Merge bk://linuxusb.bkbits.net/pci_hp-2.5

into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

arch/ppc/platforms/pmac_pic.c

@@ -492,7 +492,7 @@ pmac_pic_init(void)
  * and disables all interrupts except for the nominated one.
  * sleep_restore_intrs() restores the states of all interrupt enables.
  */
-unsigned int sleep_save_mask[2];
+unsigned long sleep_save_mask[2];
 
 void __pmac
 pmac_sleep_save_intrs(int viaint)

drivers/block/blkpg.c

@@ -237,6 +237,18 @@ int blk_ioctl(struct block_device *bdev, unsigned int cmd, unsigned long arg)
 			intval = (is_read_only(dev) != 0);
 			return put_user(intval, (int *)(arg));
 
+		case BLKRASET:
+		case BLKFRASET:
+			if(!capable(CAP_SYS_ADMIN))
+				return -EACCES;
+			return blk_set_readahead(dev, arg);
+
+		case BLKRAGET:
+		case BLKFRAGET:
+			if (!arg)
+				return -EINVAL;
+			return put_user(blk_get_readahead(dev), (long *)arg);
+
 		case BLKSECTGET:
 			if ((q = blk_get_queue(dev)) == NULL)
 				return -EINVAL;
@@ -259,11 +271,11 @@ int blk_ioctl(struct block_device *bdev, unsigned int cmd, unsigned long arg)
 
 		case BLKGETSIZE:
 			/* size in sectors, works up to 2 TB */
-			ullval = blkdev_size_in_bytes(dev);
+			ullval = bdev->bd_inode->i_size;
 			return put_user((unsigned long)(ullval >> 9), (unsigned long *) arg);
 		case BLKGETSIZE64:
 			/* size in bytes */
-			ullval = blkdev_size_in_bytes(dev);
+			ullval = bdev->bd_inode->i_size;
 			return put_user(ullval, (u64 *) arg);
 #if 0
 		case BLKRRPART: /* Re-read partition tables */

drivers/block/ll_rw_blk.c

@@ -108,6 +108,47 @@ inline request_queue_t *blk_get_queue(kdev_t dev)
 		return &blk_dev[major(dev)].request_queue;
 }
 
+/**
+ * blk_set_readahead - set a queue's readahead tunable
+ * @dev:	device
+ * @sectors:	readahead, in 512 byte sectors
+ *
+ * Returns zero on success, else negative errno
+ */
+int blk_set_readahead(kdev_t dev, unsigned sectors)
+{
+	int ret = -EINVAL;
+	request_queue_t *q = blk_get_queue(dev);
+
+	if (q) {
+		q->ra_sectors = sectors;
+		ret = 0;
+	}
+	return ret;
+}
+
+/**
+ * blk_get_readahead - query a queue's readahead tunable
+ * @dev:	device
+ *
+ * Locates the passed device's request queue and returns its
+ * readahead setting.
+ *
+ * The returned value is in units of 512 byte sectors.
+ *
+ * Will return zero if the queue has never had its readahead
+ * setting altered.
+ */
+unsigned blk_get_readahead(kdev_t dev)
+{
+	unsigned ret = 0;
+	request_queue_t *q = blk_get_queue(dev);
+
+	if (q)
+		ret = q->ra_sectors;
+	return ret;
+}
+
 void blk_queue_prep_rq(request_queue_t *q, prep_rq_fn *pfn)
 {
 	q->prep_rq_fn = pfn;
@@ -810,7 +851,8 @@ int blk_init_queue(request_queue_t *q, request_fn_proc *rfn, spinlock_t *lock)
 	q->plug_tq.data		= q;
 	q->queue_flags		= (1 << QUEUE_FLAG_CLUSTER);
 	q->queue_lock		= lock;
-	
+	q->ra_sectors		= 0;		/* Use VM default */
+
 	blk_queue_segment_boundary(q, 0xffffffff);
 
 	blk_queue_make_request(q, __make_request);

drivers/block/loop.c

@@ -160,13 +160,7 @@ compute_loop_size(struct loop_device *lo,
 {
 	loff_t size = 0;
 
-	if (S_ISREG(lo_dentry->d_inode->i_mode)) {
-		size = lo_dentry->d_inode->i_size;
-	} else {
-		size = blkdev_size_in_bytes(lodev);
-		if (size == 0)
-			return MAX_DISK_SIZE;
-	}
+	size = lo_dentry->d_inode->i_mapping->host->i_size;
 	return (size - lo->lo_offset) >> BLOCK_SIZE_BITS;
 }
 

drivers/block/rd.c

@@ -156,7 +156,6 @@ static int rd_blkdev_pagecache_IO(int rw, struct bio_vec *vec,
 
 	do {
 		int count;
-		struct page ** hash;
 		struct page * page;
 		char * src, * dst;
 		int unlock = 0;
@@ -166,8 +165,7 @@ static int rd_blkdev_pagecache_IO(int rw, struct bio_vec *vec,
 			count = size;
 		size -= count;
 
-		hash = page_hash(mapping, index);
-		page = __find_get_page(mapping, index, hash);
+		page = find_get_page(mapping, index);
 		if (!page) {
 			page = grab_cache_page(mapping, index);
 			err = -ENOMEM;

drivers/char/n_tty.c

@@ -538,7 +538,7 @@ static inline void n_tty_receive_char(struct tty_struct *tty, unsigned char c)
 	 * handle specially, do shortcut processing to speed things
 	 * up.
 	 */
-	if (!test_bit(c, &tty->process_char_map) || tty->lnext) {
+	if (!test_bit(c, tty->process_char_map) || tty->lnext) {
 		finish_erasing(tty);
 		tty->lnext = 0;
 		if (L_ECHO(tty)) {
@@ -659,7 +659,7 @@ static inline void n_tty_receive_char(struct tty_struct *tty, unsigned char c)
 
 		handle_newline:
 			spin_lock_irqsave(&tty->read_lock, flags);
-			set_bit(tty->read_head, &tty->read_flags);
+			set_bit(tty->read_head, tty->read_flags);
 			put_tty_queue_nolock(c, tty);
 			tty->canon_head = tty->read_head;
 			tty->canon_data++;
@@ -811,38 +811,38 @@ static void n_tty_set_termios(struct tty_struct *tty, struct termios * old)
 		memset(tty->process_char_map, 0, 256/8);
 
 		if (I_IGNCR(tty) || I_ICRNL(tty))
-			set_bit('\r', &tty->process_char_map);
+			set_bit('\r', tty->process_char_map);
 		if (I_INLCR(tty))
-			set_bit('\n', &tty->process_char_map);
+			set_bit('\n', tty->process_char_map);
 
 		if (L_ICANON(tty)) {
-			set_bit(ERASE_CHAR(tty), &tty->process_char_map);
-			set_bit(KILL_CHAR(tty), &tty->process_char_map);
-			set_bit(EOF_CHAR(tty), &tty->process_char_map);
-			set_bit('\n', &tty->process_char_map);
-			set_bit(EOL_CHAR(tty), &tty->process_char_map);
+			set_bit(ERASE_CHAR(tty), tty->process_char_map);
+			set_bit(KILL_CHAR(tty), tty->process_char_map);
+			set_bit(EOF_CHAR(tty), tty->process_char_map);
+			set_bit('\n', tty->process_char_map);
+			set_bit(EOL_CHAR(tty), tty->process_char_map);
 			if (L_IEXTEN(tty)) {
 				set_bit(WERASE_CHAR(tty),
-					&tty->process_char_map);
+					tty->process_char_map);
 				set_bit(LNEXT_CHAR(tty),
-					&tty->process_char_map);
+					tty->process_char_map);
 				set_bit(EOL2_CHAR(tty),
-					&tty->process_char_map);
+					tty->process_char_map);
 				if (L_ECHO(tty))
 					set_bit(REPRINT_CHAR(tty),
-						&tty->process_char_map);
+						tty->process_char_map);
 			}
 		}
 		if (I_IXON(tty)) {
-			set_bit(START_CHAR(tty), &tty->process_char_map);
-			set_bit(STOP_CHAR(tty), &tty->process_char_map);
+			set_bit(START_CHAR(tty), tty->process_char_map);
+			set_bit(STOP_CHAR(tty), tty->process_char_map);
 		}
 		if (L_ISIG(tty)) {
-			set_bit(INTR_CHAR(tty), &tty->process_char_map);
-			set_bit(QUIT_CHAR(tty), &tty->process_char_map);
-			set_bit(SUSP_CHAR(tty), &tty->process_char_map);
+			set_bit(INTR_CHAR(tty), tty->process_char_map);
+			set_bit(QUIT_CHAR(tty), tty->process_char_map);
+			set_bit(SUSP_CHAR(tty), tty->process_char_map);
 		}
-		clear_bit(__DISABLED_CHAR, &tty->process_char_map);
+		clear_bit(__DISABLED_CHAR, tty->process_char_map);
 		sti();
 		tty->raw = 0;
 		tty->real_raw = 0;
@@ -1058,7 +1058,7 @@ static ssize_t read_chan(struct tty_struct *tty, struct file *file,
  				int eol;
 
 				eol = test_and_clear_bit(tty->read_tail,
-						&tty->read_flags);
+						tty->read_flags);
 				c = tty->read_buf[tty->read_tail];
 				spin_lock_irqsave(&tty->read_lock, flags);
 				tty->read_tail = ((tty->read_tail+1) &

drivers/char/raw.c

@@ -266,6 +266,7 @@ ssize_t	rw_raw_dev(int rw, struct file *filp, char *buf,
 	unsigned long	limit;
 	int		sector_size, sector_bits, sector_mask;
 	sector_t	blocknr;
+	struct block_device *bdev;
 	
 	/*
 	 * First, a few checks on device size limits 
@@ -286,12 +287,13 @@ ssize_t	rw_raw_dev(int rw, struct file *filp, char *buf,
 		new_iobuf = 1;
 	}
 
-	dev = to_kdev_t(raw_devices[minor].binding->bd_dev);
+	bdev = raw_devices[minor].binding;
+	dev = to_kdev_t(bdev->bd_dev);
 	sector_size = raw_devices[minor].sector_size;
 	sector_bits = raw_devices[minor].sector_bits;
 	sector_mask = sector_size - 1;
 
-	limit = blkdev_size_in_bytes(dev) >> sector_bits;
+	limit = bdev->bd_inode->i_size >> sector_bits;
 	if (!limit)
 		limit = INT_MAX;
 	dprintk ("rw_raw_dev: dev %d:%d (+%d)\n",

drivers/char/tty_ioctl.c

@@ -188,7 +188,7 @@ static unsigned long inq_canon(struct tty_struct * tty)
 	nr = (head - tail) & (N_TTY_BUF_SIZE-1);
 	/* Skip EOF-chars.. */
 	while (head != tail) {
-		if (test_bit(tail, &tty->read_flags) &&
+		if (test_bit(tail, tty->read_flags) &&
 		    tty->read_buf[tail] == __DISABLED_CHAR)
 			nr--;
 		tail = (tail+1) & (N_TTY_BUF_SIZE-1);

drivers/macintosh/adb.c

@@ -77,7 +77,7 @@ struct notifier_block *adb_client_list = NULL;
 static int adb_got_sleep = 0;
 static int adb_inited = 0;
 static pid_t adb_probe_task_pid;
-static int adb_probe_task_flag;
+static unsigned long adb_probe_task_flag;
 static wait_queue_head_t adb_probe_task_wq;
 static int sleepy_trackpad;
 int __adb_probe_sync;
@@ -439,7 +439,7 @@ adb_probe_wakeup(struct adb_request *req)
 }
 
 static struct adb_request adb_sreq;
-static int adb_sreq_lock; // Use semaphore ! */ 
+static unsigned long adb_sreq_lock; // Use semaphore ! */ 
 
 int
 adb_request(struct adb_request *req, void (*done)(struct adb_request *),

drivers/md/md.c

@@ -1577,7 +1577,7 @@ static int device_size_calculation(mddev_t * mddev)
 	if (!md_size[mdidx(mddev)])
 		md_size[mdidx(mddev)] = sb->size * data_disks;
 
-	readahead = MD_READAHEAD;
+	readahead = (blk_get_readahead(rdev->dev) * 512) / PAGE_SIZE;
 	if (!sb->level || (sb->level == 4) || (sb->level == 5)) {
 		readahead = (mddev->sb->chunk_size>>PAGE_SHIFT) * 4 * data_disks;
 		if (readahead < data_disks * (MAX_SECTORS>>(PAGE_SHIFT-9))*2)
@@ -3387,7 +3387,7 @@ int md_do_sync(mddev_t *mddev, mdp_disk_t *spare)
 	/*
 	 * Tune reconstruction:
 	 */
-	window = MAX_READAHEAD*(PAGE_SIZE/512);
+	window = 32*(PAGE_SIZE/512);
 	printk(KERN_INFO "md: using %dk window, over a total of %d blocks.\n",
 	       window/2,max_sectors/2);
 
@@ -3605,7 +3605,7 @@ static void md_geninit(void)
 	for(i = 0; i < MAX_MD_DEVS; i++) {
 		md_blocksizes[i] = 1024;
 		md_size[i] = 0;
-		md_maxreadahead[i] = MD_READAHEAD;
+		md_maxreadahead[i] = 32;
 	}
 	blksize_size[MAJOR_NR] = md_blocksizes;
 	blk_size[MAJOR_NR] = md_size;

drivers/mtd/mtdblock.c

@@ -413,7 +413,7 @@ static void handle_mtdblock_request(void)
 	for (;;) {
 		INIT_REQUEST;
 		req = CURRENT;
-		spin_unlock_irq(&QUEUE->queue_lock);
+		spin_unlock_irq(QUEUE->queue_lock);
 		mtdblk = mtdblks[minor(req->rq_dev)];
 		res = 0;
 
@@ -457,7 +457,7 @@ static void handle_mtdblock_request(void)
 		}
 
 end_req:
-		spin_lock_irq(&QUEUE->queue_lock);
+		spin_lock_irq(QUEUE->queue_lock);
 		if (!end_that_request_first(req, res, req->hard_cur_sectors)) {
 			blkdev_dequeue_request(req);
 			end_that_request_last(req);
@@ -487,16 +487,16 @@ int mtdblock_thread(void *dummy)
 	while (!leaving) {
 		add_wait_queue(&thr_wq, &wait);
 		set_current_state(TASK_INTERRUPTIBLE);
-		spin_lock_irq(&QUEUE->queue_lock);
+		spin_lock_irq(QUEUE->queue_lock);
 		if (QUEUE_EMPTY || QUEUE_PLUGGED) {
-			spin_unlock_irq(&QUEUE->queue_lock);
+			spin_unlock_irq(QUEUE->queue_lock);
 			schedule();
 			remove_wait_queue(&thr_wq, &wait); 
 		} else {
 			remove_wait_queue(&thr_wq, &wait); 
 			set_current_state(TASK_RUNNING);
 			handle_mtdblock_request();
-			spin_unlock_irq(&QUEUE->queue_lock);
+			spin_unlock_irq(QUEUE->queue_lock);
 		}
 	}
 
@@ -541,7 +541,7 @@ static int mtdblock_ioctl(struct inode * inode, struct file * file,
 			return -EACCES;
 #endif
 		fsync_bdev(inode->i_bdev);
-		invalidate_bdev(inode->b_rdev, 0);
+		invalidate_bdev(inode->i_bdev, 0);
 		down(&mtdblk->cache_sem);
 		write_cached_data(mtdblk);
 		up(&mtdblk->cache_sem);

fs/block_dev.c

@@ -19,6 +19,7 @@
 #include <linux/highmem.h>
 #include <linux/blkdev.h>
 #include <linux/module.h>
+#include <linux/blkpg.h>
 
 #include <asm/uaccess.h>
 
@@ -172,7 +173,6 @@ static loff_t block_llseek(struct file *file, loff_t offset, int origin)
 	if (offset >= 0 && offset <= size) {
 		if (offset != file->f_pos) {
 			file->f_pos = offset;
-			file->f_reada = 0;
 			file->f_version = ++event;
 		}
 		retval = offset;
@@ -692,9 +692,20 @@ int blkdev_close(struct inode * inode, struct file * filp)
 static int blkdev_ioctl(struct inode *inode, struct file *file, unsigned cmd,
 			unsigned long arg)
 {
-	if (inode->i_bdev->bd_op->ioctl)
-		return inode->i_bdev->bd_op->ioctl(inode, file, cmd, arg);
-	return -EINVAL;
+	int ret = -EINVAL;
+	switch (cmd) {
+	case BLKRAGET:
+	case BLKFRAGET:
+	case BLKRASET:
+	case BLKFRASET:
+		ret = blk_ioctl(inode->i_bdev, cmd, arg);
+		break;
+	default:
+		if (inode->i_bdev->bd_op->ioctl)
+			ret =inode->i_bdev->bd_op->ioctl(inode, file, cmd, arg);
+		break;
+	}
+	return ret;
 }
 
 struct address_space_operations def_blk_aops = {

fs/cramfs/inode.c

@@ -134,7 +134,7 @@ static void *cramfs_read(struct super_block *sb, unsigned int offset, unsigned i
 		return read_buffers[i] + blk_offset;
 	}
 
-	devsize = blkdev_size_in_bytes(sb->s_dev) >> 12;
+	devsize = sb->s_bdev->bd_inode->i_size >> 12;
 	if (!devsize)
 		devsize = ~0UL;
 

fs/ext3/inode.c

@@ -1029,7 +1029,7 @@ static int ext3_prepare_write(struct file *file, struct page *page,
 		goto prepare_write_failed;
 
 	if (ext3_should_journal_data(inode)) {
-		ret = walk_page_buffers(handle, page->buffers,
+		ret = walk_page_buffers(handle, page_buffers(page),
 				from, to, NULL, do_journal_get_write_access);
 		if (ret) {
 			/*
@@ -1102,7 +1102,7 @@ static int ext3_commit_write(struct file *file, struct page *page,
 		int partial = 0;
 		loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to;
 
-		ret = walk_page_buffers(handle, page->buffers,
+		ret = walk_page_buffers(handle, page_buffers(page),
 			from, to, &partial, commit_write_fn);
 		if (!partial)
 			SetPageUptodate(page);
@@ -1112,7 +1112,7 @@ static int ext3_commit_write(struct file *file, struct page *page,
 		EXT3_I(inode)->i_state |= EXT3_STATE_JDATA;
 	} else {
 		if (ext3_should_order_data(inode)) {
-			ret = walk_page_buffers(handle, page->buffers,
+			ret = walk_page_buffers(handle, page_buffers(page),
 				from, to, NULL, journal_dirty_sync_data);
 		}
 		/* Be careful here if generic_commit_write becomes a
@@ -1252,7 +1252,7 @@ static int bget_one(handle_t *handle, struct buffer_head *bh)
 static int ext3_writepage(struct page *page)
 {
 	struct inode *inode = page->mapping->host;
-	struct buffer_head *page_buffers;
+	struct buffer_head *page_bufs;
 	handle_t *handle = NULL;
 	int ret = 0, err;
 	int needed;
@@ -1285,14 +1285,14 @@ static int ext3_writepage(struct page *page)
 
 	unlock_kernel();
 
-	page_buffers = NULL;	/* Purely to prevent compiler warning */
+	page_bufs = NULL;	/* Purely to prevent compiler warning */
 
 	/* bget() all the buffers */
 	if (order_data) {
-		if (!page->buffers)
+		if (!page_has_buffers(page))
 			create_empty_buffers(page, inode->i_sb->s_blocksize);
-		page_buffers = page->buffers;
-		walk_page_buffers(handle, page_buffers, 0,
+		page_bufs = page_buffers(page);
+		walk_page_buffers(handle, page_bufs, 0,
 				PAGE_CACHE_SIZE, NULL, bget_one);
 	}
 
@@ -1301,7 +1301,7 @@ static int ext3_writepage(struct page *page)
 	/*
 	 * The page can become unlocked at any point now, and
 	 * truncate can then come in and change things.  So we
-	 * can't touch *page from now on.  But *page_buffers is
+	 * can't touch *page from now on.  But *page_bufs is
 	 * safe due to elevated refcount.
 	 */
 
@@ -1310,7 +1310,7 @@ static int ext3_writepage(struct page *page)
 
 	/* And attach them to the current transaction */
 	if (order_data) {
-		err = walk_page_buffers(handle, page_buffers,
+		err = walk_page_buffers(handle, page_bufs,
 			0, PAGE_CACHE_SIZE, NULL, journal_dirty_async_data);
 		if (!ret)
 			ret = err;
@@ -1392,11 +1392,11 @@ static int ext3_block_truncate_page(handle_t *handle,
 	if (!page)
 		goto out;
 
-	if (!page->buffers)
+	if (!page_has_buffers(page))
 		create_empty_buffers(page, blocksize);
 
 	/* Find the buffer that contains "offset" */
-	bh = page->buffers;
+	bh = page_buffers(page);
 	pos = blocksize;
 	while (offset >= pos) {
 		bh = bh->b_this_page;

fs/hfs/file.c

@@ -166,7 +166,6 @@ static hfs_rwret_t hfs_file_read(struct file * filp, char * buf,
 	}
 	if ((read = hfs_do_read(inode, HFS_I(inode)->fork, pos, buf, left)) > 0) {
 	        *ppos += read;
-		filp->f_reada = 1;
 	}
 
 	return read;

fs/hfs/file_cap.c

@@ -105,7 +105,6 @@ static loff_t cap_info_llseek(struct file *file, loff_t offset, int origin)
 	if (offset>=0 && offset<=HFS_FORK_MAX) {
 		if (offset != file->f_pos) {
 			file->f_pos = offset;
-			file->f_reada = 0;
 		}
 		retval = offset;
 	}

fs/hfs/file_hdr.c

@@ -361,7 +361,6 @@ loff_t hdr_llseek(struct file *file, loff_t offset, int origin)
 	if (offset>=0 && offset<file->f_dentry->d_inode->i_size) {
 		if (offset != file->f_pos) {
 			file->f_pos = offset;
-			file->f_reada = 0;
 		}
 		retval = offset;
 	}
@@ -594,7 +593,6 @@ static hfs_rwret_t hdr_read(struct file * filp, char * buf,
 		} else if (fork) {
 			left = hfs_do_read(inode, fork, offset, buf, left);
 			if (left > 0) {
-				filp->f_reada = 1;
 			} else if (!read) {
 				return left;
 			} else {

fs/inode.c

@@ -143,6 +143,8 @@ void inode_init_once(struct inode *inode)
 	INIT_LIST_HEAD(&inode->i_dirty_data_buffers);
 	INIT_LIST_HEAD(&inode->i_devices);
 	sema_init(&inode->i_sem, 1);
+	INIT_RADIX_TREE(&inode->i_data.page_tree, GFP_ATOMIC);
+	rwlock_init(&inode->i_data.page_lock);
 	spin_lock_init(&inode->i_data.i_shared_lock);
 	INIT_LIST_HEAD(&inode->i_data.i_mmap);
 	INIT_LIST_HEAD(&inode->i_data.i_mmap_shared);
@@ -431,7 +433,7 @@ void sync_inodes(void)
 	}
 }
 
-static void try_to_sync_unused_inodes(void * arg)
+static void try_to_sync_unused_inodes(unsigned long pexclusive)
 {
 	struct super_block * sb;
 	int nr_inodes = inodes_stat.nr_unused;
@@ -448,10 +450,9 @@ static void try_to_sync_unused_inodes(void * arg)
 	}
 	spin_unlock(&sb_lock);
 	spin_unlock(&inode_lock);
+	clear_bit(0, (unsigned long *)pexclusive);
 }
 
-static struct tq_struct unused_inodes_flush_task;
-
 /**
  *	write_inode_now	-	write an inode to disk
  *	@inode: inode to write to disk
@@ -744,8 +745,15 @@ void prune_icache(int goal)
 	 * from here or we're either synchronously dogslow
 	 * or we deadlock with oom.
 	 */
-	if (goal)
-		schedule_task(&unused_inodes_flush_task);
+	if (goal) {
+		static unsigned long exclusive;
+
+		if (!test_and_set_bit(0, &exclusive)) {
+			if (pdflush_operation(try_to_sync_unused_inodes,
+						(unsigned long)&exclusive))
+				clear_bit(0, &exclusive);
+		}
+	}
 }
 /*
  * This is called from kswapd when we think we need some
@@ -1171,8 +1179,6 @@ void __init inode_init(unsigned long mempages)
 					 NULL);
 	if (!inode_cachep)
 		panic("cannot create inode slab cache");
-
-	unused_inodes_flush_task.routine = try_to_sync_unused_inodes;
 }
 
 static inline void do_atime_update(struct inode *inode)

fs/intermezzo/vfs.c

@@ -1884,7 +1884,6 @@ static struct file *presto_filp_dopen(struct dentry *dentry, int flags)
 
         f->f_dentry = dentry;
         f->f_pos = 0;
-        f->f_reada = 0;
         f->f_op = NULL;
         if (inode->i_op)
                 /* XXX should we set to presto ops, or leave at cache ops? */

fs/jbd/transaction.c

@@ -1661,10 +1661,11 @@ int journal_try_to_free_buffers(journal_t *journal,
 	struct buffer_head *tmp;
 	int locked_or_dirty = 0;
 	int call_ttfb = 1;
+	int ret;
 
 	J_ASSERT(PageLocked(page));
 
-	bh = page->buffers;
+	bh = page_buffers(page);
 	tmp = bh;
 	spin_lock(&journal_datalist_lock);
 	do {
@@ -1688,7 +1689,10 @@ int journal_try_to_free_buffers(journal_t *journal,
 	 */
 	call_ttfb = 1;
 out:
-	return call_ttfb;
+	ret = 0;
+	if (call_ttfb)
+		ret = try_to_free_buffers(page, gfp_mask);
+	return ret;
 }
 
 /*
@@ -1881,7 +1885,7 @@ int journal_flushpage(journal_t *journal,
 		
 	if (!PageLocked(page))
 		BUG();
-	if (!page->buffers)
+	if (!page_has_buffers(page))
 		return 1;
 
 	/* We will potentially be playing with lists other than just the
@@ -1889,7 +1893,7 @@ int journal_flushpage(journal_t *journal,
 	 * cautious in our locking. */
 	lock_journal(journal);
 
-	head = bh = page->buffers;
+	head = bh = page_buffers(page);
 	do {
 		unsigned int next_off = curr_off + bh->b_size;
 		next = bh->b_this_page;
@@ -1911,7 +1915,7 @@ int journal_flushpage(journal_t *journal,
 	if (!offset) {
 		if (!may_free || !try_to_free_buffers(page, 0))
 			return 0;
-		J_ASSERT(page->buffers == NULL);
+		J_ASSERT(!page_has_buffers(page));
 	}
 	return 1;
 }

fs/jffs2/super.c

@@ -94,12 +94,12 @@ static struct super_operations jffs2_super_operations =
 
 static int jffs2_sb_compare(struct super_block *sb, void *data)
 {
-	struct mtd_info *mtd = data;
+	struct jffs2_sb_info *p = data;
 	struct jffs2_sb_info *c = JFFS2_SB_INFO(sb);
 
 	/* The superblocks are considered to be equivalent if the underlying MTD
 	   device is the same one */
-	if (c->mtd == mtd) {
+	if (c->mtd == p->mtd) {
 		D1(printk(KERN_DEBUG "jffs2_sb_compare: match on device %d (\"%s\")\n", mtd->index, mtd->name));
 		return 1;
 	} else {
@@ -111,12 +111,14 @@ static int jffs2_sb_compare(struct super_block *sb, void *data)
 
 static int jffs2_sb_set(struct super_block *sb, void *data)
 {
-	struct mtd_info *mtd = data;
+	struct jffs2_sb_info *p = data;
 
 	/* For persistence of NFS exports etc. we use the same s_dev
 	   each time we mount the device, don't just use an anonymous
 	   device */
-	sb->s_dev = mk_kdev(MTD_BLOCK_MAJOR, mtd->index);
+	sb->u.generic_sbp = p;
+	p->os_priv = sb;
+	sb->s_dev = mk_kdev(MTD_BLOCK_MAJOR, p->mtd->index);
 
 	return 0;
 }
@@ -129,7 +131,13 @@ static struct super_block *jffs2_get_sb_mtd(struct file_system_type *fs_type,
 	struct jffs2_sb_info *c;
 	int ret;
 
-	sb = sget(fs_type, jffs2_sb_compare, jffs2_sb_set, mtd);
+	c = kmalloc(sizeof(*c), GFP_KERNEL);
+	if (!c)
+		return ERR_PTR(-ENOMEM);
+	memset(c, 0, sizeof(*c));
+	c->mtd = mtd;
+
+	sb = sget(fs_type, jffs2_sb_compare, jffs2_sb_set, c);
 
 	if (IS_ERR(sb))
 		goto out_put;
@@ -144,19 +152,8 @@ static struct super_block *jffs2_get_sb_mtd(struct file_system_type *fs_type,
 	D1(printk(KERN_DEBUG "jffs2_get_sb_mtd(): New superblock for device %d (\"%s\")\n",
 		  mtd->index, mtd->name));
 
-	c = kmalloc(sizeof(*c), GFP_KERNEL);
-	if (!c) {
-		sb = ERR_PTR(-ENOMEM);
-		goto out_put;
-	}
-
-	sb->u.generic_sbp = c;
 	sb->s_op = &jffs2_super_operations;
 
-	memset(c, 0, sizeof(*c));
-	c->os_priv = sb;
-	c->mtd = mtd;
-
 	ret = jffs2_do_fill_super(sb, data, (flags&MS_VERBOSE)?1:0);
 
 	if (ret) {
@@ -164,13 +161,15 @@ static struct super_block *jffs2_get_sb_mtd(struct file_system_type *fs_type,
 		up_write(&sb->s_umount);
 		deactivate_super(sb);
 		sb = ERR_PTR(ret);
-		goto out_put;
+		goto out_put1;
 	}
 
 	sb->s_flags |= MS_ACTIVE;
 	return sb;
 
  out_put:
+	kfree(c);
+ out_put1:
 	put_mtd_device(mtd);
 
 	return sb;
@@ -288,18 +287,23 @@ void jffs2_put_super (struct super_block *sb)
 	kfree(c->blocks);
 	if (c->mtd->sync)
 		c->mtd->sync(c->mtd);
-	put_mtd_device(c->mtd);
-
-	kfree(c);
 
 	D1(printk(KERN_DEBUG "jffs2_put_super returning\n"));
 }
+
+static void jffs2_kill_sb(struct super_block *sb)
+{
+	struct jffs2_sb_info *c = JFFS2_SB_INFO(sb);
+	generic_shutdown_super(sb);
+	put_mtd_device(c->mtd);
+	kfree(c);
+}
  
 static struct file_system_type jffs2_fs_type = {
 	owner:		THIS_MODULE,
 	name:		"jffs2",
 	get_sb:		jffs2_get_sb,
-	kill_sb:	generic_shutdown_super
+	kill_sb:	jffs2_kill_sb,
 };
 
 

fs/nfsd/vfs.c

@@ -67,11 +67,7 @@ struct raparms {
 	unsigned int		p_count;
 	ino_t			p_ino;
 	kdev_t			p_dev;
-	unsigned long		p_reada,
-				p_ramax,
-				p_raend,
-				p_ralen,
-				p_rawin;
+	struct file_ra_state	p_ra;
 };
 
 static struct raparms *		raparml;
@@ -564,11 +560,7 @@ nfsd_get_raparms(kdev_t dev, ino_t ino)
 	ra = *frap;
 	ra->p_dev = dev;
 	ra->p_ino = ino;
-	ra->p_reada = 0;
-	ra->p_ramax = 0;
-	ra->p_raend = 0;
-	ra->p_ralen = 0;
-	ra->p_rawin = 0;
+	memset(&ra->p_ra, 0, sizeof(ra->p_ra));
 found:
 	if (rap != &raparm_cache) {
 		*rap = ra->p_next;
@@ -611,31 +603,18 @@ nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t offset,
 
 	/* Get readahead parameters */
 	ra = nfsd_get_raparms(inode->i_dev, inode->i_ino);
-	if (ra) {
-		file.f_reada = ra->p_reada;
-		file.f_ramax = ra->p_ramax;
-		file.f_raend = ra->p_raend;
-		file.f_ralen = ra->p_ralen;
-		file.f_rawin = ra->p_rawin;
-	}
+	if (ra)
+		file.f_ra = ra->p_ra;
 	file.f_pos = offset;
 
-	oldfs = get_fs(); set_fs(KERNEL_DS);
+	oldfs = get_fs();
+	set_fs(KERNEL_DS);
 	err = file.f_op->read(&file, buf, *count, &file.f_pos);
 	set_fs(oldfs);
 
 	/* Write back readahead params */
-	if (ra != NULL) {
-		dprintk("nfsd: raparms %ld %ld %ld %ld %ld\n",
-			file.f_reada, file.f_ramax, file.f_raend,
-			file.f_ralen, file.f_rawin);
-		ra->p_reada = file.f_reada;
-		ra->p_ramax = file.f_ramax;
-		ra->p_raend = file.f_raend;
-		ra->p_ralen = file.f_ralen;
-		ra->p_rawin = file.f_rawin;
-		ra->p_count -= 1;
-	}
+	if (ra)
+		ra->p_ra = file.f_ra;
 
 	if (err >= 0) {
 		nfsdstats.io_read += err;

fs/open.c

@@ -635,7 +635,6 @@ struct file *dentry_open(struct dentry *dentry, struct vfsmount *mnt, int flags)
 	f->f_dentry = dentry;
 	f->f_vfsmnt = mnt;
 	f->f_pos = 0;
-	f->f_reada = 0;
 	f->f_op = fops_get(inode->i_fop);
 	file_move(f, &inode->i_sb->s_files);
 
@@ -686,7 +685,7 @@ int get_unused_fd(void)
 	write_lock(&files->file_lock);
 
 repeat:
- 	fd = find_next_zero_bit(files->open_fds, 
+ 	fd = find_next_zero_bit(files->open_fds->fds_bits, 
 				files->max_fdset, 
 				files->next_fd);
 

fs/read_write.c

@@ -37,7 +37,6 @@ loff_t generic_file_llseek(struct file *file, loff_t offset, int origin)
 	if (offset>=0 && offset<=inode->i_sb->s_maxbytes) {
 		if (offset != file->f_pos) {
 			file->f_pos = offset;
-			file->f_reada = 0;
 			file->f_version = ++event;
 		}
 		retval = offset;
@@ -62,7 +61,6 @@ loff_t remote_llseek(struct file *file, loff_t offset, int origin)
 	if (offset>=0 && offset<=file->f_dentry->d_inode->i_sb->s_maxbytes) {
 		if (offset != file->f_pos) {
 			file->f_pos = offset;
-			file->f_reada = 0;
 			file->f_version = ++event;
 		}
 		retval = offset;
@@ -92,7 +90,6 @@ loff_t default_llseek(struct file *file, loff_t offset, int origin)
 	if (offset >= 0) {
 		if (offset != file->f_pos) {
 			file->f_pos = offset;
-			file->f_reada = 0;
 			file->f_version = ++event;
 		}
 		retval = offset;

fs/reiserfs/inode.c

@@ -146,8 +146,8 @@ static void add_to_flushlist(struct inode *inode, struct buffer_head *bh) {
 static inline void fix_tail_page_for_writing(struct page *page) {
     struct buffer_head *head, *next, *bh ;
 
-    if (page && page->buffers) {
-	head = page->buffers ;
+    if (page && page_has_buffers(page)) {
+	head = page_buffers(page) ;
 	bh = head ;
 	do {
 	    next = bh->b_this_page ;
@@ -1685,7 +1685,7 @@ static int grab_tail_page(struct inode *p_s_inode,
 
     kunmap(page) ; /* mapped by block_prepare_write */
 
-    head = page->buffers ;      
+    head = page_buffers(page) ;      
     bh = head;
     do {
 	if (pos >= start) {
@@ -1930,7 +1930,7 @@ static int reiserfs_write_full_page(struct page *page) {
     struct buffer_head *arr[PAGE_CACHE_SIZE/512] ;
     int nr = 0 ;
 
-    if (!page->buffers) {
+    if (!page_has_buffers(page)) {
         block_prepare_write(page, 0, 0, NULL) ;
 	kunmap(page) ;
     }
@@ -1948,7 +1948,7 @@ static int reiserfs_write_full_page(struct page *page) {
 	flush_dcache_page(page) ;
 	kunmap(page) ;
     }
-    head = page->buffers ;
+    head = page_buffers(page) ;
     bh = head ;
     block = page->index << (PAGE_CACHE_SHIFT - inode->i_sb->s_blocksize_bits) ;
     do {

fs/reiserfs/tail_conversion.c

@@ -156,10 +156,10 @@ unmap_buffers(struct page *page, loff_t pos) {
   unsigned long cur_index ;
 
   if (page) {
-    if (page->buffers) {
+    if (page_has_buffers(page)) {
       tail_index = pos & (PAGE_CACHE_SIZE - 1) ;
       cur_index = 0 ;
-      head = page->buffers ;
+      head = page_buffers(page) ;
       bh = head ;
       do {
 	next = bh->b_this_page ;

fs/super.c

@@ -184,7 +184,7 @@ static void remove_super(struct super_block *s)
 	up_write(&s->s_umount);
 }
 
-static void generic_shutdown_super(struct super_block *sb)
+void generic_shutdown_super(struct super_block *sb)
 {
 	struct dentry *root = sb->s_root;
 	struct super_operations *sop = sb->s_op;

include/asm-i386/bitops.h

@@ -469,18 +469,23 @@ static __inline__ int ffs(int x)
 
 #ifdef __KERNEL__
 
-#define ext2_set_bit                 __test_and_set_bit
-#define ext2_clear_bit               __test_and_clear_bit
-#define ext2_test_bit                test_bit
-#define ext2_find_first_zero_bit     find_first_zero_bit
-#define ext2_find_next_zero_bit      find_next_zero_bit
+#define ext2_set_bit(nr,addr) \
+	__test_and_set_bit((nr),(unsigned long*)addr)
+#define ext2_clear_bit(nr, addr) \
+	__test_and_clear_bit((nr),(unsigned long*)addr)
+#define ext2_test_bit(nr, addr)      test_bit((nr),(unsigned long*)addr)
+#define ext2_find_first_zero_bit(addr, size) \
+	find_first_zero_bit((unsigned long*)addr, size)
+#define ext2_find_next_zero_bit(addr, size, off) \
+	find_next_zero_bit((unsigned long*)addr, size, off)
 
 /* Bitmap functions for the minix filesystem.  */
-#define minix_test_and_set_bit(nr,addr) __test_and_set_bit(nr,addr)
-#define minix_set_bit(nr,addr) __set_bit(nr,addr)
-#define minix_test_and_clear_bit(nr,addr) __test_and_clear_bit(nr,addr)
-#define minix_test_bit(nr,addr) test_bit(nr,addr)
-#define minix_find_first_zero_bit(addr,size) find_first_zero_bit(addr,size)
+#define minix_test_and_set_bit(nr,addr) __test_and_set_bit(nr,(void*)addr)
+#define minix_set_bit(nr,addr) __set_bit(nr,(void*)addr)
+#define minix_test_and_clear_bit(nr,addr) __test_and_clear_bit(nr,(void*)addr)
+#define minix_test_bit(nr,addr) test_bit(nr,(void*)addr)
+#define minix_find_first_zero_bit(addr,size) \
+	find_first_zero_bit((void*)addr,size)
 
 #endif /* __KERNEL__ */
 

include/asm-i386/pgtable.h

@@ -223,10 +223,10 @@ static inline pte_t pte_mkdirty(pte_t pte)	{ (pte).pte_low |= _PAGE_DIRTY; retur
 static inline pte_t pte_mkyoung(pte_t pte)	{ (pte).pte_low |= _PAGE_ACCESSED; return pte; }
 static inline pte_t pte_mkwrite(pte_t pte)	{ (pte).pte_low |= _PAGE_RW; return pte; }
 
-static inline  int ptep_test_and_clear_dirty(pte_t *ptep)	{ return test_and_clear_bit(_PAGE_BIT_DIRTY, ptep); }
-static inline  int ptep_test_and_clear_young(pte_t *ptep)	{ return test_and_clear_bit(_PAGE_BIT_ACCESSED, ptep); }
-static inline void ptep_set_wrprotect(pte_t *ptep)		{ clear_bit(_PAGE_BIT_RW, ptep); }
-static inline void ptep_mkdirty(pte_t *ptep)			{ set_bit(_PAGE_BIT_DIRTY, ptep); }
+static inline  int ptep_test_and_clear_dirty(pte_t *ptep)	{ return test_and_clear_bit(_PAGE_BIT_DIRTY, &ptep->pte_low); }
+static inline  int ptep_test_and_clear_young(pte_t *ptep)	{ return test_and_clear_bit(_PAGE_BIT_ACCESSED, &ptep->pte_low); }
+static inline void ptep_set_wrprotect(pte_t *ptep)		{ clear_bit(_PAGE_BIT_RW, &ptep->pte_low); }
+static inline void ptep_mkdirty(pte_t *ptep)			{ set_bit(_PAGE_BIT_DIRTY, &ptep->pte_low); }
 
 /*
  * Conversion functions: convert a page and protection to a page entry,

include/asm-i386/thread_info.h

@@ -23,7 +23,7 @@
 struct thread_info {
 	struct task_struct	*task;		/* main task structure */
 	struct exec_domain	*exec_domain;	/* execution domain */
-	__u32			flags;		/* low level flags */
+	unsigned long		flags;		/* low level flags */
 	__u32			cpu;		/* current CPU */
 	__s32			preempt_count; /* 0 => preemptable, <0 => BUG */
 

include/asm-ppc/bitops.h

@@ -394,8 +394,8 @@ static __inline__ unsigned long find_next_zero_bit(void * addr,
 
 #ifdef __KERNEL__
 
-#define ext2_set_bit(nr, addr)		__test_and_set_bit((nr) ^ 0x18, addr)
-#define ext2_clear_bit(nr, addr)	__test_and_clear_bit((nr) ^ 0x18, addr)
+#define ext2_set_bit(nr, addr)	__test_and_set_bit((nr) ^ 0x18, (unsigned long *)(addr))
+#define ext2_clear_bit(nr, addr) __test_and_clear_bit((nr) ^ 0x18, (unsigned long *)(addr))
 
 static __inline__ int ext2_test_bit(int nr, __const__ void * addr)
 {

include/asm-s390/pgtable.h

@@ -429,7 +429,7 @@ extern inline pte_t mk_pte_phys(unsigned long physpage, pgprot_t pgprot)
 	                                                                  \
 	if (__page != ZERO_PAGE(__physpage)) {                            \
 		int __users = page_count(__page);                         \
-		__users -= !!__page->buffers + !!__page->mapping;         \
+		__users -= !!PagePrivate(__page) + !!__page->mapping;     \
 	                                                                  \
 		if (__users == 1)                                         \
 			pte_val(__pte) |= _PAGE_MKCLEAR;                  \

include/asm-s390x/pgtable.h

@@ -448,7 +448,7 @@ extern inline pte_t mk_pte_phys(unsigned long physpage, pgprot_t pgprot)
 	                                                                  \
 	if (__page != ZERO_PAGE(__physpage)) {                            \
 		int __users = page_count(__page);                         \
-		__users -= !!__page->buffers + !!__page->mapping;         \
+		__users -= !!PagePrivate(page) + !!__page->mapping;       \
 	                                                                  \
 		if (__users == 1)                                         \
 			pte_val(__pte) |= _PAGE_MKCLEAR;                  \

include/linux/blkdev.h

@@ -152,6 +152,12 @@ struct request_queue
 	make_request_fn		*make_request_fn;
 	prep_rq_fn		*prep_rq_fn;
 
+	/*
+	 * The VM-level readahead tunable for this device.  In
+	 * units of 512-byte sectors.
+	 */
+	unsigned ra_sectors;
+
 	/*
 	 * The queue owner gets to use this for whatever they like.
 	 * ll_rw_blk doesn't touch it.
@@ -308,6 +314,8 @@ extern void blk_queue_hardsect_size(request_queue_t *q, unsigned short);
 extern void blk_queue_segment_boundary(request_queue_t *q, unsigned long);
 extern void blk_queue_assign_lock(request_queue_t *q, spinlock_t *);
 extern void blk_queue_prep_rq(request_queue_t *q, prep_rq_fn *pfn);
+extern int blk_set_readahead(kdev_t dev, unsigned sectors);
+extern unsigned blk_get_readahead(kdev_t dev);
 
 extern int blk_rq_map_sg(request_queue_t *, struct request *, struct scatterlist *);
 extern void blk_dump_rq_flags(struct request *, char *);
@@ -322,10 +330,6 @@ extern int * blksize_size[MAX_BLKDEV];
 
 #define MAX_SEGMENT_SIZE	65536
 
-/* read-ahead in pages.. */
-#define MAX_READAHEAD	31
-#define MIN_READAHEAD	3
-
 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
 
 extern void drive_stat_acct(struct request *, int, int);

include/linux/fs.h

@@ -21,6 +21,7 @@
 #include <linux/cache.h>
 #include <linux/stddef.h>
 #include <linux/string.h>
+#include <linux/radix-tree.h>
 
 #include <asm/atomic.h>
 #include <asm/bitops.h>
@@ -172,12 +173,10 @@ extern int leases_enable, dir_notify_enable, lease_break_time;
 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
-#if 0				/* Obsolete, these don't do anything. */
 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
-#endif
 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
@@ -286,6 +285,24 @@ extern void set_bh_page(struct buffer_head *bh, struct page *page, unsigned long
 
 #define touch_buffer(bh)	mark_page_accessed(bh->b_page)
 
+/* If we *know* page->private refers to buffer_heads */
+#define page_buffers(page)					\
+	({							\
+		if (!PagePrivate(page))				\
+			BUG();					\
+		((struct buffer_head *)(page)->private);	\
+	})
+#define page_has_buffers(page)	PagePrivate(page)
+#define set_page_buffers(page, buffers)				\
+	do {							\
+		SetPagePrivate(page);				\
+		page->private = (unsigned long)buffers;		\
+	} while (0)
+#define clear_page_buffers(page)				\
+	do {							\
+		ClearPagePrivate(page);				\
+		page->private = 0;				\
+	} while (0)
 
 #include <linux/pipe_fs_i.h>
 /* #include <linux/umsdos_fs_i.h> */
@@ -370,6 +387,8 @@ struct address_space_operations {
 };
 
 struct address_space {
+	struct radix_tree_root	page_tree;	/* radix tree of all pages */
+	rwlock_t		page_lock;	/* and rwlock protecting it */
 	struct list_head	clean_pages;	/* list of clean pages */
 	struct list_head	dirty_pages;	/* list of dirty pages */
 	struct list_head	locked_pages;	/* list of locked pages */
@@ -484,6 +503,18 @@ struct fown_struct {
 	int signum;		/* posix.1b rt signal to be delivered on IO */
 };
 
+/*
+ * Track a single file's readahead state
+ */
+struct file_ra_state {
+	unsigned long start;		/* Current window */
+	unsigned long size;
+	unsigned long next_size;	/* Next window size */
+	unsigned long prev_page;	/* Cache last read() position */
+	unsigned long ahead_start;	/* Ahead window */
+	unsigned long ahead_size;
+};
+
 struct file {
 	struct list_head	f_list;
 	struct dentry		*f_dentry;
@@ -493,10 +524,10 @@ struct file {
 	unsigned int 		f_flags;
 	mode_t			f_mode;
 	loff_t			f_pos;
-	unsigned long 		f_reada, f_ramax, f_raend, f_ralen, f_rawin;
 	struct fown_struct	f_owner;
 	unsigned int		f_uid, f_gid;
 	int			f_error;
+	struct file_ra_state	f_ra;
 
 	unsigned long		f_version;
 
@@ -925,6 +956,7 @@ struct super_block *get_sb_single(struct file_system_type *fs_type,
 struct super_block *get_sb_nodev(struct file_system_type *fs_type,
 	int flags, void *data,
 	int (*fill_super)(struct super_block *, void *, int));
+void generic_shutdown_super(struct super_block *sb);
 void kill_block_super(struct super_block *sb);
 void kill_anon_super(struct super_block *sb);
 void kill_litter_super(struct super_block *sb);

include/linux/hfs_sysdep.h

@@ -200,16 +200,16 @@ static inline void *hfs_buffer_data(const hfs_buffer buffer) {
 #endif
 
 static inline int hfs_clear_bit(int bitnr, hfs_u32 *lword) {
-	return test_and_clear_bit(BITNR(bitnr), lword);
+	return test_and_clear_bit(BITNR(bitnr), (unsigned long *)lword);
 }
 
 static inline int hfs_set_bit(int bitnr, hfs_u32 *lword) {
-	return test_and_set_bit(BITNR(bitnr), lword);
+	return test_and_set_bit(BITNR(bitnr), (unsigned long *)lword);
 }
 
 static inline int hfs_test_bit(int bitnr, const hfs_u32 *lword) {
 	/* the kernel should declare the second arg of test_bit as const */
-	return test_bit(BITNR(bitnr), (void *)lword);
+	return test_bit(BITNR(bitnr), (unsigned long *)lword);
 }
 
 #undef BITNR

include/linux/mm.h

@@ -149,15 +149,12 @@ typedef struct page {
 	struct list_head list;		/* ->mapping has some page lists. */
 	struct address_space *mapping;	/* The inode (or ...) we belong to. */
 	unsigned long index;		/* Our offset within mapping. */
-	struct page *next_hash;		/* Next page sharing our hash bucket in
-					   the pagecache hash table. */
 	atomic_t count;			/* Usage count, see below. */
 	unsigned long flags;		/* atomic flags, some possibly
 					   updated asynchronously */
 	struct list_head lru;		/* Pageout list, eg. active_list;
 					   protected by pagemap_lru_lock !! */
-	struct page **pprev_hash;	/* Complement to *next_hash. */
-	struct buffer_head * buffers;	/* Buffer maps us to a disk block. */
+	unsigned long private;		/* fs-private opaque data */
 
 	/*
 	 * On machines where all RAM is mapped into kernel address space,
@@ -180,7 +177,7 @@ typedef struct page {
  *
  * What counts for a page usage:
  * - cache mapping   (page->mapping)
- * - disk mapping    (page->buffers)
+ * - private data    (page->private)
  * - page mapped in a task's page tables, each mapping
  *   is counted separately
  *
@@ -223,22 +220,23 @@ typedef struct page {
  * page->mapping is the pointer to the inode, and page->index is the
  * file offset of the page, in units of PAGE_CACHE_SIZE.
  *
- * A page may have buffers allocated to it. In this case,
- * page->buffers is a circular list of these buffer heads. Else,
- * page->buffers == NULL.
+ * A page contains an opaque `private' member, which belongs to the
+ * page's address_space.  Usually, this is the address of a circular
+ * list of the page's disk buffers.
  *
+ * The PG_private bitflag is set if page->private contains a valid
+ * value.
  * For pages belonging to inodes, the page->count is the number of
- * attaches, plus 1 if buffers are allocated to the page, plus one
- * for the page cache itself.
+ * attaches, plus 1 if `private' contains something, plus one for
+ * the page cache itself.
  *
  * All pages belonging to an inode are in these doubly linked lists:
  * mapping->clean_pages, mapping->dirty_pages and mapping->locked_pages;
  * using the page->list list_head. These fields are also used for
  * freelist managemet (when page->count==0).
  *
- * There is also a hash table mapping (mapping,index) to the page
- * in memory if present. The lists for this hash table use the fields
- * page->next_hash and page->pprev_hash.
+ * There is also a per-mapping radix tree mapping index to the page
+ * in memory if present. The tree is rooted at mapping->root.  
  *
  * All process pages can do I/O:
  * - inode pages may need to be read from disk,
@@ -294,6 +292,8 @@ typedef struct page {
 #define PG_reserved		14
 #define PG_launder		15	/* written out by VM pressure.. */
 
+#define PG_private		16	/* Has something at ->private */
+
 /* Make it prettier to test the above... */
 #define UnlockPage(page)	unlock_page(page)
 #define Page_Uptodate(page)	test_bit(PG_uptodate, &(page)->flags)
@@ -310,6 +310,9 @@ typedef struct page {
 #define PageLaunder(page)	test_bit(PG_launder, &(page)->flags)
 #define SetPageLaunder(page)	set_bit(PG_launder, &(page)->flags)
 #define __SetPageReserved(page)	__set_bit(PG_reserved, &(page)->flags)
+#define SetPagePrivate(page)	set_bit(PG_private, &(page)->flags)
+#define ClearPagePrivate(page)	clear_bit(PG_private, &(page)->flags)
+#define PagePrivate(page)	test_bit(PG_private, &(page)->flags)
 
 /*
  * The zone field is never updated after free_area_init_core()
@@ -470,7 +473,7 @@ extern struct address_space swapper_space;
 
 static inline int is_page_cache_freeable(struct page * page)
 {
-	return page_count(page) - !!page->buffers == 1;
+	return page_count(page) - !!PagePrivate(page) == 1;
 }
 
 extern int can_share_swap_page(struct page *);
@@ -535,6 +538,13 @@ extern void truncate_inode_pages(struct address_space *, loff_t);
 extern int filemap_sync(struct vm_area_struct *, unsigned long,	size_t, unsigned int);
 extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int);
 
+/* readahead.c */
+void do_page_cache_readahead(struct file *file,
+			unsigned long offset, unsigned long nr_to_read);
+void page_cache_readahead(struct file *file, unsigned long offset);
+void page_cache_readaround(struct file *file, unsigned long offset);
+void handle_ra_thrashing(struct file *file);
+
 /* vma is the first one with  address < vma->vm_end,
  * and even  address < vma->vm_start. Have to extend vma. */
 static inline int expand_stack(struct vm_area_struct * vma, unsigned long address)
@@ -579,6 +589,9 @@ static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * m
 
 extern struct vm_area_struct *find_extend_vma(struct mm_struct *mm, unsigned long addr);
 
+extern int pdflush_operation(void (*fn)(unsigned long), unsigned long arg0);
+extern int pdflush_flush(unsigned long nr_pages);
+
 extern struct page * vmalloc_to_page(void *addr);
 
 #endif /* __KERNEL__ */

include/linux/pagemap.h

@@ -41,53 +41,39 @@ static inline struct page *page_cache_alloc(struct address_space *x)
  */
 #define page_cache_entry(x)	virt_to_page(x)
 
-extern unsigned int page_hash_bits;
-#define PAGE_HASH_BITS (page_hash_bits)
-#define PAGE_HASH_SIZE (1 << PAGE_HASH_BITS)
-
-extern atomic_t page_cache_size; /* # of pages currently in the hash table */
-extern struct page **page_hash_table;
-
-extern void page_cache_init(unsigned long);
+extern atomic_t page_cache_size; /* # of pages currently in the page cache */
+
+extern struct page * find_get_page(struct address_space *mapping,
+				unsigned long index);
+extern struct page * find_lock_page(struct address_space *mapping,
+				unsigned long index);
+extern struct page * find_trylock_page(struct address_space *mapping,
+				unsigned long index);
+extern struct page * find_or_create_page(struct address_space *mapping,
+				unsigned long index, unsigned int gfp_mask);
 
-/*
- * We use a power-of-two hash table to avoid a modulus,
- * and get a reasonable hash by knowing roughly how the
- * inode pointer and indexes are distributed (ie, we
- * roughly know which bits are "significant")
- *
- * For the time being it will work for struct address_space too (most of
- * them sitting inside the inodes). We might want to change it later.
- */
-static inline unsigned long _page_hashfn(struct address_space * mapping, unsigned long index)
+extern struct page * grab_cache_page(struct address_space *mapping,
+				unsigned long index);
+extern struct page * grab_cache_page_nowait(struct address_space *mapping,
+				unsigned long index);
+
+extern int add_to_page_cache(struct page *page,
+		struct address_space *mapping, unsigned long index);
+extern int add_to_page_cache_unique(struct page *page,
+		struct address_space *mapping, unsigned long index);
+static inline void ___add_to_page_cache(struct page *page,
+		struct address_space *mapping, unsigned long index)
 {
-#define i (((unsigned long) mapping)/(sizeof(struct inode) & ~ (sizeof(struct inode) - 1)))
-#define s(x) ((x)+((x)>>PAGE_HASH_BITS))
-	return s(i+index) & (PAGE_HASH_SIZE-1);
-#undef i
-#undef s
-}
-
-#define page_hash(mapping,index) (page_hash_table+_page_hashfn(mapping,index))
+	list_add(&page->list, &mapping->clean_pages);
+	page->mapping = mapping;
+	page->index = index;
 
-extern struct page * __find_get_page(struct address_space *mapping,
-				unsigned long index, struct page **hash);
-#define find_get_page(mapping, index) \
-	__find_get_page(mapping, index, page_hash(mapping, index))
-extern struct page * __find_lock_page (struct address_space * mapping,
-				unsigned long index, struct page **hash);
-extern struct page * find_or_create_page(struct address_space *mapping,
-				unsigned long index, unsigned int gfp_mask);
+	mapping->nrpages++;
+	atomic_inc(&page_cache_size);
+}
 
 extern void FASTCALL(lock_page(struct page *page));
 extern void FASTCALL(unlock_page(struct page *page));
-#define find_lock_page(mapping, index) \
-	__find_lock_page(mapping, index, page_hash(mapping, index))
-extern struct page *find_trylock_page(struct address_space *, unsigned long);
-
-extern void add_to_page_cache(struct page * page, struct address_space *mapping, unsigned long index);
-extern void add_to_page_cache_locked(struct page * page, struct address_space *mapping, unsigned long index);
-extern int add_to_page_cache_unique(struct page * page, struct address_space *mapping, unsigned long index, struct page **hash);
 
 extern void ___wait_on_page(struct page *);
 
@@ -99,9 +85,6 @@ static inline void wait_on_page(struct page * page)
 
 extern void wake_up_page(struct page *);
 
-extern struct page * grab_cache_page (struct address_space *, unsigned long);
-extern struct page * grab_cache_page_nowait (struct address_space *, unsigned long);
-
 typedef int filler_t(void *, struct page*);
 
 extern struct page *read_cache_page(struct address_space *, unsigned long,

include/linux/radix-tree.h

+/*
+ * Copyright (C) 2001 Momchil Velikov
+ * Portions Copyright (C) 2001 Christoph Hellwig
+ *
+ * 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, 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., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+#ifndef _LINUX_RADIX_TREE_H
+#define _LINUX_RADIX_TREE_H
+
+struct radix_tree_node;
+
+#define RADIX_TREE_SLOT_RESERVED ((void *)~0UL)
+
+struct radix_tree_root {
+	unsigned int		height;
+	int			gfp_mask;
+	struct radix_tree_node	*rnode;
+};
+
+#define RADIX_TREE_INIT(mask)	{0, (mask), NULL}
+
+#define RADIX_TREE(name, mask) \
+	struct radix_tree_root name = RADIX_TREE_INIT(mask)
+
+#define INIT_RADIX_TREE(root, mask)	\
+do {					\
+	(root)->height = 0;		\
+	(root)->gfp_mask = (mask);	\
+	(root)->rnode = NULL;		\
+} while (0)
+
+extern int radix_tree_reserve(struct radix_tree_root *, unsigned long, void ***);
+extern int radix_tree_insert(struct radix_tree_root *, unsigned long, void *);
+extern void *radix_tree_lookup(struct radix_tree_root *, unsigned long);
+extern int radix_tree_delete(struct radix_tree_root *, unsigned long);
+
+#endif /* _LINUX_RADIX_TREE_H */

include/linux/raid/md_k.h

@@ -91,7 +91,6 @@ static inline mddev_t * kdev_to_mddev (kdev_t dev)
 /*
  * default readahead
  */
-#define MD_READAHEAD	MAX_READAHEAD
 
 static inline int disk_faulty(mdp_disk_t * d)
 {

include/linux/sched.h

@@ -368,6 +368,7 @@ do { if (atomic_dec_and_test(&(tsk)->usage)) __put_task_struct(tsk); } while(0)
 #define PF_MEMDIE	0x00001000	/* Killed for out-of-memory */
 #define PF_FREE_PAGES	0x00002000	/* per process page freeing */
 #define PF_NOIO		0x00004000	/* avoid generating further I/O */
+#define PF_FLUSHER	0x00008000	/* responsible for disk writeback */
 
 /*
  * Ptrace flags

include/linux/swap.h

@@ -109,7 +109,7 @@ extern void __remove_inode_page(struct page *);
 struct task_struct;
 struct vm_area_struct;
 struct sysinfo;
-
+struct address_space;
 struct zone_t;
 
 /* linux/mm/swap.c */
@@ -139,6 +139,9 @@ extern void show_swap_cache_info(void);
 extern int add_to_swap_cache(struct page *, swp_entry_t);
 extern void __delete_from_swap_cache(struct page *page);
 extern void delete_from_swap_cache(struct page *page);
+extern int move_to_swap_cache(struct page *page, swp_entry_t entry);
+extern int move_from_swap_cache(struct page *page, unsigned long index,
+		struct address_space *mapping);
 extern void free_page_and_swap_cache(struct page *page);
 extern struct page * lookup_swap_cache(swp_entry_t);
 extern struct page * read_swap_cache_async(swp_entry_t);

init/main.c

@@ -69,6 +69,7 @@ extern void sbus_init(void);
 extern void sysctl_init(void);
 extern void signals_init(void);
 
+extern void radix_tree_init(void);
 extern void free_initmem(void);
 
 #ifdef CONFIG_TC
@@ -392,7 +393,7 @@ asmlinkage void __init start_kernel(void)
 	proc_caches_init();
 	vfs_caches_init(mempages);
 	buffer_init(mempages);
-	page_cache_init(mempages);
+	radix_tree_init();
 #if defined(CONFIG_ARCH_S390)
 	ccwcache_init();
 #endif

kernel/ksyms.c

@@ -224,8 +224,6 @@ EXPORT_SYMBOL(generic_file_write);
 EXPORT_SYMBOL(generic_file_mmap);
 EXPORT_SYMBOL(generic_ro_fops);
 EXPORT_SYMBOL(generic_buffer_fdatasync);
-EXPORT_SYMBOL(page_hash_bits);
-EXPORT_SYMBOL(page_hash_table);
 EXPORT_SYMBOL(file_lock_list);
 EXPORT_SYMBOL(locks_init_lock);
 EXPORT_SYMBOL(locks_copy_lock);
@@ -266,8 +264,8 @@ EXPORT_SYMBOL(no_llseek);
 EXPORT_SYMBOL(__pollwait);
 EXPORT_SYMBOL(poll_freewait);
 EXPORT_SYMBOL(ROOT_DEV);
-EXPORT_SYMBOL(__find_get_page);
-EXPORT_SYMBOL(__find_lock_page);
+EXPORT_SYMBOL(find_get_page);
+EXPORT_SYMBOL(find_lock_page);
 EXPORT_SYMBOL(grab_cache_page);
 EXPORT_SYMBOL(grab_cache_page_nowait);
 EXPORT_SYMBOL(read_cache_page);
@@ -295,6 +293,7 @@ EXPORT_SYMBOL(get_sb_nodev);
 EXPORT_SYMBOL(get_sb_single);
 EXPORT_SYMBOL(kill_anon_super);
 EXPORT_SYMBOL(kill_litter_super);
+EXPORT_SYMBOL(generic_shutdown_super);
 EXPORT_SYMBOL(deactivate_super);
 EXPORT_SYMBOL(sget);
 EXPORT_SYMBOL(set_anon_super);

lib/Makefile

@@ -8,9 +8,11 @@
 
 L_TARGET := lib.a
 
-export-objs := cmdline.o dec_and_lock.o rwsem-spinlock.o rwsem.o crc32.o rbtree.o
+export-objs := cmdline.o dec_and_lock.o rwsem-spinlock.o rwsem.o \
+	       crc32.o rbtree.o radix-tree.o
 
-obj-y := errno.o ctype.o string.o vsprintf.o brlock.o cmdline.o bust_spinlocks.o rbtree.o
+obj-y := errno.o ctype.o string.o vsprintf.o brlock.o cmdline.o \
+	 bust_spinlocks.o rbtree.o radix-tree.o
 
 obj-$(CONFIG_RWSEM_GENERIC_SPINLOCK) += rwsem-spinlock.o
 obj-$(CONFIG_RWSEM_XCHGADD_ALGORITHM) += rwsem.o

lib/radix-tree.c

+/*
+ * Copyright (C) 2001 Momchil Velikov
+ * Portions Copyright (C) 2001 Christoph Hellwig
+ *
+ * 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, 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., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <linux/errno.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/mempool.h>
+#include <linux/module.h>
+#include <linux/radix-tree.h>
+#include <linux/slab.h>
+
+/*
+ * Radix tree node definition.
+ */
+#define RADIX_TREE_MAP_SHIFT  7
+#define RADIX_TREE_MAP_SIZE  (1UL << RADIX_TREE_MAP_SHIFT)
+#define RADIX_TREE_MAP_MASK  (RADIX_TREE_MAP_SIZE-1)
+
+struct radix_tree_node {
+	unsigned int	count;
+	void		*slots[RADIX_TREE_MAP_SIZE];
+};
+
+struct radix_tree_path {
+	struct radix_tree_node *node, **slot;
+};
+
+#define RADIX_TREE_INDEX_BITS  (8 /* CHAR_BIT */ * sizeof(unsigned long))
+
+/*
+ * Radix tree node cache.
+ */
+static kmem_cache_t *radix_tree_node_cachep;
+static mempool_t *radix_tree_node_pool;
+
+#define radix_tree_node_alloc(root) \
+	mempool_alloc(radix_tree_node_pool, (root)->gfp_mask)
+#define radix_tree_node_free(node) \
+	mempool_free((node), radix_tree_node_pool);
+
+
+/*
+ *	Return the maximum key which can be store into a
+ *	radix tree with height HEIGHT.
+ */
+static inline unsigned long radix_tree_maxindex(unsigned int height)
+{
+	unsigned int tmp = height * RADIX_TREE_MAP_SHIFT;
+	unsigned long index = (~0UL >> (RADIX_TREE_INDEX_BITS - tmp - 1)) >> 1;
+
+	if (tmp >= RADIX_TREE_INDEX_BITS)
+		index = ~0UL;
+	return index;
+}
+
+
+/*
+ *	Extend a radix tree so it can store key @index.
+ */
+static int radix_tree_extend(struct radix_tree_root *root, unsigned long index)
+{
+	struct radix_tree_node *node;
+	unsigned int height;
+
+	/* Figure out what the height should be.  */
+	height = root->height + 1;
+	while (index > radix_tree_maxindex(height))
+		height++;
+
+	if (root->rnode) {
+		do {
+			if (!(node = radix_tree_node_alloc(root)))
+				return -ENOMEM;
+
+			/* Increase the height.  */
+			node->slots[0] = root->rnode;
+			if (root->rnode)
+				node->count = 1;
+			root->rnode = node;
+			root->height++;
+		} while (height > root->height);
+	} else 
+		root->height = height;
+
+	return 0;
+}
+
+
+/**
+ *	radix_tree_reserve    -    reserve space in a radix tree
+ *	@root:		radix tree root
+ *	@index:		index key
+ *	@pslot:		pointer to reserved slot
+ *
+ *	Reserve a slot in a radix tree for the key @index.
+ */
+int radix_tree_reserve(struct radix_tree_root *root, unsigned long index, void ***pslot)
+{
+	struct radix_tree_node *node = NULL, *tmp, **slot;
+	unsigned int height, shift;
+	int error;
+
+	/* Make sure the tree is high enough.  */
+	if (index > radix_tree_maxindex(root->height)) {
+		error = radix_tree_extend(root, index);
+		if (error)
+			return error;
+	}
+    
+	slot = &root->rnode;
+	height = root->height;
+	shift = (height-1) * RADIX_TREE_MAP_SHIFT;
+
+	while (height > 0) {
+		if (*slot == NULL) {
+			/* Have to add a child node.  */
+			if (!(tmp = radix_tree_node_alloc(root)))
+				return -ENOMEM;
+			*slot = tmp;
+			if (node)
+				node->count++;
+		}
+
+		/* Go a level down.  */
+		node = *slot;
+		slot = (struct radix_tree_node **)
+			(node->slots + ((index >> shift) & RADIX_TREE_MAP_MASK));
+		shift -= RADIX_TREE_MAP_SHIFT;
+		height--;
+	}
+
+	if (*slot != NULL)
+		return -EEXIST;
+	if (node)
+		node->count++;
+
+	*pslot = (void **)slot;
+	**pslot = RADIX_TREE_SLOT_RESERVED;
+	return 0;
+}
+
+EXPORT_SYMBOL(radix_tree_reserve);
+
+
+/**
+ *	radix_tree_insert    -    insert into a radix tree
+ *	@root:		radix tree root
+ *	@index:		index key
+ *	@item:		item to insert
+ *
+ *	Insert an item into the radix tree at position @index.
+ */
+int radix_tree_insert(struct radix_tree_root *root, unsigned long index, void *item)
+{
+	void **slot;
+	int error;
+
+	error = radix_tree_reserve(root, index, &slot);
+	if (!error)
+		*slot = item;
+	return error;
+}
+
+EXPORT_SYMBOL(radix_tree_insert);
+
+
+/**
+ *	radix_tree_lookup    -    perform lookup operation on a radix tree
+ *	@root:		radix tree root
+ *	@index:		index key
+ *
+ *	Lookup them item at the position @index in the radix tree @root.
+ */
+void *radix_tree_lookup(struct radix_tree_root *root, unsigned long index)
+{
+	unsigned int height, shift;
+	struct radix_tree_node **slot;
+
+	height = root->height;
+	if (index > radix_tree_maxindex(height))
+		return NULL;
+
+	shift = (height-1) * RADIX_TREE_MAP_SHIFT;
+	slot = &root->rnode;
+
+	while (height > 0) {
+		if (*slot == NULL)
+			return NULL;
+
+		slot = (struct radix_tree_node **)
+			((*slot)->slots + ((index >> shift) & RADIX_TREE_MAP_MASK));
+		shift -= RADIX_TREE_MAP_SHIFT;
+		height--;
+	}
+
+	return (void *) *slot;
+}
+
+EXPORT_SYMBOL(radix_tree_lookup);
+
+
+/**
+ *	radix_tree_delete    -    delete an item from a radix tree
+ *	@root:		radix tree root
+ *	@index:		index key
+ *
+ *	Remove the item at @index from the radix tree rooted at @root.
+ */
+int radix_tree_delete(struct radix_tree_root *root, unsigned long index)
+{
+	struct radix_tree_path path[RADIX_TREE_INDEX_BITS/RADIX_TREE_MAP_SHIFT + 2], *pathp = path;
+	unsigned int height, shift;
+
+	height = root->height;
+	if (index > radix_tree_maxindex(height))
+		return -ENOENT;
+
+	shift = (height-1) * RADIX_TREE_MAP_SHIFT;
+	pathp->node = NULL;
+	pathp->slot = &root->rnode;
+
+	while (height > 0) {
+		if (*pathp->slot == NULL)
+			return -ENOENT;
+
+		pathp[1].node = *pathp[0].slot;
+		pathp[1].slot = (struct radix_tree_node **)
+		    (pathp[1].node->slots + ((index >> shift) & RADIX_TREE_MAP_MASK));
+		pathp++;
+		shift -= RADIX_TREE_MAP_SHIFT;
+		height--;
+	}
+
+	if (*pathp[0].slot == NULL)
+		return -ENOENT;
+
+	*pathp[0].slot = NULL;
+	while (pathp[0].node && --pathp[0].node->count == 0) {
+		pathp--;
+		*pathp[0].slot = NULL;
+		radix_tree_node_free(pathp[1].node);
+	}
+
+	return 0;
+}
+
+EXPORT_SYMBOL(radix_tree_delete);
+
+static void radix_tree_node_ctor(void *node, kmem_cache_t *cachep, unsigned long flags)
+{
+	memset(node, 0, sizeof(struct radix_tree_node));
+}
+
+static void *radix_tree_node_pool_alloc(int gfp_mask, void *data)
+{
+	return kmem_cache_alloc(radix_tree_node_cachep, gfp_mask);
+}
+
+static void radix_tree_node_pool_free(void *node, void *data)
+{
+	kmem_cache_free(radix_tree_node_cachep, node);
+}
+
+/*
+ * FIXME!  512 nodes is 200-300k of memory.  This needs to be
+ * scaled by the amount of available memory, and hopefully
+ * reduced also.
+ */
+void __init radix_tree_init(void)
+{
+	radix_tree_node_cachep = kmem_cache_create("radix_tree_node",
+			sizeof(struct radix_tree_node), 0,
+			SLAB_HWCACHE_ALIGN, radix_tree_node_ctor, NULL);
+	if (!radix_tree_node_cachep)
+		panic ("Failed to create radix_tree_node cache\n");
+	radix_tree_node_pool = mempool_create(512, radix_tree_node_pool_alloc,
+			radix_tree_node_pool_free, NULL);
+	if (!radix_tree_node_pool)
+		panic ("Failed to create radix_tree_node pool\n");
+}

mm/Makefile

@@ -14,6 +14,7 @@ export-objs := shmem.o filemap.o mempool.o page_alloc.o
 obj-y	 := memory.o mmap.o filemap.o mprotect.o mlock.o mremap.o \
 	    vmalloc.o slab.o bootmem.o swap.o vmscan.o page_io.o \
 	    page_alloc.o swap_state.o swapfile.o numa.o oom_kill.o \
-	    shmem.o highmem.o mempool.o msync.o mincore.o
+	    shmem.o highmem.o mempool.o msync.o mincore.o readahead.o \
+	    pdflush.o
 
 include $(TOPDIR)/Rules.make

mm/mincore.c

@@ -27,9 +27,9 @@ static unsigned char mincore_page(struct vm_area_struct * vma,
 {
 	unsigned char present = 0;
 	struct address_space * as = vma->vm_file->f_dentry->d_inode->i_mapping;
-	struct page * page, ** hash = page_hash(as, pgoff);
+	struct page * page;
 
-	page = __find_get_page(as, pgoff, hash);
+	page = find_get_page(as, pgoff);
 	if (page) {
 		present = Page_Uptodate(page);
 		page_cache_release(page);

mm/page_alloc.c

@@ -97,7 +97,7 @@ static void __free_pages_ok (struct page *page, unsigned int order)
 	struct page *base;
 	zone_t *zone;
 
-	if (page->buffers)
+	if (PagePrivate(page))
 		BUG();
 	if (page->mapping)
 		BUG();
@@ -290,7 +290,7 @@ static struct page * balance_classzone(zone_t * classzone, unsigned int gfp_mask
 					set_page_count(tmp, 1);
 					page = tmp;
 
-					if (page->buffers)
+					if (PagePrivate(page))
 						BUG();
 					if (page->mapping)
 						BUG();

mm/pdflush.c

+/*
+ * mm/pdflush.c - worker threads for writing back filesystem data
+ *
+ * Copyright (C) 2002, Linus Torvalds.
+ *
+ * 09Apr2002	akpm@zip.com.au
+ *		Initial version
+ */
+
+#include <linux/sched.h>
+#include <linux/list.h>
+#include <linux/signal.h>
+#include <linux/spinlock.h>
+#include <linux/gfp.h>
+#include <linux/init.h>
+#include <linux/module.h>
+
+
+/*
+ * Minimum and maximum number of pdflush instances
+ */
+#define MIN_PDFLUSH_THREADS	2
+#define MAX_PDFLUSH_THREADS	8
+
+static void start_one_pdflush_thread(void);
+
+
+/*
+ * The pdflush threads are worker threads for writing back dirty data.
+ * Ideally, we'd like one thread per active disk spindle.  But the disk
+ * topology is very hard to divine at this level.   Instead, we take
+ * care in various places to prevent more than one pdflush thread from
+ * performing writeback against a single filesystem.  pdflush threads
+ * have the PF_FLUSHER flag set in current->flags to aid in this.
+ */
+
+/*
+ * All the pdflush threads.  Protected by pdflush_lock
+ */
+static LIST_HEAD(pdflush_list);
+static spinlock_t pdflush_lock = SPIN_LOCK_UNLOCKED;
+
+/*
+ * The count of currently-running pdflush threads.  Protected
+ * by pdflush_lock.
+ */
+static int nr_pdflush_threads = 0;
+
+/*
+ * The time at which the pdflush thread pool last went empty
+ */
+static unsigned long last_empty_jifs;
+
+/*
+ * The pdflush thread.
+ *
+ * Thread pool management algorithm:
+ * 
+ * - The minumum and maximum number of pdflush instances are bound
+ *   by MIN_PDFLUSH_THREADS and MAX_PDFLUSH_THREADS.
+ * 
+ * - If there have been no idle pdflush instances for 1 second, create
+ *   a new one.
+ * 
+ * - If the least-recently-went-to-sleep pdflush thread has been asleep
+ *   for more than one second, terminate a thread.
+ */
+
+/*
+ * A structure for passing work to a pdflush thread.  Also for passing
+ * state information between pdflush threads.  Protected by pdflush_lock.
+ */
+struct pdflush_work {
+	struct task_struct *who;	/* The thread */
+	void (*fn)(unsigned long);	/* A callback function for pdflush to work on */
+	unsigned long arg0;		/* An argument to the callback function */
+	struct list_head list;		/* On pdflush_list, when the thread is idle */
+	unsigned long when_i_went_to_sleep;
+};
+
+/*
+ * preemption is disabled in pdflush.  There was a bug in preempt
+ * which was causing pdflush to get flipped into state TASK_RUNNING
+ * when it performed a spin_unlock.  That bug is probably fixed,
+ * but play it safe.  The preempt-off paths are very short.
+ */
+static int __pdflush(struct pdflush_work *my_work)
+{
+	daemonize();
+	reparent_to_init();
+	strcpy(current->comm, "pdflush");
+
+	/* interruptible sleep, so block all signals */
+	spin_lock_irq(&current->sigmask_lock);
+	siginitsetinv(&current->blocked, 0);
+	recalc_sigpending();
+	spin_unlock_irq(&current->sigmask_lock);
+
+	current->flags |= PF_FLUSHER;
+	my_work->fn = NULL;
+	my_work->who = current;
+
+	preempt_disable();
+	spin_lock_irq(&pdflush_lock);
+	nr_pdflush_threads++;
+	for ( ; ; ) {
+		struct pdflush_work *pdf;
+
+		list_add(&my_work->list, &pdflush_list);
+		my_work->when_i_went_to_sleep = jiffies;
+		set_current_state(TASK_INTERRUPTIBLE);
+		spin_unlock_irq(&pdflush_lock);
+
+		schedule();
+
+		preempt_enable();
+		(*my_work->fn)(my_work->arg0);
+		preempt_disable();
+
+		/*
+		 * Thread creation: For how long have there been zero
+		 * available threads?
+		 */
+		if (jiffies - last_empty_jifs > 1 * HZ) {
+			/* unlocked list_empty() test is OK here */
+			if (list_empty(&pdflush_list)) {
+				/* unlocked nr_pdflush_threads test is OK here */
+				if (nr_pdflush_threads < MAX_PDFLUSH_THREADS)
+					start_one_pdflush_thread();
+			}
+		}
+
+		spin_lock_irq(&pdflush_lock);
+
+		/*
+		 * Thread destruction: For how long has the sleepiest
+		 * thread slept?
+		 */
+		if (list_empty(&pdflush_list))
+			continue;
+		if (nr_pdflush_threads <= MIN_PDFLUSH_THREADS)
+			continue;
+		pdf = list_entry(pdflush_list.prev, struct pdflush_work, list);
+		if (jiffies - pdf->when_i_went_to_sleep > 1 * HZ) {
+			pdf->when_i_went_to_sleep = jiffies;	/* Limit exit rate */
+			break;					/* exeunt */
+		}
+	}
+	nr_pdflush_threads--;
+	spin_unlock_irq(&pdflush_lock);
+	preempt_enable();
+	return 0;
+}
+
+/*
+ * Of course, my_work wants to be just a local in __pdflush().  It is
+ * separated out in this manner to hopefully prevent the compiler from
+ * performing unfortunate optimisations agains the auto variables.  Because
+ * there are visible to other tasks and CPUs.  (No problem has actually
+ * been observed.  This is just paranoia).
+ */
+static int pdflush(void *dummy)
+{
+	struct pdflush_work my_work;
+	return __pdflush(&my_work);
+}
+
+/*
+ * Attempt to wake up a pdflush thread, and get it to do some work for you.
+ * Returns zero if it indeed managed to find a worker thread, and passed your
+ * payload to it.
+ */
+int pdflush_operation(void (*fn)(unsigned long), unsigned long arg0)
+{
+	unsigned long flags;
+	int ret = 0;
+
+	if (fn == NULL)
+		BUG();		/* Hard to diagnose if it's deferred */
+
+	spin_lock_irqsave(&pdflush_lock, flags);
+	if (list_empty(&pdflush_list)) {
+		spin_unlock_irqrestore(&pdflush_lock, flags);
+		ret = -1;
+	} else {
+		struct pdflush_work *pdf;
+
+		pdf = list_entry(pdflush_list.next, struct pdflush_work, list);
+		list_del_init(&pdf->list);
+		if (list_empty(&pdflush_list))
+			last_empty_jifs = jiffies;
+		spin_unlock_irqrestore(&pdflush_lock, flags);
+		pdf->fn = fn;
+		pdf->arg0 = arg0;
+		wmb();			/* ? */
+		wake_up_process(pdf->who);
+	}
+	return ret;
+}
+
+static void start_one_pdflush_thread(void)
+{
+	kernel_thread(pdflush, NULL,
+			CLONE_FS | CLONE_FILES | CLONE_SIGNAL);
+}
+
+static int __init pdflush_init(void)
+{
+	int i;
+
+	for (i = 0; i < MIN_PDFLUSH_THREADS; i++)
+		start_one_pdflush_thread();
+	return 0;
+}
+
+module_init(pdflush_init);

mm/shmem.c

@@ -370,9 +370,10 @@ static int shmem_unuse_inode (struct shmem_inode_info *info, swp_entry_t entry,
 	swp_entry_t *ptr;
 	unsigned long idx;
 	int offset;
-	
-	idx = 0;
+
 	spin_lock (&info->lock);
+repeat:
+	idx = 0;
 	offset = shmem_clear_swp (entry, info->i_direct, SHMEM_NR_DIRECT);
 	if (offset >= 0)
 		goto found;
@@ -389,13 +390,16 @@ static int shmem_unuse_inode (struct shmem_inode_info *info, swp_entry_t entry,
 	spin_unlock (&info->lock);
 	return 0;
 found:
-	delete_from_swap_cache(page);
-	add_to_page_cache(page, info->vfs_inode.i_mapping, offset + idx);
-	SetPageDirty(page);
-	SetPageUptodate(page);
-	info->swapped--;
-	spin_unlock(&info->lock);
-	return 1;
+	if (!move_from_swap_cache (page, offset+idx, info->vfs_inode.i_mapping)) {
+		info->swapped--;
+		SetPageUptodate (page);
+		spin_unlock (&info->lock);
+		return 1;
+	}
+
+	/* Yield for kswapd, and try again */
+	yield();
+	goto repeat;
 }
 
 /*
@@ -425,6 +429,7 @@ void shmem_unuse(swp_entry_t entry, struct page *page)
  */
 static int shmem_writepage(struct page * page)
 {
+	int err;
 	struct shmem_inode_info *info;
 	swp_entry_t *entry, swap;
 	struct address_space *mapping;
@@ -442,7 +447,6 @@ static int shmem_writepage(struct page * page)
 	info = SHMEM_I(inode);
 	if (info->locked)
 		return fail_writepage(page);
-getswap:
 	swap = get_swap_page();
 	if (!swap.val)
 		return fail_writepage(page);
@@ -455,29 +459,20 @@ static int shmem_writepage(struct page * page)
 	if (entry->val)
 		BUG();
 
-	/* Remove it from the page cache */
-	remove_inode_page(page);
-	page_cache_release(page);
-
-	/* Add it to the swap cache */
-	if (add_to_swap_cache(page, swap) != 0) {
-		/*
-		 * Raced with "speculative" read_swap_cache_async.
-		 * Add page back to page cache, unref swap, try again.
-		 */
-		add_to_page_cache_locked(page, mapping, index);
+	err = move_to_swap_cache(page, swap);
+	if (!err) {
+		*entry = swap;
+		info->swapped++;
 		spin_unlock(&info->lock);
-		swap_free(swap);
-		goto getswap;
+		SetPageUptodate(page);
+		set_page_dirty(page);
+		UnlockPage(page);
+		return 0;
 	}
 
-	*entry = swap;
-	info->swapped++;
 	spin_unlock(&info->lock);
-	SetPageUptodate(page);
-	set_page_dirty(page);
-	UnlockPage(page);
-	return 0;
+	swap_free(swap);
+	return fail_writepage(page);
 }
 
 /*
@@ -493,10 +488,11 @@ static int shmem_writepage(struct page * page)
  */
 static struct page * shmem_getpage_locked(struct shmem_inode_info *info, struct inode * inode, unsigned long idx)
 {
-	struct address_space * mapping = inode->i_mapping;
+	struct address_space *mapping = inode->i_mapping;
 	struct shmem_sb_info *sbinfo;
-	struct page * page;
+	struct page *page;
 	swp_entry_t *entry;
+	int error;
 
 repeat:
 	page = find_lock_page(mapping, idx);
@@ -524,8 +520,6 @@ static struct page * shmem_getpage_locked(struct shmem_inode_info *info, struct
 	
 	shmem_recalc_inode(inode);
 	if (entry->val) {
-		unsigned long flags;
-
 		/* Look it up and read it in.. */
 		page = find_get_page(&swapper_space, entry->val);
 		if (!page) {
@@ -550,16 +544,18 @@ static struct page * shmem_getpage_locked(struct shmem_inode_info *info, struct
 			goto repeat;
 		}
 
-		/* We have to this with page locked to prevent races */
+		/* We have to do this with page locked to prevent races */
 		if (TryLockPage(page)) 
 			goto wait_retry;
 
+		error = move_from_swap_cache(page, idx, mapping);
+		if (error < 0) {
+			UnlockPage(page);
+			return ERR_PTR(error);
+		}
+
 		swap_free(*entry);
 		*entry = (swp_entry_t) {0};
-		delete_from_swap_cache(page);
-		flags = page->flags & ~((1 << PG_uptodate) | (1 << PG_error) | (1 << PG_referenced) | (1 << PG_arch_1));
-		page->flags = flags | (1 << PG_dirty);
-		add_to_page_cache_locked(page, mapping, idx);
 		info->swapped--;
 		spin_unlock (&info->lock);
 	} else {
@@ -581,9 +577,13 @@ static struct page * shmem_getpage_locked(struct shmem_inode_info *info, struct
 		page = page_cache_alloc(mapping);
 		if (!page)
 			return ERR_PTR(-ENOMEM);
+		error = add_to_page_cache(page, mapping, idx);
+		if (error < 0) {
+			page_cache_release(page);
+			return ERR_PTR(-ENOMEM);
+		}
 		clear_highpage(page);
 		inode->i_blocks += BLOCKS_PER_PAGE;
-		add_to_page_cache (page, mapping, idx);
 	}
 
 	/* We have the page */

mm/swap_state.c

@@ -37,11 +37,13 @@ static struct address_space_operations swap_aops = {
 };
 
 struct address_space swapper_space = {
-	LIST_HEAD_INIT(swapper_space.clean_pages),
-	LIST_HEAD_INIT(swapper_space.dirty_pages),
-	LIST_HEAD_INIT(swapper_space.locked_pages),
-	0,				/* nrpages	*/
-	&swap_aops,
+	page_tree:	RADIX_TREE_INIT(GFP_ATOMIC),
+	page_lock:	RW_LOCK_UNLOCKED,
+	clean_pages:	LIST_HEAD_INIT(swapper_space.clean_pages),
+	dirty_pages:	LIST_HEAD_INIT(swapper_space.dirty_pages),
+	locked_pages:	LIST_HEAD_INIT(swapper_space.locked_pages),
+	a_ops:		&swap_aops,
+	i_shared_lock:	SPIN_LOCK_UNLOCKED,
 };
 
 #ifdef SWAP_CACHE_INFO
@@ -69,17 +71,21 @@ void show_swap_cache_info(void)
 
 int add_to_swap_cache(struct page *page, swp_entry_t entry)
 {
+	int error;
+
 	if (page->mapping)
 		BUG();
 	if (!swap_duplicate(entry)) {
 		INC_CACHE_INFO(noent_race);
 		return -ENOENT;
 	}
-	if (add_to_page_cache_unique(page, &swapper_space, entry.val,
-			page_hash(&swapper_space, entry.val)) != 0) {
+
+	error = add_to_page_cache_unique(page, &swapper_space, entry.val);
+	if (error != 0) {
 		swap_free(entry);
-		INC_CACHE_INFO(exist_race);
-		return -EEXIST;
+		if (error == -EEXIST)
+			INC_CACHE_INFO(exist_race);
+		return error;
 	}
 	if (!PageLocked(page))
 		BUG();
@@ -121,14 +127,96 @@ void delete_from_swap_cache(struct page *page)
 
 	entry.val = page->index;
 
-	spin_lock(&pagecache_lock);
+	write_lock(&swapper_space.page_lock);
 	__delete_from_swap_cache(page);
-	spin_unlock(&pagecache_lock);
+	write_unlock(&swapper_space.page_lock);
 
 	swap_free(entry);
 	page_cache_release(page);
 }
 
+int move_to_swap_cache(struct page *page, swp_entry_t entry)
+{
+	struct address_space *mapping = page->mapping;
+	void **pslot;
+	int err;
+
+	if (!mapping)
+		BUG();
+
+	if (!swap_duplicate(entry)) {
+		INC_CACHE_INFO(noent_race);
+		return -ENOENT;
+	}
+
+	write_lock(&swapper_space.page_lock);
+	write_lock(&mapping->page_lock);
+
+	err = radix_tree_reserve(&swapper_space.page_tree, entry.val, &pslot);
+	if (!err) {
+		/* Remove it from the page cache */
+		__remove_inode_page (page);
+
+		/* Add it to the swap cache */
+		*pslot = page;
+		page->flags = ((page->flags & ~(1 << PG_uptodate | 1 << PG_error
+						| 1 << PG_dirty  | 1 << PG_referenced
+						| 1 << PG_arch_1 | 1 << PG_checked))
+			       | (1 << PG_locked));
+		___add_to_page_cache(page, &swapper_space, entry.val);
+	}
+
+	write_unlock(&mapping->page_lock);
+	write_unlock(&swapper_space.page_lock);
+
+	if (!err) {
+		INC_CACHE_INFO(add_total);
+		return 0;
+	}
+
+	swap_free(entry);
+
+	if (err == -EEXIST)
+		INC_CACHE_INFO(exist_race);
+
+	return err;
+}
+
+int move_from_swap_cache(struct page *page, unsigned long index,
+		struct address_space *mapping)
+{
+	void **pslot;
+	int err;
+
+	if (!PageLocked(page))
+		BUG();
+
+	write_lock(&swapper_space.page_lock);
+	write_lock(&mapping->page_lock);
+
+	err = radix_tree_reserve(&mapping->page_tree, index, &pslot);
+	if (!err) {
+		swp_entry_t entry;
+
+		block_flushpage(page, 0);
+		entry.val = page->index;
+		__delete_from_swap_cache(page);
+		swap_free(entry);
+
+		*pslot = page;
+		page->flags &= ~(1 << PG_uptodate | 1 << PG_error |
+				 1 << PG_referenced | 1 << PG_arch_1 |
+				 1 << PG_checked);
+		page->flags |= (1 << PG_dirty);
+		___add_to_page_cache(page, mapping, index);
+	}
+
+	write_unlock(&mapping->page_lock);
+	write_unlock(&swapper_space.page_lock);
+
+	return err;
+}
+
 /* 
  * Perform a free_page(), also freeing any swap cache associated with
  * this page if it is the last user of the page. Can not do a lock_page,
@@ -213,6 +301,7 @@ struct page * read_swap_cache_async(swp_entry_t entry)
 		 * swap cache: added by a racing read_swap_cache_async,
 		 * or by try_to_swap_out (or shmem_writepage) re-using
 		 * the just freed swap entry for an existing page.
+		 * May fail (-ENOMEM) if radix-tree node allocation failed.
 		 */
 		err = add_to_swap_cache(new_page, entry);
 		if (!err) {
@@ -222,7 +311,7 @@ struct page * read_swap_cache_async(swp_entry_t entry)
 			rw_swap_page(READ, new_page);
 			return new_page;
 		}
-	} while (err != -ENOENT);
+	} while (err != -ENOENT && err != -ENOMEM);
 
 	if (new_page)
 		page_cache_release(new_page);

mm/swapfile.c

@@ -239,10 +239,10 @@ static int exclusive_swap_page(struct page *page)
 		/* Is the only swap cache user the cache itself? */
 		if (p->swap_map[SWP_OFFSET(entry)] == 1) {
 			/* Recheck the page count with the pagecache lock held.. */
-			spin_lock(&pagecache_lock);
-			if (page_count(page) - !!page->buffers == 2)
+			read_lock(&swapper_space.page_lock);
+			if (page_count(page) - !!PagePrivate(page) == 2)
 				retval = 1;
-			spin_unlock(&pagecache_lock);
+			read_unlock(&swapper_space.page_lock);
 		}
 		swap_info_put(p);
 	}
@@ -265,7 +265,7 @@ int can_share_swap_page(struct page *page)
 		BUG();
 	switch (page_count(page)) {
 	case 3:
-		if (!page->buffers)
+		if (!PagePrivate(page))
 			break;
 		/* Fallthrough */
 	case 2:
@@ -295,7 +295,7 @@ int remove_exclusive_swap_page(struct page *page)
 		BUG();
 	if (!PageSwapCache(page))
 		return 0;
-	if (page_count(page) - !!page->buffers != 2)	/* 2: us + cache */
+	if (page_count(page) - !!PagePrivate(page) != 2) /* 2: us + cache */
 		return 0;
 
 	entry.val = page->index;
@@ -307,13 +307,13 @@ int remove_exclusive_swap_page(struct page *page)
 	retval = 0;
 	if (p->swap_map[SWP_OFFSET(entry)] == 1) {
 		/* Recheck the page count with the pagecache lock held.. */
-		spin_lock(&pagecache_lock);
-		if (page_count(page) - !!page->buffers == 2) {
+		read_lock(&swapper_space.page_lock);
+		if (page_count(page) - !!PagePrivate(page) == 2) {
 			__delete_from_swap_cache(page);
 			SetPageDirty(page);
 			retval = 1;
 		}
-		spin_unlock(&pagecache_lock);
+		read_unlock(&swapper_space.page_lock);
 	}
 	swap_info_put(p);
 
@@ -344,7 +344,7 @@ void free_swap_and_cache(swp_entry_t entry)
 	if (page) {
 		page_cache_get(page);
 		/* Only cache user (+us), or swap space full? Free it! */
-		if (page_count(page) - !!page->buffers == 2 || vm_swap_full()) {
+		if (page_count(page) - !!PagePrivate(page) == 2 || vm_swap_full()) {
 			delete_from_swap_cache(page);
 			SetPageDirty(page);
 		}
@@ -959,7 +959,7 @@ asmlinkage long sys_swapon(const char * specialfile, int swap_flags)
 		p->lowest_bit = 0;
 		p->highest_bit = 0;
 		for (i = 1 ; i < 8*PAGE_SIZE ; i++) {
-			if (test_bit(i,(char *) swap_header)) {
+			if (test_bit(i,(unsigned long *) swap_header)) {
 				if (!p->lowest_bit)
 					p->lowest_bit = i;
 				p->highest_bit = i;
@@ -974,7 +974,7 @@ asmlinkage long sys_swapon(const char * specialfile, int swap_flags)
 			goto bad_swap;
 		}
 		for (i = 1 ; i < maxpages ; i++) {
-			if (test_bit(i,(char *) swap_header))
+			if (test_bit(i,(unsigned long *) swap_header))
 				p->swap_map[i] = 0;
 			else
 				p->swap_map[i] = SWAP_MAP_BAD;

mm/vmscan.c

@@ -92,7 +92,8 @@ static inline int try_to_swap_out(struct mm_struct * mm, struct vm_area_struct*
 		mm->rss--;
 		UnlockPage(page);
 		{
-			int freeable = page_count(page) - !!page->buffers <= 2;
+			int freeable = page_count(page) -
+				!!PagePrivate(page) <= 2;
 			page_cache_release(page);
 			return freeable;
 		}
@@ -121,7 +122,7 @@ static inline int try_to_swap_out(struct mm_struct * mm, struct vm_area_struct*
 	 * Anonymous buffercache pages can be left behind by
 	 * concurrent truncate and pagefault.
 	 */
-	if (page->buffers)
+	if (PagePrivate(page))
 		goto preserve;
 
 	/*
@@ -138,10 +139,16 @@ static inline int try_to_swap_out(struct mm_struct * mm, struct vm_area_struct*
 		 * (adding to the page cache will clear the dirty
 		 * and uptodate bits, so we need to do it again)
 		 */
-		if (add_to_swap_cache(page, entry) == 0) {
+		switch (add_to_swap_cache(page, entry)) {
+		case 0:				/* Success */
 			SetPageUptodate(page);
 			set_page_dirty(page);
 			goto set_swap_pte;
+		case -ENOMEM:			/* radix-tree allocation */
+			swap_free(entry);
+			goto preserve;
+		default:			/* ENOENT: raced */
+			break;
 		}
 		/* Raced with "speculative" read_swap_cache_async */
 		swap_free(entry);
@@ -341,6 +348,7 @@ static int FASTCALL(shrink_cache(int nr_pages, zone_t * classzone, unsigned int
 static int shrink_cache(int nr_pages, zone_t * classzone, unsigned int gfp_mask, int priority)
 {
 	struct list_head * entry;
+	struct address_space *mapping;
 	int max_scan = nr_inactive_pages / priority;
 	int max_mapped = nr_pages << (9 - priority);
 
@@ -377,7 +385,7 @@ static int shrink_cache(int nr_pages, zone_t * classzone, unsigned int gfp_mask,
 			continue;
 
 		/* Racy check to avoid trylocking when not worthwhile */
-		if (!page->buffers && (page_count(page) != 1 || !page->mapping))
+		if (!PagePrivate(page) && (page_count(page) != 1 || !page->mapping))
 			goto page_mapped;
 
 		/*
@@ -395,7 +403,9 @@ static int shrink_cache(int nr_pages, zone_t * classzone, unsigned int gfp_mask,
 			continue;
 		}
 
-		if (PageDirty(page) && is_page_cache_freeable(page) && page->mapping) {
+		mapping = page->mapping;
+
+		if (PageDirty(page) && is_page_cache_freeable(page) && mapping) {
 			/*
 			 * It is not critical here to write it only if
 			 * the page is unmapped beause any direct writer
@@ -406,7 +416,7 @@ static int shrink_cache(int nr_pages, zone_t * classzone, unsigned int gfp_mask,
 			 */
 			int (*writepage)(struct page *);
 
-			writepage = page->mapping->a_ops->writepage;
+			writepage = mapping->a_ops->writepage;
 			if ((gfp_mask & __GFP_FS) && writepage) {
 				ClearPageDirty(page);
 				SetPageLaunder(page);
@@ -426,14 +436,14 @@ static int shrink_cache(int nr_pages, zone_t * classzone, unsigned int gfp_mask,
 		 * associated with this page. If we succeed we try to free
 		 * the page as well.
 		 */
-		if (page->buffers) {
+		if (PagePrivate(page)) {
 			spin_unlock(&pagemap_lru_lock);
 
 			/* avoid to free a locked page */
 			page_cache_get(page);
 
 			if (try_to_release_page(page, gfp_mask)) {
-				if (!page->mapping) {
+				if (!mapping) {
 					/*
 					 * We must not allow an anon page
 					 * with no buffers to be visible on
@@ -470,33 +480,35 @@ static int shrink_cache(int nr_pages, zone_t * classzone, unsigned int gfp_mask,
 			}
 		}
 
-		spin_lock(&pagecache_lock);
-
 		/*
-		 * this is the non-racy check for busy page.
+		 * This is the non-racy check for busy page.
 		 */
-		if (!page->mapping || !is_page_cache_freeable(page)) {
-			spin_unlock(&pagecache_lock);
-			UnlockPage(page);
+		if (mapping) {
+			write_lock(&mapping->page_lock);
+			if (is_page_cache_freeable(page))
+				goto page_freeable;
+			write_unlock(&mapping->page_lock);
+		}
+		UnlockPage(page);
 page_mapped:
-			if (--max_mapped >= 0)
-				continue;
+		if (--max_mapped >= 0)
+			continue;
 
-			/*
-			 * Alert! We've found too many mapped pages on the
-			 * inactive list, so we start swapping out now!
-			 */
-			spin_unlock(&pagemap_lru_lock);
-			swap_out(priority, gfp_mask, classzone);
-			return nr_pages;
-		}
+		/*
+		 * Alert! We've found too many mapped pages on the
+		 * inactive list, so we start swapping out now!
+		 */
+		spin_unlock(&pagemap_lru_lock);
+		swap_out(priority, gfp_mask, classzone);
+		return nr_pages;
 
+page_freeable:
 		/*
 		 * It is critical to check PageDirty _after_ we made sure
 		 * the page is freeable* so not in use by anybody.
 		 */
 		if (PageDirty(page)) {
-			spin_unlock(&pagecache_lock);
+			write_unlock(&mapping->page_lock);
 			UnlockPage(page);
 			continue;
 		}
@@ -504,12 +516,12 @@ static int shrink_cache(int nr_pages, zone_t * classzone, unsigned int gfp_mask,
 		/* point of no return */
 		if (likely(!PageSwapCache(page))) {
 			__remove_inode_page(page);
-			spin_unlock(&pagecache_lock);
+			write_unlock(&mapping->page_lock);
 		} else {
 			swp_entry_t swap;
 			swap.val = page->index;
 			__delete_from_swap_cache(page);
-			spin_unlock(&pagecache_lock);
+			write_unlock(&mapping->page_lock);
 			swap_free(swap);
 		}