[PATCH v9 0/8] Btrfs scrub: print path to corrupted files and trigger nodatasum fixup

[PATCH v9 0/8] Btrfs scrub: print path to corrupted files and trigger nodatasum fixup

[Jan Schmidt]

Changelog v8->v9:
- bugfix: increment extent buffer's ref-count so it can safely be used after
  releasing the path (reported by myself)
- fixed function comment for paths_from_inode()

This patch set with the recent fixes just mentioned is also available from the
scrub branch of my repository:

git://git.jan-o-sch.net/btrfs-unstable

The for-chris branch (which is the combination of scrub and raid1-repair) has
those updates, too.

Original message follows:
------------------------
This patch set introduces two new features for scrub. They share the backref
iteration code which is the reason they made it into the same patch set.

The first feature adds printk statements in case scrub finds an error which list
all affected files. You will need patch 1, 2 and 3 for that.

The second feature adds the trigger which enables us to correct i/o errors in
case the affected extent does not have a checksum (nodatasum), eventually. You
will need patch 1, 4, 5 and 6 for that.

I tried to apply all patches to the current cmason/for-linus branch and to
Arne's current for-chris branch. They do apply with no errors (some offsets
possible).

The new ioctl()s can be tested from usermode by applying the patch series
	[PATCH v2 0/3] Btrfs-progs: add the first "inspect-internal" commands
from this mailing list to the user land tools.

Please review.

Next I'm starting to make up my mind how to implement on-the-fly error
correction correctly. This will enable us to rewrite good data whenever we
encounter a bad copy. I have some preliminary patches already, the stress in the
first sentence is on "correctly". The second feature mentioned in this patch
series will then automatically use that code, too.

Changelog v1->v2:
- Various cleanup, sensible error codes as suggested by David Sterba

Changelog v2->v3:
- evaluation and iteration of shared refs
- support for in-tree refs (v2 iterated inline refs only)
- never call an interator function without releasing the path
- iterate_irefs now returns -ENOENT in case no refs are found
- some stupid bugs removed where release_path was called too early
- ioctls added to provide new functions to user mode
- bugfixes for cases where search_slot found the very end of a leaf
- bugfix: use right fs root for readpage instead of fs_root->fs_info
- based on current cmason/for-linus

Changelog v3->v4:
- fixed a regression with mirror_num that could prevent error correction
- based on current cmason/for-linus

Changelog v4->v5:
- fixed a deadlock when fixup is taking longer while scrub is about to end

Changelog v5->v6:
- fixed ioctl priviledge and input sanity checking (reported by Andi Kleen)

Changelog v6->v7:
- include everything that was stated to be in v6

Changelog v7->v8:
- bugfix for the case where refs exist but none are inline (reported by Jan
  Schubert)
- use bytes_missing and bytes_left (unsigned) instead of size (signed) (wished
  by Chris Mason)
- alloc_bytes is now size_t (reported by Li Zefan)

Please try it and report errors (or confirm there are none, of course). I can
provide a place to pull from if anyone likes.

-Jan


Jan Schmidt (8):
  btrfs: added helper functions to iterate backrefs
  btrfs scrub: added unverified_errors
  btrfs scrub: print paths of corrupted files
  btrfs scrub: bugfix: mirror_num off by one
  btrfs: add mirror_num to extent_read_full_page
  btrfs scrub: use int for mirror_num, not u64
  btrfs scrub: add fixup code for errors on nodatasum files
  btrfs: new ioctls to do logical->inode and inode->path resolving

 fs/btrfs/Makefile    |    3 +-
 fs/btrfs/backref.c   |  776 ++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/backref.h   |   62 ++++
 fs/btrfs/disk-io.c   |    2 +-
 fs/btrfs/extent_io.c |    6 +-
 fs/btrfs/extent_io.h |    3 +-
 fs/btrfs/inode.c     |    2 +-
 fs/btrfs/ioctl.c     |  143 +++++++++
 fs/btrfs/ioctl.h     |   30 ++
 fs/btrfs/scrub.c     |  414 +++++++++++++++++++++++++---
 10 files changed, 1401 insertions(+), 40 deletions(-)
 create mode 100644 fs/btrfs/backref.c
 create mode 100644 fs/btrfs/backref.h

-- 
1.7.3.4

[PATCH v9 1/8] btrfs: added helper functions to iterate backrefs

[Jan Schmidt]

These helper functions iterate back references and call a function for each
backref. There is also a function to resolve an inode to a path in the
file system.

Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
---
 fs/btrfs/Makefile  |    3 +-
 fs/btrfs/backref.c |  776 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/backref.h |   62 +++++
 fs/btrfs/ioctl.h   |   11 +
 4 files changed, 851 insertions(+), 1 deletions(-)

diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile
index 40e6ac0..89b6ce3 100644
--- a/fs/btrfs/Makefile
+++ b/fs/btrfs/Makefile
@@ -7,6 +7,7 @@ btrfs-y += super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \
 	   extent_map.o sysfs.o struct-funcs.o xattr.o ordered-data.o \
 	   extent_io.o volumes.o async-thread.o ioctl.o locking.o orphan.o \
 	   export.o tree-log.o free-space-cache.o zlib.o lzo.o \
-	   compression.o delayed-ref.o relocation.o delayed-inode.o scrub.o
+	   compression.o delayed-ref.o relocation.o delayed-inode.o backref.o \
+	   scrub.o
 
 btrfs-$(CONFIG_BTRFS_FS_POSIX_ACL) += acl.o
diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
new file mode 100644
index 0000000..2351df0
--- /dev/null
+++ b/fs/btrfs/backref.c
@@ -0,0 +1,776 @@
+/*
+ * Copyright (C) 2011 STRATO.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License v2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#include "ctree.h"
+#include "disk-io.h"
+#include "backref.h"
+
+struct __data_ref {
+	struct list_head list;
+	u64 inum;
+	u64 root;
+	u64 extent_data_item_offset;
+};
+
+struct __shared_ref {
+	struct list_head list;
+	u64 disk_byte;
+};
+
+static int __inode_info(u64 inum, u64 ioff, u8 key_type,
+			struct btrfs_root *fs_root, struct btrfs_path *path,
+			struct btrfs_key *found_key)
+{
+	int ret;
+	struct btrfs_key key;
+	struct extent_buffer *eb;
+
+	key.type = key_type;
+	key.objectid = inum;
+	key.offset = ioff;
+
+	ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+
+	eb = path->nodes[0];
+	if (ret && path->slots[0] >= btrfs_header_nritems(eb)) {
+		ret = btrfs_next_leaf(fs_root, path);
+		if (ret)
+			return ret;
+		eb = path->nodes[0];
+	}
+
+	btrfs_item_key_to_cpu(eb, found_key, path->slots[0]);
+	if (found_key->type != key.type || found_key->objectid != key.objectid)
+		return 1;
+
+	return 0;
+}
+
+/*
+ * this makes the path point to (inum INODE_ITEM ioff)
+ */
+int inode_item_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
+			struct btrfs_path *path)
+{
+	struct btrfs_key key;
+	return __inode_info(inum, ioff, BTRFS_INODE_ITEM_KEY, fs_root, path,
+				&key);
+}
+
+static int inode_ref_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
+				struct btrfs_path *path,
+				struct btrfs_key *found_key)
+{
+	return __inode_info(inum, ioff, BTRFS_INODE_REF_KEY, fs_root, path,
+				found_key);
+}
+
+/*
+ * this iterates to turn a btrfs_inode_ref into a full filesystem path. elements
+ * of the path are separated by '/' and the path is guaranteed to be
+ * 0-terminated. the path is only given within the current file system.
+ * Therefore, it never starts with a '/'. the caller is responsible to provide
+ * "size" bytes in "dest". the dest buffer will be filled backwards. finally,
+ * the start point of the resulting string is returned. this pointer is within
+ * dest, normally.
+ * in case the path buffer would overflow, the pointer is decremented further
+ * as if output was written to the buffer, though no more output is actually
+ * generated. that way, the caller can determine how much space would be
+ * required for the path to fit into the buffer. in that case, the returned
+ * value will be smaller than dest. callers must check this!
+ */
+static char *iref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path,
+				struct btrfs_inode_ref *iref,
+				struct extent_buffer *eb_in, u64 parent,
+				char *dest, u32 size)
+{
+	u32 len;
+	int slot;
+	u64 next_inum;
+	int ret;
+	s64 bytes_left = size - 1;
+	struct extent_buffer *eb = eb_in;
+	struct btrfs_key found_key;
+
+	if (bytes_left >= 0)
+		dest[bytes_left] = '\0';
+
+	while (1) {
+		len = btrfs_inode_ref_name_len(eb, iref);
+		bytes_left -= len;
+		if (bytes_left >= 0)
+			read_extent_buffer(eb, dest + bytes_left,
+						(unsigned long)(iref + 1), len);
+		if (eb != eb_in)
+			free_extent_buffer(eb);
+		ret = inode_ref_info(parent, 0, fs_root, path, &found_key);
+		if (ret)
+			break;
+		next_inum = found_key.offset;
+
+		/* regular exit ahead */
+		if (parent == next_inum)
+			break;
+
+		slot = path->slots[0];
+		eb = path->nodes[0];
+		/* make sure we can use eb after releasing the path */
+		if (eb != eb_in)
+			atomic_inc(&eb->refs);
+		btrfs_release_path(path);
+
+		iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
+		parent = next_inum;
+		--bytes_left;
+		if (bytes_left >= 0)
+			dest[bytes_left] = '/';
+	}
+
+	btrfs_release_path(path);
+
+	if (ret)
+		return ERR_PTR(ret);
+
+	return dest + bytes_left;
+}
+
+/*
+ * this makes the path point to (logical EXTENT_ITEM *)
+ * returns BTRFS_EXTENT_FLAG_DATA for data, BTRFS_EXTENT_FLAG_TREE_BLOCK for
+ * tree blocks and <0 on error.
+ */
+int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
+			struct btrfs_path *path, struct btrfs_key *found_key)
+{
+	int ret;
+	u64 flags;
+	u32 item_size;
+	struct extent_buffer *eb;
+	struct btrfs_extent_item *ei;
+	struct btrfs_key key;
+
+	key.type = BTRFS_EXTENT_ITEM_KEY;
+	key.objectid = logical;
+	key.offset = (u64)-1;
+
+	ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+	ret = btrfs_previous_item(fs_info->extent_root, path,
+					0, BTRFS_EXTENT_ITEM_KEY);
+	if (ret < 0)
+		return ret;
+
+	btrfs_item_key_to_cpu(path->nodes[0], found_key, path->slots[0]);
+	if (found_key->type != BTRFS_EXTENT_ITEM_KEY ||
+	    found_key->objectid > logical ||
+	    found_key->objectid + found_key->offset <= logical)
+		return -ENOENT;
+
+	eb = path->nodes[0];
+	item_size = btrfs_item_size_nr(eb, path->slots[0]);
+	BUG_ON(item_size < sizeof(*ei));
+
+	ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
+	flags = btrfs_extent_flags(eb, ei);
+
+	if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+		return BTRFS_EXTENT_FLAG_TREE_BLOCK;
+	if (flags & BTRFS_EXTENT_FLAG_DATA)
+		return BTRFS_EXTENT_FLAG_DATA;
+
+	return -EIO;
+}
+
+/*
+ * helper function to iterate extent inline refs. ptr must point to a 0 value
+ * for the first call and may be modified. it is used to track state.
+ * if more refs exist, 0 is returned and the next call to
+ * __get_extent_inline_ref must pass the modified ptr parameter to get the
+ * next ref. after the last ref was processed, 1 is returned.
+ * returns <0 on error
+ */
+static int __get_extent_inline_ref(unsigned long *ptr, struct extent_buffer *eb,
+				struct btrfs_extent_item *ei, u32 item_size,
+				struct btrfs_extent_inline_ref **out_eiref,
+				int *out_type)
+{
+	unsigned long end;
+	u64 flags;
+	struct btrfs_tree_block_info *info;
+
+	if (!*ptr) {
+		/* first call */
+		flags = btrfs_extent_flags(eb, ei);
+		if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+			info = (struct btrfs_tree_block_info *)(ei + 1);
+			*out_eiref =
+				(struct btrfs_extent_inline_ref *)(info + 1);
+		} else {
+			*out_eiref = (struct btrfs_extent_inline_ref *)(ei + 1);
+		}
+		*ptr = (unsigned long)*out_eiref;
+		if ((void *)*ptr >= (void *)ei + item_size)
+			return -ENOENT;
+	}
+
+	end = (unsigned long)ei + item_size;
+	*out_eiref = (struct btrfs_extent_inline_ref *)*ptr;
+	*out_type = btrfs_extent_inline_ref_type(eb, *out_eiref);
+
+	*ptr += btrfs_extent_inline_ref_size(*out_type);
+	WARN_ON(*ptr > end);
+	if (*ptr == end)
+		return 1; /* last */
+
+	return 0;
+}
+
+/*
+ * reads the tree block backref for an extent. tree level and root are returned
+ * through out_level and out_root. ptr must point to a 0 value for the first
+ * call and may be modified (see __get_extent_inline_ref comment).
+ * returns 0 if data was provided, 1 if there was no more data to provide or
+ * <0 on error.
+ */
+int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
+				struct btrfs_extent_item *ei, u32 item_size,
+				u64 *out_root, u8 *out_level)
+{
+	int ret;
+	int type;
+	struct btrfs_tree_block_info *info;
+	struct btrfs_extent_inline_ref *eiref;
+
+	if (*ptr == (unsigned long)-1)
+		return 1;
+
+	while (1) {
+		ret = __get_extent_inline_ref(ptr, eb, ei, item_size,
+						&eiref, &type);
+		if (ret < 0)
+			return ret;
+
+		if (type == BTRFS_TREE_BLOCK_REF_KEY ||
+		    type == BTRFS_SHARED_BLOCK_REF_KEY)
+			break;
+
+		if (ret == 1)
+			return 1;
+	}
+
+	/* we can treat both ref types equally here */
+	info = (struct btrfs_tree_block_info *)(ei + 1);
+	*out_root = btrfs_extent_inline_ref_offset(eb, eiref);
+	*out_level = btrfs_tree_block_level(eb, info);
+
+	if (ret == 1)
+		*ptr = (unsigned long)-1;
+
+	return 0;
+}
+
+static int __data_list_add(struct list_head *head, u64 inum,
+				u64 extent_data_item_offset, u64 root)
+{
+	struct __data_ref *ref;
+
+	ref = kmalloc(sizeof(*ref), GFP_NOFS);
+	if (!ref)
+		return -ENOMEM;
+
+	ref->inum = inum;
+	ref->extent_data_item_offset = extent_data_item_offset;
+	ref->root = root;
+	list_add_tail(&ref->list, head);
+
+	return 0;
+}
+
+static int __data_list_add_eb(struct list_head *head, struct extent_buffer *eb,
+				struct btrfs_extent_data_ref *dref)
+{
+	return __data_list_add(head, btrfs_extent_data_ref_objectid(eb, dref),
+				btrfs_extent_data_ref_offset(eb, dref),
+				btrfs_extent_data_ref_root(eb, dref));
+}
+
+static int __shared_list_add(struct list_head *head, u64 disk_byte)
+{
+	struct __shared_ref *ref;
+
+	ref = kmalloc(sizeof(*ref), GFP_NOFS);
+	if (!ref)
+		return -ENOMEM;
+
+	ref->disk_byte = disk_byte;
+	list_add_tail(&ref->list, head);
+
+	return 0;
+}
+
+static int __iter_shared_inline_ref_inodes(struct btrfs_fs_info *fs_info,
+					   u64 logical, u64 inum,
+					   u64 extent_data_item_offset,
+					   u64 extent_offset,
+					   struct btrfs_path *path,
+					   struct list_head *data_refs,
+					   iterate_extent_inodes_t *iterate,
+					   void *ctx)
+{
+	u64 ref_root;
+	u32 item_size;
+	struct btrfs_key key;
+	struct extent_buffer *eb;
+	struct btrfs_extent_item *ei;
+	struct btrfs_extent_inline_ref *eiref;
+	struct __data_ref *ref;
+	int ret;
+	int type;
+	int last;
+	unsigned long ptr = 0;
+
+	WARN_ON(!list_empty(data_refs));
+	ret = extent_from_logical(fs_info, logical, path, &key);
+	if (ret & BTRFS_EXTENT_FLAG_DATA)
+		ret = -EIO;
+	if (ret < 0)
+		goto out;
+
+	eb = path->nodes[0];
+	ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
+	item_size = btrfs_item_size_nr(eb, path->slots[0]);
+
+	ret = 0;
+	ref_root = 0;
+	/*
+	 * as done in iterate_extent_inodes, we first build a list of refs to
+	 * iterate, then free the path and then iterate them to avoid deadlocks.
+	 */
+	do {
+		last = __get_extent_inline_ref(&ptr, eb, ei, item_size,
+						&eiref, &type);
+		if (last < 0) {
+			ret = last;
+			goto out;
+		}
+		if (type == BTRFS_TREE_BLOCK_REF_KEY ||
+		    type == BTRFS_SHARED_BLOCK_REF_KEY) {
+			ref_root = btrfs_extent_inline_ref_offset(eb, eiref);
+			ret = __data_list_add(data_refs, inum,
+						extent_data_item_offset,
+						ref_root);
+		}
+	} while (!ret && !last);
+
+	btrfs_release_path(path);
+
+	if (ref_root == 0) {
+		printk(KERN_ERR "btrfs: failed to find tree block ref "
+			"for shared data backref %llu\n", logical);
+		WARN_ON(1);
+		ret = -EIO;
+	}
+
+out:
+	while (!list_empty(data_refs)) {
+		ref = list_first_entry(data_refs, struct __data_ref, list);
+		list_del(&ref->list);
+		if (!ret)
+			ret = iterate(ref->inum, extent_offset +
+					ref->extent_data_item_offset,
+					ref->root, ctx);
+		kfree(ref);
+	}
+
+	return ret;
+}
+
+static int __iter_shared_inline_ref(struct btrfs_fs_info *fs_info,
+				    u64 logical, u64 orig_extent_item_objectid,
+				    u64 extent_offset, struct btrfs_path *path,
+				    struct list_head *data_refs,
+				    iterate_extent_inodes_t *iterate,
+				    void *ctx)
+{
+	u64 disk_byte;
+	struct btrfs_key key;
+	struct btrfs_file_extent_item *fi;
+	struct extent_buffer *eb;
+	int slot;
+	int nritems;
+	int ret;
+	int found = 0;
+
+	eb = read_tree_block(fs_info->tree_root, logical,
+				fs_info->tree_root->leafsize, 0);
+	if (!eb)
+		return -EIO;
+
+	/*
+	 * from the shared data ref, we only have the leaf but we need
+	 * the key. thus, we must look into all items and see that we
+	 * find one (some) with a reference to our extent item.
+	 */
+	nritems = btrfs_header_nritems(eb);
+	for (slot = 0; slot < nritems; ++slot) {
+		btrfs_item_key_to_cpu(eb, &key, slot);
+		if (key.type != BTRFS_EXTENT_DATA_KEY)
+			continue;
+		fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
+		if (!fi) {
+			free_extent_buffer(eb);
+			return -EIO;
+		}
+		disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
+		if (disk_byte != orig_extent_item_objectid) {
+			if (found)
+				break;
+			else
+				continue;
+		}
+		++found;
+		ret = __iter_shared_inline_ref_inodes(fs_info, logical,
+							key.objectid,
+							key.offset,
+							extent_offset, path,
+							data_refs,
+							iterate, ctx);
+		if (ret)
+			break;
+	}
+
+	if (!found) {
+		printk(KERN_ERR "btrfs: failed to follow shared data backref "
+			"to parent %llu\n", logical);
+		WARN_ON(1);
+		ret = -EIO;
+	}
+
+	free_extent_buffer(eb);
+	return ret;
+}
+
+/*
+ * calls iterate() for every inode that references the extent identified by
+ * the given parameters. will use the path given as a parameter and return it
+ * released.
+ * when the iterator function returns a non-zero value, iteration stops.
+ */
+int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
+				struct btrfs_path *path,
+				u64 extent_item_objectid,
+				u64 extent_offset,
+				iterate_extent_inodes_t *iterate, void *ctx)
+{
+	unsigned long ptr = 0;
+	int last;
+	int ret;
+	int type;
+	u64 logical;
+	u32 item_size;
+	struct btrfs_extent_inline_ref *eiref;
+	struct btrfs_extent_data_ref *dref;
+	struct extent_buffer *eb;
+	struct btrfs_extent_item *ei;
+	struct btrfs_key key;
+	struct list_head data_refs = LIST_HEAD_INIT(data_refs);
+	struct list_head shared_refs = LIST_HEAD_INIT(shared_refs);
+	struct __data_ref *ref_d;
+	struct __shared_ref *ref_s;
+
+	eb = path->nodes[0];
+	ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
+	item_size = btrfs_item_size_nr(eb, path->slots[0]);
+
+	/* first we iterate the inline refs, ... */
+	do {
+		last = __get_extent_inline_ref(&ptr, eb, ei, item_size,
+						&eiref, &type);
+		if (last == -ENOENT) {
+			ret = 0;
+			break;
+		}
+		if (last < 0) {
+			ret = last;
+			break;
+		}
+
+		if (type == BTRFS_EXTENT_DATA_REF_KEY) {
+			dref = (struct btrfs_extent_data_ref *)(&eiref->offset);
+			ret = __data_list_add_eb(&data_refs, eb, dref);
+		} else if (type == BTRFS_SHARED_DATA_REF_KEY) {
+			logical = btrfs_extent_inline_ref_offset(eb, eiref);
+			ret = __shared_list_add(&shared_refs, logical);
+		}
+	} while (!ret && !last);
+
+	/* ... then we proceed to in-tree references and ... */
+	while (!ret) {
+		++path->slots[0];
+		if (path->slots[0] > btrfs_header_nritems(eb)) {
+			ret = btrfs_next_leaf(fs_info->extent_root, path);
+			if (ret) {
+				if (ret == 1)
+					ret = 0; /* we're done */
+				break;
+			}
+			eb = path->nodes[0];
+		}
+		btrfs_item_key_to_cpu(eb, &key, path->slots[0]);
+		if (key.objectid != extent_item_objectid)
+			break;
+		if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
+			dref = btrfs_item_ptr(eb, path->slots[0],
+						struct btrfs_extent_data_ref);
+			ret = __data_list_add_eb(&data_refs, eb, dref);
+		} else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
+			ret = __shared_list_add(&shared_refs, key.offset);
+		}
+	}
+
+	btrfs_release_path(path);
+
+	/*
+	 * ... only at the very end we can process the refs we found. this is
+	 * because the iterator function we call is allowed to make tree lookups
+	 * and we have to avoid deadlocks. additionally, we need more tree
+	 * lookups ourselves for shared data refs.
+	 */
+	while (!list_empty(&data_refs)) {
+		ref_d = list_first_entry(&data_refs, struct __data_ref, list);
+		list_del(&ref_d->list);
+		if (!ret)
+			ret = iterate(ref_d->inum, extent_offset +
+					ref_d->extent_data_item_offset,
+					ref_d->root, ctx);
+		kfree(ref_d);
+	}
+
+	while (!list_empty(&shared_refs)) {
+		ref_s = list_first_entry(&shared_refs, struct __shared_ref,
+					list);
+		list_del(&ref_s->list);
+		if (!ret)
+			ret = __iter_shared_inline_ref(fs_info,
+							ref_s->disk_byte,
+							extent_item_objectid,
+							extent_offset, path,
+							&data_refs,
+							iterate, ctx);
+		kfree(ref_s);
+	}
+
+	return ret;
+}
+
+int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
+				struct btrfs_path *path,
+				iterate_extent_inodes_t *iterate, void *ctx)
+{
+	int ret;
+	u64 offset;
+	struct btrfs_key found_key;
+
+	ret = extent_from_logical(fs_info, logical, path,
+					&found_key);
+	if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+		ret = -EINVAL;
+	if (ret < 0)
+		return ret;
+
+	offset = logical - found_key.objectid;
+	ret = iterate_extent_inodes(fs_info, path, found_key.objectid,
+					offset, iterate, ctx);
+
+	return ret;
+}
+
+static int iterate_irefs(u64 inum, struct btrfs_root *fs_root,
+				struct btrfs_path *path,
+				iterate_irefs_t *iterate, void *ctx)
+{
+	int ret;
+	int slot;
+	u32 cur;
+	u32 len;
+	u32 name_len;
+	u64 parent = 0;
+	int found = 0;
+	struct extent_buffer *eb;
+	struct btrfs_item *item;
+	struct btrfs_inode_ref *iref;
+	struct btrfs_key found_key;
+
+	while (1) {
+		ret = inode_ref_info(inum, parent ? parent+1 : 0, fs_root, path,
+					&found_key);
+		if (ret < 0)
+			break;
+		if (ret) {
+			ret = found ? 0 : -ENOENT;
+			break;
+		}
+		++found;
+
+		parent = found_key.offset;
+		slot = path->slots[0];
+		eb = path->nodes[0];
+		/* make sure we can use eb after releasing the path */
+		atomic_inc(&eb->refs);
+		btrfs_release_path(path);
+
+		item = btrfs_item_nr(eb, slot);
+		iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref);
+
+		for (cur = 0; cur < btrfs_item_size(eb, item); cur += len) {
+			name_len = btrfs_inode_ref_name_len(eb, iref);
+			/* path must be released before calling iterate()! */
+			ret = iterate(parent, iref, eb, ctx);
+			if (ret) {
+				free_extent_buffer(eb);
+				break;
+			}
+			len = sizeof(*iref) + name_len;
+			iref = (struct btrfs_inode_ref *)((char *)iref + len);
+		}
+		free_extent_buffer(eb);
+	}
+
+	btrfs_release_path(path);
+
+	return ret;
+}
+
+/*
+ * returns 0 if the path could be dumped (probably truncated)
+ * returns <0 in case of an error
+ */
+static int inode_to_path(u64 inum, struct btrfs_inode_ref *iref,
+				struct extent_buffer *eb, void *ctx)
+{
+	struct inode_fs_paths *ipath = ctx;
+	char *fspath;
+	char *fspath_min;
+	int i = ipath->fspath->elem_cnt;
+	const int s_ptr = sizeof(char *);
+	u32 bytes_left;
+
+	bytes_left = ipath->fspath->bytes_left > s_ptr ?
+					ipath->fspath->bytes_left - s_ptr : 0;
+
+	fspath_min = (char *)ipath->fspath->str + (i + 1) * s_ptr;
+	fspath = iref_to_path(ipath->fs_root, ipath->btrfs_path, iref, eb,
+				inum, fspath_min, bytes_left);
+	if (IS_ERR(fspath))
+		return PTR_ERR(fspath);
+
+	if (fspath > fspath_min) {
+		ipath->fspath->str[i] = fspath;
+		++ipath->fspath->elem_cnt;
+		ipath->fspath->bytes_left = fspath - fspath_min;
+	} else {
+		++ipath->fspath->elem_missed;
+		ipath->fspath->bytes_missing += fspath_min - fspath;
+		ipath->fspath->bytes_left = 0;
+	}
+
+	return 0;
+}
+
+/*
+ * this dumps all file system paths to the inode into the ipath struct, provided
+ * is has been created large enough. each path is zero-terminated and accessed
+ * from ipath->fspath->str[i].
+ * when it returns, there are ipath->fspath->elem_cnt number of paths available
+ * in ipath->fspath->str[]. when the allocated space wasn't sufficient, the
+ * number of missed paths in recored in ipath->fspath->elem_missed, otherwise,
+ * it's zero. ipath->fspath->bytes_missing holds the number of bytes that would
+ * have been needed to return all paths.
+ */
+int paths_from_inode(u64 inum, struct inode_fs_paths *ipath)
+{
+	return iterate_irefs(inum, ipath->fs_root, ipath->btrfs_path,
+				inode_to_path, ipath);
+}
+
+/*
+ * allocates space to return multiple file system paths for an inode.
+ * total_bytes to allocate are passed, note that space usable for actual path
+ * information will be total_bytes - sizeof(struct inode_fs_paths).
+ * the returned pointer must be freed with free_ipath() in the end.
+ */
+struct btrfs_data_container *init_data_container(u32 total_bytes)
+{
+	struct btrfs_data_container *data;
+	size_t alloc_bytes;
+
+	alloc_bytes = max_t(size_t, total_bytes, sizeof(*data));
+	data = kmalloc(alloc_bytes, GFP_NOFS);
+	if (!data)
+		return ERR_PTR(-ENOMEM);
+
+	if (total_bytes >= sizeof(*data)) {
+		data->bytes_left = total_bytes - sizeof(*data);
+		data->bytes_missing = 0;
+	} else {
+		data->bytes_missing = sizeof(*data) - total_bytes;
+		data->bytes_left = 0;
+	}
+
+	data->elem_cnt = 0;
+	data->elem_missed = 0;
+
+	return data;
+}
+
+/*
+ * allocates space to return multiple file system paths for an inode.
+ * total_bytes to allocate are passed, note that space usable for actual path
+ * information will be total_bytes - sizeof(struct inode_fs_paths).
+ * the returned pointer must be freed with free_ipath() in the end.
+ */
+struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
+					struct btrfs_path *path)
+{
+	struct inode_fs_paths *ifp;
+	struct btrfs_data_container *fspath;
+
+	fspath = init_data_container(total_bytes);
+	if (IS_ERR(fspath))
+		return (void *)fspath;
+
+	ifp = kmalloc(sizeof(*ifp), GFP_NOFS);
+	if (!ifp) {
+		kfree(fspath);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	ifp->btrfs_path = path;
+	ifp->fspath = fspath;
+	ifp->fs_root = fs_root;
+
+	return ifp;
+}
+
+void free_ipath(struct inode_fs_paths *ipath)
+{
+	kfree(ipath);
+}
diff --git a/fs/btrfs/backref.h b/fs/btrfs/backref.h
new file mode 100644
index 0000000..9261883
--- /dev/null
+++ b/fs/btrfs/backref.h
@@ -0,0 +1,62 @@
+/*
+ * Copyright (C) 2011 STRATO.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License v2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#ifndef __BTRFS_BACKREF__
+#define __BTRFS_BACKREF__
+
+#include "ioctl.h"
+
+struct inode_fs_paths {
+	struct btrfs_path		*btrfs_path;
+	struct btrfs_root		*fs_root;
+	struct btrfs_data_container	*fspath;
+};
+
+typedef int (iterate_extent_inodes_t)(u64 inum, u64 offset, u64 root,
+		void *ctx);
+typedef int (iterate_irefs_t)(u64 parent, struct btrfs_inode_ref *iref,
+				struct extent_buffer *eb, void *ctx);
+
+int inode_item_info(u64 inum, u64 ioff, struct btrfs_root *fs_root,
+			struct btrfs_path *path);
+
+int extent_from_logical(struct btrfs_fs_info *fs_info, u64 logical,
+			struct btrfs_path *path, struct btrfs_key *found_key);
+
+int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
+				struct btrfs_extent_item *ei, u32 item_size,
+				u64 *out_root, u8 *out_level);
+
+int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
+				struct btrfs_path *path,
+				u64 extent_item_objectid,
+				u64 extent_offset,
+				iterate_extent_inodes_t *iterate, void *ctx);
+
+int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
+				struct btrfs_path *path,
+				iterate_extent_inodes_t *iterate, void *ctx);
+
+int paths_from_inode(u64 inum, struct inode_fs_paths *ipath);
+
+struct btrfs_data_container *init_data_container(u32 total_bytes);
+struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,
+					struct btrfs_path *path);
+void free_ipath(struct inode_fs_paths *ipath);
+
+#endif
diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h
index ad1ea78..1857e38 100644
--- a/fs/btrfs/ioctl.h
+++ b/fs/btrfs/ioctl.h
@@ -193,6 +193,17 @@ struct btrfs_ioctl_space_args {
 	struct btrfs_ioctl_space_info spaces[0];
 };
 
+struct btrfs_data_container {
+	__u32	bytes_left;	/* out -- bytes not needed to deliver output */
+	__u32	bytes_missing;	/* out -- additional bytes needed for result */
+	__u32	elem_cnt;	/* out */
+	__u32	elem_missed;	/* out */
+	union {
+		char	*str[0];	/* out */
+		__u64	val[0];		/* out */
+	};
+};
+
 #define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \
 				   struct btrfs_ioctl_vol_args)
 #define BTRFS_IOC_DEFRAG _IOW(BTRFS_IOCTL_MAGIC, 2, \
-- 
1.7.3.4

[PATCH v9 2/8] btrfs scrub: added unverified_errors

[Jan Schmidt]

In normal operation, scrub is reading data sequentially in large portions.
In case of an i/o error, we try to find the corrupted area(s) by issuing
page sized read requests. With this commit we increment the
unverified_errors counter if all of the small size requests succeed.

Userland patches carrying such conspicous events to the administrator should
already be around.

Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
---
 fs/btrfs/scrub.c |   37 ++++++++++++++++++++++++++-----------
 1 files changed, 26 insertions(+), 11 deletions(-)

diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index a8d03d5..35099fa 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -201,18 +201,25 @@ nomem:
  * recheck_error gets called for every page in the bio, even though only
  * one may be bad
  */
-static void scrub_recheck_error(struct scrub_bio *sbio, int ix)
+static int scrub_recheck_error(struct scrub_bio *sbio, int ix)
 {
+	struct scrub_dev *sdev = sbio->sdev;
+	u64 sector = (sbio->physical + ix * PAGE_SIZE) >> 9;
+
 	if (sbio->err) {
-		if (scrub_fixup_io(READ, sbio->sdev->dev->bdev,
-				   (sbio->physical + ix * PAGE_SIZE) >> 9,
+		if (scrub_fixup_io(READ, sbio->sdev->dev->bdev, sector,
 				   sbio->bio->bi_io_vec[ix].bv_page) == 0) {
 			if (scrub_fixup_check(sbio, ix) == 0)
-				return;
+				return 0;
 		}
 	}
 
+	spin_lock(&sdev->stat_lock);
+	++sdev->stat.read_errors;
+	spin_unlock(&sdev->stat_lock);
+
 	scrub_fixup(sbio, ix);
+	return 1;
 }
 
 static int scrub_fixup_check(struct scrub_bio *sbio, int ix)
@@ -382,8 +389,14 @@ static void scrub_checksum(struct btrfs_work *work)
 	int ret;
 
 	if (sbio->err) {
+		ret = 0;
 		for (i = 0; i < sbio->count; ++i)
-			scrub_recheck_error(sbio, i);
+			ret |= scrub_recheck_error(sbio, i);
+		if (!ret) {
+			spin_lock(&sdev->stat_lock);
+			++sdev->stat.unverified_errors;
+			spin_unlock(&sdev->stat_lock);
+		}
 
 		sbio->bio->bi_flags &= ~(BIO_POOL_MASK - 1);
 		sbio->bio->bi_flags |= 1 << BIO_UPTODATE;
@@ -396,10 +409,6 @@ static void scrub_checksum(struct btrfs_work *work)
 			bi->bv_offset = 0;
 			bi->bv_len = PAGE_SIZE;
 		}
-
-		spin_lock(&sdev->stat_lock);
-		++sdev->stat.read_errors;
-		spin_unlock(&sdev->stat_lock);
 		goto out;
 	}
 	for (i = 0; i < sbio->count; ++i) {
@@ -420,8 +429,14 @@ static void scrub_checksum(struct btrfs_work *work)
 			WARN_ON(1);
 		}
 		kunmap_atomic(buffer, KM_USER0);
-		if (ret)
-			scrub_recheck_error(sbio, i);
+		if (ret) {
+			ret = scrub_recheck_error(sbio, i);
+			if (!ret) {
+				spin_lock(&sdev->stat_lock);
+				++sdev->stat.unverified_errors;
+				spin_unlock(&sdev->stat_lock);
+			}
+		}
 	}
 
 out:
-- 
1.7.3.4

[PATCH v9 3/8] btrfs scrub: print paths of corrupted files

[Jan Schmidt]

While scrubbing, we may encounter various errors. Previously, a logical
address was printed to the log only. Now, all paths belonging to that
address are resolved and printed separately. That should work for hardlinks
as well as reflinks.

Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
---
 fs/btrfs/scrub.c |  169 ++++++++++++++++++++++++++++++++++++++++++++++++++++--
 1 files changed, 163 insertions(+), 6 deletions(-)

diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 35099fa..221fd5c 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -17,10 +17,12 @@
  */
 
 #include <linux/blkdev.h>
+#include <linux/ratelimit.h>
 #include "ctree.h"
 #include "volumes.h"
 #include "disk-io.h"
 #include "ordered-data.h"
+#include "backref.h"
 
 /*
  * This is only the first step towards a full-features scrub. It reads all
@@ -100,6 +102,19 @@ struct scrub_dev {
 	spinlock_t		stat_lock;
 };
 
+struct scrub_warning {
+	struct btrfs_path	*path;
+	u64			extent_item_size;
+	char			*scratch_buf;
+	char			*msg_buf;
+	const char		*errstr;
+	sector_t		sector;
+	u64			logical;
+	struct btrfs_device	*dev;
+	int			msg_bufsize;
+	int			scratch_bufsize;
+};
+
 static void scrub_free_csums(struct scrub_dev *sdev)
 {
 	while (!list_empty(&sdev->csum_list)) {
@@ -195,6 +210,143 @@ nomem:
 	return ERR_PTR(-ENOMEM);
 }
 
+static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root, void *ctx)
+{
+	u64 isize;
+	u32 nlink;
+	int ret;
+	int i;
+	struct extent_buffer *eb;
+	struct btrfs_inode_item *inode_item;
+	struct scrub_warning *swarn = ctx;
+	struct btrfs_fs_info *fs_info = swarn->dev->dev_root->fs_info;
+	struct inode_fs_paths *ipath = NULL;
+	struct btrfs_root *local_root;
+	struct btrfs_key root_key;
+
+	root_key.objectid = root;
+	root_key.type = BTRFS_ROOT_ITEM_KEY;
+	root_key.offset = (u64)-1;
+	local_root = btrfs_read_fs_root_no_name(fs_info, &root_key);
+	if (IS_ERR(local_root)) {
+		ret = PTR_ERR(local_root);
+		goto err;
+	}
+
+	ret = inode_item_info(inum, 0, local_root, swarn->path);
+	if (ret) {
+		btrfs_release_path(swarn->path);
+		goto err;
+	}
+
+	eb = swarn->path->nodes[0];
+	inode_item = btrfs_item_ptr(eb, swarn->path->slots[0],
+					struct btrfs_inode_item);
+	isize = btrfs_inode_size(eb, inode_item);
+	nlink = btrfs_inode_nlink(eb, inode_item);
+	btrfs_release_path(swarn->path);
+
+	ipath = init_ipath(4096, local_root, swarn->path);
+	ret = paths_from_inode(inum, ipath);
+
+	if (ret < 0)
+		goto err;
+
+	/*
+	 * we deliberately ignore the bit ipath might have been too small to
+	 * hold all of the paths here
+	 */
+	for (i = 0; i < ipath->fspath->elem_cnt; ++i)
+		printk(KERN_WARNING "btrfs: %s at logical %llu on dev "
+			"%s, sector %llu, root %llu, inode %llu, offset %llu, "
+			"length %llu, links %u (path: %s)\n", swarn->errstr,
+			swarn->logical, swarn->dev->name,
+			(unsigned long long)swarn->sector, root, inum, offset,
+			min(isize - offset, (u64)PAGE_SIZE), nlink,
+			ipath->fspath->str[i]);
+
+	free_ipath(ipath);
+	return 0;
+
+err:
+	printk(KERN_WARNING "btrfs: %s at logical %llu on dev "
+		"%s, sector %llu, root %llu, inode %llu, offset %llu: path "
+		"resolving failed with ret=%d\n", swarn->errstr,
+		swarn->logical, swarn->dev->name,
+		(unsigned long long)swarn->sector, root, inum, offset, ret);
+
+	free_ipath(ipath);
+	return 0;
+}
+
+static void scrub_print_warning(const char *errstr, struct scrub_bio *sbio,
+				int ix)
+{
+	struct btrfs_device *dev = sbio->sdev->dev;
+	struct btrfs_fs_info *fs_info = dev->dev_root->fs_info;
+	struct btrfs_path *path;
+	struct btrfs_key found_key;
+	struct extent_buffer *eb;
+	struct btrfs_extent_item *ei;
+	struct scrub_warning swarn;
+	u32 item_size;
+	int ret;
+	u64 ref_root;
+	u8 ref_level;
+	unsigned long ptr = 0;
+	const int bufsize = 4096;
+	u64 extent_offset;
+
+	path = btrfs_alloc_path();
+
+	swarn.scratch_buf = kmalloc(bufsize, GFP_NOFS);
+	swarn.msg_buf = kmalloc(bufsize, GFP_NOFS);
+	swarn.sector = (sbio->physical + ix * PAGE_SIZE) >> 9;
+	swarn.logical = sbio->logical + ix * PAGE_SIZE;
+	swarn.errstr = errstr;
+	swarn.dev = dev;
+	swarn.msg_bufsize = bufsize;
+	swarn.scratch_bufsize = bufsize;
+
+	if (!path || !swarn.scratch_buf || !swarn.msg_buf)
+		goto out;
+
+	ret = extent_from_logical(fs_info, swarn.logical, path, &found_key);
+	if (ret < 0)
+		goto out;
+
+	extent_offset = swarn.logical - found_key.objectid;
+	swarn.extent_item_size = found_key.offset;
+
+	eb = path->nodes[0];
+	ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
+	item_size = btrfs_item_size_nr(eb, path->slots[0]);
+
+	if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
+		do {
+			ret = tree_backref_for_extent(&ptr, eb, ei, item_size,
+							&ref_root, &ref_level);
+			printk(KERN_WARNING "%s at logical %llu on dev %s, "
+				"sector %llu: metadata %s (level %d) in tree "
+				"%llu\n", errstr, swarn.logical, dev->name,
+				(unsigned long long)swarn.sector,
+				ref_level ? "node" : "leaf",
+				ret < 0 ? -1 : ref_level,
+				ret < 0 ? -1 : ref_root);
+		} while (ret != 1);
+	} else {
+		swarn.path = path;
+		iterate_extent_inodes(fs_info, path, found_key.objectid,
+					extent_offset,
+					scrub_print_warning_inode, &swarn);
+	}
+
+out:
+	btrfs_free_path(path);
+	kfree(swarn.scratch_buf);
+	kfree(swarn.msg_buf);
+}
+
 /*
  * scrub_recheck_error gets called when either verification of the page
  * failed or the bio failed to read, e.g. with EIO. In the latter case,
@@ -205,6 +357,8 @@ static int scrub_recheck_error(struct scrub_bio *sbio, int ix)
 {
 	struct scrub_dev *sdev = sbio->sdev;
 	u64 sector = (sbio->physical + ix * PAGE_SIZE) >> 9;
+	static DEFINE_RATELIMIT_STATE(_rs, DEFAULT_RATELIMIT_INTERVAL,
+					DEFAULT_RATELIMIT_BURST);
 
 	if (sbio->err) {
 		if (scrub_fixup_io(READ, sbio->sdev->dev->bdev, sector,
@@ -212,6 +366,11 @@ static int scrub_recheck_error(struct scrub_bio *sbio, int ix)
 			if (scrub_fixup_check(sbio, ix) == 0)
 				return 0;
 		}
+		if (__ratelimit(&_rs))
+			scrub_print_warning("i/o error", sbio, ix);
+	} else {
+		if (__ratelimit(&_rs))
+			scrub_print_warning("checksum error", sbio, ix);
 	}
 
 	spin_lock(&sdev->stat_lock);
@@ -326,9 +485,8 @@ static void scrub_fixup(struct scrub_bio *sbio, int ix)
 	++sdev->stat.corrected_errors;
 	spin_unlock(&sdev->stat_lock);
 
-	if (printk_ratelimit())
-		printk(KERN_ERR "btrfs: fixed up at %llu\n",
-		       (unsigned long long)logical);
+	printk_ratelimited(KERN_ERR "btrfs: fixed up error at logical %llu\n",
+			       (unsigned long long)logical);
 	return;
 
 uncorrectable:
@@ -337,9 +495,8 @@ uncorrectable:
 	++sdev->stat.uncorrectable_errors;
 	spin_unlock(&sdev->stat_lock);
 
-	if (printk_ratelimit())
-		printk(KERN_ERR "btrfs: unable to fixup at %llu\n",
-			 (unsigned long long)logical);
+	printk_ratelimited(KERN_ERR "btrfs: unable to fixup (regular) error at "
+				"logical %llu\n", (unsigned long long)logical);
 }
 
 static int scrub_fixup_io(int rw, struct block_device *bdev, sector_t sector,
-- 
1.7.3.4

[PATCH v9 4/8] btrfs scrub: bugfix: mirror_num off by one

[Jan Schmidt]

Fix the mirror_num determination in scrub_stripe. The rest of the scrub code
did not use mirror_num for anything important and that error went unnoticed.
The nodatasum fixup patch of this set depends on a correct mirror_num.

Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
---
 fs/btrfs/scrub.c |   12 ++++++------
 1 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 221fd5c..59caf8f 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -452,7 +452,7 @@ static void scrub_fixup(struct scrub_bio *sbio, int ix)
 	 * first find a good copy
 	 */
 	for (i = 0; i < multi->num_stripes; ++i) {
-		if (i == sbio->spag[ix].mirror_num)
+		if (i + 1 == sbio->spag[ix].mirror_num)
 			continue;
 
 		if (scrub_fixup_io(READ, multi->stripes[i].dev->bdev,
@@ -930,21 +930,21 @@ static noinline_for_stack int scrub_stripe(struct scrub_dev *sdev,
 	if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
 		offset = map->stripe_len * num;
 		increment = map->stripe_len * map->num_stripes;
-		mirror_num = 0;
+		mirror_num = 1;
 	} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
 		int factor = map->num_stripes / map->sub_stripes;
 		offset = map->stripe_len * (num / map->sub_stripes);
 		increment = map->stripe_len * factor;
-		mirror_num = num % map->sub_stripes;
+		mirror_num = num % map->sub_stripes + 1;
 	} else if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
 		increment = map->stripe_len;
-		mirror_num = num % map->num_stripes;
+		mirror_num = num % map->num_stripes + 1;
 	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
 		increment = map->stripe_len;
-		mirror_num = num % map->num_stripes;
+		mirror_num = num % map->num_stripes + 1;
 	} else {
 		increment = map->stripe_len;
-		mirror_num = 0;
+		mirror_num = 1;
 	}
 
 	path = btrfs_alloc_path();
-- 
1.7.3.4

[PATCH v9 5/8] btrfs: add mirror_num to extent_read_full_page

[Jan Schmidt]

Currently, extent_read_full_page always assumes we are trying to read mirror
0, which generally is the best we can do. To add flexibility, pass it as a
parameter. This will be needed by scrub fixup code.

Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
---
 fs/btrfs/disk-io.c   |    2 +-
 fs/btrfs/extent_io.c |    6 +++---
 fs/btrfs/extent_io.h |    2 +-
 fs/btrfs/inode.c     |    2 +-
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 94ecac3..89ad5ef 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -908,7 +908,7 @@ static int btree_readpage(struct file *file, struct page *page)
 {
 	struct extent_io_tree *tree;
 	tree = &BTRFS_I(page->mapping->host)->io_tree;
-	return extent_read_full_page(tree, page, btree_get_extent);
+	return extent_read_full_page(tree, page, btree_get_extent, 0);
 }
 
 static int btree_releasepage(struct page *page, gfp_t gfp_flags)
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 8491712..edf1c29 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -2076,16 +2076,16 @@ out:
 }
 
 int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
-			    get_extent_t *get_extent)
+			    get_extent_t *get_extent, int mirror_num)
 {
 	struct bio *bio = NULL;
 	unsigned long bio_flags = 0;
 	int ret;
 
-	ret = __extent_read_full_page(tree, page, get_extent, &bio, 0,
+	ret = __extent_read_full_page(tree, page, get_extent, &bio, mirror_num,
 				      &bio_flags);
 	if (bio)
-		ret = submit_one_bio(READ, bio, 0, bio_flags);
+		ret = submit_one_bio(READ, bio, mirror_num, bio_flags);
 	return ret;
 }
 
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index 7b2f0c3..a9dd994 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -185,7 +185,7 @@ int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end,
 int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
 		    gfp_t mask);
 int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
-			  get_extent_t *get_extent);
+			  get_extent_t *get_extent, int mirror_num);
 int __init extent_io_init(void);
 void extent_io_exit(void);
 
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 34195f9..2a01e32 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -6244,7 +6244,7 @@ int btrfs_readpage(struct file *file, struct page *page)
 {
 	struct extent_io_tree *tree;
 	tree = &BTRFS_I(page->mapping->host)->io_tree;
-	return extent_read_full_page(tree, page, btrfs_get_extent);
+	return extent_read_full_page(tree, page, btrfs_get_extent, 0);
 }
 
 static int btrfs_writepage(struct page *page, struct writeback_control *wbc)
-- 
1.7.3.4

[PATCH v9 6/8] btrfs scrub: use int for mirror_num, not u64

[Jan Schmidt]

the rest of the code uses int mirror_num, and so should scrub

Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
---
 fs/btrfs/scrub.c |    8 ++++----
 1 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 59caf8f..41a0114 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -65,7 +65,7 @@ static void scrub_fixup(struct scrub_bio *sbio, int ix);
 struct scrub_page {
 	u64			flags;  /* extent flags */
 	u64			generation;
-	u64			mirror_num;
+	int			mirror_num;
 	int			have_csum;
 	u8			csum[BTRFS_CSUM_SIZE];
 };
@@ -776,7 +776,7 @@ nomem:
 }
 
 static int scrub_page(struct scrub_dev *sdev, u64 logical, u64 len,
-		      u64 physical, u64 flags, u64 gen, u64 mirror_num,
+		      u64 physical, u64 flags, u64 gen, int mirror_num,
 		      u8 *csum, int force)
 {
 	struct scrub_bio *sbio;
@@ -873,7 +873,7 @@ static int scrub_find_csum(struct scrub_dev *sdev, u64 logical, u64 len,
 
 /* scrub extent tries to collect up to 64 kB for each bio */
 static int scrub_extent(struct scrub_dev *sdev, u64 logical, u64 len,
-			u64 physical, u64 flags, u64 gen, u64 mirror_num)
+			u64 physical, u64 flags, u64 gen, int mirror_num)
 {
 	int ret;
 	u8 csum[BTRFS_CSUM_SIZE];
@@ -919,7 +919,7 @@ static noinline_for_stack int scrub_stripe(struct scrub_dev *sdev,
 	u64 physical;
 	u64 logical;
 	u64 generation;
-	u64 mirror_num;
+	int mirror_num;
 
 	u64 increment = map->stripe_len;
 	u64 offset;
-- 
1.7.3.4

[PATCH v9 7/8] btrfs scrub: add fixup code for errors on nodatasum files

[Jan Schmidt]

This removes a FIXME comment and introduces the first part of nodatasum
fixup: It gets the corresponding inode for a logical address and triggers a
regular readpage for the corrupted sector.

Once we have on-the-fly error correction our error will be automatically
corrected. The correction code is expected to clear the newly introduced
EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead
of "uncorrectable" eventually.

Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
---
 fs/btrfs/extent_io.h |    1 +
 fs/btrfs/scrub.c     |  188 ++++++++++++++++++++++++++++++++++++++++++++++++--
 2 files changed, 183 insertions(+), 6 deletions(-)

diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index a9dd994..435d454 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -17,6 +17,7 @@
 #define EXTENT_NODATASUM (1 << 10)
 #define EXTENT_DO_ACCOUNTING (1 << 11)
 #define EXTENT_FIRST_DELALLOC (1 << 12)
+#define EXTENT_DAMAGED (1 << 13)
 #define EXTENT_IOBITS (EXTENT_LOCKED | EXTENT_WRITEBACK)
 #define EXTENT_CTLBITS (EXTENT_DO_ACCOUNTING | EXTENT_FIRST_DELALLOC)
 
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 41a0114..db09f01 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -22,6 +22,7 @@
 #include "volumes.h"
 #include "disk-io.h"
 #include "ordered-data.h"
+#include "transaction.h"
 #include "backref.h"
 
 /*
@@ -89,6 +90,7 @@ struct scrub_dev {
 	int			first_free;
 	int			curr;
 	atomic_t		in_flight;
+	atomic_t		fixup_cnt;
 	spinlock_t		list_lock;
 	wait_queue_head_t	list_wait;
 	u16			csum_size;
@@ -102,6 +104,14 @@ struct scrub_dev {
 	spinlock_t		stat_lock;
 };
 
+struct scrub_fixup_nodatasum {
+	struct scrub_dev	*sdev;
+	u64			logical;
+	struct btrfs_root	*root;
+	struct btrfs_work	work;
+	int			mirror_num;
+};
+
 struct scrub_warning {
 	struct btrfs_path	*path;
 	u64			extent_item_size;
@@ -190,12 +200,13 @@ struct scrub_dev *scrub_setup_dev(struct btrfs_device *dev)
 
 		if (i != SCRUB_BIOS_PER_DEV-1)
 			sdev->bios[i]->next_free = i + 1;
-		 else
+		else
 			sdev->bios[i]->next_free = -1;
 	}
 	sdev->first_free = 0;
 	sdev->curr = -1;
 	atomic_set(&sdev->in_flight, 0);
+	atomic_set(&sdev->fixup_cnt, 0);
 	atomic_set(&sdev->cancel_req, 0);
 	sdev->csum_size = btrfs_super_csum_size(&fs_info->super_copy);
 	INIT_LIST_HEAD(&sdev->csum_list);
@@ -347,6 +358,151 @@ out:
 	kfree(swarn.msg_buf);
 }
 
+static int scrub_fixup_readpage(u64 inum, u64 offset, u64 root, void *ctx)
+{
+	struct page *page;
+	unsigned long index;
+	struct scrub_fixup_nodatasum *fixup = ctx;
+	int ret;
+	int corrected;
+	struct btrfs_key key;
+	struct inode *inode;
+	u64 end = offset + PAGE_SIZE - 1;
+	struct btrfs_root *local_root;
+
+	key.objectid = root;
+	key.type = BTRFS_ROOT_ITEM_KEY;
+	key.offset = (u64)-1;
+	local_root = btrfs_read_fs_root_no_name(fixup->root->fs_info, &key);
+	if (IS_ERR(local_root))
+		return PTR_ERR(local_root);
+
+	key.type = BTRFS_INODE_ITEM_KEY;
+	key.objectid = inum;
+	key.offset = 0;
+	inode = btrfs_iget(fixup->root->fs_info->sb, &key, local_root, NULL);
+	if (IS_ERR(inode))
+		return PTR_ERR(inode);
+
+	ret = set_extent_bit(&BTRFS_I(inode)->io_tree, offset, end,
+				EXTENT_DAMAGED, 0, NULL, NULL, GFP_NOFS);
+
+	/* set_extent_bit should either succeed or give proper error */
+	WARN_ON(ret > 0);
+	if (ret)
+		return ret < 0 ? ret : -EFAULT;
+
+	index = offset >> PAGE_CACHE_SHIFT;
+
+	page = find_or_create_page(inode->i_mapping, index, GFP_NOFS);
+	if (!page)
+		return -ENOMEM;
+
+	ret = extent_read_full_page(&BTRFS_I(inode)->io_tree, page,
+					btrfs_get_extent, fixup->mirror_num);
+	wait_on_page_locked(page);
+	corrected = !test_range_bit(&BTRFS_I(inode)->io_tree, offset, end,
+					EXTENT_DAMAGED, 0, NULL);
+
+	if (corrected)
+		WARN_ON(!PageUptodate(page));
+	else
+		clear_extent_bit(&BTRFS_I(inode)->io_tree, offset, end,
+					EXTENT_DAMAGED, 0, 0, NULL, GFP_NOFS);
+
+	put_page(page);
+	iput(inode);
+
+	if (ret < 0)
+		return ret;
+
+	if (ret == 0 && corrected) {
+		/*
+		 * we only need to call readpage for one of the inodes belonging
+		 * to this extent. so make iterate_extent_inodes stop
+		 */
+		return 1;
+	}
+
+	return -EIO;
+}
+
+static void scrub_fixup_nodatasum(struct btrfs_work *work)
+{
+	int ret;
+	struct scrub_fixup_nodatasum *fixup;
+	struct scrub_dev *sdev;
+	struct btrfs_trans_handle *trans = NULL;
+	struct btrfs_fs_info *fs_info;
+	struct btrfs_path *path;
+	int uncorrectable = 0;
+
+	fixup = container_of(work, struct scrub_fixup_nodatasum, work);
+	sdev = fixup->sdev;
+	fs_info = fixup->root->fs_info;
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		spin_lock(&sdev->stat_lock);
+		++sdev->stat.malloc_errors;
+		spin_unlock(&sdev->stat_lock);
+		uncorrectable = 1;
+		goto out;
+	}
+
+	trans = btrfs_join_transaction(fixup->root);
+	if (IS_ERR(trans)) {
+		uncorrectable = 1;
+		goto out;
+	}
+
+	/*
+	 * the idea is to trigger a regular read through the standard path. we
+	 * read a page from the (failed) logical address by specifying the
+	 * corresponding copynum of the failed sector. thus, that readpage is
+	 * expected to fail.
+	 * that is the point where on-the-fly error correction will kick in
+	 * (once it's finished) and rewrite the failed sector if a good copy
+	 * can be found.
+	 */
+	ret = iterate_inodes_from_logical(fixup->logical, fixup->root->fs_info,
+						path, scrub_fixup_readpage,
+						fixup);
+	if (ret < 0) {
+		uncorrectable = 1;
+		goto out;
+	}
+	WARN_ON(ret != 1);
+
+	spin_lock(&sdev->stat_lock);
+	++sdev->stat.corrected_errors;
+	spin_unlock(&sdev->stat_lock);
+
+out:
+	if (trans && !IS_ERR(trans))
+		btrfs_end_transaction(trans, fixup->root);
+	if (uncorrectable) {
+		spin_lock(&sdev->stat_lock);
+		++sdev->stat.uncorrectable_errors;
+		spin_unlock(&sdev->stat_lock);
+		printk_ratelimited(KERN_ERR "btrfs: unable to fixup "
+					"(nodatasum) error at logical %llu\n",
+					fixup->logical);
+	}
+
+	btrfs_free_path(path);
+	kfree(fixup);
+
+	/* see caller why we're pretending to be paused in the scrub counters */
+	mutex_lock(&fs_info->scrub_lock);
+	atomic_dec(&fs_info->scrubs_running);
+	atomic_dec(&fs_info->scrubs_paused);
+	mutex_unlock(&fs_info->scrub_lock);
+	atomic_dec(&sdev->fixup_cnt);
+	wake_up(&fs_info->scrub_pause_wait);
+	wake_up(&sdev->list_wait);
+}
+
 /*
  * scrub_recheck_error gets called when either verification of the page
  * failed or the bio failed to read, e.g. with EIO. In the latter case,
@@ -417,6 +573,7 @@ static void scrub_fixup(struct scrub_bio *sbio, int ix)
 	struct btrfs_fs_info *fs_info = sdev->dev->dev_root->fs_info;
 	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
 	struct btrfs_multi_bio *multi = NULL;
+	struct scrub_fixup_nodatasum *fixup;
 	u64 logical = sbio->logical + ix * PAGE_SIZE;
 	u64 length;
 	int i;
@@ -425,12 +582,30 @@ static void scrub_fixup(struct scrub_bio *sbio, int ix)
 
 	if ((sbio->spag[ix].flags & BTRFS_EXTENT_FLAG_DATA) &&
 	    (sbio->spag[ix].have_csum == 0)) {
+		fixup = kzalloc(sizeof(*fixup), GFP_NOFS);
+		if (!fixup)
+			goto uncorrectable;
+		fixup->sdev = sdev;
+		fixup->logical = logical;
+		fixup->root = fs_info->extent_root;
+		fixup->mirror_num = sbio->spag[ix].mirror_num;
 		/*
-		 * nodatasum, don't try to fix anything
-		 * FIXME: we can do better, open the inode and trigger a
-		 * writeback
+		 * increment scrubs_running to prevent cancel requests from
+		 * completing as long as a fixup worker is running. we must also
+		 * increment scrubs_paused to prevent deadlocking on pause
+		 * requests used for transactions commits (as the worker uses a
+		 * transaction context). it is safe to regard the fixup worker
+		 * as paused for all matters practical. effectively, we only
+		 * avoid cancellation requests from completing.
 		 */
-		goto uncorrectable;
+		mutex_lock(&fs_info->scrub_lock);
+		atomic_inc(&fs_info->scrubs_running);
+		atomic_inc(&fs_info->scrubs_paused);
+		mutex_unlock(&fs_info->scrub_lock);
+		atomic_inc(&sdev->fixup_cnt);
+		fixup->work.func = scrub_fixup_nodatasum;
+		btrfs_queue_worker(&fs_info->scrub_workers, &fixup->work);
+		return;
 	}
 
 	length = PAGE_SIZE;
@@ -1425,10 +1600,11 @@ int btrfs_scrub_dev(struct btrfs_root *root, u64 devid, u64 start, u64 end,
 		ret = scrub_enumerate_chunks(sdev, start, end);
 
 	wait_event(sdev->list_wait, atomic_read(&sdev->in_flight) == 0);
-
 	atomic_dec(&fs_info->scrubs_running);
 	wake_up(&fs_info->scrub_pause_wait);
 
+	wait_event(sdev->list_wait, atomic_read(&sdev->fixup_cnt) == 0);
+
 	if (progress)
 		memcpy(progress, &sdev->stat, sizeof(*progress));
 
-- 
1.7.3.4

[PATCH v9 8/8] btrfs: new ioctls to do logical->inode and inode->path resolving

[Jan Schmidt]

these ioctls make use of the new functions initially added for scrub. they
return all inodes belonging to a logical address (BTRFS_IOC_LOGICAL_INO) and
all paths belonging to an inode (BTRFS_IOC_INO_PATHS).

Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
---
 fs/btrfs/ioctl.c |  143 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/ioctl.h |   19 +++++++
 2 files changed, 162 insertions(+), 0 deletions(-)

diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 2bb0886..cf67219 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -51,6 +51,7 @@
 #include "volumes.h"
 #include "locking.h"
 #include "inode-map.h"
+#include "backref.h"
 
 /* Mask out flags that are inappropriate for the given type of inode. */
 static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags)
@@ -2837,6 +2838,144 @@ static long btrfs_ioctl_scrub_progress(struct btrfs_root *root,
 	return ret;
 }
 
+static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg)
+{
+	int ret = 0;
+	int i;
+	unsigned long rel_ptr;
+	int size;
+	struct btrfs_ioctl_ino_path_args *ipa;
+	struct inode_fs_paths *ipath = NULL;
+	struct btrfs_path *path;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	ipa = memdup_user(arg, sizeof(*ipa));
+	if (IS_ERR(ipa)) {
+		ret = PTR_ERR(ipa);
+		ipa = NULL;
+		goto out;
+	}
+
+	size = min_t(u32, ipa->size, 4096);
+	ipath = init_ipath(size, root, path);
+	if (IS_ERR(ipath)) {
+		ret = PTR_ERR(ipath);
+		ipath = NULL;
+		goto out;
+	}
+
+	ret = paths_from_inode(ipa->inum, ipath);
+	if (ret < 0)
+		goto out;
+
+	for (i = 0; i < ipath->fspath->elem_cnt; ++i) {
+		rel_ptr = ipath->fspath->str[i] - (char *)ipath->fspath->str;
+		ipath->fspath->str[i] = (void *)rel_ptr;
+	}
+
+	ret = copy_to_user(ipa->fspath, ipath->fspath, size);
+	if (ret) {
+		ret = -EFAULT;
+		goto out;
+	}
+
+out:
+	btrfs_free_path(path);
+	free_ipath(ipath);
+	kfree(ipa);
+
+	return ret;
+}
+
+static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
+{
+	struct btrfs_data_container *inodes = ctx;
+	const size_t c = 3 * sizeof(u64);
+
+	if (inodes->bytes_left >= c) {
+		inodes->bytes_left -= c;
+		inodes->val[inodes->elem_cnt] = inum;
+		inodes->val[inodes->elem_cnt + 1] = offset;
+		inodes->val[inodes->elem_cnt + 2] = root;
+		inodes->elem_cnt += 3;
+	} else {
+		inodes->bytes_missing += c - inodes->bytes_left;
+		inodes->bytes_left = 0;
+		inodes->elem_missed += 3;
+	}
+
+	return 0;
+}
+
+static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root,
+					void __user *arg)
+{
+	int ret = 0;
+	int size;
+	u64 extent_offset;
+	struct btrfs_ioctl_logical_ino_args *loi;
+	struct btrfs_data_container *inodes = NULL;
+	struct btrfs_path *path = NULL;
+	struct btrfs_key key;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	loi = memdup_user(arg, sizeof(*loi));
+	if (IS_ERR(loi)) {
+		ret = PTR_ERR(loi);
+		loi = NULL;
+		goto out;
+	}
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	size = min_t(u32, loi->size, 4096);
+	inodes = init_data_container(size);
+	if (IS_ERR(inodes)) {
+		ret = PTR_ERR(inodes);
+		inodes = NULL;
+		goto out;
+	}
+
+	ret = extent_from_logical(root->fs_info, loi->logical, path, &key);
+
+	if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+		ret = -ENOENT;
+	if (ret < 0)
+		goto out;
+
+	extent_offset = loi->logical - key.objectid;
+	ret = iterate_extent_inodes(root->fs_info, path, key.objectid,
+					extent_offset, build_ino_list, inodes);
+
+	if (ret < 0)
+		goto out;
+
+	ret = copy_to_user(loi->inodes, inodes, size);
+	if (ret)
+		ret = -EFAULT;
+
+out:
+	btrfs_free_path(path);
+	kfree(inodes);
+	kfree(loi);
+
+	return ret;
+}
+
 long btrfs_ioctl(struct file *file, unsigned int
 		cmd, unsigned long arg)
 {
@@ -2894,6 +3033,10 @@ long btrfs_ioctl(struct file *file, unsigned int
 		return btrfs_ioctl_tree_search(file, argp);
 	case BTRFS_IOC_INO_LOOKUP:
 		return btrfs_ioctl_ino_lookup(file, argp);
+	case BTRFS_IOC_INO_PATHS:
+		return btrfs_ioctl_ino_to_path(root, argp);
+	case BTRFS_IOC_LOGICAL_INO:
+		return btrfs_ioctl_logical_to_ino(root, argp);
 	case BTRFS_IOC_SPACE_INFO:
 		return btrfs_ioctl_space_info(root, argp);
 	case BTRFS_IOC_SYNC:
diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h
index 1857e38..2da30d4 100644
--- a/fs/btrfs/ioctl.h
+++ b/fs/btrfs/ioctl.h
@@ -204,6 +204,20 @@ struct btrfs_data_container {
 	};
 };
 
+struct btrfs_ioctl_ino_path_args {
+	__u64				inum;		/* in */
+	__u32				size;		/* in */
+	__u64				reserved[4];
+	struct btrfs_data_container	*fspath;	/* out */
+};
+
+struct btrfs_ioctl_logical_ino_args {
+	__u64				logical;	/* in */
+	__u32				size;		/* in */
+	__u64				reserved[4];
+	struct btrfs_data_container	*inodes;	/* out */
+};
+
 #define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \
 				   struct btrfs_ioctl_vol_args)
 #define BTRFS_IOC_DEFRAG _IOW(BTRFS_IOCTL_MAGIC, 2, \
@@ -259,4 +273,9 @@ struct btrfs_data_container {
 				 struct btrfs_ioctl_dev_info_args)
 #define BTRFS_IOC_FS_INFO _IOR(BTRFS_IOCTL_MAGIC, 31, \
 			       struct btrfs_ioctl_fs_info_args)
+#define BTRFS_IOC_INO_PATHS _IOWR(BTRFS_IOCTL_MAGIC, 35, \
+					struct btrfs_ioctl_ino_path_args)
+#define BTRFS_IOC_LOGICAL_INO _IOWR(BTRFS_IOCTL_MAGIC, 36, \
+					struct btrfs_ioctl_ino_path_args)
+
 #endif
-- 
1.7.3.4