Merge branch 'master' into for-2.6.33

72 files changed, 1896 insertions, 449 deletions

diff --git a/fs/9p/cache.c b/fs/9p/cache.c
index 51c94e26a34..e777961939f 100644
--- a/fs/9p/cache.c
+++ b/fs/9p/cache.c
@@ -343,18 +343,7 @@ int __v9fs_fscache_release_page(struct page *page, gfp_t gfp)
 
 	BUG_ON(!vcookie->fscache);
 
-	if (PageFsCache(page)) {
-		if (fscache_check_page_write(vcookie->fscache, page)) {
-			if (!(gfp & __GFP_WAIT))
-				return 0;
-			fscache_wait_on_page_write(vcookie->fscache, page);
-		}
-
-		fscache_uncache_page(vcookie->fscache, page);
-		ClearPageFsCache(page);
-	}
-
-	return 1;
+	return fscache_maybe_release_page(vcookie->fscache, page, gfp);
 }
 
 void __v9fs_fscache_invalidate_page(struct page *page)
@@ -368,7 +357,6 @@ void __v9fs_fscache_invalidate_page(struct page *page)
 		fscache_wait_on_page_write(vcookie->fscache, page);
 		BUG_ON(!PageLocked(page));
 		fscache_uncache_page(vcookie->fscache, page);
-		ClearPageFsCache(page);
 	}
 }
 
diff --git a/fs/9p/vfs_dir.c b/fs/9p/vfs_dir.c
index 873cd31baa4..15cce53bf61 100644
--- a/fs/9p/vfs_dir.c
+++ b/fs/9p/vfs_dir.c
@@ -40,6 +40,24 @@
 #include "fid.h"
 
 /**
+ * struct p9_rdir - readdir accounting
+ * @mutex: mutex protecting readdir
+ * @head: start offset of current dirread buffer
+ * @tail: end offset of current dirread buffer
+ * @buf: dirread buffer
+ *
+ * private structure for keeping track of readdir
+ * allocated on demand
+ */
+
+struct p9_rdir {
+	struct mutex mutex;
+	int head;
+	int tail;
+	uint8_t *buf;
+};
+
+/**
  * dt_type - return file type
  * @mistat: mistat structure
  *
@@ -70,56 +88,79 @@ static int v9fs_dir_readdir(struct file *filp, void *dirent, filldir_t filldir)
 {
 	int over;
 	struct p9_wstat st;
-	int err;
+	int err = 0;
 	struct p9_fid *fid;
 	int buflen;
-	char *statbuf;
-	int n, i = 0;
+	int reclen = 0;
+	struct p9_rdir *rdir;
 
 	P9_DPRINTK(P9_DEBUG_VFS, "name %s\n", filp->f_path.dentry->d_name.name);
 	fid = filp->private_data;
 
 	buflen = fid->clnt->msize - P9_IOHDRSZ;
-	statbuf = kmalloc(buflen, GFP_KERNEL);
-	if (!statbuf)
-		return -ENOMEM;
-
-	while (1) {
-		err = v9fs_file_readn(filp, statbuf, NULL, buflen,
-								fid->rdir_fpos);
-		if (err <= 0)
-			break;
-
-		n = err;
-		while (i < n) {
-			err = p9stat_read(statbuf + i, buflen-i, &st,
-							fid->clnt->dotu);
+
+	/* allocate rdir on demand */
+	if (!fid->rdir) {
+		rdir = kmalloc(sizeof(struct p9_rdir) + buflen, GFP_KERNEL);
+
+		if (rdir == NULL) {
+			err = -ENOMEM;
+			goto exit;
+		}
+		spin_lock(&filp->f_dentry->d_lock);
+		if (!fid->rdir) {
+			rdir->buf = (uint8_t *)rdir + sizeof(struct p9_rdir);
+			mutex_init(&rdir->mutex);
+			rdir->head = rdir->tail = 0;
+			fid->rdir = (void *) rdir;
+			rdir = NULL;
+		}
+		spin_unlock(&filp->f_dentry->d_lock);
+		kfree(rdir);
+	}
+	rdir = (struct p9_rdir *) fid->rdir;
+
+	err = mutex_lock_interruptible(&rdir->mutex);
+	while (err == 0) {
+		if (rdir->tail == rdir->head) {
+			err = v9fs_file_readn(filp, rdir->buf, NULL,
+							buflen, filp->f_pos);
+			if (err <= 0)
+				goto unlock_and_exit;
+
+			rdir->head = 0;
+			rdir->tail = err;
+		}
+
+		while (rdir->head < rdir->tail) {
+			err = p9stat_read(rdir->buf + rdir->head,
+						buflen - rdir->head, &st,
+						fid->clnt->dotu);
 			if (err) {
 				P9_DPRINTK(P9_DEBUG_VFS, "returned %d\n", err);
 				err = -EIO;
 				p9stat_free(&st);
-				goto free_and_exit;
+				goto unlock_and_exit;
 			}
-
-			i += st.size+2;
-			fid->rdir_fpos += st.size+2;
+			reclen = st.size+2;
 
 			over = filldir(dirent, st.name, strlen(st.name),
 			    filp->f_pos, v9fs_qid2ino(&st.qid), dt_type(&st));
 
-			filp->f_pos += st.size+2;
-
 			p9stat_free(&st);
 
 			if (over) {
 				err = 0;
-				goto free_and_exit;
+				goto unlock_and_exit;
 			}
+			rdir->head += reclen;
+			filp->f_pos += reclen;
 		}
 	}
 
-free_and_exit:
-	kfree(statbuf);
+unlock_and_exit:
+	mutex_unlock(&rdir->mutex);
+exit:
 	return err;
 }
 
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 5947628aefe..18f74ec4dce 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -994,8 +994,7 @@ static int v9fs_readlink(struct dentry *dentry, char *buffer, int buflen)
 	P9_DPRINTK(P9_DEBUG_VFS,
 		"%s -> %s (%s)\n", dentry->d_name.name, st->extension, buffer);
 
-	retval = buflen;
-
+	retval = strnlen(buffer, buflen);
 done:
 	kfree(st);
 	return retval;
@@ -1062,7 +1061,7 @@ static void *v9fs_vfs_follow_link(struct dentry *dentry, struct nameidata *nd)
 			__putname(link);
 			link = ERR_PTR(len);
 		} else
-			link[len] = 0;
+			link[min(len, PATH_MAX-1)] = 0;
 	}
 	nd_set_link(nd, link);
 
diff --git a/fs/Kconfig b/fs/Kconfig
index 2126078a38e..64d44efad7a 100644
--- a/fs/Kconfig
+++ b/fs/Kconfig
@@ -135,7 +135,7 @@ config TMPFS_POSIX_ACL
 
 config HUGETLBFS
 	bool "HugeTLB file system support"
-	depends on X86 || IA64 || PPC_BOOK3S_64 || SPARC64 || (S390 && 64BIT) || \
+	depends on X86 || IA64 || SPARC64 || (S390 && 64BIT) || \
 		   SYS_SUPPORTS_HUGETLBFS || BROKEN
 	help
 	  hugetlbfs is a filesystem backing for HugeTLB pages, based on
diff --git a/fs/afs/file.c b/fs/afs/file.c
index 681c2a7b013..39b301662f2 100644
--- a/fs/afs/file.c
+++ b/fs/afs/file.c
@@ -315,7 +315,6 @@ static void afs_invalidatepage(struct page *page, unsigned long offset)
 			struct afs_vnode *vnode = AFS_FS_I(page->mapping->host);
 			fscache_wait_on_page_write(vnode->cache, page);
 			fscache_uncache_page(vnode->cache, page);
-			ClearPageFsCache(page);
 		}
 #endif
 
@@ -349,17 +348,9 @@ static int afs_releasepage(struct page *page, gfp_t gfp_flags)
 	/* deny if page is being written to the cache and the caller hasn't
 	 * elected to wait */
 #ifdef CONFIG_AFS_FSCACHE
-	if (PageFsCache(page)) {
-		if (fscache_check_page_write(vnode->cache, page)) {
-			if (!(gfp_flags & __GFP_WAIT)) {
-				_leave(" = F [cache busy]");
-				return 0;
-			}
-			fscache_wait_on_page_write(vnode->cache, page);
-		}
-
-		fscache_uncache_page(vnode->cache, page);
-		ClearPageFsCache(page);
+	if (!fscache_maybe_release_page(vnode->cache, page, gfp_flags)) {
+		_leave(" = F [cache busy]");
+		return 0;
 	}
 #endif
 
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index e238a0cdac6..94627c4cc19 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2977,10 +2977,10 @@ static int maybe_allocate_chunk(struct btrfs_root *root,
 
 	free_space = btrfs_super_total_bytes(disk_super);
 	/*
-	 * we allow the metadata to grow to a max of either 5gb or 5% of the
+	 * we allow the metadata to grow to a max of either 10gb or 5% of the
 	 * space in the volume.
 	 */
-	min_metadata = min((u64)5 * 1024 * 1024 * 1024,
+	min_metadata = min((u64)10 * 1024 * 1024 * 1024,
 			     div64_u64(free_space * 5, 100));
 	if (info->total_bytes >= min_metadata) {
 		spin_unlock(&info->lock);
@@ -4102,7 +4102,7 @@ wait_block_group_cache_done(struct btrfs_block_group_cache *cache)
 }
 
 enum btrfs_loop_type {
-	LOOP_CACHED_ONLY = 0,
+	LOOP_FIND_IDEAL = 0,
 	LOOP_CACHING_NOWAIT = 1,
 	LOOP_CACHING_WAIT = 2,
 	LOOP_ALLOC_CHUNK = 3,
@@ -4131,12 +4131,15 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 	struct btrfs_block_group_cache *block_group = NULL;
 	int empty_cluster = 2 * 1024 * 1024;
 	int allowed_chunk_alloc = 0;
+	int done_chunk_alloc = 0;
 	struct btrfs_space_info *space_info;
 	int last_ptr_loop = 0;
 	int loop = 0;
 	bool found_uncached_bg = false;
 	bool failed_cluster_refill = false;
 	bool failed_alloc = false;
+	u64 ideal_cache_percent = 0;
+	u64 ideal_cache_offset = 0;
 
 	WARN_ON(num_bytes < root->sectorsize);
 	btrfs_set_key_type(ins, BTRFS_EXTENT_ITEM_KEY);
@@ -4172,14 +4175,19 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 		empty_cluster = 0;
 
 	if (search_start == hint_byte) {
+ideal_cache:
 		block_group = btrfs_lookup_block_group(root->fs_info,
 						       search_start);
 		/*
 		 * we don't want to use the block group if it doesn't match our
 		 * allocation bits, or if its not cached.
+		 *
+		 * However if we are re-searching with an ideal block group
+		 * picked out then we don't care that the block group is cached.
 		 */
 		if (block_group && block_group_bits(block_group, data) &&
-		    block_group_cache_done(block_group)) {
+		    (block_group->cached != BTRFS_CACHE_NO ||
+		     search_start == ideal_cache_offset)) {
 			down_read(&space_info->groups_sem);
 			if (list_empty(&block_group->list) ||
 			    block_group->ro) {
@@ -4191,13 +4199,13 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 				 */
 				btrfs_put_block_group(block_group);
 				up_read(&space_info->groups_sem);
-			} else
+			} else {
 				goto have_block_group;
+			}
 		} else if (block_group) {
 			btrfs_put_block_group(block_group);
 		}
 	}
-
 search:
 	down_read(&space_info->groups_sem);
 	list_for_each_entry(block_group, &space_info->block_groups, list) {
@@ -4209,28 +4217,45 @@ search:
 
 have_block_group:
 		if (unlikely(block_group->cached == BTRFS_CACHE_NO)) {
+			u64 free_percent;
+
+			free_percent = btrfs_block_group_used(&block_group->item);
+			free_percent *= 100;
+			free_percent = div64_u64(free_percent,
+						 block_group->key.offset);
+			free_percent = 100 - free_percent;
+			if (free_percent > ideal_cache_percent &&
+			    likely(!block_group->ro)) {
+				ideal_cache_offset = block_group->key.objectid;
+				ideal_cache_percent = free_percent;
+			}
+
 			/*
-			 * we want to start caching kthreads, but not too many
-			 * right off the bat so we don't overwhelm the system,
-			 * so only start them if there are less than 2 and we're
-			 * in the initial allocation phase.
+			 * We only want to start kthread caching if we are at
+			 * the point where we will wait for caching to make
+			 * progress, or if our ideal search is over and we've
+			 * found somebody to start caching.
 			 */
 			if (loop > LOOP_CACHING_NOWAIT ||
-			    atomic_read(&space_info->caching_threads) < 2) {
+			    (loop > LOOP_FIND_IDEAL &&
+			     atomic_read(&space_info->caching_threads) < 2)) {
 				ret = cache_block_group(block_group);
 				BUG_ON(ret);
 			}
-		}
-
-		cached = block_group_cache_done(block_group);
-		if (unlikely(!cached)) {
 			found_uncached_bg = true;
 
-			/* if we only want cached bgs, loop */
-			if (loop == LOOP_CACHED_ONLY)
+			/*
+			 * If loop is set for cached only, try the next block
+			 * group.
+			 */
+			if (loop == LOOP_FIND_IDEAL)
 				goto loop;
 		}
 
+		cached = block_group_cache_done(block_group);
+		if (unlikely(!cached))
+			found_uncached_bg = true;
+
 		if (unlikely(block_group->ro))
 			goto loop;
 
@@ -4410,9 +4435,11 @@ loop:
 	}
 	up_read(&space_info->groups_sem);
 
-	/* LOOP_CACHED_ONLY, only search fully cached block groups
-	 * LOOP_CACHING_NOWAIT, search partially cached block groups, but
-	 *			dont wait foR them to finish caching
+	/* LOOP_FIND_IDEAL, only search caching/cached bg's, and don't wait for
+	 *			for them to make caching progress.  Also
+	 *			determine the best possible bg to cache
+	 * LOOP_CACHING_NOWAIT, search partially cached block groups, kicking
+	 *			caching kthreads as we move along
 	 * LOOP_CACHING_WAIT, search everything, and wait if our bg is caching
 	 * LOOP_ALLOC_CHUNK, force a chunk allocation and try again
 	 * LOOP_NO_EMPTY_SIZE, set empty_size and empty_cluster to 0 and try
@@ -4421,12 +4448,47 @@ loop:
 	if (!ins->objectid && loop < LOOP_NO_EMPTY_SIZE &&
 	    (found_uncached_bg || empty_size || empty_cluster ||
 	     allowed_chunk_alloc)) {
-		if (found_uncached_bg) {
+		if (loop == LOOP_FIND_IDEAL && found_uncached_bg) {
 			found_uncached_bg = false;
-			if (loop < LOOP_CACHING_WAIT) {
-				loop++;
+			loop++;
+			if (!ideal_cache_percent &&
+			    atomic_read(&space_info->caching_threads))
 				goto search;
-			}
+
+			/*
+			 * 1 of the following 2 things have happened so far
+			 *
+			 * 1) We found an ideal block group for caching that
+			 * is mostly full and will cache quickly, so we might
+			 * as well wait for it.
+			 *
+			 * 2) We searched for cached only and we didn't find
+			 * anything, and we didn't start any caching kthreads
+			 * either, so chances are we will loop through and
+			 * start a couple caching kthreads, and then come back
+			 * around and just wait for them.  This will be slower
+			 * because we will have 2 caching kthreads reading at
+			 * the same time when we could have just started one
+			 * and waited for it to get far enough to give us an
+			 * allocation, so go ahead and go to the wait caching
+			 * loop.
+			 */
+			loop = LOOP_CACHING_WAIT;
+			search_start = ideal_cache_offset;
+			ideal_cache_percent = 0;
+			goto ideal_cache;
+		} else if (loop == LOOP_FIND_IDEAL) {
+			/*
+			 * Didn't find a uncached bg, wait on anything we find
+			 * next.
+			 */
+			loop = LOOP_CACHING_WAIT;
+			goto search;
+		}
+
+		if (loop < LOOP_CACHING_WAIT) {
+			loop++;
+			goto search;
 		}
 
 		if (loop == LOOP_ALLOC_CHUNK) {
@@ -4438,7 +4500,8 @@ loop:
 			ret = do_chunk_alloc(trans, root, num_bytes +
 					     2 * 1024 * 1024, data, 1);
 			allowed_chunk_alloc = 0;
-		} else {
+			done_chunk_alloc = 1;
+		} else if (!done_chunk_alloc) {
 			space_info->force_alloc = 1;
 		}
 
diff --git a/fs/btrfs/extent_map.c b/fs/btrfs/extent_map.c
index 2c726b7b9fa..ccbdcb54ec5 100644
--- a/fs/btrfs/extent_map.c
+++ b/fs/btrfs/extent_map.c
@@ -208,7 +208,7 @@ int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len)
 	write_lock(&tree->lock);
 	em = lookup_extent_mapping(tree, start, len);
 
-	WARN_ON(em->start != start || !em);
+	WARN_ON(!em || em->start != start);
 
 	if (!em)
 		goto out;
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 5c2caad7621..cb2849f0325 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -1296,7 +1296,7 @@ again:
 			window_start = entry->offset;
 			window_free = entry->bytes;
 			last = entry;
-			max_extent = 0;
+			max_extent = entry->bytes;
 		} else {
 			last = next;
 			window_free += next->bytes;
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index dae12dc7e15..b3ad168a0bf 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -538,7 +538,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	struct extent_io_tree *io_tree;
-	int ret;
+	int ret = 0;
 
 	if (list_empty(&async_cow->extents))
 		return 0;
@@ -552,6 +552,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
 
 		io_tree = &BTRFS_I(inode)->io_tree;
 
+retry:
 		/* did the compression code fall back to uncompressed IO? */
 		if (!async_extent->pages) {
 			int page_started = 0;
@@ -562,11 +563,11 @@ static noinline int submit_compressed_extents(struct inode *inode,
 				    async_extent->ram_size - 1, GFP_NOFS);
 
 			/* allocate blocks */
-			cow_file_range(inode, async_cow->locked_page,
-				       async_extent->start,
-				       async_extent->start +
-				       async_extent->ram_size - 1,
-				       &page_started, &nr_written, 0);
+			ret = cow_file_range(inode, async_cow->locked_page,
+					     async_extent->start,
+					     async_extent->start +
+					     async_extent->ram_size - 1,
+					     &page_started, &nr_written, 0);
 
 			/*
 			 * if page_started, cow_file_range inserted an
@@ -574,7 +575,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
 			 * and IO for us.  Otherwise, we need to submit
 			 * all those pages down to the drive.
 			 */
-			if (!page_started)
+			if (!page_started && !ret)
 				extent_write_locked_range(io_tree,
 						  inode, async_extent->start,
 						  async_extent->start +
@@ -602,7 +603,21 @@ static noinline int submit_compressed_extents(struct inode *inode,
 					   async_extent->compressed_size,
 					   0, alloc_hint,
 					   (u64)-1, &ins, 1);
-		BUG_ON(ret);
+		if (ret) {
+			int i;
+			for (i = 0; i < async_extent->nr_pages; i++) {
+				WARN_ON(async_extent->pages[i]->mapping);
+				page_cache_release(async_extent->pages[i]);
+			}
+			kfree(async_extent->pages);
+			async_extent->nr_pages = 0;
+			async_extent->pages = NULL;
+			unlock_extent(io_tree, async_extent->start,
+				      async_extent->start +
+				      async_extent->ram_size - 1, GFP_NOFS);
+			goto retry;
+		}
+
 		em = alloc_extent_map(GFP_NOFS);
 		em->