_flags)
{
	struct buffer_head *ret = kmem_cache_zalloc(bh_cachep, gfp_flags);
	if (ret) {
		INIT_LIST_HEAD(&ret->b_assoc_buffers);
		get_cpu_var(bh_accounting).nr++;
		recalc_bh_state();
		put_cpu_var(bh_accounting);
	}
	return ret;
}
EXPORT_SYMBOL(alloc_buffer_head);

void free_buffer_head(struct buffer_head *bh)
{
	BUG_ON(!list_empty(&bh->b_assoc_buffers));
	kmem_cache_free(bh_cachep, bh);
	get_cpu_var(bh_accounting).nr--;
	recalc_bh_state();
	put_cpu_var(bh_accounting);
}
EXPORT_SYMBOL(free_buffer_head);

static void buffer_exit_cpu(int cpu)
{
	int i;
	struct bh_lru *b = &per_cpu(bh_lrus, cpu);

	for (i = 0; i < BH_LRU_SIZE; i++) {
		brelse(b->bhs[i]);
		b->bhs[i] = NULL;
	}
	get_cpu_var(bh_accounting).nr += per_cpu(bh_accounting, cpu).nr;
	per_cpu(bh_accounting, cpu).nr = 0;
	put_cpu_var(bh_accounting);
}

static int buffer_cpu_notify(struct notifier_block *self,
			      unsigned long action, void *hcpu)
{
	if (action == CPU_DEAD || action == CPU_DEAD_FROZEN)
		buffer_exit_cpu((unsigned long)hcpu);
	return NOTIFY_OK;
}

/**
 * bh_uptodate_or_lock - Test whether the buffer is uptodate
 * @bh: struct buffer_head
 *
 * Return true if the buffer is up-to-date and false,
 * with the buffer locked, if not.
 */
int bh_uptodate_or_lock(struct buffer_head *bh)
{
	if (!buffer_uptodate(bh)) {
		lock_buffer(bh);
		if (!buffer_uptodate(bh))
			return 0;
		unlock_buffer(bh);
	}
	return 1;
}
EXPORT_SYMBOL(bh_uptodate_or_lock);

/**
 * bh_submit_read - Submit a locked buffer for reading
 * @bh: struct buffer_head
 *
 * Returns zero on success and -EIO on error.
 */
int bh_submit_read(struct buffer_head *bh)
{
	BUG_ON(!buffer_locked(bh));

	if (buffer_uptodate(bh)) {
		unlock_buffer(bh);
		return 0;
	}

	get_bh(bh);
	bh->b_end_io = end_buffer_read_sync;
	submit_bh(READ, bh);
	wait_on_buffer(bh);
	if (buffer_uptodate(bh))
		return 0;
	return -EIO;
}
EXPORT_SYMBOL(bh_submit_read);

void __init buffer_init(void)
{
	int nrpages;

	bh_cachep = kmem_cache_create("buffer_head",
			sizeof(struct buffer_head), 0,
				(SLAB_RECLAIM_ACCOUNT|SLAB_PANIC|
				SLAB_MEM_SPREAD),
				NULL);

	/*
	 * Limit the bh occupancy to 10% of ZONE_NORMAL
	 */
	nrpages = (nr_free_buffer_pages() * 10) / 100;
	max_buffer_heads = nrpages * (PAGE_SIZE / sizeof(struct buffer_head));
	hotcpu_notifier(buffer_cpu_notify, 0);
}