xfs_buf.c

linux2.6.16版本

/*
 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
 * All Rights Reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it would be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write the Free Software Foundation,
 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */
#include <linux/stddef.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/pagemap.h>
#include <linux/init.h>
#include <linux/vmalloc.h>
#include <linux/bio.h>
#include <linux/sysctl.h>
#include <linux/proc_fs.h>
#include <linux/workqueue.h>
#include <linux/percpu.h>
#include <linux/blkdev.h>
#include <linux/hash.h>
#include <linux/kthread.h>
#include "xfs_linux.h"

STATIC kmem_zone_t *xfs_buf_zone;
STATIC kmem_shaker_t xfs_buf_shake;
STATIC int xfsbufd(void *);
STATIC int xfsbufd_wakeup(int, gfp_t);
STATIC void xfs_buf_delwri_queue(xfs_buf_t *, int);

STATIC struct workqueue_struct *xfslogd_workqueue;
struct workqueue_struct *xfsdatad_workqueue;

#ifdef XFS_BUF_TRACE
void
xfs_buf_trace(
	xfs_buf_t	*bp,
	char		*id,
	void		*data,
	void		*ra)
{
	ktrace_enter(xfs_buf_trace_buf,
		bp, id,
		(void *)(unsigned long)bp->b_flags,
		(void *)(unsigned long)bp->b_hold.counter,
		(void *)(unsigned long)bp->b_sema.count.counter,
		(void *)current,
		data, ra,
		(void *)(unsigned long)((bp->b_file_offset>>32) & 0xffffffff),
		(void *)(unsigned long)(bp->b_file_offset & 0xffffffff),
		(void *)(unsigned long)bp->b_buffer_length,
		NULL, NULL, NULL, NULL, NULL);
}
ktrace_t *xfs_buf_trace_buf;
#define XFS_BUF_TRACE_SIZE	4096
#define XB_TRACE(bp, id, data)	\
	xfs_buf_trace(bp, id, (void *)data, (void *)__builtin_return_address(0))
#else
#define XB_TRACE(bp, id, data)	do { } while (0)
#endif

#ifdef XFS_BUF_LOCK_TRACKING
# define XB_SET_OWNER(bp)	((bp)->b_last_holder = current->pid)
# define XB_CLEAR_OWNER(bp)	((bp)->b_last_holder = -1)
# define XB_GET_OWNER(bp)	((bp)->b_last_holder)
#else
# define XB_SET_OWNER(bp)	do { } while (0)
# define XB_CLEAR_OWNER(bp)	do { } while (0)
# define XB_GET_OWNER(bp)	do { } while (0)
#endif

#define xb_to_gfp(flags) \
	((((flags) & XBF_READ_AHEAD) ? __GFP_NORETRY : \
	  ((flags) & XBF_DONT_BLOCK) ? GFP_NOFS : GFP_KERNEL) | __GFP_NOWARN)

#define xb_to_km(flags) \
	 (((flags) & XBF_DONT_BLOCK) ? KM_NOFS : KM_SLEEP)

#define xfs_buf_allocate(flags) \
	kmem_zone_alloc(xfs_buf_zone, xb_to_km(flags))
#define xfs_buf_deallocate(bp) \
	kmem_zone_free(xfs_buf_zone, (bp));

/*
 *	Page Region interfaces.
 *
 *	For pages in filesystems where the blocksize is smaller than the
 *	pagesize, we use the page->private field (long) to hold a bitmap
 * 	of uptodate regions within the page.
 *
 *	Each such region is "bytes per page / bits per long" bytes long.
 *
 *	NBPPR == number-of-bytes-per-page-region
 *	BTOPR == bytes-to-page-region (rounded up)
 *	BTOPRT == bytes-to-page-region-truncated (rounded down)
 */
#if (BITS_PER_LONG == 32)
#define PRSHIFT		(PAGE_CACHE_SHIFT - 5)	/* (32 == 1<<5) */
#elif (BITS_PER_LONG == 64)
#define PRSHIFT		(PAGE_CACHE_SHIFT - 6)	/* (64 == 1<<6) */
#else
#error BITS_PER_LONG must be 32 or 64
#endif
#define NBPPR		(PAGE_CACHE_SIZE/BITS_PER_LONG)
#define BTOPR(b)	(((unsigned int)(b) + (NBPPR - 1)) >> PRSHIFT)
#define BTOPRT(b)	(((unsigned int)(b) >> PRSHIFT))

STATIC unsigned long
page_region_mask(
	size_t		offset,
	size_t		length)
{
	unsigned long	mask;
	int		first, final;

	first = BTOPR(offset);
	final = BTOPRT(offset + length - 1);
	first = min(first, final);

	mask = ~0UL;
	mask <<= BITS_PER_LONG - (final - first);
	mask >>= BITS_PER_LONG - (final);

	ASSERT(offset + length <= PAGE_CACHE_SIZE);
	ASSERT((final - first) < BITS_PER_LONG && (final - first) >= 0);

	return mask;
}

STATIC inline void
set_page_region(
	struct page	*page,
	size_t		offset,
	size_t		length)
{
	set_page_private(page,
		page_private(page) | page_region_mask(offset, length));
	if (page_private(page) == ~0UL)
		SetPageUptodate(page);
}

STATIC inline int
test_page_region(
	struct page	*page,
	size_t		offset,
	size_t		length)
{
	unsigned long	mask = page_region_mask(offset, length);

	return (mask && (page_private(page) & mask) == mask);
}

/*
 *	Mapping of multi-page buffers into contiguous virtual space
 */

typedef struct a_list {
	void		*vm_addr;
	struct a_list	*next;
} a_list_t;

STATIC a_list_t		*as_free_head;
STATIC int		as_list_len;
STATIC DEFINE_SPINLOCK(as_lock);

/*
 *	Try to batch vunmaps because they are costly.
 */
STATIC void
free_address(
	void		*addr)
{
	a_list_t	*aentry;

	aentry = kmalloc(sizeof(a_list_t), GFP_ATOMIC & ~__GFP_HIGH);
	if (likely(aentry)) {
		spin_lock(&as_lock);
		aentry->next = as_free_head;
		aentry->vm_addr = addr;
		as_free_head = aentry;
		as_list_len++;
		spin_unlock(&as_lock);
	} else {
		vunmap(addr);
	}
}

STATIC void
purge_addresses(void)
{
	a_list_t	*aentry, *old;

	if (as_free_head == NULL)
		return;

	spin_lock(&as_lock);
	aentry = as_free_head;
	as_free_head = NULL;
	as_list_len = 0;
	spin_unlock(&as_lock);

	while ((old = aentry) != NULL) {
		vunmap(aentry->vm_addr);
		aentry = aentry->next;
		kfree(old);
	}
}

/*
 *	Internal xfs_buf_t object manipulation
 */

STATIC void
_xfs_buf_initialize(
	xfs_buf_t		*bp,
	xfs_buftarg_t		*target,
	xfs_off_t		range_base,
	size_t			range_length,
	xfs_buf_flags_t		flags)
{
	/*
	 * We don't want certain flags to appear in b_flags.
	 */
	flags &= ~(XBF_LOCK|XBF_MAPPED|XBF_DONT_BLOCK|XBF_READ_AHEAD);

	memset(bp, 0, sizeof(xfs_buf_t));
	atomic_set(&bp->b_hold, 1);
	init_MUTEX_LOCKED(&bp->b_iodonesema);
	INIT_LIST_HEAD(&bp->b_list);
	INIT_LIST_HEAD(&bp->b_hash_list);
	init_MUTEX_LOCKED(&bp->b_sema); /* held, no waiters */
	XB_SET_OWNER(bp);
	bp->b_target = target;
	bp->b_file_offset = range_base;
	/*
	 * Set buffer_length and count_desired to the same value initially.
	 * I/O routines should use count_desired, which will be the same in
	 * most cases but may be reset (e.g. XFS recovery).
	 */
	bp->b_buffer_length = bp->b_count_desired = range_length;
	bp->b_flags = flags;
	bp->b_bn = XFS_BUF_DADDR_NULL;
	atomic_set(&bp->b_pin_count, 0);
	init_waitqueue_head(&bp->b_waiters);

	XFS_STATS_INC(xb_create);
	XB_TRACE(bp, "initialize", target);
}

/*
 *	Allocate a page array capable of holding a specified number
 *	of pages, and point the page buf at it.
 */
STATIC int
_xfs_buf_get_pages(
	xfs_buf_t		*bp,
	int			page_count,
	xfs_buf_flags_t		flags)
{
	/* Make sure that we have a page list */
	if (bp->b_pages == NULL) {
		bp->b_offset = xfs_buf_poff(bp->b_file_offset);
		bp->b_page_count = page_count;
		if (page_count <= XB_PAGES) {
			bp->b_pages = bp->b_page_array;
		} else {
			bp->b_pages = kmem_alloc(sizeof(struct page *) *
					page_count, xb_to_km(flags));
			if (bp->b_pages == NULL)
				return -ENOMEM;
		}
		memset(bp->b_pages, 0, sizeof(struct page *) * page_count);
	}
	return 0;
}

/*
 *	Frees b_pages if it was allocated.
 */
STATIC void
_xfs_buf_free_pages(
	xfs_buf_t	*bp)
{
	if (bp->b_pages != bp->b_page_array) {
		kmem_free(bp->b_pages,
			  bp->b_page_count * sizeof(struct page *));
	}
}

/*
 *	Releases the specified buffer.
 *
 * 	The modification state of any associated pages is left unchanged.
 * 	The buffer most not be on any hash - use xfs_buf_rele instead for
 * 	hashed and refcounted buffers
 */
void
xfs_buf_free(
	xfs_buf_t		*bp)
{
	XB_TRACE(bp, "free", 0);

	ASSERT(list_empty(&bp->b_hash_list));

	if (bp->b_flags & _XBF_PAGE_CACHE) {
		uint		i;

		if ((bp->b_flags & XBF_MAPPED) && (bp->b_page_count > 1))
			free_address(bp->b_addr - bp->b_offset);

		for (i = 0; i < bp->b_page_count; i++)
			page_cache_release(bp->b_pages[i]);
		_xfs_buf_free_pages(bp);
	} else if (bp->b_flags & _XBF_KMEM_ALLOC) {
		 /*
		  * XXX(hch): bp->b_count_desired might be incorrect (see
		  * xfs_buf_associate_memory for details), but fortunately
		  * the Linux version of kmem_free ignores the len argument..
		  */
		kmem_free(bp->b_addr, bp->b_count_desired);
		_xfs_buf_free_pages(bp);
	}

	xfs_buf_deallocate(bp);
}

/*
 *	Finds all pages for buffer in question and builds it's page list.
 */
STATIC int
_xfs_buf_lookup_pages(
	xfs_buf_t		*bp,
	uint			flags)
{
	struct address_space	*mapping = bp->b_target->bt_mapping;
	size_t			blocksize = bp->b_target->bt_bsize;
	size_t			size = bp->b_count_desired;
	size_t			nbytes, offset;
	gfp_t			gfp_mask = xb_to_gfp(flags);
	unsigned short		page_count, i;
	pgoff_t			first;
	xfs_off_t		end;
	int			error;

	end = bp->b_file_offset + bp->b_buffer_length;
	page_count = xfs_buf_btoc(end) - xfs_buf_btoct(bp->b_file_offset);

	error = _xfs_buf_get_pages(bp, page_count, flags);
	if (unlikely(error))
		return error;
	bp->b_flags |= _XBF_PAGE_CACHE;

	offset = bp->b_offset;
	first = bp->b_file_offset >> PAGE_CACHE_SHIFT;

	for (i = 0; i < bp->b_page_count; i++) {
		struct page	*page;
		uint		retries = 0;

	      retry:
		page = find_or_create_page(mapping, first + i, gfp_mask);
		if (unlikely(page == NULL)) {
			if (flags & XBF_READ_AHEAD) {
				bp->b_page_count = i;
				for (i = 0; i < bp->b_page_count; i++)
					unlock_page(bp->b_pages[i]);
				return -ENOMEM;
			}

			/*
			 * This could deadlock.
			 *
			 * But until all the XFS lowlevel code is revamped to
			 * handle buffer allocation failures we can't do much.
			 */
			if (!(++retries % 100))
				printk(KERN_ERR
					"XFS: possible memory allocation "
					"deadlock in %s (mode:0x%x)\n",
					__FUNCTION__, gfp_mask);

			XFS_STATS_INC(xb_page_retries);
			xfsbufd_wakeup(0, gfp_mask);
			blk_congestion_wait(WRITE, HZ/50);
			goto retry;
		}

		XFS_STATS_INC(xb_page_found);

		nbytes = min_t(size_t, size, PAGE_CACHE_SIZE - offset);
		size -= nbytes;

		if (!PageUptodate(page)) {
			page_count--;
			if (blocksize >= PAGE_CACHE_SIZE) {
				if (flags & XBF_READ)
					bp->b_locked = 1;
			} else if (!PagePrivate(page)) {
				if (test_page_region(page, offset, nbytes))
					page_count++;
			}
		}

		bp->b_pages[i] = page;
		offset = 0;
	}

	if (!bp->b_locked) {
		for (i = 0; i < bp->b_page_count; i++)
			unlock_page(bp->b_pages[i]);
	}

	if (page_count == bp->b_page_count)
		bp->b_flags |= XBF_DONE;

	XB_TRACE(bp, "lookup_pages", (long)page_count);
	return error;
}

/*
 *	Map buffer into kernel address-space if nessecary.
 */
STATIC int
_xfs_buf_map_pages(
	xfs_buf_t		*bp,
	uint			flags)
{
	/* A single page buffer is always mappable */
	if (bp->b_page_count == 1) {
		bp->b_addr = page_address(bp->b_pages[0]) + bp->b_offset;
		bp->b_flags |= XBF_MAPPED;
	} else if (flags & XBF_MAPPED) {
		if (as_list_len > 64)
			purge_addresses();
		bp->b_addr = vmap(bp->b_pages, bp->b_page_count,
					VM_MAP, PAGE_KERNEL);
		if (unlikely(bp->b_addr == NULL))
			return -ENOMEM;
		bp->b_addr += bp->b_offset;
		bp->b_flags |= XBF_MAPPED;
	}

	return 0;
}

/*
 *	Finding and Reading Buffers
 */

/*
 *	Look up, and creates if absent, a lockable buffer for
 *	a given range of an inode.  The buffer is returned
 *	locked.	 If other overlapping buffers exist, they are
 *	released before the new buffer is created and locked,
 *	which may imply that this call will block until those buffers
 *	are unlocked.  No I/O is implied by this call.
 */
xfs_buf_t *
_xfs_buf_find(
	xfs_buftarg_t		*btp,	/* block device target		*/
	xfs_off_t		ioff,	/* starting offset of range	*/
	size_t			isize,	/* length of range		*/
	xfs_buf_flags_t		flags,
	xfs_buf_t		*new_bp)
{
	xfs_off_t		range_base;
	size_t			range_length;
	xfs_bufhash_t		*hash;
	xfs_buf_t		*bp, *n;

	range_base = (ioff << BBSHIFT);
	range_length = (isize << BBSHIFT);

	/* Check for IOs smaller than the sector size / not sector aligned */
	ASSERT(!(range_length < (1 << btp->bt_sshift)));
	ASSERT(!(range_base & (xfs_off_t)btp->bt_smask));

	hash = &btp->bt_hash[hash_long((unsigned long)ioff, btp->bt_hashshift)];

	spin_lock(&hash->bh_lock);

	list_for_each_entry_safe(bp, n, &hash->bh_list, b_hash_list) {
		ASSERT(btp == bp->b_target);
		if (bp->b_file_offset == range_base &&
		    bp->b_buffer_length == range_length) {
			/*
			 * If we look at something, bring it to the
			 * front of the list for next time.
			 */
			atomic_inc(&bp->b_hold);
			list_move(&bp->b_hash_list, &hash->bh_list);
			goto found;
		}
	}

	/* No match found */
	if (new_bp) {
		_xfs_buf_initialize(new_bp, btp, range_base,
				range_length, flags);
		new_bp->b_hash = hash;
		list_add(&new_bp->b_hash_list, &hash->bh_list);
	} else {
		XFS_STATS_INC(xb_miss_locked);
	}

	spin_unlock(&hash->bh_lock);
	return new_bp;

found:
	spin_unlock(&hash->bh_lock);

	/* Attempt to get the semaphore without sleeping,
	 * if this does not work then we need to drop the
	 * spinlock and do a hard attempt on the semaphore.
	 */
	if (down_trylock(&bp->b_sema)) {
		if (!(flags & XBF_TRYLOCK)) {
			/* wait for buffer ownership */
			XB_TRACE(bp, "get_lock", 0);
			xfs_buf_lock(bp);
			XFS_STATS_INC(xb_get_locked_waited);
		} else {
			/* We asked for a trylock and failed, no need
			 * to look at file offset and length here, we
			 * know that this buffer at least overlaps our
			 * buffer and is locked, therefore our buffer
			 * either does not exist, or is this buffer.
			 */
			xfs_buf_rele(bp);
			XFS_STATS_INC(xb_busy_locked);
			return NULL;
		}
	} else {
		/* trylock worked */
		XB_SET_OWNER(bp);
	}

	if (bp->b_flags & XBF_STALE) {
		ASSERT((bp->b_flags & _XBF_DELWRI_Q) == 0);
		bp->b_flags &= XBF_MAPPED;
	}
	XB_TRACE(bp, "got_lock", 0);
	XFS_STATS_INC(xb_get_locked);
	return bp;
}

/*
 *	Assembles a buffer covering the specified range.
 *	Storage in memory for all portions of the buffer will be allocated,
 *	although backing storage may not be.
 */
xfs_buf_t *
xfs_buf_get_flags(
	xfs_buftarg_t		*target,/* target for buffer		*/
	xfs_off_t		ioff,	/* starting offset of range	*/
	size_t			isize,	/* length of range		*/
	xfs_buf_flags_t		flags)
{
	xfs_buf_t		*bp, *new_bp;
	int			error = 0, i;

	new_bp = xfs_buf_allocate(flags);
	if (unlikely(!new_bp))
		return NULL;

	bp = _xfs_buf_find(target, ioff, isize, flags, new_bp);
	if (bp == new_bp) {
		error = _xfs_buf_lookup_pages(bp, flags);
		if (error)
			goto no_buffer;
	} else {
		xfs_buf_deallocate(new_bp);
		if (unlikely(bp == NULL))
			return NULL;
	}

	for (i = 0; i < bp->b_page_count; i++)
		mark_page_accessed(bp->b_pages[i]);

	if (!(bp->b_flags & XBF_MAPPED)) {
		error = _xfs_buf_map_pages(bp, flags);
		if (unlikely(error)) {
			printk(KERN_WARNING "%s: failed to map pages\n",
					__FUNCTION__);
			goto no_buffer;
		}
	}

	XFS_STATS_INC(xb_get);

	/*
	 * Always fill in the block number now, the mapped cases can do
	 * their own overlay of this later.
	 */
	bp->b_bn = ioff;
	bp->b_count_desired = bp->b_buffer_length;

	XB_TRACE(bp, "get", (unsigned long)flags);
	return bp;

 no_buffer:
	if (flags & (XBF_LOCK | XBF_TRYLOCK))
		xfs_buf_unlock(bp);
	xfs_buf_rele(bp);
	return NULL;
}

xfs_buf_t *
xfs_buf_read_flags(
	xfs_buftarg_t		*target,