1 files changed, 151 insertions, 349 deletions

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 31513860508..067b1747421 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -10,6 +10,8 @@
 #include <linux/swap.h>
 #include <linux/writeback.h>
 #include <linux/pagevec.h>
+#include <linux/prefetch.h>
+#include <linux/cleancache.h>
 #include "extent_io.h"
 #include "extent_map.h"
 #include "compat.h"
@@ -101,7 +103,7 @@ void extent_io_exit(void)
 }
 
 void extent_io_tree_init(struct extent_io_tree *tree,
-			  struct address_space *mapping, gfp_t mask)
+			 struct address_space *mapping)
 {
 	tree->state = RB_ROOT;
 	INIT_RADIX_TREE(&tree->buffer, GFP_ATOMIC);
@@ -279,11 +281,10 @@ static int merge_state(struct extent_io_tree *tree,
 		if (other->start == state->end + 1 &&
 		    other->state == state->state) {
 			merge_cb(tree, state, other);
-			other->start = state->start;
-			state->tree = NULL;
-			rb_erase(&state->rb_node, &tree->state);
-			free_extent_state(state);
-			state = NULL;
+			state->end = other->end;
+			other->tree = NULL;
+			rb_erase(&other->rb_node, &tree->state);
+			free_extent_state(other);
 		}
 	}
 
@@ -349,7 +350,6 @@ static int insert_state(struct extent_io_tree *tree,
 		       "%llu %llu\n", (unsigned long long)found->start,
 		       (unsigned long long)found->end,
 		       (unsigned long long)start, (unsigned long long)end);
-		free_extent_state(state);
 		return -EEXIST;
 	}
 	state->tree = tree;
@@ -439,6 +439,15 @@ static int clear_state_bit(struct extent_io_tree *tree,
 	return ret;
 }
 
+static struct extent_state *
+alloc_extent_state_atomic(struct extent_state *prealloc)
+{
+	if (!prealloc)
+		prealloc = alloc_extent_state(GFP_ATOMIC);
+
+	return prealloc;
+}
+
 /*
  * clear some bits on a range in the tree.  This may require splitting
  * or inserting elements in the tree, so the gfp mask is used to
@@ -489,7 +498,8 @@ again:
 			cached_state = NULL;
 		}
 
-		if (cached && cached->tree && cached->start == start) {
+		if (cached && cached->tree && cached->start <= start &&
+		    cached->end > start) {
 			if (clear)
 				atomic_dec(&cached->refs);
 			state = cached;
@@ -529,8 +539,8 @@ hit_next:
 	 */
 
 	if (state->start < start) {
-		if (!prealloc)
-			prealloc = alloc_extent_state(GFP_ATOMIC);
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
 		err = split_state(tree, state, prealloc, start);
 		BUG_ON(err == -EEXIST);
 		prealloc = NULL;
@@ -551,8 +561,8 @@ hit_next:
 	 * on the first half
 	 */
 	if (state->start <= end && state->end > end) {
-		if (!prealloc)
-			prealloc = alloc_extent_state(GFP_ATOMIC);
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
 		err = split_state(tree, state, prealloc, end + 1);
 		BUG_ON(err == -EEXIST);
 		if (wake)
@@ -725,14 +735,14 @@ int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 again:
 	if (!prealloc && (mask & __GFP_WAIT)) {
 		prealloc = alloc_extent_state(mask);
-		if (!prealloc)
-			return -ENOMEM;
+		BUG_ON(!prealloc);
 	}
 
 	spin_lock(&tree->lock);
 	if (cached_state && *cached_state) {
 		state = *cached_state;
-		if (state->start == start && state->tree) {
+		if (state->start <= start && state->end > start &&
+		    state->tree) {
 			node = &state->rb_node;
 			goto hit_next;
 		}
@@ -743,6 +753,8 @@ again:
 	 */
 	node = tree_search(tree, start);
 	if (!node) {
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
 		err = insert_state(tree, prealloc, start, end, &bits);
 		prealloc = NULL;
 		BUG_ON(err == -EEXIST);
@@ -777,14 +789,12 @@ hit_next:
 			goto out;
 
 		start = last_end + 1;
-		if (start < end && prealloc && !need_resched()) {
-			next_node = rb_next(node);
-			if (next_node) {
-				state = rb_entry(next_node, struct extent_state,
-						 rb_node);
-				if (state->start == start)
-					goto hit_next;
-			}
+		next_node = rb_next(&state->rb_node);
+		if (next_node && start < end && prealloc && !need_resched()) {
+			state = rb_entry(next_node, struct extent_state,
+					 rb_node);
+			if (state->start == start)
+				goto hit_next;
 		}
 		goto search_again;
 	}
@@ -811,6 +821,9 @@ hit_next:
 			err = -EEXIST;
 			goto out;
 		}
+
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
 		err = split_state(tree, state, prealloc, start);
 		BUG_ON(err == -EEXIST);
 		prealloc = NULL;
@@ -841,10 +854,19 @@ hit_next:
 			this_end = end;
 		else
 			this_end = last_start - 1;
+
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
+
+		/*
+		 * Avoid to free 'prealloc' if it can be merged with
+		 * the later extent.
+		 */
 		err = insert_state(tree, prealloc, start, this_end,
 				   &bits);
 		BUG_ON(err == -EEXIST);
 		if (err) {
+			free_extent_state(prealloc);
 			prealloc = NULL;
 			goto out;
 		}
@@ -865,6 +887,9 @@ hit_next:
 			err = -EEXIST;
 			goto out;
 		}
+
+		prealloc = alloc_extent_state_atomic(prealloc);
+		BUG_ON(!prealloc);
 		err = split_state(tree, state, prealloc, end + 1);
 		BUG_ON(err == -EEXIST);
 
@@ -941,13 +966,6 @@ int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
 			      NULL, mask);
 }
 
-static int clear_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
-		       gfp_t mask)
-{
-	return clear_extent_bit(tree, start, end, EXTENT_NEW, 0, 0,
-				NULL, mask);
-}
-
 int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
 			struct extent_state **cached_state, gfp_t mask)
 {
@@ -963,11 +981,6 @@ static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
 				cached_state, mask);
 }
 
-int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end)
-{
-	return wait_extent_bit(tree, start, end, EXTENT_WRITEBACK);
-}
-
 /*
  * either insert or lock state struct between start and end use mask to tell
  * us if waiting is desired.
@@ -1028,25 +1041,6 @@ int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
 }
 
 /*
- * helper function to set pages and extents in the tree dirty
- */
-int set_range_dirty(struct extent_io_tree *tree, u64 start, u64 end)
-{
-	unsigned long index = start >> PAGE_CACHE_SHIFT;
-	unsigned long end_index = end >> PAGE_CACHE_SHIFT;
-	struct page *page;
-
-	while (index <= end_index) {
-		page = find_get_page(tree->mapping, index);
-		BUG_ON(!page);
-		__set_page_dirty_nobuffers(page);
-		page_cache_release(page);
-		index++;
-	}
-	return 0;
-}
-
-/*
  * helper function to set both pages and extents in the tree writeback
  */
 static int set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end)
@@ -1480,7 +1474,7 @@ u64 count_range_bits(struct extent_io_tree *tree,
 			if (total_bytes >= max_bytes)
 				break;
 			if (!found) {
-				*start = state->start;
+				*start = max(cur_start, state->start);
 				found = 1;
 			}
 			last = state->end;
@@ -1568,7 +1562,8 @@ int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
 	int bitset = 0;
 
 	spin_lock(&tree->lock);
-	if (cached && cached->tree && cached->start == start)
+	if (cached && cached->tree && cached->start <= start &&
+	    cached->end > start)
 		node = &cached->rb_node;
 	else
 		node = tree_search(tree, start);
@@ -1819,46 +1814,6 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 	bio_put(bio);
 }
 
-/*
- * IO done from prepare_write is pretty simple, we just unlock
- * the structs in the extent tree when done, and set the uptodate bits
- * as appropriate.
- */
-static void end_bio_extent_preparewrite(struct bio *bio, int err)
-{
-	const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
-	struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
-	struct extent_io_tree *tree;
-	u64 start;
-	u64 end;
-
-	do {
-		struct page *page = bvec->bv_page;
-		struct extent_state *cached = NULL;
-		tree = &BTRFS_I(page->mapping->host)->io_tree;
-
-		start = ((u64)page->index << PAGE_CACHE_SHIFT) +
-			bvec->bv_offset;
-		end = start + bvec->bv_len - 1;
-
-		if (--bvec >= bio->bi_io_vec)
-			prefetchw(&bvec->bv_page->flags);
-
-		if (uptodate) {
-			set_extent_uptodate(tree, start, end, &cached,
-					    GFP_ATOMIC);
-		} else {
-			ClearPageUptodate(page);
-			SetPageError(page);
-		}
-
-		unlock_extent_cached(tree, start, end, &cached, GFP_ATOMIC);
-
-	} while (bvec >= bio->bi_io_vec);
-
-	bio_put(bio);
-}
-
 struct bio *
 btrfs_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
 		gfp_t gfp_flags)
@@ -2007,7 +1962,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 	struct btrfs_ordered_extent *ordered;
 	int ret;
 	int nr = 0;
-	size_t page_offset = 0;
+	size_t pg_offset = 0;
 	size_t iosize;
 	size_t disk_io_size;
 	size_t blocksize = inode->i_sb->s_blocksize;
@@ -2015,6 +1970,13 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 
 	set_page_extent_mapped(page);
 
+	if (!PageUptodate(page)) {
+		if (cleancache_get_page(page) == 0) {
+			BUG_ON(blocksize != PAGE_SIZE);
+			goto out;
+		}
+	}
+
 	end = page_end;
 	while (1) {
 		lock_extent(tree, start, end, GFP_NOFS);
@@ -2043,9 +2005,9 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 			char *userpage;
 			struct extent_state *cached = NULL;
 
-			iosize = PAGE_CACHE_SIZE - page_offset;
+			iosize = PAGE_CACHE_SIZE - pg_offset;
 			userpage = kmap_atomic(page, KM_USER0);
-			memset(userpage + page_offset, 0, iosize);
+			memset(userpage + pg_offset, 0, iosize);
 			flush_dcache_page(page);
 			kunmap_atomic(userpage, KM_USER0);
 			set_extent_uptodate(tree, cur, cur + iosize - 1,
@@ -2054,9 +2016,9 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 					     &cached, GFP_NOFS);
 			break;
 		}
-		em = get_extent(inode, page, page_offset, cur,
+		em = get_extent(inode, page, pg_offset, cur,
 				end - cur + 1, 0);
-		if (IS_ERR(em) || !em) {
+		if (IS_ERR_OR_NULL(em)) {
 			SetPageError(page);
 			unlock_extent(tree, cur, end, GFP_NOFS);
 			break;
@@ -2094,7 +2056,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 			struct extent_state *cached = NULL;
 
 			userpage = kmap_atomic(page, KM_USER0);
-			memset(userpage + page_offset, 0, iosize);
+			memset(userpage + pg_offset, 0, iosize);
 			flush_dcache_page(page);
 			kunmap_atomic(userpage, KM_USER0);
 
@@ -2103,7 +2065,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 			unlock_extent_cached(tree, cur, cur + iosize - 1,
 			                     &cached, GFP_NOFS);
 			cur = cur + iosize;
-			page_offset += iosize;
+			pg_offset += iosize;
 			continue;
 		}
 		/* the get_extent function already copied into the page */
@@ -2112,7 +2074,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 			check_page_uptodate(tree, page);
 			unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
 			cur = cur + iosize;
-			page_offset += iosize;
+			pg_offset += iosize;
 			continue;
 		}
 		/* we have an inline extent but it didn't get marked up
@@ -2122,7 +2084,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 			SetPageError(page);
 			unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
 			cur = cur + iosize;
-			page_offset += iosize;
+			pg_offset += iosize;
 			continue;
 		}
 
@@ -2135,7 +2097,7 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 			unsigned long pnr = (last_byte >> PAGE_CACHE_SHIFT) + 1;
 			pnr -= page->index;
 			ret = submit_extent_page(READ, tree, page,
-					 sector, disk_io_size, page_offset,
+					 sector, disk_io_size, pg_offset,
 					 bdev, bio, pnr,
 					 end_bio_extent_readpage, mirror_num,
 					 *bio_flags,
@@ -2146,8 +2108,9 @@ static int __extent_read_full_page(struct extent_io_tree *tree,
 		if (ret)
 			SetPageError(page);
 		cur = cur + iosize;
-		page_offset += iosize;
+		pg_offset += iosize;
 	}
+out:
 	if (!nr) {
 		if (!PageError(page))
 			SetPageUptodate(page);
@@ -2341,7 +2304,7 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 		}
 		em = epd->get_extent(inode, page, pg_offset, cur,
 				     end - cur + 1, 1);
-		if (IS_ERR(em) || !em) {
+		if (IS_ERR_OR_NULL(em)) {
 			SetPageError(page);
 			break;
 		}
@@ -2468,6 +2431,7 @@ static int extent_write_cache_pages(struct extent_io_tree *tree,
 	pgoff_t index;
 	pgoff_t end;		/* Inclusive */
 	int scanned = 0;
+	int tag;
 
 	pagevec_init(&pvec, 0);
 	if (wbc->range_cyclic) {
@@ -2478,11 +2442,16 @@ static int extent_write_cache_pages(struct extent_io_tree *tree,
 		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,
-			      PAGECACHE_TAG_DIRTY, min(end - index,
-				  (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
+	       (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
+			min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
 		unsigned i;
 
 		scanned = 1;
@@ -2587,7 +2556,6 @@ int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
 	};
 	struct writeback_control wbc_writepages = {
 		.sync_mode	= wbc->sync_mode,
-		.older_than_this = NULL,
 		.nr_to_write	= 64,
 		.range_start	= page_offset(page) + PAGE_CACHE_SIZE,
 		.range_end	= (loff_t)-1,
@@ -2620,7 +2588,6 @@ int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode,
 	};
 	struct writeback_control wbc_writepages = {
 		.sync_mode	= mode,
-		.older_than_this = NULL,
 		.nr_to_write	= nr_pages * 2,
 		.range_start	= start,
 		.range_end	= end + 1,
@@ -2681,7 +2648,7 @@ int extent_readpages(struct extent_io_tree *tree,
 		prefetchw(&page->flags);
 		list_del(&page->lru);
 		if (!add_to_page_cache_lru(page, mapping,
-					page->index, GFP_KERNEL)) {
+					page->index, GFP_NOFS)) {
 			__extent_read_full_page(tree, page, get_extent,
 						&bio, 0, &bio_flags);
 		}
@@ -2720,128 +2687,6 @@ int extent_invalidatepage(struct extent_io_tree *tree,
 }
 
 /*
- * simple commit_write call, set_range_dirty is used to mark both
- * the pages and the extent records as dirty
- */
-int extent_commit_write(struct extent_io_tree *tree,
-			struct inode *inode, struct page *page,
-			unsigned from, unsigned to)
-{
-	loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to;
-
-	set_page_extent_mapped(page);
-	set_page_dirty(page);
-
-	if (pos > inode->i_size) {
-		i_size_write(inode, pos);
-		mark_inode_dirty(inode);
-	}
-	return 0;
-}
-
-int extent_prepare_write(struct extent_io_tree *tree,
-			 struct inode *inode, struct page *page,
-			 unsigned from, unsigned to, get_extent_t *get_extent)
-{
-	u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
-	u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
-	u64 block_start;
-	u64 orig_block_start;
-	u64 block_end;
-	u64 cur_end;
-	struct extent_map *em;
-	unsigned blocksize = 1 << inode->i_blkbits;
-	size_t page_offset = 0;
-	size_t block_off_start;
-	size_t block_off_end;
-	int err = 0;
-	int iocount = 0;
-	int ret = 0;
-	int isnew;
-
-	set_page_extent_mapped(page);
-
-	block_start = (page_start + from) & ~((u64)blocksize - 1);
-	block_end = (page_start + to - 1) | (blocksize - 1);
-	orig_block_start = block_start;
-
-	lock_extent(tree, page_start, page_end, GFP_NOFS);
-	while (block_start <= block_end) {
-		em = get_extent(inode, page, page_offset, block_start,
-				block_end - block_start + 1, 1);
-		if (IS_ERR(em) || !em)
-			goto err;
-
-		cur_end = min(block_end, extent_map_end(em) - 1);
-		block_off_start = block_start & (PAGE_CACHE_SIZE - 1);
-		block_off_end = block_off_start + blocksize;
-		isnew = clear_extent_new(tree, block_start, cur_end, GFP_NOFS);
-
-		if (!PageUptodate(page) && isnew &&
-		    (block_off_end > to || block_off_start < from)) {
-			void *kaddr;
-
-			kaddr = kmap_atomic(page, KM_USER0);
-			if (block_off_end > to)
-				memset(kaddr + to, 0, block_off_end - to);
-			if (block_off_start < from)
-				memset(kaddr + block_off_start, 0,
-				       from - block_off_start);
-			flush_dcache_page(page);
-			kunmap_atomic(kaddr, KM_USER0);
-		}
-		if ((em->block_start != EXTENT_MAP_HOLE &&
-		     em->block_start != EXTENT_MAP_INLINE) &&
-		    !isnew && !PageUptodate(page) &&
-		    (block_off_end > to || block_off_start < from) &&
-		    !test_range_bit(tree, block_start, cur_end,
-				    EXTENT_UPTODATE, 1, NULL)) {
-			u64 sector;
-			u64 extent_offset = block_start - em->start;
-			size_t iosize;
-			sector = (em->block_start + extent_offset) >> 9;
-			iosize = (cur_end - block_start + blocksize) &
-				~((u64)blocksize - 1);
-			/*
-			 * we've already got the extent locked, but we
-			 * need to split the state such that our end_bio
-			 * handler can clear the lock.
-			 */
-			set_extent_bit(tree, block_start,
-				       block_start + iosize - 1,
-				       EXTENT_LOCKED, 0, NULL, NULL, GFP_NOFS);
-			ret = submit_extent_page(READ, tree, page,
-					 sector, iosize, page_offset, em->bdev,
-					 NULL, 1,
-					 end_bio_extent_preparewrite, 0,
-					 0, 0);
-			if (ret && !err)
-				err = ret;
-			iocount++;
-			block_start = block_start + iosize;
-		} else {
-			struct extent_state *cached = NULL;
-
-			set_extent_uptodate(tree, block_start, cur_end, &cached,
-					    GFP_NOFS);
-			unlock_extent_cached(tree, block_start, cur_end,
-					     &cached, GFP_NOFS);
-			block_start = cur_end + 1;
-		}
-		page_offset = block_start & (PAGE_CACHE_SIZE - 1);
-		free_extent_map(em);
-	}
-	if (iocount) {
-		wait_extent_bit(tree, orig_block_start,
-				block_end, EXTENT_LOCKED);
-	}
-	check_page_uptodate(tree, page);
-err:
-	/* FIXME, zero out newly allocated blocks on error */
-	return err;
-}
-
-/*
  * a helper for releasepage, this tests for areas of the page that
  * are locked or under IO and drops the related state bits if it is safe
  * to drop the page.
@@ -2899,7 +2744,7 @@ int try_release_extent_mapping(struct extent_map_tree *map,
 			len = end - start + 1;
 			write_lock(&map->lock);
 			em = lookup_extent_mapping(map, start, len);
-			if (!em || IS_ERR(em)) {
+			if (IS_ERR_OR_NULL(em)) {
 				write_unlock(&map->lock);
 				break;
 			}
@@ -2927,33 +2772,6 @@ int try_release_extent_mapping(struct extent_map_tree *map,
 	return try_release_extent_state(map, tree, page, mask);
 }
 
-sector_t extent_bmap(struct address_space *mapping, sector_t iblock,
-		get_extent_t *get_extent)
-{
-	struct inode *inode = mapping->host;
-	struct extent_state *cached_state = NULL;
-	u64 start = iblock << inode->i_blkbits;
-	sector_t sector = 0;
-	size_t blksize = (1 << inode->i_blkbits);
-	struct extent_map *em;
-
-	lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + blksize - 1,
-			 0, &cached_state, GFP_NOFS);
-	em = get_extent(inode, NULL, 0, start, blksize, 0);
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, start,
-			     start + blksize - 1, &cached_state, GFP_NOFS);
-	if (!em || IS_ERR(em))
-		return 0;
-
-	if (em->block_start > EXTENT_MAP_LAST_BYTE)
-		goto out;
-
-	sector = (em->block_start + start - em->start) >> inode->i_blkbits;
-out:
-	free_extent_map(em);
-	return sector;
-}
-
 /*
  * helper function for fiemap, which doesn't want to see any holes.
  * This maps until we find something past 'last'
@@ -2976,7 +2794,7 @@ static struct extent_map *get_extent_skip_holes(struct inode *inode,
 			break;
 		len = (len + sectorsize - 1) & ~(sectorsize - 1);
 		em = get_extent(inode, NULL, 0, offset, len, 0);
-		if (!em || IS_ERR(em))
+		if (IS_ERR_OR_NULL(em))
 			return em;
 
 		/* if this isn't a hole return it */
@@ -3030,7 +2848,7 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 	 * because there might be preallocation past i_size
 	 */
 	ret = btrfs_lookup_file_extent(NULL, BTRFS_I(inode)->root,
-				       path, inode->i_ino, -1, 0);
+				       path, btrfs_ino(inode), -1, 0);
 	if (ret < 0) {
 		btrfs_free_path(path);
 		return ret;
@@ -3043,7 +2861,7 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 	found_type = btrfs_key_type(&found_key);
 
 	/* No extents, but there might be delalloc bits */
-	if (found_key.objectid != inode->i_ino ||
+	if (found_key.objectid != btrfs_ino(inode) ||
 	    found_type != BTRFS_EXTENT_DATA_KEY) {
 		/* have to trust i_size as the end */
 		last = (u64)-1;
@@ -3207,8 +3025,15 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
 		return NULL;
 	eb->start = start;
 	eb->len = len;
-	spin_lock_init(&eb->lock);
-	init_waitqueue_head(&eb->lock_wq);
+	rwlock_init(&eb->lock);
+	atomic_set(&eb->write_locks, 0);
+	atomic_set(&eb->read_locks, 0);
+	atomic_set(&eb->blocking_readers, 0);
+	atomic_set(&eb->blocking_writers, 0);
+	atomic_set(&eb->spinning_readers, 0);
+	atomic_set(&eb->spinning_writers, 0);
+	init_waitqueue_head(&eb->write_lock_wq);
+	init_waitqueue_head(&eb->read_lock_wq);
 
 #if LEAK_DEBUG
 	spin_lock_irqsave(&leak_lock, flags);
@@ -3266,8 +3091,7 @@ static inline void btrfs_release_extent_buffer(struct extent_buffer *eb)
 
 struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 					  u64 start, unsigned long len,
-					  struct page *page0,
-					  gfp_t mask)
+					  struct page *page0)
 {
 	unsigned long num_pages = num_extent_pages(start, len);
 	unsigned long i;
@@ -3288,7 +3112,7 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 	}
 	rcu_read_unlock();
 
-	eb = __alloc_extent_buffer(tree, start, len, mask);
+	eb = __alloc_extent_buffer(tree, start, len, GFP_NOFS);
 	if (!eb)
 		return NULL;
 
@@ -3305,7 +3129,7 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
 		i = 0;
 	}
 	for (; i < num_pages; i++, index++) {
-		p = find_or_create_page(mapping, index, mask | __GFP_HIGHMEM);
+		p = find_or_create_page(mapping, index, GFP_NOFS);
 		if (!p) {
 			WARN_ON(1);
 			goto free_eb;
@@ -3377,8 +3201,7 @@ free_eb:
 }
 
 struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree,
-					 u64 start, unsigned long len,
-					  gfp_t mask)
+					 u64 start, unsigned long len)
 {
 	struct extent_buffer *eb;
 
@@ -3439,13 +3262,6 @@ int clear_extent_buffer_dirty(struct extent_io_tree *tree,
 	return 0;
 }
 
-int wait_on_extent_buffer_writeback(struct extent_io_tree *tree,
-				    struct extent_buffer *eb)
-{
-	return wait_on_extent_writeback(tree, eb->start,
-					eb->start + eb->len - 1);
-}
-
 int set_extent_buffer_dirty(struct extent_io_tree *tree,
 			     struct extent_buffer *eb)
 {
@@ -3460,6 +3276,22 @@ int set_extent_buffer_dirty(struct extent_io_tree *tree,
 	return was_dirty;
 }
 
+static int __eb_straddles_pages(u64 start, u64 len)
+{
+	if (len < PAGE_CACHE_SIZE)
+		return 1;
+	if (start & (PAGE_CACHE_SIZE - 1))
+		return 1;
+	if ((start + len) & (PAGE_CACHE_SIZE - 1))
+		return 1;
+	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)
@@ -3471,8 +3303,10 @@ int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
 	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);
+	if (eb_straddles_pages(eb)) {
+		clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
+				      cached_state, GFP_NOFS);
+	}
 	for (i = 0; i < num_pages; i++) {
 		page = extent_buffer_page(eb, i);
 		if (page)
@@ -3490,8 +3324,10 @@ int set_extent_buffer_uptodate(struct extent_io_tree *tree,
 
 	num_pages = num_extent_pages(eb->start, eb->len);
 
-	set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
-			    NULL, GFP_NOFS);
+	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))) ||
@@ -3514,9 +3350,12 @@ int extent_range_uptodate(struct extent_io_tree *tree,
 	int uptodate;
 	unsigned long index;
 
-	ret = test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL);
-	if (ret)
-		return 1;
+	if (__eb_straddles_pages(start, end - start + 1)) {
+		ret = test_range_bit(tree, start, end,
+				     EXTENT_UPTODATE, 1, NULL);
+		if (ret)
+			return 1;
+	}
 	while (start <= end) {
 		index = start >> PAGE_CACHE_SHIFT;
 		page = find_get_page(tree->mapping, index);
@@ -3544,10 +3383,12 @@ int extent_buffer_uptodate(struct extent_io_tree *tree,
 	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
 		return 1;
 
-	ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1,
-			   EXTENT_UPTODATE, 1, cached_state);
-	if (ret)
-		return ret;
+	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++) {
@@ -3580,9 +3421,11 @@ int read_extent_buffer_pages(struct extent_io_tree *tree,
 	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
 		return 0;
 
-	if (test_range_bit(tree, eb->start, eb->start + eb->len - 1,
-			   EXTENT_UPTODATE, 1, NULL)) {
-		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) {
@@ -3686,9 +3529,8 @@ void read_extent_buffer(struct extent_buffer *eb, void *dstv,
 		page = extent_buffer_page(eb, i);
 
 		cur = min(len, (PAGE_CACHE_SIZE - offset));
-		kaddr = kmap_atomic(page, KM_USER1);
+		kaddr = page_address(page);
 		memcpy(dst, kaddr + offset, cur);
-		kunmap_atomic(kaddr, KM_USER1);
 
 		dst += cur;
 		len -= cur;
@@ -3698,9 +3540,9 @@ void read_extent_buffer(struct extent_buffer *eb, void *dstv,
 }
 
 int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start,
-			       unsigned long min_len, char **token, char **map,
+			       unsigned long min_len, char **map,
 			       unsigned long *map_start,
-			       unsigned long *map_len, int km)
+			       unsigned long *map_len)
 {
 	size_t offset = start & (PAGE_CACHE_SIZE - 1);
 	char *kaddr;
@@ -3730,42 +3572,12 @@ int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start,
 	}
 
 	p = extent_buffer_page(eb, i);
-	kaddr = kmap_atomic(p, km);
-	*token = kaddr;
+	kaddr = page_address(p);
 	*map = kaddr + offset;
 	*map_len = PAGE_CACHE_SIZE - offset;
 	return 0;
 }
 
-int map_extent_buffer(struct extent_buffer *eb, unsigned long start,
-		      unsigned long min_len,
-		      char **token, char **map,
-		      unsigned long *map_start,
-		      unsigned long *map_len, int km)
-{
-	int err;
-	int save = 0;
-	if (eb->map_token) {
-		unmap_extent_buffer(eb, eb->map_token, km);
-		eb->map_token = NULL;
-		save = 1;
-	}
-	err = map_private_extent_buffer(eb, start, min_len, token, map,
-				       map_start, map_len, km);
-	if (!err && save) {
-		eb->map_token = *token;
-		eb->kaddr = *map;
-		eb->map_start = *map_start;
-		eb->map_len = *map_len;
-	}
-	return err;
-}
-
-void unmap_extent_buffer(struct extent_buffer *eb, char *token, int km)
-{
-	kunmap_atomic(token, km);
-}
-
 int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv,
 			  unsigned long start,
 			  unsigned long len)
@@ -3789,9 +3601,8 @@ int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv,
 
 		cur = min(len, (PAGE_CACHE_SIZE - offset));
 
-		kaddr = kmap_atomic(page, KM_USER0);
+		kaddr = page_address(page);
 		ret = memcmp(ptr, kaddr + offset, cur);
-		kunmap_atomic(kaddr, KM_USER0);
 		if (ret)
 			break;
 
@@ -3824,9 +3635,8 @@ void write_extent_buffer(struct extent_buffer *eb, const void *srcv,
 		WARN_ON(!PageUptodate(page));
 
 		cur = min(len, PAGE_CACHE_SIZE - offset);
-		kaddr = kmap_atomic(page, KM_USER1);
+		kaddr = page_address(page);
 		memcpy(kaddr + offset, src, cur);
-		kunmap_atomic(kaddr, KM_USER1);
 
 		src += cur;
 		len -= cur;
@@ -3855,9 +3665,8 @@ void memset_extent_buffer(struct extent_buffer *eb, char c,
 		WARN_ON(!PageUptodate(page));
 
 		cur = min(len, PAGE_CACHE_SIZE - offset);
-		kaddr = kmap_atomic(page, KM_USER0);
+		kaddr = page_address(page);
 		memset(kaddr + offset, c, cur);
-		kunmap_atomic(kaddr, KM_USER0);
 
 		len -= cur;
 		offset = 0;
@@ -3888,9 +3697,8 @@ void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
 
 		cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - offset));
 
-		kaddr = kmap_atomic(page, KM_USER0);
+		kaddr = page_address(page);
 		read_extent_buffer(src, kaddr + offset, src_offset, cur);
-		kunmap_atomic(kaddr, KM_USER0);
 
 		src_offset += cur;
 		len -= cur;
@@ -3903,20 +3711,17 @@ static void move_pages(struct page *dst_page, struct page *src_page,
 		       unsigned long dst_off, unsigned long src_off,
 		       unsigned long len)
 {
-	char *dst_kaddr = kmap_atomic(dst_page, KM_USER0);
+	char *dst_kaddr = page_address(dst_page);
 	if (dst_page == src_page) {
 		memmove(dst_kaddr + dst_off, dst_kaddr + src_off, len);
 	} else {
-		char *src_kaddr = kmap_atomic(src_page, KM_USER1);
+		char *src_kaddr = page_address(src_page);
 		char *p = dst_kaddr + dst_off + len;
 		char *s = src_kaddr + src_off + len;
 
 		while (len--)
 			*--p = *--s;
-
-		kunmap_atomic(src_kaddr, KM_USER1);
 	}
-	kunmap_atomic(dst_kaddr, KM_USER0);
 }
 
 static inline bool areas_overlap(unsigned long src, unsigned long dst, unsigned long len)
@@ -3929,20 +3734,17 @@ static void copy_pages(struct page *dst_page, struct page *src_page,
 		       unsigned long dst_off, unsigned long src_off,
 		       unsigned long len)
 {
-	char *dst_kaddr = kmap_atomic(dst_page, KM_USER0);
+	char *dst_kaddr = page_address(dst_page);
 	char *src_kaddr;
 
 	if (dst_page != src_page) {
-		src_kaddr = kmap_atomic(src_page, KM_USER1);
+		src_kaddr = page_address(src_page);
 	} else {
 		src_kaddr = dst_kaddr;
 		BUG_ON(areas_overlap(src_off, dst_off, len));
 	}
 
 	memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
-	kunmap_atomic(dst_kaddr, KM_USER0);
-	if (dst_page != src_page)
-		kunmap_atomic(src_kaddr, KM_USER1);
 }
 
 void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,