Btrfs: ensure an entire eb is written at once

This patch simplifies how we track our extent buffers.  Previously we could exit
writepages with only having written half of an extent buffer, which meant we had
to track the state of the pages and the state of the extent buffers differently.
Now we only read in entire extent buffers and write out entire extent buffers,
this allows us to simply set bits in our bflags to indicate the state of the eb
and we no longer have to do things like track uptodate with our iotree.  Thanks,
Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

fs/btrfs/disk-io.c

@@ -333,7 +333,7 @@ static int verify_parent_transid(struct extent_io_tree *io_tree,
 
 	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
 			 0, &cached_state, GFP_NOFS);
-	if (extent_buffer_uptodate(io_tree, eb, cached_state) &&
+	if (extent_buffer_uptodate(eb) &&
 	    btrfs_header_generation(eb) == parent_transid) {
 		ret = 0;
 		goto out;
@@ -344,7 +344,7 @@ static int verify_parent_transid(struct extent_io_tree *io_tree,
 		       (unsigned long long)parent_transid,
 		       (unsigned long long)btrfs_header_generation(eb));
 	ret = 1;
-	clear_extent_buffer_uptodate(io_tree, eb, &cached_state);
+	clear_extent_buffer_uptodate(eb);
 out:
 	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
 			     &cached_state, GFP_NOFS);
@@ -566,7 +566,12 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 	tree = &BTRFS_I(page->mapping->host)->io_tree;
 	eb = (struct extent_buffer *)page->private;
 
-	reads_done = atomic_dec_and_test(&eb->pages_reading);
+	/* the pending IO might have been the only thing that kept this buffer
+	 * in memory.  Make sure we have a ref for all this other checks
+	 */
+	extent_buffer_get(eb);
+
+	reads_done = atomic_dec_and_test(&eb->io_pages);
 	if (!reads_done)
 		goto err;
 
@@ -606,14 +611,17 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 		ret = -EIO;
 	}
 
+	if (!ret)
+		set_extent_buffer_uptodate(eb);
 err:
 	if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) {
 		clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags);
 		btree_readahead_hook(root, eb, eb->start, ret);
 	}
 
-	if (ret && eb)
-		clear_extent_buffer_uptodate(tree, eb, NULL);
+	if (ret)
+		clear_extent_buffer_uptodate(eb);
+	free_extent_buffer(eb);
 out:
 	return ret;
 }
@@ -878,20 +886,6 @@ static int btree_migratepage(struct address_space *mapping,
 }
 #endif
 
-static int btree_writepage(struct page *page, struct writeback_control *wbc)
-{
-	struct extent_io_tree *tree;
-	tree = &BTRFS_I(page->mapping->host)->io_tree;
-
-	if (!(current->flags & PF_MEMALLOC)) {
-		return extent_write_full_page(tree, page,
-					      btree_get_extent, wbc);
-	}
-
-	redirty_page_for_writepage(wbc, page);
-	unlock_page(page);
-	return 0;
-}
 
 static int btree_writepages(struct address_space *mapping,
 			    struct writeback_control *wbc)
@@ -911,7 +905,7 @@ static int btree_writepages(struct address_space *mapping,
 		if (num_dirty < thresh)
 			return 0;
 	}
-	return extent_writepages(tree, mapping, btree_get_extent, wbc);
+	return btree_write_cache_pages(mapping, wbc);
 }
 
 static int btree_readpage(struct file *file, struct page *page)
@@ -950,15 +944,28 @@ static void btree_invalidatepage(struct page *page, unsigned long offset)
 	}
 }
 
+static int btree_set_page_dirty(struct page *page)
+{
+	struct extent_buffer *eb;
+
+	BUG_ON(!PagePrivate(page));
+	eb = (struct extent_buffer *)page->private;
+	BUG_ON(!eb);
+	BUG_ON(!test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
+	BUG_ON(!atomic_read(&eb->refs));
+	btrfs_assert_tree_locked(eb);
+	return __set_page_dirty_nobuffers(page);
+}
+
 static const struct address_space_operations btree_aops = {
 	.readpage	= btree_readpage,
-	.writepage	= btree_writepage,
 	.writepages	= btree_writepages,
 	.releasepage	= btree_releasepage,
 	.invalidatepage = btree_invalidatepage,
 #ifdef CONFIG_MIGRATION
 	.migratepage	= btree_migratepage,
 #endif
+	.set_page_dirty = btree_set_page_dirty,
 };
 
 int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize,
@@ -1001,7 +1008,7 @@ int reada_tree_block_flagged(struct btrfs_root *root, u64 bytenr, u32 blocksize,
 	if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) {
 		free_extent_buffer(buf);
 		return -EIO;
-	} else if (extent_buffer_uptodate(io_tree, buf, NULL)) {
+	} else if (extent_buffer_uptodate(buf)) {
 		*eb = buf;
 	} else {
 		free_extent_buffer(buf);
@@ -1054,9 +1061,6 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
 		return NULL;
 
 	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
-
-	if (ret == 0)
-		set_bit(EXTENT_BUFFER_UPTODATE, &buf->bflags);
 	return buf;
 
 }
@@ -1064,7 +1068,6 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
 int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 		     struct extent_buffer *buf)
 {
-	struct inode *btree_inode = root->fs_info->btree_inode;
 	if (btrfs_header_generation(buf) ==
 	    root->fs_info->running_transaction->transid) {
 		btrfs_assert_tree_locked(buf);
@@ -1080,8 +1083,7 @@ int clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 
 		/* ugh, clear_extent_buffer_dirty needs to lock the page */
 		btrfs_set_lock_blocking(buf);
-		clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree,
-					  buf);
+		clear_extent_buffer_dirty(buf);
 	}
 	return 0;
 }
@@ -1948,6 +1950,7 @@ int open_ctree(struct super_block *sb,
 	RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node);
 	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
 			     fs_info->btree_inode->i_mapping);
+	BTRFS_I(fs_info->btree_inode)->io_tree.track_uptodate = 0;
 	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree);
 
 	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
@@ -3058,8 +3061,7 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid)
 	int ret;
 	struct inode *btree_inode = buf->pages[0]->mapping->host;
 
-	ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf,
-				     NULL);
+	ret = extent_buffer_uptodate(buf);
 	if (!ret)
 		return ret;
 
@@ -3070,16 +3072,13 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid)
 
 int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
 {
-	struct inode *btree_inode = buf->pages[0]->mapping->host;
-	return set_extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree,
-					  buf);
+	return set_extent_buffer_uptodate(buf);
 }
 
 void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 {
 	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
 	u64 transid = btrfs_header_generation(buf);
-	struct inode *btree_inode = root->fs_info->btree_inode;
 	int was_dirty;
 
 	btrfs_assert_tree_locked(buf);
@@ -3091,8 +3090,7 @@ void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
 			(unsigned long long)root->fs_info->generation);
 		WARN_ON(1);
 	}
-	was_dirty = set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree,
-					    buf);
+	was_dirty = set_extent_buffer_dirty(buf);
 	if (!was_dirty) {
 		spin_lock(&root->fs_info->delalloc_lock);
 		root->fs_info->dirty_metadata_bytes += buf->len;
@@ -3147,11 +3145,7 @@ void __btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr)
 int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
 {
 	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
-	int ret;
-	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
-	if (ret == 0)
-		set_bit(EXTENT_BUFFER_UPTODATE, &buf->bflags);
-	return ret;
+	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
 }
 
 static int btree_lock_page_hook(struct page *page, void *data,

fs/btrfs/extent_io.c

@@ -19,6 +19,7 @@
 #include "btrfs_inode.h"
 #include "volumes.h"
 #include "check-integrity.h"
+#include "locking.h"
 
 static struct kmem_cache *extent_state_cache;
 static struct kmem_cache *extent_buffer_cache;
@@ -53,6 +54,8 @@ struct extent_page_data {
 	unsigned int sync_io:1;
 };
 
+static noinline void flush_write_bio(void *data);
+
 int __init extent_io_init(void)
 {
 	extent_state_cache = kmem_cache_create("extent_state",
@@ -2337,7 +2340,7 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 			}
 		}
 
-		if (uptodate) {
+		if (uptodate && tree->track_uptodate) {
 			set_extent_uptodate(tree, start, end, &cached,
 					    GFP_ATOMIC);
 		}
@@ -2973,6 +2976,275 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 	return 0;
 }
 
+static int eb_wait(void *word)
+{
+	io_schedule();
+	return 0;
+}
+
+static void wait_on_extent_buffer_writeback(struct extent_buffer *eb)
+{
+	wait_on_bit(&eb->bflags, EXTENT_BUFFER_WRITEBACK, eb_wait,
+		    TASK_UNINTERRUPTIBLE);
+}
+
+static int lock_extent_buffer_for_io(struct extent_buffer *eb,
+				     struct btrfs_fs_info *fs_info,
+				     struct extent_page_data *epd)
+{
+	unsigned long i, num_pages;
+	int flush = 0;
+	int ret = 0;
+
+	if (!btrfs_try_tree_write_lock(eb)) {
+		flush = 1;
+		flush_write_bio(epd);
+		btrfs_tree_lock(eb);
+	}
+
+	if (test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags)) {
+		btrfs_tree_unlock(eb);
+		if (!epd->sync_io)
+			return 0;
+		if (!flush) {
+			flush_write_bio(epd);
+			flush = 1;
+		}
+		wait_on_extent_buffer_writeback(eb);
+		btrfs_tree_lock(eb);
+		if (test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags)) {
+			printk(KERN_ERR "Um, ok?\n");
+			btrfs_tree_unlock(eb);
+			return 0;
+		}
+	}
+
+	if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
+		set_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags);
+		btrfs_set_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN);
+		spin_lock(&fs_info->delalloc_lock);
+		if (fs_info->dirty_metadata_bytes >= eb->len)
+			fs_info->dirty_metadata_bytes -= eb->len;
+		else
+			WARN_ON(1);
+		spin_unlock(&fs_info->delalloc_lock);
+		ret = 1;
+	}
+
+	btrfs_tree_unlock(eb);
+
+	if (!ret)
+		return ret;
+
+	num_pages = num_extent_pages(eb->start, eb->len);
+	for (i = 0; i < num_pages; i++) {
+		struct page *p = extent_buffer_page(eb, i);
+
+		if (!trylock_page(p)) {
+			if (!flush) {
+				flush_write_bio(epd);
+				flush = 1;
+			}
+			lock_page(p);
+		}
+	}
+
+	return ret;
+}
+
+static void end_extent_buffer_writeback(struct extent_buffer *eb)
+{
+	clear_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags);
+	smp_mb__after_clear_bit();
+	wake_up_bit(&eb->bflags, EXTENT_BUFFER_WRITEBACK);
+}
+
+static void end_bio_extent_buffer_writepage(struct bio *bio, int err)
+{
+	int uptodate = err == 0;
+	struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
+	struct extent_buffer *eb;
+	int done;
+
+	do {
+		struct page *page = bvec->bv_page;
+
+		bvec--;
+		eb = (struct extent_buffer *)page->private;
+		BUG_ON(!eb);
+		done = atomic_dec_and_test(&eb->io_pages);
+
+		if (!uptodate || test_bit(EXTENT_BUFFER_IOERR, &eb->bflags)) {
+			set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+			ClearPageUptodate(page);
+			SetPageError(page);
+		}
+
+		end_page_writeback(page);
+
+		if (!done)
+			continue;
+
+		end_extent_buffer_writeback(eb);
+	} while (bvec >= bio->bi_io_vec);
+
+	bio_put(bio);
+
+}
+
+static int write_one_eb(struct extent_buffer *eb,
+			struct btrfs_fs_info *fs_info,
+			struct writeback_control *wbc,
+			struct extent_page_data *epd)
+{
+	struct block_device *bdev = fs_info->fs_devices->latest_bdev;
+	u64 offset = eb->start;
+	unsigned long i, num_pages;
+	int rw = (epd->sync_io ? WRITE_SYNC : WRITE);
+	int ret;
+
+	clear_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+	num_pages = num_extent_pages(eb->start, eb->len);
+	atomic_set(&eb->io_pages, num_pages);
+	for (i = 0; i < num_pages; i++) {
+		struct page *p = extent_buffer_page(eb, i);
+
+		clear_page_dirty_for_io(p);
+		set_page_writeback(p);
+		ret = submit_extent_page(rw, eb->tree, p, offset >> 9,
+					 PAGE_CACHE_SIZE, 0, bdev, &epd->bio,
+					 -1, end_bio_extent_buffer_writepage,
+					 0, 0, 0);
+		if (ret) {
+			set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
+			SetPageError(p);
+			if (atomic_sub_and_test(num_pages - i, &eb->io_pages))
+				end_extent_buffer_writeback(eb);
+			ret = -EIO;
+			break;
+		}
+		offset += PAGE_CACHE_SIZE;
+		update_nr_written(p, wbc, 1);
+		unlock_page(p);
+	}
+
+	if (unlikely(ret)) {
+		for (; i < num_pages; i++) {
+			struct page *p = extent_buffer_page(eb, i);
+			unlock_page(p);
+		}
+	}
+
+	return ret;
+}
+
+int btree_write_cache_pages(struct address_space *mapping,
+				   struct writeback_control *wbc)
+{
+	struct extent_io_tree *tree = &BTRFS_I(mapping->host)->io_tree;
+	struct btrfs_fs_info *fs_info = BTRFS_I(mapping->host)->root->fs_info;
+	struct extent_buffer *eb, *prev_eb = NULL;
+	struct extent_page_data epd = {
+		.bio = NULL,
+		.tree = tree,
+		.extent_locked = 0,
+		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
+	};
+	int ret = 0;
+	int done = 0;
+	int nr_to_write_done = 0;
+	struct pagevec pvec;
+	int nr_pages;
+	pgoff_t index;
+	pgoff_t end;		/* Inclusive */
+	int scanned = 0;
+	int tag;
+
+	pagevec_init(&pvec, 0);
+	if (wbc->range_cyclic) {
+		index = mapping->writeback_index; /* Start from prev offset */
+		end = -1;
+	} else {
+		index = wbc->range_start >> PAGE_CACHE_SHIFT;
+		end = wbc->range_end >> PAGE_CACHE_SHIFT;
+		scanned = 1;
+	}
+	if (wbc->sync_mode == WB_SYNC_ALL)
+		tag = PAGECACHE_TAG_TOWRITE;
+	else
+		tag = PAGECACHE_TAG_DIRTY;
+retry:
+	if (wbc->sync_mode == WB_SYNC_ALL)
+		tag_pages_for_writeback(mapping, index, end);
+	while (!done && !nr_to_write_done && (index <= end) &&
+	       (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
+			min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
+		unsigned i;
+
+		scanned = 1;
+		for (i = 0; i < nr_pages; i++) {
+			struct page *page = pvec.pages[i];
+
+			if (!PagePrivate(page))
+				continue;
+
+			if (!wbc->range_cyclic && page->index > end) {
+				done = 1;
+				break;
+			}
+
+			eb = (struct extent_buffer *)page->private;
+			if (!eb) {
+				WARN_ON(1);
+				continue;
+			}
+
+			if (eb == prev_eb)
+				continue;
+
+			if (!atomic_inc_not_zero(&eb->refs)) {
+				WARN_ON(1);
+				continue;
+			}
+
+			prev_eb = eb;
+			ret = lock_extent_buffer_for_io(eb, fs_info, &epd);
+			if (!ret) {
+				free_extent_buffer(eb);
+				continue;
+			}
+
+			ret = write_one_eb(eb, fs_info, wbc, &epd);
+			if (ret) {
+				done = 1;
+				free_extent_buffer(eb);
+				break;
+			}
+			free_extent_buffer(eb);
+
+			/*
+			 * the filesystem may choose to bump up nr_to_write.
+			 * We have to make sure to honor the new nr_to_write
+			 * at any time
+			 */
+			nr_to_write_done = wbc->nr_to_write <= 0;
+		}
+		pagevec_release(&pvec);
+		cond_resched();
+	}
+	if (!scanned && !done) {
+		/*
+		 * We hit the last page and there is more work to be done: wrap
+		 * back to the start of the file
+		 */
+		scanned = 1;
+		index = 0;
+		goto retry;
+	}
+	flush_write_bio(&epd);
+	return ret;
+}
+
 /**
  * write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
  * @mapping: address space structure to write
@@ -3609,7 +3881,7 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
 #endif
 	spin_lock_init(&eb->refs_lock);
 	atomic_set(&eb->refs, 1);
-	atomic_set(&eb->pages_reading, 0);
+	atomic_set(&eb->io_pages, 0);
 
 	if (len > MAX_INLINE_EXTENT_BUFFER_SIZE) {
 		struct page **pages;
@@ -3628,6 +3900,13 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
 	return eb;
 }
 
+static int extent_buffer_under_io(struct extent_buffer *eb)
+{
+	return (atomic_read(&eb->io_pages) ||
+		test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags) ||
+		test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
+}
+
 /*
  * Helper for releasing extent buffer page.
  */
@@ -3637,6 +3916,8 @@ static void btrfs_release_extent_buffer_page(struct extent_buffer *eb,
 	unsigned long index;
 	struct page *page;
 
+	BUG_ON(extent_buffer_under_io(eb));
+
 	index = num_extent_pages(eb->start, eb->len);
 	if (start_idx >= index)
 		return;
@@ -3655,6 +3936,7 @@ static void btrfs_release_extent_buffer_page(struct extent_buffer *eb,
 			 */
 			if (PagePrivate(page) &&
 			    page->private == (unsigned long)eb) {
+				BUG_ON(test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
 				BUG_ON(PageDirty(page));
 				BUG_ON(PageWriteback(page));
 				/*
@@ -3683,10 +3965,41 @@ static inline void btrfs_release_extent_buffer(struct extent_buffer *eb)
 	__free_extent_buffer(eb);
 }
 
+static void check_buffer_tree_ref(struct extent_buffer *eb)
+{
+	/* the ref bit is tricky.  We have to make sure it is set
+	 * if we have the buffer dirty.   Otherwise the
+	 * code to free a buffer can end up dropping a dirty
+	 * page
+	 *
+	 * Once the ref bit is set, it won't go away while the
+	 * buffer is dirty or in writeback, and it also won't
+	 * go away while we have the reference count on the
+	 * eb bumped.
+	 *
+	 * We can't just set the ref bit without bumping the
+	 * ref on the eb because free_extent_buffer might
+	 * see the ref bit and try to clear it.  If this happens
+	 * free_extent_buffer might end up dropping our original
+	 * ref by mistake and freeing the page before we are able
+	 * to add one more ref.
+	 *
+	 * So bump the ref count first, then set the bit.  If someone
+	 * beat us to it, drop the ref we added.
+	 */
+	if (!test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags)) {
+		atomic_inc(&eb->refs);
+		if (test_and_set_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
+			atomic_dec(&eb->refs);
+	}
+}
+
 static void mark_extent_buffer_accessed(struct extent_buffer *eb)
 {
 	unsigned long num_pages, i;
 
+	check_buffer_tree_ref(eb);
+
 	num_pages = num_extent_pages(eb->start, eb->len);
 	for (i = 0; i < num_pages; i++) {
 		struct page *p = extent_buffer_page(eb, i);
@@ -3744,15 +4057,17 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 				goto free_eb;
 			}
 
-			/* 
+			/*
 			 * Do this so attach doesn't complain and we need to
 			 * drop the ref the old guy had.
 			 */
 			ClearPagePrivate(p);
+			WARN_ON(PageDirty(p));
 			page_cache_release(p);
 		}
 		attach_extent_buffer_page(eb, p);
 		spin_unlock(&mapping->private_lock);
+		WARN_ON(PageDirty(p));
 		mark_page_accessed(p);
 		eb->pages[i] = p;
 		if (!PageUptodate(p))
@@ -3788,8 +4103,7 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 	}
 	/* add one reference for the tree */
 	spin_lock(&eb->refs_lock);
-	atomic_inc(&eb->refs);
-	set_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags);
+	check_buffer_tree_ref(eb);
 	spin_unlock(&eb->refs_lock);
 	spin_unlock(&tree->buffer_lock);
 	radix_tree_preload_end();
@@ -3849,90 +4163,15 @@ static inline void btrfs_release_extent_buffer_rcu(struct rcu_head *head)
 	__free_extent_buffer(eb);
 }
 
-static int extent_buffer_under_io(struct extent_buffer *eb,
-				  struct page *locked_page)
-{
-	unsigned long num_pages, i;
-
-	num_pages = num_extent_pages(eb->start, eb->len);
-	for (i = 0; i < num_pages; i++) {
-		struct page *page = eb->pages[i];
-		int need_unlock = 0;
-
-		if (!page)
-			continue;
-
-		if (page != locked_page) {
-			if (!trylock_page(page))
-				return 1;
-			need_unlock = 1;
-		}
-
-		if (PageDirty(page) || PageWriteback(page)) {
-			if (need_unlock)
-				unlock_page(page);
-			return 1;
-		}
-		if (need_unlock)
-			unlock_page(page);
-	}
-
-	return 0;
-}
-
 /* Expects to have eb->eb_lock already held */
 static void release_extent_buffer(struct extent_buffer *eb, gfp_t mask)
 {
 	WARN_ON(atomic_read(&eb->refs) == 0);
 	if (atomic_dec_and_test(&eb->refs)) {
 		struct extent_io_tree *tree = eb->tree;
-		int ret;
 
 		spin_unlock(&eb->refs_lock);
 
-		might_sleep_if(mask & __GFP_WAIT);
-		ret = clear_extent_bit(tree, eb->start,
-				       eb->start + eb->len - 1, -1, 0, 0,
-				       NULL, mask);
-		if (ret < 0) {
-			unsigned long num_pages, i;
-
-			num_pages = num_extent_pages(eb->start, eb->len);
-			/*
-			 * We failed to clear the state bits which likely means
-			 * ENOMEM, so just re-up the eb ref and continue, we
-			 * will get freed later on via releasepage or something
-			 * else and will be ok.
-			 */
-			spin_lock(&eb->tree->mapping->private_lock);
-			spin_lock(&eb->refs_lock);
-			set_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags);
-			atomic_inc(&eb->refs);
-
-			/*
-			 * We may have started to reclaim the pages for a newly
-			 * allocated eb, make sure we own all of them again.
-			 */
-			for (i = 0; i < num_pages; i++) {
-				struct page *page = eb->pages[i];
-
-				if (!page) {
-					WARN_ON(1);
-					continue;
-				}
-
-				BUG_ON(!PagePrivate(page));
-				if (page->private != (unsigned long)eb) {
-					ClearPagePrivate(page);
-					page_cache_release(page);
-					attach_extent_buffer_page(eb, page);
-				}
-			}
-			spin_unlock(&eb->refs_lock);
-			spin_unlock(&eb->tree->mapping->private_lock);
-			return;
-		}
-
 		spin_lock(&tree->buffer_lock);
 		radix_tree_delete(&tree->buffer,
 				  eb->start >> PAGE_CACHE_SHIFT);
@@ -3955,7 +4194,7 @@ void free_extent_buffer(struct extent_buffer *eb)
 	spin_lock(&eb->refs_lock);
 	if (atomic_read(&eb->refs) == 2 &&
 	    test_bit(EXTENT_BUFFER_STALE, &eb->bflags) &&
-	    !extent_buffer_under_io(eb, NULL) &&
+	    !extent_buffer_under_io(eb) &&
 	    test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
 		atomic_dec(&eb->refs);
 
@@ -3974,20 +4213,20 @@ void free_extent_buffer_stale(struct extent_buffer *eb)
 	spin_lock(&eb->refs_lock);
 	set_bit(EXTENT_BUFFER_STALE, &eb->bflags);
 
-	if (atomic_read(&eb->refs) == 2 && !extent_buffer_under_io(eb, NULL) &&
+	if (atomic_read(&eb->refs) == 2 && !extent_buffer_under_io(eb) &&
 	    test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
 		atomic_dec(&eb->refs);
 	release_extent_buffer(eb, GFP_NOFS);
 }
 
-int clear_extent_buffer_dirty(struct extent_io_tree *tree,
-			      struct extent_buffer *eb)
+int clear_extent_buffer_dirty(struct extent_buffer *eb)
 {
 	unsigned long i;
 	unsigned long num_pages;
 	struct page *page;
 
 	num_pages = num_extent_pages(eb->start, eb->len);
+	WARN_ON(atomic_read(&eb->refs) == 0);
 
 	for (i = 0; i < num_pages; i++) {
 		page = extent_buffer_page(eb, i);
@@ -4008,25 +4247,30 @@ int clear_extent_buffer_dirty(struct extent_io_tree *tree,
 		ClearPageError(page);
 		unlock_page(page);
 	}
+	WARN_ON(atomic_read(&eb->refs) == 0);
 	return 0;
 }
 
-int set_extent_buffer_dirty(struct extent_io_tree *tree,
-			     struct extent_buffer *eb)
+int set_extent_buffer_dirty(struct extent_buffer *eb)
 {
 	unsigned long i;
 	unsigned long num_pages;
 	int was_dirty = 0;
 
+	check_buffer_tree_ref(eb);
+
 	was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
+
 	num_pages = num_extent_pages(eb->start, eb->len);
 	WARN_ON(atomic_read(&eb->refs) == 0);
+	WARN_ON(!test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags));
+
 	for (i = 0; i < num_pages; i++)
-		__set_page_dirty_nobuffers(extent_buffer_page(eb, i));
+		set_page_dirty(extent_buffer_page(eb, i));
 	return was_dirty;
 }
 
-static int __eb_straddles_pages(u64 start, u64 len)
+static int range_straddles_pages(u64 start, u64 len)
 {
 	if (len < PAGE_CACHE_SIZE)
 		return 1;
@@ -4037,25 +4281,14 @@ static int __eb_straddles_pages(u64 start, u64 len)
 	return 0;
 }
 
-static int eb_straddles_pages(struct extent_buffer *eb)
-{
-	return __eb_straddles_pages(eb->start, eb->len);
-}
-
-int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
-				struct extent_buffer *eb,
-				struct extent_state **cached_state)
+int clear_extent_buffer_uptodate(struct extent_buffer *eb)
 {
 	unsigned long i;
 	struct page *page;
 	unsigned long num_pages;
 
-	num_pages = num_extent_pages(eb->start, eb->len);
 	clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
-
-	clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
-			      cached_state, GFP_NOFS);
-
+	num_pages = num_extent_pages(eb->start, eb->len);
 	for (i = 0; i < num_pages; i++) {
 		page = extent_buffer_page(eb, i);
 		if (page)
@@ -4064,27 +4297,16 @@ int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
 	return 0;
 }
 
-int set_extent_buffer_uptodate(struct extent_io_tree *tree,
-				struct extent_buffer *eb)
+int set_extent_buffer_uptodate(struct extent_buffer *eb)
 {
 	unsigned long i;
 	struct page *page;
 	unsigned long num_pages;
 
+	set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
 	num_pages = num_extent_pages(eb->start, eb->len);
-
-	if (eb_straddles_pages(eb)) {
-		set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
-				    NULL, GFP_NOFS);
-	}
 	for (i = 0; i < num_pages; i++) {
 		page = extent_buffer_page(eb, i);
-		if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) ||
-		    ((i == num_pages - 1) &&
-		     ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) {
-			check_page_uptodate(tree, page);
-			continue;
-		}
 		SetPageUptodate(page);
 	}
 	return 0;
@@ -4099,7 +4321,7 @@ int extent_range_uptodate(struct extent_io_tree *tree,
 	int uptodate;
 	unsigned long index;
 
-	if (__eb_straddles_pages(start, end - start + 1)) {
+	if (range_straddles_pages(start, end - start + 1)) {
 		ret = test_range_bit(tree, start, end,
 				     EXTENT_UPTODATE, 1, NULL);
 		if (ret)
@@ -4121,35 +4343,9 @@ int extent_range_uptodate(struct extent_io_tree *tree,
 	return pg_uptodate;
 }
 
-int extent_buffer_uptodate(struct extent_io_tree *tree,
-			   struct extent_buffer *eb,
-			   struct extent_state *cached_state)
+int extent_buffer_uptodate(struct extent_buffer *eb)
 {
-	int ret = 0;
-	unsigned long num_pages;
-	unsigned long i;
-	struct page *page;
-	int pg_uptodate = 1;
-
-	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
-		return 1;
-
-	if (eb_straddles_pages(eb)) {
-		ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1,
-				   EXTENT_UPTODATE, 1, cached_state);
-		if (ret)
-			return ret;
-	}
-
-	num_pages = num_extent_pages(eb->start, eb->len);
-	for (i = 0; i < num_pages; i++) {
-		page = extent_buffer_page(eb, i);
-		if (!PageUptodate(page)) {
-			pg_uptodate = 0;
-			break;
-		}
-	}
-	return pg_uptodate;
+	return test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
 }
 
 int read_extent_buffer_pages(struct extent_io_tree *tree,
@@ -4171,13 +4367,6 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
 		return 0;
 
-	if (eb_straddles_pages(eb)) {
-		if (test_range_bit(tree, eb->start, eb->start + eb->len - 1,
-				   EXTENT_UPTODATE, 1, NULL)) {
-			return 0;
-		}
-	}
-
 	if (start) {
 		WARN_ON(start < eb->start);
 		start_i = (start >> PAGE_CACHE_SHIFT) -
@@ -4207,7 +4396,7 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 		goto unlock_exit;
 	}
 
-	atomic_set(&eb->pages_reading, num_reads);
+	atomic_set(&eb->io_pages, num_reads);
 	for (i = start_i; i < num_pages; i++) {
 		page = extent_buffer_page(eb, i);
 		if (!PageUptodate(page)) {
@@ -4235,8 +4424,6 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 			ret = -EIO;
 	}
 
-	if (!ret)
-		set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
 	return ret;
 
 unlock_exit:
@@ -4604,13 +4791,13 @@ int try_release_extent_buffer(struct page *page, gfp_t mask)
 	eb = (struct extent_buffer *)page->private;
 	BUG_ON(!eb);
 
-	/* 
+	/*
 	 * This is a little awful but should be ok, we need to make sure that
 	 * the eb doesn't disappear out from under us while we're looking at
 	 * this page.
 	 */
 	spin_lock(&eb->refs_lock);
-	if (atomic_read(&eb->refs) != 1 || extent_buffer_under_io(eb, page)) {
+	if (atomic_read(&eb->refs) != 1 || extent_buffer_under_io(eb)) {
 		spin_unlock(&eb->refs_lock);
 		spin_unlock(&page->mapping->private_lock);
 		return 0;

fs/btrfs/extent_io.h

@@ -37,6 +37,8 @@
 #define EXTENT_BUFFER_READAHEAD 4	/* this got triggered by readahead */
 #define EXTENT_BUFFER_TREE_REF 5
 #define EXTENT_BUFFER_STALE 6
+#define EXTENT_BUFFER_WRITEBACK 7
+#define EXTENT_BUFFER_IOERR 8
 
 /* these are flags for extent_clear_unlock_delalloc */
 #define EXTENT_CLEAR_UNLOCK_PAGE 0x1
@@ -99,6 +101,7 @@ struct extent_io_tree {
 	struct radix_tree_root buffer;
 	struct address_space *mapping;
 	u64 dirty_bytes;
+	int track_uptodate;
 	spinlock_t lock;
 	spinlock_t buffer_lock;
 	struct extent_io_ops *ops;
@@ -132,7 +135,7 @@ struct extent_buffer {
 	struct extent_io_tree *tree;
 	spinlock_t refs_lock;
 	atomic_t refs;
-	atomic_t pages_reading;
+	atomic_t io_pages;
 	struct list_head leak_list;
 	struct rcu_head rcu_head;
 	pid_t lock_owner;
@@ -249,6 +252,8 @@ int extent_writepages(struct extent_io_tree *tree,
 		      struct address_space *mapping,
 		      get_extent_t *get_extent,
 		      struct writeback_control *wbc);
+int btree_write_cache_pages(struct address_space *mapping,
+			    struct writeback_control *wbc);
 int extent_readpages(struct extent_io_tree *tree,
 		     struct address_space *mapping,
 		     struct list_head *pages, unsigned nr_pages,
@@ -297,18 +302,11 @@ void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
 void memset_extent_buffer(struct extent_buffer *eb, char c,
 			  unsigned long start, unsigned long len);
 int wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits);
-int clear_extent_buffer_dirty(struct extent_io_tree *tree,
-			      struct extent_buffer *eb);
-int set_extent_buffer_dirty(struct extent_io_tree *tree,
-			     struct extent_buffer *eb);
-int set_extent_buffer_uptodate(struct extent_io_tree *tree,
-			       struct extent_buffer *eb);
-int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
-				struct extent_buffer *eb,
-				struct extent_state **cached_state);
-int extent_buffer_uptodate(struct extent_io_tree *tree,
-			   struct extent_buffer *eb,
-			   struct extent_state *cached_state);
+int clear_extent_buffer_dirty(struct extent_buffer *eb);
+int set_extent_buffer_dirty(struct extent_buffer *eb);
+int set_extent_buffer_uptodate(struct extent_buffer *eb);
+int clear_extent_buffer_uptodate(struct extent_buffer *eb);
+int extent_buffer_uptodate(struct extent_buffer *eb);
 int map_private_extent_buffer(struct extent_buffer *eb, unsigned long offset,
 		      unsigned long min_len, char **map,
 		      unsigned long *map_start,

fs/btrfs/inode.c

@@ -6782,6 +6782,8 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	extent_map_tree_init(&ei->extent_tree);
 	extent_io_tree_init(&ei->io_tree, &inode->i_data);
 	extent_io_tree_init(&ei->io_failure_tree, &inode->i_data);
+	ei->io_tree.track_uptodate = 1;
+	ei->io_failure_tree.track_uptodate = 1;
 	mutex_init(&ei->log_mutex);
 	mutex_init(&ei->delalloc_mutex);
 	btrfs_ordered_inode_tree_init(&ei->ordered_tree);