]> nv-tegra.nvidia Code Review - linux-2.6.git/blobdiff - fs/buffer.c
fs: buffer don't PageUptodate without page locked
[linux-2.6.git] / fs / buffer.c
index a9b3994020074a800c4480bb72fed0ac0bc64f5f..dcc5faa573bae6677964453cf5a2392310c73c32 100644 (file)
@@ -18,7 +18,6 @@
  * async buffer flushing, 1999 Andrea Arcangeli <andrea@suse.de>
  */
 
-#include <linux/config.h>
 #include <linux/kernel.h>
 #include <linux/syscalls.h>
 #include <linux/fs.h>
@@ -36,6 +35,7 @@
 #include <linux/hash.h>
 #include <linux/suspend.h>
 #include <linux/buffer_head.h>
+#include <linux/task_io_accounting_ops.h>
 #include <linux/bio.h>
 #include <linux/notifier.h>
 #include <linux/cpu.h>
@@ -78,6 +78,7 @@ EXPORT_SYMBOL(__lock_buffer);
 
 void fastcall unlock_buffer(struct buffer_head *bh)
 {
+       smp_mb__before_clear_bit();
        clear_buffer_locked(bh);
        smp_mb__after_clear_bit();
        wake_up_bit(&bh->b_state, BH_Lock);
@@ -160,27 +161,6 @@ int sync_blockdev(struct block_device *bdev)
 }
 EXPORT_SYMBOL(sync_blockdev);
 
-/*
- * Write out and wait upon all dirty data associated with this
- * superblock.  Filesystem data as well as the underlying block
- * device.  Takes the superblock lock.
- */
-int fsync_super(struct super_block *sb)
-{
-       sync_inodes_sb(sb, 0);
-       DQUOT_SYNC(sb);
-       lock_super(sb);
-       if (sb->s_dirt && sb->s_op->write_super)
-               sb->s_op->write_super(sb);
-       unlock_super(sb);
-       if (sb->s_op->sync_fs)
-               sb->s_op->sync_fs(sb, 1);
-       sync_blockdev(sb->s_bdev);
-       sync_inodes_sb(sb, 1);
-
-       return sync_blockdev(sb->s_bdev);
-}
-
 /*
  * Write out and wait upon all dirty data associated with this
  * device.   Filesystem data as well as the underlying block
@@ -216,19 +196,7 @@ struct super_block *freeze_bdev(struct block_device *bdev)
                sb->s_frozen = SB_FREEZE_WRITE;
                smp_wmb();
 
-               sync_inodes_sb(sb, 0);
-               DQUOT_SYNC(sb);
-
-               lock_super(sb);
-               if (sb->s_dirt && sb->s_op->write_super)
-                       sb->s_op->write_super(sb);
-               unlock_super(sb);
-
-               if (sb->s_op->sync_fs)
-                       sb->s_op->sync_fs(sb, 1);
-
-               sync_blockdev(sb->s_bdev);
-               sync_inodes_sb(sb, 1);
+               __fsync_super(sb);
 
                sb->s_frozen = SB_FREEZE_TRANS;
                smp_wmb();
@@ -268,117 +236,6 @@ void thaw_bdev(struct block_device *bdev, struct super_block *sb)
 }
 EXPORT_SYMBOL(thaw_bdev);
 
-/*
- * sync everything.  Start out by waking pdflush, because that writes back
- * all queues in parallel.
- */
-static void do_sync(unsigned long wait)
-{
-       wakeup_pdflush(0);
-       sync_inodes(0);         /* All mappings, inodes and their blockdevs */
-       DQUOT_SYNC(NULL);
-       sync_supers();          /* Write the superblocks */
-       sync_filesystems(0);    /* Start syncing the filesystems */
-       sync_filesystems(wait); /* Waitingly sync the filesystems */
-       sync_inodes(wait);      /* Mappings, inodes and blockdevs, again. */
-       if (!wait)
-               printk("Emergency Sync complete\n");
-       if (unlikely(laptop_mode))
-               laptop_sync_completion();
-}
-
-asmlinkage long sys_sync(void)
-{
-       do_sync(1);
-       return 0;
-}
-
-void emergency_sync(void)
-{
-       pdflush_operation(do_sync, 0);
-}
-
-/*
- * Generic function to fsync a file.
- *
- * filp may be NULL if called via the msync of a vma.
- */
-int file_fsync(struct file *filp, struct dentry *dentry, int datasync)
-{
-       struct inode * inode = dentry->d_inode;
-       struct super_block * sb;
-       int ret, err;
-
-       /* sync the inode to buffers */
-       ret = write_inode_now(inode, 0);
-
-       /* sync the superblock to buffers */
-       sb = inode->i_sb;
-       lock_super(sb);
-       if (sb->s_op->write_super)
-               sb->s_op->write_super(sb);
-       unlock_super(sb);
-
-       /* .. finally sync the buffers to disk */
-       err = sync_blockdev(sb->s_bdev);
-       if (!ret)
-               ret = err;
-       return ret;
-}
-
-static long do_fsync(unsigned int fd, int datasync)
-{
-       struct file * file;
-       struct address_space *mapping;
-       int ret, err;
-
-       ret = -EBADF;
-       file = fget(fd);
-       if (!file)
-               goto out;
-
-       ret = -EINVAL;
-       if (!file->f_op || !file->f_op->fsync) {
-               /* Why?  We can still call filemap_fdatawrite */
-               goto out_putf;
-       }
-
-       mapping = file->f_mapping;
-
-       current->flags |= PF_SYNCWRITE;
-       ret = filemap_fdatawrite(mapping);
-
-       /*
-        * We need to protect against concurrent writers,
-        * which could cause livelocks in fsync_buffers_list
-        */
-       mutex_lock(&mapping->host->i_mutex);
-       err = file->f_op->fsync(file, file->f_dentry, datasync);
-       if (!ret)
-               ret = err;
-       mutex_unlock(&mapping->host->i_mutex);
-       err = filemap_fdatawait(mapping);
-       if (!ret)
-               ret = err;
-       current->flags &= ~PF_SYNCWRITE;
-
-out_putf:
-       fput(file);
-out:
-       return ret;
-}
-
-asmlinkage long sys_fsync(unsigned int fd)
-{
-       return do_fsync(fd, 0);
-}
-
-asmlinkage long sys_fdatasync(unsigned int fd)
-{
-       return do_fsync(fd, 1);
-}
-
 /*
  * Various filesystems appear to want __find_get_block to be non-blocking.
  * But it's the page lock which protects the buffers.  To get around this,
@@ -431,8 +288,10 @@ __find_get_block_slow(struct block_device *bdev, sector_t block)
        if (all_mapped) {
                printk("__find_get_block_slow() failed. "
                        "block=%llu, b_blocknr=%llu\n",
-                       (unsigned long long)block, (unsigned long long)bh->b_blocknr);
-               printk("b_state=0x%08lx, b_size=%u\n", bh->b_state, bh->b_size);
+                       (unsigned long long)block,
+                       (unsigned long long)bh->b_blocknr);
+               printk("b_state=0x%08lx, b_size=%zu\n",
+                       bh->b_state, bh->b_size);
                printk("device blocksize: %d\n", 1 << bd_inode->i_blkbits);
        }
 out_unlock:
@@ -476,13 +335,18 @@ out:
    pass does the actual I/O. */
 void invalidate_bdev(struct block_device *bdev, int destroy_dirty_buffers)
 {
+       struct address_space *mapping = bdev->bd_inode->i_mapping;
+
+       if (mapping->nrpages == 0)
+               return;
+
        invalidate_bh_lrus();
        /*
         * FIXME: what about destroy_dirty_buffers?
         * We really want to use invalidate_inode_pages2() for
         * that, but not until that's cleaned up.
         */
-       invalidate_inode_pages(bdev->bd_inode->i_mapping);
+       invalidate_mapping_pages(mapping, 0, -1);
 }
 
 /*
@@ -496,7 +360,7 @@ static void free_more_memory(void)
        wakeup_pdflush(1024);
        yield();
 
-       for_each_pgdat(pgdat) {
+       for_each_online_pgdat(pgdat) {
                zones = pgdat->node_zonelists[gfp_zone(GFP_NOFS)].zones;
                if (*zones)
                        try_to_free_pages(zones, GFP_NOFS);
@@ -569,7 +433,7 @@ still_busy:
  * Completion handler for block_write_full_page() - pages which are unlocked
  * during I/O, and which have PageWriteback cleared upon I/O completion.
  */
-void end_buffer_async_write(struct buffer_head *bh, int uptodate)
+static void end_buffer_async_write(struct buffer_head *bh, int uptodate)
 {
        char b[BDEVNAME_SIZE];
        unsigned long flags;
@@ -590,6 +454,7 @@ void end_buffer_async_write(struct buffer_head *bh, int uptodate)
                               bdevname(bh->b_bdev, b));
                }
                set_bit(AS_EIO, &page->mapping->flags);
+               set_buffer_write_io_error(bh);
                clear_buffer_uptodate(bh);
                SetPageError(page);
        }
@@ -709,6 +574,10 @@ EXPORT_SYMBOL(mark_buffer_async_write);
 static inline void __remove_assoc_queue(struct buffer_head *bh)
 {
        list_del_init(&bh->b_assoc_buffers);
+       WARN_ON(!bh->b_assoc_map);
+       if (buffer_write_io_error(bh))
+               set_bit(AS_EIO, &bh->b_assoc_map->flags);
+       bh->b_assoc_map = NULL;
 }
 
 int inode_has_buffers(struct inode *inode)
@@ -801,13 +670,13 @@ void mark_buffer_dirty_inode(struct buffer_head *bh, struct inode *inode)
        if (!mapping->assoc_mapping) {
                mapping->assoc_mapping = buffer_mapping;
        } else {
-               if (mapping->assoc_mapping != buffer_mapping)
-                       BUG();
+               BUG_ON(mapping->assoc_mapping != buffer_mapping);
        }
        if (list_empty(&bh->b_assoc_buffers)) {
                spin_lock(&buffer_mapping->private_lock);
                list_move_tail(&bh->b_assoc_buffers,
                                &mapping->private_list);
+               bh->b_assoc_map = mapping;
                spin_unlock(&buffer_mapping->private_lock);
        }
 }
@@ -840,7 +709,10 @@ EXPORT_SYMBOL(mark_buffer_dirty_inode);
  */
 int __set_page_dirty_buffers(struct page *page)
 {
-       struct address_space * const mapping = page->mapping;
+       struct address_space * const mapping = page_mapping(page);
+
+       if (unlikely(!mapping))
+               return !TestSetPageDirty(page);
 
        spin_lock(&mapping->private_lock);
        if (page_has_buffers(page)) {
@@ -854,20 +726,21 @@ int __set_page_dirty_buffers(struct page *page)
        }
        spin_unlock(&mapping->private_lock);
 
-       if (!TestSetPageDirty(page)) {
-               write_lock_irq(&mapping->tree_lock);
-               if (page->mapping) {    /* Race with truncate? */
-                       if (mapping_cap_account_dirty(mapping))
-                               inc_page_state(nr_dirty);
-                       radix_tree_tag_set(&mapping->page_tree,
-                                               page_index(page),
-                                               PAGECACHE_TAG_DIRTY);
+       if (TestSetPageDirty(page))
+               return 0;
+
+       write_lock_irq(&mapping->tree_lock);
+       if (page->mapping) {    /* Race with truncate? */
+               if (mapping_cap_account_dirty(mapping)) {
+                       __inc_zone_page_state(page, NR_FILE_DIRTY);
+                       task_io_account_write(PAGE_CACHE_SIZE);
                }
-               write_unlock_irq(&mapping->tree_lock);
-               __mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
+               radix_tree_tag_set(&mapping->page_tree,
+                               page_index(page), PAGECACHE_TAG_DIRTY);
        }
-       
-       return 0;
+       write_unlock_irq(&mapping->tree_lock);
+       __mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
+       return 1;
 }
 EXPORT_SYMBOL(__set_page_dirty_buffers);
 
@@ -901,7 +774,7 @@ static int fsync_buffers_list(spinlock_t *lock, struct list_head *list)
        spin_lock(lock);
        while (!list_empty(list)) {
                bh = BH_ENTRY(list->next);
-               list_del_init(&bh->b_assoc_buffers);
+               __remove_assoc_queue(bh);
                if (buffer_dirty(bh) || buffer_locked(bh)) {
                        list_add(&bh->b_assoc_buffers, &tmp);
                        if (buffer_dirty(bh)) {
@@ -922,7 +795,7 @@ static int fsync_buffers_list(spinlock_t *lock, struct list_head *list)
 
        while (!list_empty(&tmp)) {
                bh = BH_ENTRY(tmp.prev);
-               __remove_assoc_queue(bh);
+               list_del_init(&bh->b_assoc_buffers);
                get_bh(bh);
                spin_unlock(lock);
                wait_on_buffer(bh);
@@ -1119,8 +992,7 @@ grow_dev_page(struct block_device *bdev, sector_t block,
        if (!page)
                return NULL;
 
-       if (!PageLocked(page))
-               BUG();
+       BUG_ON(!PageLocked(page));
 
        if (page_has_buffers(page)) {
                bh = page_buffers(page);
@@ -1179,8 +1051,21 @@ grow_buffers(struct block_device *bdev, sector_t block, int size)
        } while ((size << sizebits) < PAGE_SIZE);
 
        index = block >> sizebits;
-       block = index << sizebits;
 
+       /*
+        * Check for a block which wants to lie outside our maximum possible
+        * pagecache index.  (this comparison is done using sector_t types).
+        */
+       if (unlikely(index != block >> sizebits)) {
+               char b[BDEVNAME_SIZE];
+
+               printk(KERN_ERR "%s: requested out-of-range block %llu for "
+                       "device %s\n",
+                       __FUNCTION__, (unsigned long long)block,
+                       bdevname(bdev, b));
+               return -EIO;
+       }
+       block = index << sizebits;
        /* Create a page with the proper size buffers.. */
        page = grow_dev_page(bdev, block, index, size);
        if (!page)
@@ -1207,12 +1092,16 @@ __getblk_slow(struct block_device *bdev, sector_t block, int size)
 
        for (;;) {
                struct buffer_head * bh;
+               int ret;
 
                bh = __find_get_block(bdev, block, size);
                if (bh)
                        return bh;
 
-               if (!grow_buffers(bdev, block, size))
+               ret = grow_buffers(bdev, block, size);
+               if (ret < 0)
+                       return NULL;
+               if (ret == 0)
                        free_more_memory();
        }
 }
@@ -1287,6 +1176,7 @@ void __bforget(struct buffer_head *bh)
 
                spin_lock(&buffer_mapping->private_lock);
                list_del_init(&bh->b_assoc_buffers);
+               bh->b_assoc_map = NULL;
                spin_unlock(&buffer_mapping->private_lock);
        }
        __brelse(bh);
@@ -1393,11 +1283,11 @@ static void bh_lru_install(struct buffer_head *bh)
  * Look up the bh in this cpu's LRU.  If it's there, move it to the head.
  */
 static struct buffer_head *
-lookup_bh_lru(struct block_device *bdev, sector_t block, int size)
+lookup_bh_lru(struct block_device *bdev, sector_t block, unsigned size)
 {
        struct buffer_head *ret = NULL;
        struct bh_lru *lru;
-       int i;
+       unsigned int i;
 
        check_irqs_on();
        bh_lru_lock();
@@ -1429,7 +1319,7 @@ lookup_bh_lru(struct block_device *bdev, sector_t block, int size)
  * NULL
  */
 struct buffer_head *
-__find_get_block(struct block_device *bdev, sector_t block, int size)
+__find_get_block(struct block_device *bdev, sector_t block, unsigned size)
 {
        struct buffer_head *bh = lookup_bh_lru(bdev, block, size);
 
@@ -1457,7 +1347,7 @@ EXPORT_SYMBOL(__find_get_block);
  * attempt is failing.  FIXME, perhaps?
  */
 struct buffer_head *
-__getblk(struct block_device *bdev, sector_t block, int size)
+__getblk(struct block_device *bdev, sector_t block, unsigned size)
 {
        struct buffer_head *bh = __find_get_block(bdev, block, size);
 
@@ -1471,7 +1361,7 @@ EXPORT_SYMBOL(__getblk);
 /*
  * Do async read-ahead on a buffer..
  */
-void __breadahead(struct block_device *bdev, sector_t block, int size)
+void __breadahead(struct block_device *bdev, sector_t block, unsigned size)
 {
        struct buffer_head *bh = __getblk(bdev, block, size);
        if (likely(bh)) {
@@ -1491,7 +1381,7 @@ EXPORT_SYMBOL(__breadahead);
  *  It returns NULL if the block was unreadable.
  */
 struct buffer_head *
-__bread(struct block_device *bdev, sector_t block, int size)
+__bread(struct block_device *bdev, sector_t block, unsigned size)
 {
        struct buffer_head *bh = __getblk(bdev, block, size);
 
@@ -1527,8 +1417,7 @@ void set_bh_page(struct buffer_head *bh,
                struct page *page, unsigned long offset)
 {
        bh->b_page = page;
-       if (offset >= PAGE_SIZE)
-               BUG();
+       BUG_ON(offset >= PAGE_SIZE);
        if (PageHighMem(page))
                /*
                 * This catches illegal uses and preserves the offset:
@@ -1551,38 +1440,10 @@ static void discard_buffer(struct buffer_head * bh)
        clear_buffer_req(bh);
        clear_buffer_new(bh);
        clear_buffer_delay(bh);
+       clear_buffer_unwritten(bh);
        unlock_buffer(bh);
 }
 
-/**
- * try_to_release_page() - release old fs-specific metadata on a page
- *
- * @page: the page which the kernel is trying to free
- * @gfp_mask: memory allocation flags (and I/O mode)
- *
- * The address_space is to try to release any data against the page
- * (presumably at page->private).  If the release was successful, return `1'.
- * Otherwise return zero.
- *
- * The @gfp_mask argument specifies whether I/O may be performed to release
- * this page (__GFP_IO), and whether the call may block (__GFP_WAIT).
- *
- * NOTE: @gfp_mask may go away, and this function may become non-blocking.
- */
-int try_to_release_page(struct page *page, gfp_t gfp_mask)
-{
-       struct address_space * const mapping = page->mapping;
-
-       BUG_ON(!PageLocked(page));
-       if (PageWriteback(page))
-               return 0;
-       
-       if (mapping && mapping->a_ops->releasepage)
-               return mapping->a_ops->releasepage(page, gfp_mask);
-       return try_to_free_buffers(page);
-}
-EXPORT_SYMBOL(try_to_release_page);
-
 /**
  * block_invalidatepage - invalidate part of all of a buffer-backed page
  *
@@ -1598,11 +1459,10 @@ EXPORT_SYMBOL(try_to_release_page);
  * point.  Because the caller is about to free (and possibly reuse) those
  * blocks on-disk.
  */
-int block_invalidatepage(struct page *page, unsigned long offset)
+void block_invalidatepage(struct page *page, unsigned long offset)
 {
        struct buffer_head *head, *bh, *next;
        unsigned int curr_off = 0;
-       int ret = 1;
 
        BUG_ON(!PageLocked(page));
        if (!page_has_buffers(page))
@@ -1629,21 +1489,12 @@ int block_invalidatepage(struct page *page, unsigned long offset)
         * so real IO is not possible anymore.
         */
        if (offset == 0)
-               ret = try_to_release_page(page, 0);
+               try_to_release_page(page, 0);
 out:
-       return ret;
+       return;
 }
 EXPORT_SYMBOL(block_invalidatepage);
 
-int do_invalidatepage(struct page *page, unsigned long offset)
-{
-       int (*invalidatepage)(struct page *, unsigned long);
-       invalidatepage = page->mapping->a_ops->invalidatepage;
-       if (invalidatepage == NULL)
-               invalidatepage = block_invalidatepage;
-       return (*invalidatepage)(page, offset);
-}
-
 /*
  * We attach and possibly dirty the buffers atomically wrt
  * __set_page_dirty_buffers() via private_lock.  try_to_free_buffers
@@ -1743,6 +1594,7 @@ static int __block_write_full_page(struct inode *inode, struct page *page,
        sector_t block;
        sector_t last_block;
        struct buffer_head *bh, *head;
+       const unsigned blocksize = 1 << inode->i_blkbits;
        int nr_underway = 0;
 
        BUG_ON(!PageLocked(page));
@@ -1750,7 +1602,7 @@ static int __block_write_full_page(struct inode *inode, struct page *page,
        last_block = (i_size_read(inode) - 1) >> inode->i_blkbits;
 
        if (!page_has_buffers(page)) {
-               create_empty_buffers(page, 1 << inode->i_blkbits,
+               create_empty_buffers(page, blocksize,
                                        (1 << BH_Dirty)|(1 << BH_Uptodate));
        }
 
@@ -1785,6 +1637,7 @@ static int __block_write_full_page(struct inode *inode, struct page *page,
                        clear_buffer_dirty(bh);
                        set_buffer_uptodate(bh);
                } else if (!buffer_mapped(bh) && buffer_dirty(bh)) {
+                       WARN_ON(bh->b_size != blocksize);
                        err = get_block(inode, block, bh, 1);
                        if (err)
                                goto recover;
@@ -1847,17 +1700,8 @@ done:
                 * clean.  Someone wrote them back by hand with
                 * ll_rw_block/submit_bh.  A rare case.
                 */
-               int uptodate = 1;
-               do {
-                       if (!buffer_uptodate(bh)) {
-                               uptodate = 0;
-                               break;
-                       }
-                       bh = bh->b_this_page;
-               } while (bh != head);
-               if (uptodate)
-                       SetPageUptodate(page);
                end_page_writeback(page);
+
                /*
                 * The page and buffer_heads can be released at any time from
                 * here on.
@@ -1890,7 +1734,6 @@ recover:
        SetPageError(page);
        BUG_ON(PageWriteback(page));
        set_page_writeback(page);
-       unlock_page(page);
        do {
                struct buffer_head *next = bh->b_this_page;
                if (buffer_async_write(bh)) {
@@ -1900,6 +1743,7 @@ recover:
                }
                bh = next;
        } while (bh != head);
+       unlock_page(page);
        goto done;
 }
 
@@ -1938,6 +1782,7 @@ static int __block_prepare_write(struct inode *inode, struct page *page,
                if (buffer_new(bh))
                        clear_buffer_new(bh);
                if (!buffer_mapped(bh)) {
+                       WARN_ON(bh->b_size != blocksize);
                        err = get_block(inode, block, bh, 1);
                        if (err)
                                break;
@@ -1970,6 +1815,7 @@ static int __block_prepare_write(struct inode *inode, struct page *page,
                        continue; 
                }
                if (!buffer_uptodate(bh) && !buffer_delay(bh) &&
+                   !buffer_unwritten(bh) &&
                     (block_start < from || block_end > to)) {
                        ll_rw_block(READ, 1, &bh);
                        *wait_bh++=bh;
@@ -2011,6 +1857,7 @@ static int __block_prepare_write(struct inode *inode, struct page *page,
                        clear_buffer_new(bh);
                        kaddr = kmap_atomic(page, KM_USER0);
                        memset(kaddr+block_start, 0, bh->b_size);
+                       flush_dcache_page(page);
                        kunmap_atomic(kaddr, KM_USER0);
                        set_buffer_uptodate(bh);
                        mark_buffer_dirty(bh);
@@ -2093,6 +1940,7 @@ int block_read_full_page(struct page *page, get_block_t *get_block)
 
                        fully_mapped = 0;
                        if (iblock < lblock) {
+                               WARN_ON(bh->b_size != blocksize);
                                err = get_block(inode, iblock, bh, 0);
                                if (err)
                                        SetPageError(page);
@@ -2391,7 +2239,6 @@ int nobh_prepare_write(struct page *page, unsigned from, unsigned to,
        int i;
        int ret = 0;
        int is_mapped_to_disk = 1;
-       int dirtied_it = 0;
 
        if (PageMappedToDisk(page))
                return 0;
@@ -2414,6 +2261,7 @@ int nobh_prepare_write(struct page *page, unsigned from, unsigned to,
                create = 1;
                if (block_start >= to)
                        create = 0;
+               map_bh.b_size = blocksize;
                ret = get_block(inode, block_in_file + block_in_page,
                                        &map_bh, create);
                if (ret)
@@ -2427,14 +2275,10 @@ int nobh_prepare_write(struct page *page, unsigned from, unsigned to,
                        continue;
                if (buffer_new(&map_bh) || !buffer_mapped(&map_bh)) {
                        kaddr = kmap_atomic(page, KM_USER0);
-                       if (block_start < from) {
+                       if (block_start < from)
                                memset(kaddr+block_start, 0, from-block_start);
-                               dirtied_it = 1;
-                       }
-                       if (block_end > to) {
+                       if (block_end > to)
                                memset(kaddr + to, 0, block_end - to);
-                               dirtied_it = 1;
-                       }
                        flush_dcache_page(page);
                        kunmap_atomic(kaddr, KM_USER0);
                        continue;
@@ -2489,17 +2333,6 @@ int nobh_prepare_write(struct page *page, unsigned from, unsigned to,
 
        if (is_mapped_to_disk)
                SetPageMappedToDisk(page);
-       SetPageUptodate(page);
-
-       /*
-        * Setting the page dirty here isn't necessary for the prepare_write
-        * function - commit_write will do that.  But if/when this function is
-        * used within the pagefault handler to ensure that all mmapped pages
-        * have backing space in the filesystem, we will need to dirty the page
-        * if its contents were altered.
-        */
-       if (dirtied_it)
-               set_page_dirty(page);
 
        return 0;
 
@@ -2515,6 +2348,7 @@ failed:
         */
        kaddr = kmap_atomic(page, KM_USER0);
        memset(kaddr, 0, PAGE_CACHE_SIZE);
+       flush_dcache_page(page);
        kunmap_atomic(kaddr, KM_USER0);
        SetPageUptodate(page);
        set_page_dirty(page);
@@ -2522,12 +2356,17 @@ failed:
 }
 EXPORT_SYMBOL(nobh_prepare_write);
 
+/*
+ * Make sure any changes to nobh_commit_write() are reflected in
+ * nobh_truncate_page(), since it doesn't call commit_write().
+ */
 int nobh_commit_write(struct file *file, struct page *page,
                unsigned from, unsigned to)
 {
        struct inode *inode = page->mapping->host;
        loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to;
 
+       SetPageUptodate(page);
        set_page_dirty(page);
        if (pos > inode->i_size) {
                i_size_write(inode, pos);
@@ -2603,7 +2442,7 @@ int nobh_truncate_page(struct address_space *mapping, loff_t from)
        unsigned offset = from & (PAGE_CACHE_SIZE-1);
        unsigned to;
        struct page *page;
-       struct address_space_operations *a_ops = mapping->a_ops;
+       const struct address_space_operations *a_ops = mapping->a_ops;
        char *kaddr;
        int ret = 0;
 
@@ -2622,6 +2461,11 @@ int nobh_truncate_page(struct address_space *mapping, loff_t from)
                memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset);
                flush_dcache_page(page);
                kunmap_atomic(kaddr, KM_USER0);
+               /*
+                * It would be more correct to call aops->commit_write()
+                * here, but this is more efficient.
+                */
+               SetPageUptodate(page);
                set_page_dirty(page);
        }
        unlock_page(page);
@@ -2674,6 +2518,7 @@ int block_truncate_page(struct address_space *mapping,
 
        err = 0;
        if (!buffer_mapped(bh)) {
+               WARN_ON(bh->b_size != blocksize);
                err = get_block(inode, iblock, bh, 0);
                if (err)
                        goto unlock;
@@ -2686,7 +2531,7 @@ int block_truncate_page(struct address_space *mapping,
        if (PageUptodate(page))
                set_buffer_uptodate(bh);
 
-       if (!buffer_uptodate(bh) && !buffer_delay(bh)) {
+       if (!buffer_uptodate(bh) && !buffer_delay(bh) && !buffer_unwritten(bh)) {
                err = -EIO;
                ll_rw_block(READ, 1, &bh);
                wait_on_buffer(bh);
@@ -2760,6 +2605,7 @@ sector_t generic_block_bmap(struct address_space *mapping, sector_t block,
        struct inode *inode = mapping->host;
        tmp.b_state = 0;
        tmp.b_blocknr = 0;
+       tmp.b_size = 1 << inode->i_blkbits;
        get_block(inode, block, &tmp, 0);
        return tmp.b_blocknr;
 }
@@ -2986,17 +2832,23 @@ int try_to_free_buffers(struct page *page)
 
        spin_lock(&mapping->private_lock);
        ret = drop_buffers(page, &buffers_to_free);
-       if (ret) {
-               /*
-                * If the filesystem writes its buffers by hand (eg ext3)
-                * then we can have clean buffers against a dirty page.  We
-                * clean the page here; otherwise later reattachment of buffers
-                * could encounter a non-uptodate page, which is unresolvable.
-                * This only applies in the rare case where try_to_free_buffers
-                * succeeds but the page is not freed.
-                */
-               clear_page_dirty(page);
-       }
+
+       /*
+        * If the filesystem writes its buffers by hand (eg ext3)
+        * then we can have clean buffers against a dirty page.  We
+        * clean the page here; otherwise the VM will never notice
+        * that the filesystem did any IO at all.
+        *
+        * Also, during truncate, discard_buffer will have marked all
+        * the page's buffers clean.  We discover that here and clean
+        * the page also.
+        *
+        * private_lock must be held over this entire operation in order
+        * to synchronise against __set_page_dirty_buffers and prevent the
+        * dirty bit from being lost.
+        */
+       if (ret)
+               cancel_dirty_page(page, PAGE_CACHE_SIZE);
        spin_unlock(&mapping->private_lock);
 out:
        if (buffers_to_free) {
@@ -3012,7 +2864,7 @@ out:
 }
 EXPORT_SYMBOL(try_to_free_buffers);
 
-int block_sync_page(struct page *page)
+void block_sync_page(struct page *page)
 {
        struct address_space *mapping;
 
@@ -3020,7 +2872,6 @@ int block_sync_page(struct page *page)
        mapping = page_mapping(page);
        if (mapping)
                blk_run_backing_dev(mapping->backing_dev_info, page);
-       return 0;
 }
 
 /*
@@ -3050,72 +2901,10 @@ asmlinkage long sys_bdflush(int func, long data)
        return 0;
 }
 
-/*
- * Migration function for pages with buffers. This function can only be used
- * if the underlying filesystem guarantees that no other references to "page"
- * exist.
- */
-#ifdef CONFIG_MIGRATION
-int buffer_migrate_page(struct page *newpage, struct page *page)
-{
-       struct address_space *mapping = page->mapping;
-       struct buffer_head *bh, *head;
-       int rc;
-
-       if (!mapping)
-               return -EAGAIN;
-
-       if (!page_has_buffers(page))
-               return migrate_page(newpage, page);
-
-       head = page_buffers(page);
-
-       rc = migrate_page_remove_references(newpage, page, 3);
-       if (rc)
-               return rc;
-
-       bh = head;
-       do {
-               get_bh(bh);
-               lock_buffer(bh);
-               bh = bh->b_this_page;
-
-       } while (bh != head);
-
-       ClearPagePrivate(page);
-       set_page_private(newpage, page_private(page));
-       set_page_private(page, 0);
-       put_page(page);
-       get_page(newpage);
-
-       bh = head;
-       do {
-               set_bh_page(bh, newpage, bh_offset(bh));
-               bh = bh->b_this_page;
-
-       } while (bh != head);
-
-       SetPagePrivate(newpage);
-
-       migrate_page_copy(newpage, page);
-
-       bh = head;
-       do {
-               unlock_buffer(bh);
-               put_bh(bh);
-               bh = bh->b_this_page;
-
-       } while (bh != head);
-
-       return 0;
-}
-EXPORT_SYMBOL(buffer_migrate_page);
-#endif
-
 /*
  * Buffer-head allocation
  */
-static kmem_cache_t *bh_cachep;
+static struct kmem_cache *bh_cachep;
 
 /*
  * Once the number of bh's in the machine exceeds this level, we start
@@ -3140,7 +2929,7 @@ static void recalc_bh_state(void)
        if (__get_cpu_var(bh_accounting).ratelimit++ < 4096)
                return;
        __get_cpu_var(bh_accounting).ratelimit = 0;
-       for_each_cpu(i)
+       for_each_online_cpu(i)
                tot += per_cpu(bh_accounting, i).nr;
        buffer_heads_over_limit = (tot > max_buffer_heads);
 }
@@ -3168,7 +2957,7 @@ void free_buffer_head(struct buffer_head *bh)
 EXPORT_SYMBOL(free_buffer_head);
 
 static void
-init_buffer_head(void *data, kmem_cache_t *cachep, unsigned long flags)
+init_buffer_head(void *data, struct kmem_cache *cachep, unsigned long flags)
 {
        if ((flags & (SLAB_CTOR_VERIFY|SLAB_CTOR_CONSTRUCTOR)) ==
                            SLAB_CTOR_CONSTRUCTOR) {
@@ -3179,7 +2968,6 @@ init_buffer_head(void *data, kmem_cache_t *cachep, unsigned long flags)
        }
 }
 
-#ifdef CONFIG_HOTPLUG_CPU
 static void buffer_exit_cpu(int cpu)
 {
        int i;
@@ -3189,6 +2977,9 @@ static void buffer_exit_cpu(int cpu)
                brelse(b->bhs[i]);
                b->bhs[i] = NULL;
        }
+       get_cpu_var(bh_accounting).nr += per_cpu(bh_accounting, cpu).nr;
+       per_cpu(bh_accounting, cpu).nr = 0;
+       put_cpu_var(bh_accounting);
 }
 
 static int buffer_cpu_notify(struct notifier_block *self,
@@ -3198,15 +2989,17 @@ static int buffer_cpu_notify(struct notifier_block *self,
                buffer_exit_cpu((unsigned long)hcpu);
        return NOTIFY_OK;
 }
-#endif /* CONFIG_HOTPLUG_CPU */
 
 void __init buffer_init(void)
 {
        int nrpages;
 
        bh_cachep = kmem_cache_create("buffer_head",
-                       sizeof(struct buffer_head), 0,
-                       SLAB_RECLAIM_ACCOUNT|SLAB_PANIC, init_buffer_head, NULL);
+                                       sizeof(struct buffer_head), 0,
+                                       (SLAB_RECLAIM_ACCOUNT|SLAB_PANIC|
+                                       SLAB_MEM_SPREAD),
+                                       init_buffer_head,
+                                       NULL);
 
        /*
         * Limit the bh occupancy to 10% of ZONE_NORMAL
@@ -3226,7 +3019,6 @@ EXPORT_SYMBOL(block_sync_page);
 EXPORT_SYMBOL(block_truncate_page);
 EXPORT_SYMBOL(block_write_full_page);
 EXPORT_SYMBOL(cont_prepare_write);
-EXPORT_SYMBOL(end_buffer_async_write);
 EXPORT_SYMBOL(end_buffer_read_sync);
 EXPORT_SYMBOL(end_buffer_write_sync);
 EXPORT_SYMBOL(file_fsync);