Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull more btrfs updates from Chris Mason:
 "This has a few fixes since our last pull and a new ioctl for doing
  btree searches from userland.  It's very similar to the existing
  ioctl, but lets us return larger items back down to the app"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  btrfs: fix error handling in create_pending_snapshot
  btrfs: fix use of uninit "ret" in end_extent_writepage()
  btrfs: free ulist in qgroup_shared_accounting() error path
  Btrfs: fix qgroups sanity test crash or hang
  btrfs: prevent RCU warning when dereferencing radix tree slot
  Btrfs: fix unfinished readahead thread for raid5/6 degraded mounting
  btrfs: new ioctl TREE_SEARCH_V2
  btrfs: tree_search, search_ioctl: direct copy to userspace
  btrfs: new function read_extent_buffer_to_user
  btrfs: tree_search, copy_to_sk: return needed size on EOVERFLOW
  btrfs: tree_search, copy_to_sk: return EOVERFLOW for too small buffer
  btrfs: tree_search, search_ioctl: accept varying buffer
  btrfs: tree_search: eliminate redundant nr_items check

8 files changed, 182 insertions, 36 deletions

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index f25a9092b94..a389820d158 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -2354,7 +2354,7 @@ int end_extent_writepage(struct page *page, int err, u64 start, u64 end)
 {
 	int uptodate = (err == 0);
 	struct extent_io_tree *tree;
-	int ret;
+	int ret = 0;
 
 	tree = &BTRFS_I(page->mapping->host)->io_tree;
 
@@ -5068,6 +5068,43 @@ void read_extent_buffer(struct extent_buffer *eb, void *dstv,
 	}
 }
 
+int read_extent_buffer_to_user(struct extent_buffer *eb, void __user *dstv,
+			unsigned long start,
+			unsigned long len)
+{
+	size_t cur;
+	size_t offset;
+	struct page *page;
+	char *kaddr;
+	char __user *dst = (char __user *)dstv;
+	size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
+	unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
+	int ret = 0;
+
+	WARN_ON(start > eb->len);
+	WARN_ON(start + len > eb->start + eb->len);
+
+	offset = (start_offset + start) & (PAGE_CACHE_SIZE - 1);
+
+	while (len > 0) {
+		page = extent_buffer_page(eb, i);
+
+		cur = min(len, (PAGE_CACHE_SIZE - offset));
+		kaddr = page_address(page);
+		if (copy_to_user(dst, kaddr + offset, cur)) {
+			ret = -EFAULT;
+			break;
+		}
+
+		dst += cur;
+		len -= cur;
+		offset = 0;
+		i++;
+	}
+
+	return ret;
+}
+
 int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start,
 			       unsigned long min_len, char **map,
 			       unsigned long *map_start,
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index 8b63f2d4651..15ce5f2a2b6 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -304,6 +304,9 @@ int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv,
 void read_extent_buffer(struct extent_buffer *eb, void *dst,
 			unsigned long start,
 			unsigned long len);
+int read_extent_buffer_to_user(struct extent_buffer *eb, void __user *dst,
+			       unsigned long start,
+			       unsigned long len);
 void write_extent_buffer(struct extent_buffer *eb, const void *src,
 			 unsigned long start, unsigned long len);
 void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 82c18ba12e3..0d321c23069 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -1957,7 +1957,8 @@ static noinline int copy_to_sk(struct btrfs_root *root,
 			       struct btrfs_path *path,
 			       struct btrfs_key *key,
 			       struct btrfs_ioctl_search_key *sk,
-			       char *buf,
+			       size_t *buf_size,
+			       char __user *ubuf,
 			       unsigned long *sk_offset,
 			       int *num_found)
 {
@@ -1989,13 +1990,25 @@ static noinline int copy_to_sk(struct btrfs_root *root,
 		if (!key_in_sk(key, sk))
 			continue;
 
-		if (sizeof(sh) + item_len > BTRFS_SEARCH_ARGS_BUFSIZE)
+		if (sizeof(sh) + item_len > *buf_size) {
+			if (*num_found) {
+				ret = 1;
+				goto out;
+			}
+
+			/*
+			 * return one empty item back for v1, which does not
+			 * handle -EOVERFLOW
+			 */
+
+			*buf_size = sizeof(sh) + item_len;
 			item_len = 0;
+			ret = -EOVERFLOW;
+		}
 
-		if (sizeof(sh) + item_len + *sk_offset >
-		    BTRFS_SEARCH_ARGS_BUFSIZE) {
+		if (sizeof(sh) + item_len + *sk_offset > *buf_size) {
 			ret = 1;
-			goto overflow;
+			goto out;
 		}
 
 		sh.objectid = key->objectid;
@@ -2005,20 +2018,33 @@ static noinline int copy_to_sk(struct btrfs_root *root,
 		sh.transid = found_transid;
 
 		/* copy search result header */
-		memcpy(buf + *sk_offset, &sh, sizeof(sh));
+		if (copy_to_user(ubuf + *sk_offset, &sh, sizeof(sh))) {
+			ret = -EFAULT;
+			goto out;
+		}
+
 		*sk_offset += sizeof(sh);
 
 		if (item_len) {
-			char *p = buf + *sk_offset;
+			char __user *up = ubuf + *sk_offset;
 			/* copy the item */
-			read_extent_buffer(leaf, p,
-					   item_off, item_len);
+			if (read_extent_buffer_to_user(leaf, up,
+						       item_off, item_len)) {
+				ret = -EFAULT;
+				goto out;
+			}
+
 			*sk_offset += item_len;
 		}
 		(*num_found)++;
 
-		if (*num_found >= sk->nr_items)
-			break;
+		if (ret) /* -EOVERFLOW from above */
+			goto out;
+
+		if (*num_found >= sk->nr_items) {
+			ret = 1;
+			goto out;
+		}
 	}
 advance_key:
 	ret = 0;
@@ -2033,22 +2059,37 @@ advance_key:
 		key->objectid++;
 	} else
 		ret = 1;
-overflow:
+out:
+	/*
+	 *  0: all items from this leaf copied, continue with next
+	 *  1: * more items can be copied, but unused buffer is too small
+	 *     * all items were found
+	 *     Either way, it will stops the loop which iterates to the next
+	 *     leaf
+	 *  -EOVERFLOW: item was to large for buffer
+	 *  -EFAULT: could not copy extent buffer back to userspace
+	 */
 	return ret;
 }
 
 static noinline int search_ioctl(struct inode *inode,
-				 struct btrfs_ioctl_search_args *args)
+				 struct btrfs_ioctl_search_key *sk,
+				 size_t *buf_size,
+				 char __user *ubuf)
 {
 	struct btrfs_root *root;
 	struct btrfs_key key;
 	struct btrfs_path *path;
-	struct btrfs_ioctl_search_key *sk = &args->key;
 	struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info;
 	int ret;
 	int num_found = 0;
 	unsigned long sk_offset = 0;
 
+	if (*buf_size < sizeof(struct btrfs_ioctl_search_header)) {
+		*buf_size = sizeof(struct btrfs_ioctl_search_header);
+		return -EOVERFLOW;
+	}
+
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
@@ -2082,14 +2123,15 @@ static noinline int search_ioctl(struct inode *inode,
 				ret = 0;
 			goto err;
 		}
-		ret = copy_to_sk(root, path, &key, sk, args->buf,
+		ret = copy_to_sk(root, path, &key, sk, buf_size, ubuf,
 				 &sk_offset, &num_found);
 		btrfs_release_path(path);
-		if (ret || num_found >= sk->nr_items)
+		if (ret)
 			break;
 
 	}
-	ret = 0;
+	if (ret > 0)
+		ret = 0;
 err:
 	sk->nr_items = num_found;
 	btrfs_free_path(path);
@@ -2099,22 +2141,73 @@ err:
 static noinline int btrfs_ioctl_tree_search(struct file *file,
 					   void __user *argp)
 {
-	 struct btrfs_ioctl_search_args *args;
-	 struct inode *inode;
-	 int ret;
+	struct btrfs_ioctl_search_args __user *uargs;
+	struct btrfs_ioctl_search_key sk;
+	struct inode *inode;
+	int ret;
+	size_t buf_size;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	args = memdup_user(argp, sizeof(*args));
-	if (IS_ERR(args))
-		return PTR_ERR(args);
+	uargs = (struct btrfs_ioctl_search_args __user *)argp;
+
+	if (copy_from_user(&sk, &uargs->key, sizeof(sk)))
+		return -EFAULT;
+
+	buf_size = sizeof(uargs->buf);
 
 	inode = file_inode(file);
-	ret = search_ioctl(inode, args);
-	if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
+	ret = search_ioctl(inode, &sk, &buf_size, uargs->buf);
+
+	/*
+	 * In the origin implementation an overflow is handled by returning a
+	 * search header with a len of zero, so reset ret.
+	 */
+	if (ret == -EOVERFLOW)
+		ret = 0;
+
+	if (ret == 0 && copy_to_user(&uargs->key, &sk, sizeof(sk)))
 		ret = -EFAULT;
-	kfree(args);
+	return ret;
+}
+
+static noinline int btrfs_ioctl_tree_search_v2(struct file *file,
+					       void __user *argp)
+{
+	struct btrfs_ioctl_search_args_v2 __user *uarg;
+	struct btrfs_ioctl_search_args_v2 args;
+	struct inode *inode;
+	int ret;
+	size_t buf_size;
+	const size_t buf_limit = 16 * 1024 * 1024;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	/* copy search header and buffer size */
+	uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp;
+	if (copy_from_user(&args, uarg, sizeof(args)))
+		return -EFAULT;
+
+	buf_size = args.buf_size;
+
+	if (buf_size < sizeof(struct btrfs_ioctl_search_header))
+		return -EOVERFLOW;
+
+	/* limit result size to 16MB */
+	if (buf_size > buf_limit)
+		buf_size = buf_limit;
+
+	inode = file_inode(file);
+	ret = search_ioctl(inode, &args.key, &buf_size,
+			   (char *)(&uarg->buf[0]));
+	if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key)))
+		ret = -EFAULT;
+	else if (ret == -EOVERFLOW &&
+		copy_to_user(&uarg->buf_size, &buf_size, sizeof(buf_size)))
+		ret = -EFAULT;
+
 	return ret;
 }
 
@@ -5198,6 +5291,8 @@ long btrfs_ioctl(struct file *file, unsigned int
 		return btrfs_ioctl_trans_end(file);
 	case BTRFS_IOC_TREE_SEARCH:
 		return btrfs_ioctl_tree_search(file, argp);
+	case BTRFS_IOC_TREE_SEARCH_V2:
+		return btrfs_ioctl_tree_search_v2(file, argp);
 	case BTRFS_IOC_INO_LOOKUP:
 		return btrfs_ioctl_ino_lookup(file, argp);
 	case BTRFS_IOC_INO_PATHS:
diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index cf5aead95a7..98cb6b2630f 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -1798,8 +1798,10 @@ static int qgroup_shared_accounting(struct btrfs_trans_handle *trans,
 		return -ENOMEM;
 
 	tmp = ulist_alloc(GFP_NOFS);
-	if (!tmp)
+	if (!tmp) {
+		ulist_free(qgroups);
 		return -ENOMEM;
+	}
 
 	btrfs_get_tree_mod_seq(fs_info, &elem);
 	ret = btrfs_find_all_roots(trans, fs_info, oper->bytenr, elem.seq,
diff --git a/fs/btrfs/reada.c b/fs/btrfs/reada.c
index 30947f92362..09230cf3a24 100644
--- a/fs/btrfs/reada.c
+++ b/fs/btrfs/reada.c
@@ -428,8 +428,13 @@ static struct reada_extent *reada_find_extent(struct btrfs_root *root,
 			continue;
 		}
 		if (!dev->bdev) {
-			/* cannot read ahead on missing device */
-			continue;
+			/*
+			 * cannot read ahead on missing device, but for RAID5/6,
+			 * REQ_GET_READ_MIRRORS return 1. So don't skip missing
+			 * device for such case.
+			 */
+			if (nzones > 1)
+				continue;
 		}
 		if (dev_replace_is_ongoing &&
 		    dev == fs_info->dev_replace.tgtdev) {
diff --git a/fs/btrfs/tests/btrfs-tests.c b/fs/btrfs/tests/btrfs-tests.c
index a5dcacb5df9..9626252ee6b 100644
--- a/fs/btrfs/tests/btrfs-tests.c
+++ b/fs/btrfs/tests/btrfs-tests.c
@@ -135,7 +135,7 @@ restart:
 	radix_tree_for_each_slot(slot, &fs_info->buffer_radix, &iter, 0) {
 		struct extent_buffer *eb;
 
-		eb = radix_tree_deref_slot(slot);
+		eb = radix_tree_deref_slot_protected(slot, &fs_info->buffer_lock);
 		if (!eb)
 			continue;
 		/* Shouldn't happen but that kind of thinking creates CVE's */
diff --git a/fs/btrfs/tests/qgroup-tests.c b/fs/btrfs/tests/qgroup-tests.c
index fa691b754aa..ec3dcb20235 100644
--- a/fs/btrfs/tests/qgroup-tests.c
+++ b/fs/btrfs/tests/qgroup-tests.c
@@ -415,6 +415,8 @@ int btrfs_test_qgroups(void)
 		ret = -ENOMEM;
 		goto out;
 	}
+	btrfs_set_header_level(root->node, 0);
+	btrfs_set_header_nritems(root->node, 0);
 	root->alloc_bytenr += 8192;
 
 	tmp_root = btrfs_alloc_dummy_root();
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 9630f10f8e1..511839c04f1 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -1284,11 +1284,13 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 		goto fail;
 	}
 
-	pending->error = btrfs_qgroup_inherit(trans, fs_info,
-					      root->root_key.objectid,
-					      objectid, pending->inherit);
-	if (pending->error)
-		goto no_free_objectid;
+	ret = btrfs_qgroup_inherit(trans, fs_info,
+				   root->root_key.objectid,
+				   objectid, pending->inherit);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
 
 	/* see comments in should_cow_block() */
 	set_bit(BTRFS_ROOT_FORCE_COW, &root->state);