jfs_metapage.c

linux 内核源代码

/*
 *   Copyright (C) International Business Machines Corp., 2000-2005
 *   Portions Copyright (C) Christoph Hellwig, 2001-2002
 *
 *   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; either version 2 of the License, or
 *   (at your option) any later version.
 *
 *   This program is distributed in the hope that it will 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 to the Free Software
 *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
 */

#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/bio.h>
#include <linux/init.h>
#include <linux/buffer_head.h>
#include <linux/mempool.h>
#include "jfs_incore.h"
#include "jfs_superblock.h"
#include "jfs_filsys.h"
#include "jfs_metapage.h"
#include "jfs_txnmgr.h"
#include "jfs_debug.h"

#ifdef CONFIG_JFS_STATISTICS
static struct {
	uint	pagealloc;	/* # of page allocations */
	uint	pagefree;	/* # of page frees */
	uint	lockwait;	/* # of sleeping lock_metapage() calls */
} mpStat;
#endif

#define metapage_locked(mp) test_bit(META_locked, &(mp)->flag)
#define trylock_metapage(mp) test_and_set_bit(META_locked, &(mp)->flag)

static inline void unlock_metapage(struct metapage *mp)
{
	clear_bit(META_locked, &mp->flag);
	wake_up(&mp->wait);
}

static inline void __lock_metapage(struct metapage *mp)
{
	DECLARE_WAITQUEUE(wait, current);
	INCREMENT(mpStat.lockwait);
	add_wait_queue_exclusive(&mp->wait, &wait);
	do {
		set_current_state(TASK_UNINTERRUPTIBLE);
		if (metapage_locked(mp)) {
			unlock_page(mp->page);
			io_schedule();
			lock_page(mp->page);
		}
	} while (trylock_metapage(mp));
	__set_current_state(TASK_RUNNING);
	remove_wait_queue(&mp->wait, &wait);
}

/*
 * Must have mp->page locked
 */
static inline void lock_metapage(struct metapage *mp)
{
	if (trylock_metapage(mp))
		__lock_metapage(mp);
}

#define METAPOOL_MIN_PAGES 32
static struct kmem_cache *metapage_cache;
static mempool_t *metapage_mempool;

#define MPS_PER_PAGE (PAGE_CACHE_SIZE >> L2PSIZE)

#if MPS_PER_PAGE > 1

struct meta_anchor {
	int mp_count;
	atomic_t io_count;
	struct metapage *mp[MPS_PER_PAGE];
};
#define mp_anchor(page) ((struct meta_anchor *)page_private(page))

static inline struct metapage *page_to_mp(struct page *page, uint offset)
{
	if (!PagePrivate(page))
		return NULL;
	return mp_anchor(page)->mp[offset >> L2PSIZE];
}

static inline int insert_metapage(struct page *page, struct metapage *mp)
{
	struct meta_anchor *a;
	int index;
	int l2mp_blocks;	/* log2 blocks per metapage */

	if (PagePrivate(page))
		a = mp_anchor(page);
	else {
		a = kzalloc(sizeof(struct meta_anchor), GFP_NOFS);
		if (!a)
			return -ENOMEM;
		set_page_private(page, (unsigned long)a);
		SetPagePrivate(page);
		kmap(page);
	}

	if (mp) {
		l2mp_blocks = L2PSIZE - page->mapping->host->i_blkbits;
		index = (mp->index >> l2mp_blocks) & (MPS_PER_PAGE - 1);
		a->mp_count++;
		a->mp[index] = mp;
	}

	return 0;
}

static inline void remove_metapage(struct page *page, struct metapage *mp)
{
	struct meta_anchor *a = mp_anchor(page);
	int l2mp_blocks = L2PSIZE - page->mapping->host->i_blkbits;
	int index;

	index = (mp->index >> l2mp_blocks) & (MPS_PER_PAGE - 1);

	BUG_ON(a->mp[index] != mp);

	a->mp[index] = NULL;
	if (--a->mp_count == 0) {
		kfree(a);
		set_page_private(page, 0);
		ClearPagePrivate(page);
		kunmap(page);
	}
}

static inline void inc_io(struct page *page)
{
	atomic_inc(&mp_anchor(page)->io_count);
}

static inline void dec_io(struct page *page, void (*handler) (struct page *))
{
	if (atomic_dec_and_test(&mp_anchor(page)->io_count))
		handler(page);
}

#else
static inline struct metapage *page_to_mp(struct page *page, uint offset)
{
	return PagePrivate(page) ? (struct metapage *)page_private(page) : NULL;
}

static inline int insert_metapage(struct page *page, struct metapage *mp)
{
	if (mp) {
		set_page_private(page, (unsigned long)mp);
		SetPagePrivate(page);
		kmap(page);
	}
	return 0;
}

static inline void remove_metapage(struct page *page, struct metapage *mp)
{
	set_page_private(page, 0);
	ClearPagePrivate(page);
	kunmap(page);
}

#define inc_io(page) do {} while(0)
#define dec_io(page, handler) handler(page)

#endif

static void init_once(struct kmem_cache *cachep, void *foo)
{
	struct metapage *mp = (struct metapage *)foo;

	mp->lid = 0;
	mp->lsn = 0;
	mp->flag = 0;
	mp->data = NULL;
	mp->clsn = 0;
	mp->log = NULL;
	set_bit(META_free, &mp->flag);
	init_waitqueue_head(&mp->wait);
}

static inline struct metapage *alloc_metapage(gfp_t gfp_mask)
{
	return mempool_alloc(metapage_mempool, gfp_mask);
}

static inline void free_metapage(struct metapage *mp)
{
	mp->flag = 0;
	set_bit(META_free, &mp->flag);

	mempool_free(mp, metapage_mempool);
}

int __init metapage_init(void)
{
	/*
	 * Allocate the metapage structures
	 */
	metapage_cache = kmem_cache_create("jfs_mp", sizeof(struct metapage),
					   0, 0, init_once);
	if (metapage_cache == NULL)
		return -ENOMEM;

	metapage_mempool = mempool_create_slab_pool(METAPOOL_MIN_PAGES,
						    metapage_cache);

	if (metapage_mempool == NULL) {
		kmem_cache_destroy(metapage_cache);
		return -ENOMEM;
	}

	return 0;
}

void metapage_exit(void)
{
	mempool_destroy(metapage_mempool);
	kmem_cache_destroy(metapage_cache);
}

static inline void drop_metapage(struct page *page, struct metapage *mp)
{
	if (mp->count || mp->nohomeok || test_bit(META_dirty, &mp->flag) ||
	    test_bit(META_io, &mp->flag))
		return;
	remove_metapage(page, mp);
	INCREMENT(mpStat.pagefree);
	free_metapage(mp);
}

/*
 * Metapage address space operations
 */

static sector_t metapage_get_blocks(struct inode *inode, sector_t lblock,
				    unsigned int *len)
{
	int rc = 0;
	int xflag;
	s64 xaddr;
	sector_t file_blocks = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
			       inode->i_blkbits;

	if (lblock >= file_blocks)
		return 0;
	if (lblock + *len > file_blocks)
		*len = file_blocks - lblock;

	if (inode->i_ino) {
		rc = xtLookup(inode, (s64)lblock, *len, &xflag, &xaddr, len, 0);
		if ((rc == 0) && *len)
			lblock = (sector_t)xaddr;
		else
			lblock = 0;
	} /* else no mapping */

	return lblock;
}

static void last_read_complete(struct page *page)
{
	if (!PageError(page))
		SetPageUptodate(page);
	unlock_page(page);
}

static void metapage_read_end_io(struct bio *bio, int err)
{
	struct page *page = bio->bi_private;

	if (!test_bit(BIO_UPTODATE, &bio->bi_flags)) {
		printk(KERN_ERR "metapage_read_end_io: I/O error\n");
		SetPageError(page);
	}

	dec_io(page, last_read_complete);
	bio_put(bio);
}

static void remove_from_logsync(struct metapage *mp)
{
	struct jfs_log *log = mp->log;
	unsigned long flags;
/*
 * This can race.  Recheck that log hasn't been set to null, and after
 * acquiring logsync lock, recheck lsn
 */
	if (!log)
		return;

	LOGSYNC_LOCK(log, flags);
	if (mp->lsn) {
		mp->log = NULL;
		mp->lsn = 0;
		mp->clsn = 0;
		log->count--;
		list_del(&mp->synclist);
	}
	LOGSYNC_UNLOCK(log, flags);
}

static void last_write_complete(struct page *page)
{
	struct metapage *mp;
	unsigned int offset;

	for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
		mp = page_to_mp(page, offset);
		if (mp && test_bit(META_io, &mp->flag)) {
			if (mp->lsn)
				remove_from_logsync(mp);
			clear_bit(META_io, &mp->flag);
		}
		/*
		 * I'd like to call drop_metapage here, but I don't think it's
		 * safe unless I have the page locked
		 */
	}
	end_page_writeback(page);
}

static void metapage_write_end_io(struct bio *bio, int err)
{
	struct page *page = bio->bi_private;

	BUG_ON(!PagePrivate(page));

	if (! test_bit(BIO_UPTODATE, &bio->bi_flags)) {
		printk(KERN_ERR "metapage_write_end_io: I/O error\n");
		SetPageError(page);
	}
	dec_io(page, last_write_complete);
	bio_put(bio);
}

static int metapage_writepage(struct page *page, struct writeback_control *wbc)
{
	struct bio *bio = NULL;
	unsigned int block_offset;	/* block offset of mp within page */
	struct inode *inode = page->mapping->host;
	unsigned int blocks_per_mp = JFS_SBI(inode->i_sb)->nbperpage;
	unsigned int len;
	unsigned int xlen;
	struct metapage *mp;
	int redirty = 0;
	sector_t lblock;
	sector_t pblock;
	sector_t next_block = 0;
	sector_t page_start;
	unsigned long bio_bytes = 0;
	unsigned long bio_offset = 0;
	unsigned int offset;

	page_start = (sector_t)page->index <<
		     (PAGE_CACHE_SHIFT - inode->i_blkbits);
	BUG_ON(!PageLocked(page));
	BUG_ON(PageWriteback(page));

	for (offset = 0; offset < PAGE_CACHE_SIZE; offset += PSIZE) {
		mp = page_to_mp(page, offset);

		if (!mp || !test_bit(META_dirty, &mp->flag))
			continue;

		if (mp->nohomeok && !test_bit(META_forcewrite, &mp->flag)) {
			redirty = 1;
			/*
			 * Make sure this page isn't blocked indefinitely.
			 * If the journal isn't undergoing I/O, push it
			 */
			if (mp->log && !(mp->log->cflag & logGC_PAGEOUT))
				jfs_flush_journal(mp->log, 0);
			continue;
		}

		clear_bit(META_dirty, &mp->flag);
		block_offset = offset >> inode->i_blkbits;
		lblock = page_start + block_offset;
		if (bio) {
			if (xlen && lblock == next_block) {
				/* Contiguous, in memory & on disk */
				len = min(xlen, blocks_per_mp);
				xlen -= len;
				bio_bytes += len << inode->i_blkbits;
				set_bit(META_io, &mp->flag);
				continue;
			}
			/* Not contiguous */
			if (bio_add_page(bio, page, bio_bytes, bio_offset) <
			    bio_bytes)
				goto add_failed;
			/*
			 * Increment counter before submitting i/o to keep
			 * count from hitting zero before we're through
			 */
			inc_io(page);
			if (!bio->bi_size)
				goto dump_bio;
			submit_bio(WRITE, bio);
			bio = NULL;
		} else {