From 60b62978bc5e903cd487de34972fb30f76c74a2e Mon Sep 17 00:00:00 2001
From: David Sterba
Date: Tue, 30 Apr 2013 17:29:29 +0000
Subject: btrfs: annotate quota tree for lockdep

Quota tree has been missing from lockdep annotations, though no warning
has been seen in the wild.

There's currently one entry that does not belong there,
BTRFS_ORPHAN_OBJECTID.  No such tree exists, it's probably a copy &
paste mistake, the id is defined among tree ids.

Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/ctree.h   | 6 +++---
 fs/btrfs/disk-io.c | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 63c328a9ce95..2720d5555883 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -88,12 +88,12 @@ struct btrfs_ordered_sum;
 /* holds checksums of all the data extents */
 #define BTRFS_CSUM_TREE_OBJECTID 7ULL
 
-/* for storing balance parameters in the root tree */
-#define BTRFS_BALANCE_OBJECTID -4ULL
-
 /* holds quota configuration and tracking */
 #define BTRFS_QUOTA_TREE_OBJECTID 8ULL
 
+/* for storing balance parameters in the root tree */
+#define BTRFS_BALANCE_OBJECTID -4ULL
+
 /* orhpan objectid for tracking unlinked/truncated files */
 #define BTRFS_ORPHAN_OBJECTID -5ULL
 
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 4e9ebe1f1827..72b17276c255 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -152,7 +152,7 @@ static struct btrfs_lockdep_keyset {
 	{ .id = BTRFS_DEV_TREE_OBJECTID,	.name_stem = "dev"	},
 	{ .id = BTRFS_FS_TREE_OBJECTID,		.name_stem = "fs"	},
 	{ .id = BTRFS_CSUM_TREE_OBJECTID,	.name_stem = "csum"	},
-	{ .id = BTRFS_ORPHAN_OBJECTID,		.name_stem = "orphan"	},
+	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
 	{ .id = BTRFS_TREE_LOG_OBJECTID,	.name_stem = "log"	},
 	{ .id = BTRFS_TREE_RELOC_OBJECTID,	.name_stem = "treloc"	},
 	{ .id = BTRFS_DATA_RELOC_TREE_OBJECTID,	.name_stem = "dreloc"	},
-- 
cgit v1.2.3


From a52f4cd2b1a863a42c1cb268b1cddad451cdfede Mon Sep 17 00:00:00 2001
From: Liu Bo
Date: Wed, 1 May 2013 16:23:41 +0000
Subject: Btrfs: fix off-by-one in fiemap

lock_extent/unlock_extent expect an exclusive end.

Tested-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/extent_io.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index d2ac518f90e4..3e6e410002e5 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -3989,7 +3989,7 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 		last_for_get_extent = isize;
 	}
 
-	lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len, 0,
+	lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len - 1, 0,
 			 &cached_state);
 
 	em = get_extent_skip_holes(inode, start, last_for_get_extent,
@@ -4076,7 +4076,7 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 out_free:
 	free_extent_map(em);
 out:
-	unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, start + len,
+	unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, start + len - 1,
 			     &cached_state, GFP_NOFS);
 	return ret;
 }
-- 
cgit v1.2.3


From 03b71c6ca6286625d8f1ed44aabab9b5bf5dac10 Mon Sep 17 00:00:00 2001
From: Gabriel de Perthuis
Date: Mon, 6 May 2013 17:40:18 +0000
Subject: btrfs: don't stop searching after encountering the wrong item

The search ioctl skips items that are too large for a result buffer, but
inline items of a certain size occuring before any search result is
found would trigger an overflow and stop the search entirely.

Bug: https://bugzilla.kernel.org/show_bug.cgi?id=57641

Cc: stable@vger.kernel.org
Signed-off-by: Gabriel de Perthuis <g2p.code+btrfs@gmail.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/ioctl.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 0de4a2fcfb24..0f81d67cdc8d 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -1801,7 +1801,11 @@ static noinline int copy_to_sk(struct btrfs_root *root,
 		item_off = btrfs_item_ptr_offset(leaf, i);
 		item_len = btrfs_item_size_nr(leaf, i);
 
-		if (item_len > BTRFS_SEARCH_ARGS_BUFSIZE)
+		btrfs_item_key_to_cpu(leaf, key, i);
+		if (!key_in_sk(key, sk))
+			continue;
+
+		if (sizeof(sh) + item_len > BTRFS_SEARCH_ARGS_BUFSIZE)
 			item_len = 0;
 
 		if (sizeof(sh) + item_len + *sk_offset >
@@ -1810,10 +1814,6 @@ static noinline int copy_to_sk(struct btrfs_root *root,
 			goto overflow;
 		}
 
-		btrfs_item_key_to_cpu(leaf, key, i);
-		if (!key_in_sk(key, sk))
-			continue;
-
 		sh.objectid = key->objectid;
 		sh.offset = key->offset;
 		sh.type = key->type;
-- 
cgit v1.2.3


From 69a85bd87cc81bcbd36730d4a1214c12fdb8a548 Mon Sep 17 00:00:00 2001
From: Josef Bacik
Date: Wed, 8 May 2013 13:30:11 -0400
Subject: Btrfs: don't null pointer deref on abort

I'm sorry, theres no excuse for this sort of work.  We need to use
root->leafsize since eb may be NULL.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/disk-io.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs')

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 72b17276c255..e8b29da30154 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -3808,7 +3808,7 @@ static int btrfs_destroy_marked_extents(struct btrfs_root *root,
 		while (start <= end) {
 			eb = btrfs_find_tree_block(root, start,
 						   root->leafsize);
-			start += eb->len;
+			start += root->leafsize;
 			if (!eb)
 				continue;
 			wait_on_extent_buffer_writeback(eb);
-- 
cgit v1.2.3


From 73e1e61fb85ab206854b6d87ff31733628bb8d72 Mon Sep 17 00:00:00 2001
From: Josef Bacik
Date: Wed, 8 May 2013 16:44:57 -0400
Subject: Btrfs: remove warn on in free space cache writeout

This catches block groups that are too large to properly cache.  We deal with
this case fine, so the warning just confuses users.  Remove the warning.
Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/free-space-cache.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index ecca6c7375a6..6a8bb9c79674 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -920,10 +920,8 @@ static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
 
 	/* Make sure we can fit our crcs into the first page */
 	if (io_ctl.check_crcs &&
-	    (io_ctl.num_pages * sizeof(u32)) >= PAGE_CACHE_SIZE) {
-		WARN_ON(1);
+	    (io_ctl.num_pages * sizeof(u32)) >= PAGE_CACHE_SIZE)
 		goto out_nospc;
-	}
 
 	io_ctl_set_generation(&io_ctl, trans->transid);
 
-- 
cgit v1.2.3


From b1c79e0947e0c190f865e2eb7b84a0fea0021cec Mon Sep 17 00:00:00 2001
From: Josef Bacik
Date: Thu, 9 May 2013 13:49:30 -0400
Subject: Btrfs: handle running extent ops with skinny metadata

Chris hit a bug where we weren't finding extent records when running extent ops.
This is because we use the delayed_ref_head when running the extent op, which
means we can't use the ->type checks to see if we are metadata.  We also lose
the level of the metadata we are working on.  So to fix this we can just check
the ->is_data section of the extent_op, and we can store the level of the buffer
we were modifying in the extent_op.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/ctree.c       |  4 +++-
 fs/btrfs/ctree.h       |  2 +-
 fs/btrfs/delayed-ref.h |  1 +
 fs/btrfs/extent-tree.c | 15 +++++++--------
 4 files changed, 12 insertions(+), 10 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index de6de8e60b46..02fae7f7e42c 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -951,10 +951,12 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
 			BUG_ON(ret); /* -ENOMEM */
 		}
 		if (new_flags != 0) {
+			int level = btrfs_header_level(buf);
+
 			ret = btrfs_set_disk_extent_flags(trans, root,
 							  buf->start,
 							  buf->len,
-							  new_flags, 0);
+							  new_flags, level, 0);
 			if (ret)
 				return ret;
 		}
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 2720d5555883..d6dd49b51ba8 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -3075,7 +3075,7 @@ int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
 				u64 bytenr, u64 num_bytes, u64 flags,
-				int is_data);
+				int level, int is_data);
 int btrfs_free_extent(struct btrfs_trans_handle *trans,
 		      struct btrfs_root *root,
 		      u64 bytenr, u64 num_bytes, u64 parent, u64 root_objectid,
diff --git a/fs/btrfs/delayed-ref.h b/fs/btrfs/delayed-ref.h
index f75fcaf79aeb..70b962cc177d 100644
--- a/fs/btrfs/delayed-ref.h
+++ b/fs/btrfs/delayed-ref.h
@@ -60,6 +60,7 @@ struct btrfs_delayed_ref_node {
 struct btrfs_delayed_extent_op {
 	struct btrfs_disk_key key;
 	u64 flags_to_set;
+	int level;
 	unsigned int update_key:1;
 	unsigned int update_flags:1;
 	unsigned int is_data:1;
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 2305b5c5cf00..c4c94b30c729 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2070,8 +2070,7 @@ static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
 	u32 item_size;
 	int ret;
 	int err = 0;
-	int metadata = (node->type == BTRFS_TREE_BLOCK_REF_KEY ||
-			node->type == BTRFS_SHARED_BLOCK_REF_KEY);
+	int metadata = !extent_op->is_data;
 
 	if (trans->aborted)
 		return 0;
@@ -2086,11 +2085,8 @@ static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
 	key.objectid = node->bytenr;
 
 	if (metadata) {
-		struct btrfs_delayed_tree_ref *tree_ref;
-
-		tree_ref = btrfs_delayed_node_to_tree_ref(node);
 		key.type = BTRFS_METADATA_ITEM_KEY;
-		key.offset = tree_ref->level;
+		key.offset = extent_op->level;
 	} else {
 		key.type = BTRFS_EXTENT_ITEM_KEY;
 		key.offset = node->num_bytes;
@@ -2719,7 +2715,7 @@ out:
 int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
 				u64 bytenr, u64 num_bytes, u64 flags,
-				int is_data)
+				int level, int is_data)
 {
 	struct btrfs_delayed_extent_op *extent_op;
 	int ret;
@@ -2732,6 +2728,7 @@ int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
 	extent_op->update_flags = 1;
 	extent_op->update_key = 0;
 	extent_op->is_data = is_data ? 1 : 0;
+	extent_op->level = level;
 
 	ret = btrfs_add_delayed_extent_op(root->fs_info, trans, bytenr,
 					  num_bytes, extent_op);
@@ -6763,6 +6760,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 			extent_op->update_key = 1;
 		extent_op->update_flags = 1;
 		extent_op->is_data = 0;
+		extent_op->level = level;
 
 		ret = btrfs_add_delayed_tree_ref(root->fs_info, trans,
 					ins.objectid,
@@ -6934,7 +6932,8 @@ static noinline int walk_down_proc(struct btrfs_trans_handle *trans,
 		ret = btrfs_dec_ref(trans, root, eb, 0, wc->for_reloc);
 		BUG_ON(ret); /* -ENOMEM */
 		ret = btrfs_set_disk_extent_flags(trans, root, eb->start,
-						  eb->len, flag, 0);
+						  eb->len, flag,
+						  btrfs_header_level(eb), 0);
 		BUG_ON(ret); /* -ENOMEM */
 		wc->flags[level] |= flag;
 	}
-- 
cgit v1.2.3


From 49688107527a24b0ed3780576257a1225902180b Mon Sep 17 00:00:00 2001
From: Stefan Behrens
Date: Tue, 7 May 2013 17:28:03 +0000
Subject: Btrfs: don't allow device replace on RAID5/RAID6

This is not yet supported and causes crashes. One sad user reported
that it destroyed his filesystem.

One failure is in __btrfs_map_block+0xc1f calling kmalloc(0).

0x5f21f is in __btrfs_map_block (fs/btrfs/volumes.c:4923).
4918                            num_stripes = map->num_stripes;
4919                            max_errors = nr_parity_stripes(map);
4920
4921                            raid_map = kmalloc(sizeof(u64) * num_stripes,
4922                                               GFP_NOFS);
4923                            if (!raid_map) {
4924                                    ret = -ENOMEM;
4925                                    goto out;
4926                            }
4927

There might be more issues. Until this is really tested, don't allow
users to start the procedure on RAID5/RAID6 filesystems.

Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/dev-replace.c | 5 +++++
 1 file changed, 5 insertions(+)

(limited to 'fs')

diff --git a/fs/btrfs/dev-replace.c b/fs/btrfs/dev-replace.c
index 7ba7b3900cb8..65241f32d3f8 100644
--- a/fs/btrfs/dev-replace.c
+++ b/fs/btrfs/dev-replace.c
@@ -313,6 +313,11 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 	struct btrfs_device *tgt_device = NULL;
 	struct btrfs_device *src_device = NULL;
 
+	if (btrfs_fs_incompat(fs_info, RAID56)) {
+		pr_warn("btrfs: dev_replace cannot yet handle RAID5/RAID6\n");
+		return -EINVAL;
+	}
+
 	switch (args->start.cont_reading_from_srcdev_mode) {
 	case BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_ALWAYS:
 	case BTRFS_IOCTL_DEV_REPLACE_CONT_READING_FROM_SRCDEV_MODE_AVOID:
-- 
cgit v1.2.3


From c16c2e2e51c2f0951fffa73c343b8fcb641108ba Mon Sep 17 00:00:00 2001
From: Wang Shilong
Date: Wed, 8 May 2013 08:10:25 +0000
Subject: Btrfs: fix possible memory leak in the find_parent_nodes()

In the find_parent_nodes(), if read_tree_block() fails, we can
not return directly, we should free some allocated memory otherwise
memory leak happens.

Signed-off-by: Wang Shilong <wangsl-fnst@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/backref.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'fs')

diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index b4fb41558111..290e347b6db3 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -918,7 +918,8 @@ again:
 							   ref->parent, bsz, 0);
 				if (!eb || !extent_buffer_uptodate(eb)) {
 					free_extent_buffer(eb);
-					return -EIO;
+					ret = -EIO;
+					goto out;
 				}
 				ret = find_extent_in_eb(eb, bytenr,
 							*extent_item_pos, &eie);
-- 
cgit v1.2.3


From 379cde741b220091d2124fb500b178b90ad7f460 Mon Sep 17 00:00:00 2001
From: Stefan Behrens
Date: Wed, 8 May 2013 08:56:09 +0000
Subject: Btrfs: fix possible memory leak in replace_path()

In replace_path(), if read_tree_block() fails, we cannot return
directly, we should free some allocated memory otherwise memory
leak happens.

Similar to Wang's "Btrfs: fix possible memory leak in the
find_parent_nodes()" patch, the current commit fixes an issue that
is related to the "Btrfs: fix all callers of read_tree_block"
commit.

Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/relocation.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs')

diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 704a1b8d2a2b..5c5b8bb44ee5 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -1773,7 +1773,7 @@ again:
 			if (!eb || !extent_buffer_uptodate(eb)) {
 				ret = (!eb) ? -ENOMEM : -EIO;
 				free_extent_buffer(eb);
-				return ret;
+				break;
 			}
 			btrfs_tree_lock(eb);
 			if (cow) {
-- 
cgit v1.2.3


From 8250dabedb633e162bce89f2aacf5e65fa9e6464 Mon Sep 17 00:00:00 2001
From: Andreas Philipp
Date: Sat, 11 May 2013 11:13:03 +0000
Subject: Correct allowed raid levels on balance.

Raid5 with 3 devices is well defined while the old logic allowed
raid5 only with a minimum of 4 devices when converting the block group
profile via btrfs balance. Creating a raid5 with just three devices
using mkfs.btrfs worked always as expected. This is now fixed and the
whole logic is rewritten.

Signed-off-by: Andreas Philipp <philipp.andreas@gmail.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/volumes.c | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index a191bac31d85..062e93069c5e 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -3120,14 +3120,13 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 	allowed = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
 	if (num_devices == 1)
 		allowed |= BTRFS_BLOCK_GROUP_DUP;
-	else if (num_devices < 4)
+	else if (num_devices > 1)
 		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1);
-	else
-		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 |
-				BTRFS_BLOCK_GROUP_RAID10 |
-				BTRFS_BLOCK_GROUP_RAID5 |
-				BTRFS_BLOCK_GROUP_RAID6);
-
+	if (num_devices > 2)
+		allowed |= BTRFS_BLOCK_GROUP_RAID5;
+	if (num_devices > 3)
+		allowed |= (BTRFS_BLOCK_GROUP_RAID10 |
+			    BTRFS_BLOCK_GROUP_RAID6);
 	if ((bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
 	    (!alloc_profile_is_valid(bctl->data.target, 1) ||
 	     (bctl->data.target & ~allowed))) {
-- 
cgit v1.2.3


From 7cfa9e51d2948ae90e7599cc114dcce2c7c2b1fc Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Mon, 13 May 2013 13:55:08 +0000
Subject: Btrfs: don't abort the current transaction if there is no enough
 space for inode cache

The filesystem with inode cache was forced to be read-only when we umounted it.

Steps to reproduce:
 # mkfs.btrfs -f ${DEV}
 # mount -o inode_cache ${DEV} ${MNT}
 # dd if=/dev/zero of=${MNT}/file1 bs=1M count=8192
 # btrfs fi syn ${MNT}
 # dd if=${MNT}/file1 of=/dev/null bs=1M
 # rm -f ${MNT}/file1
 # btrfs fi syn ${MNT}
 # umount ${MNT}

It is because there was no enough space to do inode cache truncation, and then
we aborted the current transaction.

But no space error is not a serious problem when we write out the inode cache,
and it is safe that we just skip this step if we meet this problem. So we need
not abort the current transaction.

Reported-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Tested-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/inode-map.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'fs')

diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c
index d26f67a59e36..9818d4a3f829 100644
--- a/fs/btrfs/inode-map.c
+++ b/fs/btrfs/inode-map.c
@@ -468,7 +468,8 @@ again:
 	if (i_size_read(inode) > 0) {
 		ret = btrfs_truncate_free_space_cache(root, trans, path, inode);
 		if (ret) {
-			btrfs_abort_transaction(trans, root, ret);
+			if (ret != -ENOSPC)
+				btrfs_abort_transaction(trans, root, ret);
 			goto out_put;
 		}
 	}
-- 
cgit v1.2.3


From 7b61cd92242542944fc27024900c495a6a7b3396 Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Mon, 13 May 2013 13:55:09 +0000
Subject: Btrfs: don't use global block reservation for inode cache truncation

It is very likely that there are lots of subvolumes/snapshots in the filesystem,
so if we use global block reservation to do inode cache truncation, we may hog
all the free space that is reserved in global rsv. So it is better that we do
the free space reservation for inode cache truncation by ourselves.

Cc: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/extent-tree.c      |  5 +++++
 fs/btrfs/free-space-cache.c | 39 +++++++++++++++++++--------------------
 fs/btrfs/free-space-cache.h |  2 ++
 fs/btrfs/inode-map.c        |  5 +++--
 fs/btrfs/relocation.c       |  5 +++++
 5 files changed, 34 insertions(+), 22 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index c4c94b30c729..162a66bfbffd 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -3106,6 +3106,11 @@ again:
 	WARN_ON(ret);
 
 	if (i_size_read(inode) > 0) {
+		ret = btrfs_check_trunc_cache_free_space(root,
+					&root->fs_info->global_block_rsv);
+		if (ret)
+			goto out_put;
+
 		ret = btrfs_truncate_free_space_cache(root, trans, path,
 						      inode);
 		if (ret)
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 6a8bb9c79674..e53009657f0e 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -197,30 +197,32 @@ int create_free_space_inode(struct btrfs_root *root,
 					 block_group->key.objectid);
 }
 
-int btrfs_truncate_free_space_cache(struct btrfs_root *root,
-				    struct btrfs_trans_handle *trans,
-				    struct btrfs_path *path,
-				    struct inode *inode)
+int btrfs_check_trunc_cache_free_space(struct btrfs_root *root,
+				       struct btrfs_block_rsv *rsv)
 {
-	struct btrfs_block_rsv *rsv;
 	u64 needed_bytes;
-	loff_t oldsize;
-	int ret = 0;
-
-	rsv = trans->block_rsv;
-	trans->block_rsv = &root->fs_info->global_block_rsv;
+	int ret;
 
 	/* 1 for slack space, 1 for updating the inode */
 	needed_bytes = btrfs_calc_trunc_metadata_size(root, 1) +
 		btrfs_calc_trans_metadata_size(root, 1);
 
-	spin_lock(&trans->block_rsv->lock);
-	if (trans->block_rsv->reserved < needed_bytes) {
-		spin_unlock(&trans->block_rsv->lock);
-		trans->block_rsv = rsv;
-		return -ENOSPC;
-	}
-	spin_unlock(&trans->block_rsv->lock);
+	spin_lock(&rsv->lock);
+	if (rsv->reserved < needed_bytes)
+		ret = -ENOSPC;
+	else
+		ret = 0;
+	spin_unlock(&rsv->lock);
+	return 0;
+}
+
+int btrfs_truncate_free_space_cache(struct btrfs_root *root,
+				    struct btrfs_trans_handle *trans,
+				    struct btrfs_path *path,
+				    struct inode *inode)
+{
+	loff_t oldsize;
+	int ret = 0;
 
 	oldsize = i_size_read(inode);
 	btrfs_i_size_write(inode, 0);
@@ -232,9 +234,7 @@ int btrfs_truncate_free_space_cache(struct btrfs_root *root,
 	 */
 	ret = btrfs_truncate_inode_items(trans, root, inode,
 					 0, BTRFS_EXTENT_DATA_KEY);
-
 	if (ret) {
-		trans->block_rsv = rsv;
 		btrfs_abort_transaction(trans, root, ret);
 		return ret;
 	}
@@ -242,7 +242,6 @@ int btrfs_truncate_free_space_cache(struct btrfs_root *root,
 	ret = btrfs_update_inode(trans, root, inode);
 	if (ret)
 		btrfs_abort_transaction(trans, root, ret);
-	trans->block_rsv = rsv;
 
 	return ret;
 }
diff --git a/fs/btrfs/free-space-cache.h b/fs/btrfs/free-space-cache.h
index 4dc17d8809c7..8b7f19f44961 100644
--- a/fs/btrfs/free-space-cache.h
+++ b/fs/btrfs/free-space-cache.h
@@ -54,6 +54,8 @@ int create_free_space_inode(struct btrfs_root *root,
 			    struct btrfs_block_group_cache *block_group,
 			    struct btrfs_path *path);
 
+int btrfs_check_trunc_cache_free_space(struct btrfs_root *root,
+				       struct btrfs_block_rsv *rsv);
 int btrfs_truncate_free_space_cache(struct btrfs_root *root,
 				    struct btrfs_trans_handle *trans,
 				    struct btrfs_path *path,
diff --git a/fs/btrfs/inode-map.c b/fs/btrfs/inode-map.c
index 9818d4a3f829..2c66ddbbe670 100644
--- a/fs/btrfs/inode-map.c
+++ b/fs/btrfs/inode-map.c
@@ -429,11 +429,12 @@ int btrfs_save_ino_cache(struct btrfs_root *root,
 	num_bytes = trans->bytes_reserved;
 	/*
 	 * 1 item for inode item insertion if need
-	 * 3 items for inode item update (in the worst case)
+	 * 4 items for inode item update (in the worst case)
+	 * 1 items for slack space if we need do truncation
 	 * 1 item for free space object
 	 * 3 items for pre-allocation
 	 */
-	trans->bytes_reserved = btrfs_calc_trans_metadata_size(root, 8);
+	trans->bytes_reserved = btrfs_calc_trans_metadata_size(root, 10);
 	ret = btrfs_block_rsv_add(root, trans->block_rsv,
 				  trans->bytes_reserved,
 				  BTRFS_RESERVE_NO_FLUSH);
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 5c5b8bb44ee5..395b82031a42 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -3350,6 +3350,11 @@ static int delete_block_group_cache(struct btrfs_fs_info *fs_info,
 	}
 
 truncate:
+	ret = btrfs_check_trunc_cache_free_space(root,
+						 &fs_info->global_block_rsv);
+	if (ret)
+		goto out;
+
 	path = btrfs_alloc_path();
 	if (!path) {
 		ret = -ENOMEM;
-- 
cgit v1.2.3


From b586b32374909863311b7c916c7c0c709141e35a Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Mon, 13 May 2013 13:55:10 +0000
Subject: Btrfs: optimize the error handle of use_block_rsv()

cc: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/extent-tree.c | 65 ++++++++++++++++++++++----------------------------
 1 file changed, 28 insertions(+), 37 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 162a66bfbffd..5e2c0bf84075 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -6656,48 +6656,39 @@ use_block_rsv(struct btrfs_trans_handle *trans,
 
 	block_rsv = get_block_rsv(trans, root);
 
-	if (block_rsv->size == 0) {
-		ret = reserve_metadata_bytes(root, block_rsv, blocksize,
-					     BTRFS_RESERVE_NO_FLUSH);
-		/*
-		 * If we couldn't reserve metadata bytes try and use some from
-		 * the global reserve.
-		 */
-		if (ret && block_rsv != global_rsv) {
-			ret = block_rsv_use_bytes(global_rsv, blocksize);
-			if (!ret)
-				return global_rsv;
-			return ERR_PTR(ret);
-		} else if (ret) {
-			return ERR_PTR(ret);
-		}
-		return block_rsv;
-	}
+	if (unlikely(block_rsv->size == 0))
+		goto try_reserve;
 
 	ret = block_rsv_use_bytes(block_rsv, blocksize);
 	if (!ret)
 		return block_rsv;
-	if (ret && !block_rsv->failfast) {
-		if (btrfs_test_opt(root, ENOSPC_DEBUG)) {
-			static DEFINE_RATELIMIT_STATE(_rs,
-					DEFAULT_RATELIMIT_INTERVAL * 10,
-					/*DEFAULT_RATELIMIT_BURST*/ 1);
-			if (__ratelimit(&_rs))
-				WARN(1, KERN_DEBUG
-					"btrfs: block rsv returned %d\n", ret);
-		}
-		ret = reserve_metadata_bytes(root, block_rsv, blocksize,
-					     BTRFS_RESERVE_NO_FLUSH);
-		if (!ret) {
-			return block_rsv;
-		} else if (ret && block_rsv != global_rsv) {
-			ret = block_rsv_use_bytes(global_rsv, blocksize);
-			if (!ret)
-				return global_rsv;
-		}
-	}
 
-	return ERR_PTR(-ENOSPC);
+	if (block_rsv->failfast)
+		return ERR_PTR(ret);
+
+	if (btrfs_test_opt(root, ENOSPC_DEBUG)) {
+		static DEFINE_RATELIMIT_STATE(_rs,
+				DEFAULT_RATELIMIT_INTERVAL * 10,
+				/*DEFAULT_RATELIMIT_BURST*/ 1);
+		if (__ratelimit(&_rs))
+			WARN(1, KERN_DEBUG
+				"btrfs: block rsv returned %d\n", ret);
+	}
+try_reserve:
+	ret = reserve_metadata_bytes(root, block_rsv, blocksize,
+				     BTRFS_RESERVE_NO_FLUSH);
+	if (!ret)
+		return block_rsv;
+	/*
+	 * If we couldn't reserve metadata bytes try and use some from
+	 * the global reserve.
+	 */
+	if (block_rsv->type != BTRFS_BLOCK_RSV_GLOBAL) {
+		ret = block_rsv_use_bytes(global_rsv, blocksize);
+		if (!ret)
+			return global_rsv;
+	}
+	return ERR_PTR(ret);
 }
 
 static void unuse_block_rsv(struct btrfs_fs_info *fs_info,
-- 
cgit v1.2.3


From 5881cfc924c8143dbc3e1f343516fc6527eb8311 Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Mon, 13 May 2013 13:55:11 +0000
Subject: Btrfs: don't steal the reserved space from the global reserve if
 their space type is different

If the type of the space we need is different with the global reserve, we
can not steal the space from the global reserve, because we can not allocate
the space from the free space cache that the global reserve points to.

Cc: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/extent-tree.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 5e2c0bf84075..54e63b273a64 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -6681,9 +6681,11 @@ try_reserve:
 		return block_rsv;
 	/*
 	 * If we couldn't reserve metadata bytes try and use some from
-	 * the global reserve.
+	 * the global reserve if its space type is the same as the global
+	 * reservation.
 	 */
-	if (block_rsv->type != BTRFS_BLOCK_RSV_GLOBAL) {
+	if (block_rsv->type != BTRFS_BLOCK_RSV_GLOBAL &&
+	    block_rsv->space_info == global_rsv->space_info) {
 		ret = block_rsv_use_bytes(global_rsv, blocksize);
 		if (!ret)
 			return global_rsv;
-- 
cgit v1.2.3


From d88033dbf4c23279b012725876f1e164e09644ff Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Mon, 13 May 2013 13:55:12 +0000
Subject: Btrfs: update the global reserve if it is empty

Before applying this patch, we reserved the space for the global reserve
by the minimum unit if we found it is empty, it was unreasonable and
inefficient, because if the global reserve space was depleted, it implied
that the size of the global reserve was too small. In this case, we shoud
update the global reserve and fill it.

Cc: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/extent-tree.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

(limited to 'fs')

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 54e63b273a64..42f5e6196021 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -6653,12 +6653,13 @@ use_block_rsv(struct btrfs_trans_handle *trans,
 	struct btrfs_block_rsv *block_rsv;
 	struct btrfs_block_rsv *global_rsv = &root->fs_info->global_block_rsv;
 	int ret;
+	bool global_updated = false;
 
 	block_rsv = get_block_rsv(trans, root);
 
 	if (unlikely(block_rsv->size == 0))
 		goto try_reserve;
-
+again:
 	ret = block_rsv_use_bytes(block_rsv, blocksize);
 	if (!ret)
 		return block_rsv;
@@ -6666,6 +6667,12 @@ use_block_rsv(struct btrfs_trans_handle *trans,
 	if (block_rsv->failfast)
 		return ERR_PTR(ret);
 
+	if (block_rsv->type == BTRFS_BLOCK_RSV_GLOBAL && !global_updated) {
+		global_updated = true;
+		update_global_block_rsv(root->fs_info);
+		goto again;
+	}
+
 	if (btrfs_test_opt(root, ENOSPC_DEBUG)) {
 		static DEFINE_RATELIMIT_STATE(_rs,
 				DEFAULT_RATELIMIT_INTERVAL * 10,
-- 
cgit v1.2.3


From b9aa55bed1c1a3a329da31884b643c62d57ebb21 Mon Sep 17 00:00:00 2001
From: Liu Bo
Date: Tue, 14 May 2013 02:12:15 +0000
Subject: Btrfs: return errno if possible when we fail to allocate memory

We need to set return value explicitly, otherwise we'll lose the error
value.

Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/inode.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 1669c3b4be2f..99a9c25d36a6 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -714,8 +714,10 @@ retry:
 					async_extent->ram_size - 1, 0);
 
 		em = alloc_extent_map();
-		if (!em)
+		if (!em) {
+			ret = -ENOMEM;
 			goto out_free_reserve;
+		}
 		em->start = async_extent->start;
 		em->len = async_extent->ram_size;
 		em->orig_start = em->start;
@@ -922,8 +924,10 @@ static noinline int __cow_file_range(struct btrfs_trans_handle *trans,
 		}
 
 		em = alloc_extent_map();
-		if (!em)
+		if (!em) {
+			ret = -ENOMEM;
 			goto out_reserve;
+		}
 		em->start = start;
 		em->orig_start = em->start;
 		ram_size = ins.offset;
-- 
cgit v1.2.3


From 89042e5ad23d50449691141334f30d53d6271266 Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Wed, 15 May 2013 07:48:15 +0000
Subject: Btrfs: fix accessing a freed tree root

inode_tree_del() will move the tree root into the dead root list, and
then the tree will be destroyed by the cleaner. So if we remove the
delayed node which is cached in the inode after inode_tree_del(),
we may access a freed tree root. Fix it.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/inode.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs')

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 99a9c25d36a6..790eceb48fb0 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4727,6 +4727,7 @@ void btrfs_evict_inode(struct inode *inode)
 	btrfs_end_transaction(trans, root);
 	btrfs_btree_balance_dirty(root);
 no_delete:
+	btrfs_remove_delayed_node(inode);
 	clear_inode(inode);
 	return;
 }
@@ -7982,7 +7983,6 @@ void btrfs_destroy_inode(struct inode *inode)
 	inode_tree_del(inode);
 	btrfs_drop_extent_cache(inode, 0, (u64)-1, 0);
 free:
-	btrfs_remove_delayed_node(inode);
 	call_rcu(&inode->i_rcu, btrfs_i_callback);
 }
 
-- 
cgit v1.2.3


From e1409cef85894f96f4bddc6633d64d1c5275e2a3 Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Wed, 15 May 2013 07:48:16 +0000
Subject: Btrfs: fix unprotected root node of the subvolume's inode rb-tree

The root node of the rb-tree may be changed, so we should get it under
the lock. Fix it.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/inode.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 790eceb48fb0..19eef3e852b0 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4843,14 +4843,13 @@ static void inode_tree_add(struct inode *inode)
 	struct rb_node **p;
 	struct rb_node *parent;
 	u64 ino = btrfs_ino(inode);
-again:
-	p = &root->inode_tree.rb_node;
-	parent = NULL;
 
 	if (inode_unhashed(inode))
 		return;
-
+again:
+	parent = NULL;
 	spin_lock(&root->inode_lock);
+	p = &root->inode_tree.rb_node;
 	while (*p) {
 		parent = *p;
 		entry = rb_entry(parent, struct btrfs_inode, rb_node);
-- 
cgit v1.2.3


From 061594ef171a5ba52b5786688ae766907b0bda2b Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Wed, 15 May 2013 07:48:17 +0000
Subject: Btrfs: pause the space balance when remounting to R/O

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/super.c | 1 +
 1 file changed, 1 insertion(+)

(limited to 'fs')

diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index a4807ced23cc..f0857e092a3c 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -1263,6 +1263,7 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 
 		btrfs_dev_replace_suspend_for_unmount(fs_info);
 		btrfs_scrub_cancel(fs_info);
+		btrfs_pause_balance(fs_info);
 
 		ret = btrfs_commit_super(root);
 		if (ret)
-- 
cgit v1.2.3


From 314297c2a3fbcbda992507f70cd04cc82084e434 Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Wed, 15 May 2013 07:48:18 +0000
Subject: Btrfs: remove BUG_ON() in btrfs_read_fs_tree_no_radix()

We have checked if ->node is NULL or not, so it is unnecessary to
use BUG_ON() to check again. Remove it.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/disk-io.c | 1 -
 1 file changed, 1 deletion(-)

(limited to 'fs')

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index e8b29da30154..9b9f28664b4f 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1513,7 +1513,6 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
 	}
 
 	root->commit_root = btrfs_root_node(root);
-	BUG_ON(!root->node); /* -ENOMEM */
 out:
 	if (location->objectid != BTRFS_TREE_LOG_OBJECTID) {
 		root->ref_cows = 1;
-- 
cgit v1.2.3


From b216cbfb52c08300c203abf06ea9519d15d10045 Mon Sep 17 00:00:00 2001
From: Miao Xie
Date: Wed, 15 May 2013 07:48:21 +0000
Subject: Btrfs: don't invoke btrfs_invalidate_inodes() in the spin lock
 context

btrfs_invalidate_inodes() may sleep, so we should not invoke it in the
spin lock context. Fix it.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/disk-io.c | 6 ++++++
 1 file changed, 6 insertions(+)

(limited to 'fs')

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 9b9f28664b4f..1b03f8393a69 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -3658,8 +3658,11 @@ static void btrfs_destroy_ordered_operations(struct btrfs_transaction *t,
 					 ordered_operations);
 
 		list_del_init(&btrfs_inode->ordered_operations);
+		spin_unlock(&root->fs_info->ordered_extent_lock);
 
 		btrfs_invalidate_inodes(btrfs_inode->root);
+
+		spin_lock(&root->fs_info->ordered_extent_lock);
 	}
 
 	spin_unlock(&root->fs_info->ordered_extent_lock);
@@ -3781,8 +3784,11 @@ static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
 		list_del_init(&btrfs_inode->delalloc_inodes);
 		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
 			  &btrfs_inode->runtime_flags);
+		spin_unlock(&root->fs_info->delalloc_lock);
 
 		btrfs_invalidate_inodes(btrfs_inode->root);
+
+		spin_lock(&root->fs_info->delalloc_lock);
 	}
 
 	spin_unlock(&root->fs_info->delalloc_lock);
-- 
cgit v1.2.3


From 17a5adccf3fd01added91f3bf9aa7ee9aa28843b Mon Sep 17 00:00:00 2001
From: Alexandre Oliva
Date: Wed, 15 May 2013 11:38:55 -0400
Subject: btrfs: do away with non-whole_page extent I/O

end_bio_extent_readpage computes whole_page based on bv_offset and
bv_len, without taking into account that blk_update_request may modify
them when some of the blocks to be read into a page produce a read
error.  This would cause the read to unlock only part of the file
range associated with the page, which would in turn leave the entire
page locked, which would not only keep the process blocked instead of
returning -EIO to it, but also prevent any further access to the file.

It turns out that btrfs always issues whole-page reads and writes.
The special handling of non-whole_page appears to be a mistake or a
left-over from a time when this wasn't the case.  Indeed,
end_bio_extent_writepage distinguished between whole_page and
non-whole_page writes but behaved identically in both cases!

I've replaced the whole_page computations with warnings, just to be
sure that we're not issuing partial page reads or writes.  The
warnings should probably just go away some time.

Signed-off-by: Alexandre Oliva <oliva@gnu.org>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/extent_io.c | 85 +++++++++++++++++++---------------------------------
 1 file changed, 30 insertions(+), 55 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 3e6e410002e5..ca4355ddea06 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -1947,28 +1947,6 @@ static void check_page_uptodate(struct extent_io_tree *tree, struct page *page)
 		SetPageUptodate(page);
 }
 
-/*
- * helper function to unlock a page if all the extents in the tree
- * for that page are unlocked
- */
-static void check_page_locked(struct extent_io_tree *tree, struct page *page)
-{
-	u64 start = page_offset(page);
-	u64 end = start + PAGE_CACHE_SIZE - 1;
-	if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0, NULL))
-		unlock_page(page);
-}
-
-/*
- * helper function to end page writeback if all the extents
- * in the tree for that page are done with writeback
- */
-static void check_page_writeback(struct extent_io_tree *tree,
-				 struct page *page)
-{
-	end_page_writeback(page);
-}
-
 /*
  * When IO fails, either with EIO or csum verification fails, we
  * try other mirrors that might have a good copy of the data.  This
@@ -2398,19 +2376,24 @@ static void end_bio_extent_writepage(struct bio *bio, int err)
 	struct extent_io_tree *tree;
 	u64 start;
 	u64 end;
-	int whole_page;
 
 	do {
 		struct page *page = bvec->bv_page;
 		tree = &BTRFS_I(page->mapping->host)->io_tree;
 
-		start = page_offset(page) + bvec->bv_offset;
-		end = start + bvec->bv_len - 1;
+		/* We always issue full-page reads, but if some block
+		 * in a page fails to read, blk_update_request() will
+		 * advance bv_offset and adjust bv_len to compensate.
+		 * Print a warning for nonzero offsets, and an error
+		 * if they don't add up to a full page.  */
+		if (bvec->bv_offset || bvec->bv_len != PAGE_CACHE_SIZE)
+			printk("%s page write in btrfs with offset %u and length %u\n",
+			       bvec->bv_offset + bvec->bv_len != PAGE_CACHE_SIZE
+			       ? KERN_ERR "partial" : KERN_INFO "incomplete",
+			       bvec->bv_offset, bvec->bv_len);
 
-		if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE)
-			whole_page = 1;
-		else
-			whole_page = 0;
+		start = page_offset(page);
+		end = start + bvec->bv_offset + bvec->bv_len - 1;
 
 		if (--bvec >= bio->bi_io_vec)
 			prefetchw(&bvec->bv_page->flags);
@@ -2418,10 +2401,7 @@ static void end_bio_extent_writepage(struct bio *bio, int err)
 		if (end_extent_writepage(page, err, start, end))
 			continue;
 
-		if (whole_page)
-			end_page_writeback(page);
-		else
-			check_page_writeback(tree, page);
+		end_page_writeback(page);
 	} while (bvec >= bio->bi_io_vec);
 
 	bio_put(bio);
@@ -2446,7 +2426,6 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 	struct extent_io_tree *tree;
 	u64 start;
 	u64 end;
-	int whole_page;
 	int mirror;
 	int ret;
 
@@ -2463,13 +2442,19 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 			 (long int)bio->bi_bdev);
 		tree = &BTRFS_I(page->mapping->host)->io_tree;
 
-		start = page_offset(page) + bvec->bv_offset;
-		end = start + bvec->bv_len - 1;
+		/* We always issue full-page reads, but if some block
+		 * in a page fails to read, blk_update_request() will
+		 * advance bv_offset and adjust bv_len to compensate.
+		 * Print a warning for nonzero offsets, and an error
+		 * if they don't add up to a full page.  */
+		if (bvec->bv_offset || bvec->bv_len != PAGE_CACHE_SIZE)
+			printk("%s page read in btrfs with offset %u and length %u\n",
+			       bvec->bv_offset + bvec->bv_len != PAGE_CACHE_SIZE
+			       ? KERN_ERR "partial" : KERN_INFO "incomplete",
+			       bvec->bv_offset, bvec->bv_len);
 
-		if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE)
-			whole_page = 1;
-		else
-			whole_page = 0;
+		start = page_offset(page);
+		end = start + bvec->bv_offset + bvec->bv_len - 1;
 
 		if (++bvec <= bvec_end)
 			prefetchw(&bvec->bv_page->flags);
@@ -2528,23 +2513,13 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 		}
 		unlock_extent_cached(tree, start, end, &cached, GFP_ATOMIC);
 
-		if (whole_page) {
-			if (uptodate) {
-				SetPageUptodate(page);
-			} else {
-				ClearPageUptodate(page);
-				SetPageError(page);
-			}
-			unlock_page(page);
+		if (uptodate) {
+			SetPageUptodate(page);
 		} else {
-			if (uptodate) {
-				check_page_uptodate(tree, page);
-			} else {
-				ClearPageUptodate(page);
-				SetPageError(page);
-			}
-			check_page_locked(tree, page);
+			ClearPageUptodate(page);
+			SetPageError(page);
 		}
+		unlock_page(page);
 	} while (bvec <= bvec_end);
 
 	bio_put(bio);
-- 
cgit v1.2.3


From 3a6cad9009c85e29e83aafc8ac00b1dd5067fc5f Mon Sep 17 00:00:00 2001
From: Stefan Behrens
Date: Thu, 16 May 2013 14:48:19 +0000
Subject: Btrfs: explicitly use global_block_rsv for quota_tree

The quota_tree was set up to use the empty_block_rsv before
which would be problematic when the filesystem is filled up
and ENOSPC happens during internal operations while the quota
tree is updated and COWed (when the btrfs_qgroup_info_item
items) are written. In fact, use_block_rsv() which is used
in btrfs_cow_block() falls back to the global_block_rsv in
this case. But just in order to make it more clear what is
happening, change it to explicitly use the global_block_rsv.

Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/extent-tree.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'fs')

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 42f5e6196021..df472ab1b5ac 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -4564,6 +4564,8 @@ static void init_global_block_rsv(struct btrfs_fs_info *fs_info)
 	fs_info->csum_root->block_rsv = &fs_info->global_block_rsv;
 	fs_info->dev_root->block_rsv = &fs_info->global_block_rsv;
 	fs_info->tree_root->block_rsv = &fs_info->global_block_rsv;
+	if (fs_info->quota_root)
+		fs_info->quota_root->block_rsv = &fs_info->global_block_rsv;
 	fs_info->chunk_root->block_rsv = &fs_info->chunk_block_rsv;
 
 	update_global_block_rsv(fs_info);
-- 
cgit v1.2.3


From 655b09fe540b73edeaabfb4c2d700be51a1f8bce Mon Sep 17 00:00:00 2001
From: Josef Bacik
Date: Fri, 17 May 2013 14:06:51 -0400
Subject: Btrfs: make sure roots are assigned before freeing their nodes

If we fail to load the chunk tree we'll call free_root_pointers, except we may
not have assigned the roots for the dev_root/extent_root/csum_root yet, so we
could NULL pointer deref at this point.  Just add checks to make sure these
roots are set to keep us from panicing.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/disk-io.c | 39 +++++++++++++++++++++------------------
 1 file changed, 21 insertions(+), 18 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 1b03f8393a69..4bdb0528688a 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1987,30 +1987,33 @@ static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
 {
 	free_extent_buffer(info->tree_root->node);
 	free_extent_buffer(info->tree_root->commit_root);
-	free_extent_buffer(info->dev_root->node);
-	free_extent_buffer(info->dev_root->commit_root);
-	free_extent_buffer(info->extent_root->node);
-	free_extent_buffer(info->extent_root->commit_root);
-	free_extent_buffer(info->csum_root->node);
-	free_extent_buffer(info->csum_root->commit_root);
-	if (info->quota_root) {
-		free_extent_buffer(info->quota_root->node);
-		free_extent_buffer(info->quota_root->commit_root);
-	}
-
 	info->tree_root->node = NULL;
 	info->tree_root->commit_root = NULL;
-	info->dev_root->node = NULL;
-	info->dev_root->commit_root = NULL;
-	info->extent_root->node = NULL;
-	info->extent_root->commit_root = NULL;
-	info->csum_root->node = NULL;
-	info->csum_root->commit_root = NULL;
+
+	if (info->dev_root) {
+		free_extent_buffer(info->dev_root->node);
+		free_extent_buffer(info->dev_root->commit_root);
+		info->dev_root->node = NULL;
+		info->dev_root->commit_root = NULL;
+	}
+	if (info->extent_root) {
+		free_extent_buffer(info->extent_root->node);
+		free_extent_buffer(info->extent_root->commit_root);
+		info->extent_root->node = NULL;
+		info->extent_root->commit_root = NULL;
+	}
+	if (info->csum_root) {
+		free_extent_buffer(info->csum_root->node);
+		free_extent_buffer(info->csum_root->commit_root);
+		info->csum_root->node = NULL;
+		info->csum_root->commit_root = NULL;
+	}
 	if (info->quota_root) {
+		free_extent_buffer(info->quota_root->node);
+		free_extent_buffer(info->quota_root->commit_root);
 		info->quota_root->node = NULL;
 		info->quota_root->commit_root = NULL;
 	}
-
 	if (chunk_root) {
 		free_extent_buffer(info->chunk_root->node);
 		free_extent_buffer(info->chunk_root->commit_root);
-- 
cgit v1.2.3


From 9be3395bcd4ad4af76476ac38152b4cafa6b6159 Mon Sep 17 00:00:00 2001
From: Chris Mason
Date: Fri, 17 May 2013 18:30:14 -0400
Subject: Btrfs: use a btrfs bioset instead of abusing bio internals

Btrfs has been pointer tagging bi_private and using bi_bdev
to store the stripe index and mirror number of failed IOs.

As bios bubble back up through the call chain, we use these
to decide if and how to retry our IOs.  They are also used
to count IO failures on a per device basis.

Recently a bio tracepoint was added lead to crashes because
we were abusing bi_bdev.

This commit adds a btrfs bioset, and creates explicit fields
for the mirror number and stripe index.  The plan is to
extend this structure for all of the fields currently in
struct btrfs_bio, which will mean one less kmalloc in
our IO path.

Signed-off-by: Chris Mason <chris.mason@fusionio.com>
Reported-by: Tejun Heo <tj@kernel.org>
---
 fs/btrfs/check-integrity.c |  2 +-
 fs/btrfs/disk-io.c         |  2 +-
 fs/btrfs/extent_io.c       | 49 +++++++++++++++++++++++++++++------
 fs/btrfs/extent_io.h       |  2 ++
 fs/btrfs/inode.c           | 64 +++++++++++++++++++++++++++++++---------------
 fs/btrfs/raid56.c          |  2 +-
 fs/btrfs/scrub.c           | 10 ++++----
 fs/btrfs/volumes.c         | 41 +++++------------------------
 fs/btrfs/volumes.h         | 20 +++++++++++++++
 9 files changed, 120 insertions(+), 72 deletions(-)

(limited to 'fs')

diff --git a/fs/btrfs/check-integrity.c b/fs/btrfs/check-integrity.c
index 18af6f48781a..1431a6965017 100644
--- a/fs/btrfs/check-integrity.c
+++ b/fs/btrfs/check-integrity.c
@@ -1700,7 +1700,7 @@ static int btrfsic_read_block(struct btrfsic_state *state,
 		unsigned int j;
 		DECLARE_COMPLETION_ONSTACK(complete);
 
-		bio = bio_alloc(GFP_NOFS, num_pages - i);
+		bio = btrfs_io_bio_alloc(GFP_NOFS, num_pages - i);
 		if (!bio) {
 			printk(KERN_INFO
 			       "btrfsic: bio_alloc() for %u pages failed!\n",
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 4e9ebe1f1827..ca0ea9928210 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -3128,7 +3128,7 @@ static int write_dev_flush(struct btrfs_device *device, int wait)
 	 * caller
 	 */
 	device->flush_bio = NULL;
-	bio = bio_alloc(GFP_NOFS, 0);
+	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
 	if (!bio)
 		return -ENOMEM;
 
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index d2ac518f90e4..fe1d6c3424a5 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -23,6 +23,7 @@
 
 static struct kmem_cache *extent_state_cache;
 static struct kmem_cache *extent_buffer_cache;
+static struct bio_set *btrfs_bioset;
 
 #ifdef CONFIG_BTRFS_DEBUG
 static LIST_HEAD(buffers);
@@ -125,10 +126,20 @@ int __init extent_io_init(void)
 			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!extent_buffer_cache)
 		goto free_state_cache;
+
+	btrfs_bioset = bioset_create(BIO_POOL_SIZE,
+				     offsetof(struct btrfs_io_bio, bio));
+	if (!btrfs_bioset)
+		goto free_buffer_cache;
 	return 0;
 
+free_buffer_cache:
+	kmem_cache_destroy(extent_buffer_cache);
+	extent_buffer_cache = NULL;
+
 free_state_cache:
 	kmem_cache_destroy(extent_state_cache);
+	extent_state_cache = NULL;
 	return -ENOMEM;
 }
 
@@ -145,6 +156,8 @@ void extent_io_exit(void)
 		kmem_cache_destroy(extent_state_cache);
 	if (extent_buffer_cache)
 		kmem_cache_destroy(extent_buffer_cache);
+	if (btrfs_bioset)
+		bioset_free(btrfs_bioset);
 }
 
 void extent_io_tree_init(struct extent_io_tree *tree,
@@ -2046,7 +2059,7 @@ int repair_io_failure(struct btrfs_fs_info *fs_info, u64 start,
 	if (btrfs_is_parity_mirror(map_tree, logical, length, mirror_num))
 		return 0;
 
-	bio = bio_alloc(GFP_NOFS, 1);
+	bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
 	if (!bio)
 		return -EIO;
 	bio->bi_private = &compl;
@@ -2336,7 +2349,7 @@ static int bio_readpage_error(struct bio *failed_bio, struct page *page,
 		return -EIO;
 	}
 
-	bio = bio_alloc(GFP_NOFS, 1);
+	bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
 	if (!bio) {
 		free_io_failure(inode, failrec, 0);
 		return -EIO;
@@ -2457,10 +2470,11 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 		struct page *page = bvec->bv_page;
 		struct extent_state *cached = NULL;
 		struct extent_state *state;
+		struct btrfs_io_bio *io_bio = btrfs_io_bio(bio);
 
 		pr_debug("end_bio_extent_readpage: bi_sector=%llu, err=%d, "
-			 "mirror=%ld\n", (u64)bio->bi_sector, err,
-			 (long int)bio->bi_bdev);
+			 "mirror=%lu\n", (u64)bio->bi_sector, err,
+			 io_bio->mirror_num);
 		tree = &BTRFS_I(page->mapping->host)->io_tree;
 
 		start = page_offset(page) + bvec->bv_offset;
@@ -2485,7 +2499,7 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 		}
 		spin_unlock(&tree->lock);
 
-		mirror = (int)(unsigned long)bio->bi_bdev;
+		mirror = io_bio->mirror_num;
 		if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) {
 			ret = tree->ops->readpage_end_io_hook(page, start, end,
 							      state, mirror);
@@ -2550,17 +2564,23 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 	bio_put(bio);
 }
 
+/*
+ * this allocates from the btrfs_bioset.  We're returning a bio right now
+ * but you can call btrfs_io_bio for the appropriate container_of magic
+ */
 struct bio *
 btrfs_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
 		gfp_t gfp_flags)
 {
 	struct bio *bio;
 
-	bio = bio_alloc(gfp_flags, nr_vecs);
+	bio = bio_alloc_bioset(gfp_flags, nr_vecs, btrfs_bioset);
 
 	if (bio == NULL && (current->flags & PF_MEMALLOC)) {
-		while (!bio && (nr_vecs /= 2))
-			bio = bio_alloc(gfp_flags, nr_vecs);
+		while (!bio && (nr_vecs /= 2)) {
+			bio = bio_alloc_bioset(gfp_flags,
+					       nr_vecs, btrfs_bioset);
+		}
 	}
 
 	if (bio) {
@@ -2571,6 +2591,19 @@ btrfs_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
 	return bio;
 }
 
+struct bio *btrfs_bio_clone(struct bio *bio, gfp_t gfp_mask)
+{
+	return bio_clone_bioset(bio, gfp_mask, btrfs_bioset);
+}
+
+
+/* this also allocates from the btrfs_bioset */
+struct bio *btrfs_io_bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
+{
+	return bio_alloc_bioset(gfp_mask, nr_iovecs, btrfs_bioset);
+}
+
+
 static int __must_check submit_one_bio(int rw, struct bio *bio,
 				       int mirror_num, unsigned long bio_flags)
 {
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index a2c03a175009..41fb81e7ec53 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -336,6 +336,8 @@ int extent_clear_unlock_delalloc(struct inode *inode,
 struct bio *
 btrfs_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
 		gfp_t gfp_flags);
+struct bio *btrfs_io_bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs);
+struct bio *btrfs_bio_clone(struct bio *bio, gfp_t gfp_mask);
 
 struct btrfs_fs_info;
 
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 1669c3b4be2f..d59dddfb1f96 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -6927,7 +6927,11 @@ struct btrfs_dio_private {
 	/* IO errors */
 	int errors;
 
+	/* orig_bio is our btrfs_io_bio */
 	struct bio *orig_bio;
+
+	/* dio_bio came from fs/direct-io.c */
+	struct bio *dio_bio;
 };
 
 static void btrfs_endio_direct_read(struct bio *bio, int err)
@@ -6937,6 +6941,7 @@ static void btrfs_endio_direct_read(struct bio *bio, int err)
 	struct bio_vec *bvec = bio->bi_io_vec;
 	struct inode *inode = dip->inode;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct bio *dio_bio;
 	u64 start;
 
 	start = dip->logical_offset;
@@ -6976,14 +6981,15 @@ failed:
 
 	unlock_extent(&BTRFS_I(inode)->io_tree, dip->logical_offset,
 		      dip->logical_offset + dip->bytes - 1);
-	bio->bi_private = dip->private;
+	dio_bio = dip->dio_bio;
 
 	kfree(dip);
 
 	/* If we had a csum failure make sure to clear the uptodate flag */
 	if (err)
-		clear_bit(BIO_UPTODATE, &bio->bi_flags);
-	dio_end_io(bio, err);
+		clear_bit(BIO_UPTODATE, &dio_bio->bi_flags);
+	dio_end_io(dio_bio, err);
+	bio_put(bio);
 }
 
 static void btrfs_endio_direct_write(struct bio *bio, int err)
@@ -6994,6 +7000,7 @@ static void btrfs_endio_direct_write(struct bio *bio, int err)
 	struct btrfs_ordered_extent *ordered = NULL;
 	u64 ordered_offset = dip->logical_offset;
 	u64 ordered_bytes = dip->bytes;
+	struct bio *dio_bio;
 	int ret;
 
 	if (err)
@@ -7021,14 +7028,15 @@ out_test:
 		goto again;
 	}
 out_done:
-	bio->bi_private = dip->private;
+	dio_bio = dip->dio_bio;
 
 	kfree(dip);
 
 	/* If we had an error make sure to clear the uptodate flag */
 	if (err)
-		clear_bit(BIO_UPTODATE, &bio->bi_flags);
-	dio_end_io(bio, err);
+		clear_bit(BIO_UPTODATE, &dio_bio->bi_flags);
+	dio_end_io(dio_bio, err);
+	bio_put(bio);
 }
 
 static int __btrfs_submit_bio_start_direct_io(struct inode *inode, int rw,
@@ -7064,10 +7072,10 @@ static void btrfs_end_dio_bio(struct bio *bio, int err)
 	if (!atomic_dec_and_test(&dip->pending_bios))
 		goto out;
 
-	if (dip->errors)
+	if (dip->errors) {
 		bio_io_error(dip->orig_bio);
-	else {
-		set_bit(BIO_UPTODATE, &dip->orig_bio->bi_flags);
+	} else {
+		set_bit(BIO_UPTODATE, &dip->dio_bio->bi_flags);
 		bio_endio(dip->orig_bio, 0);
 	}
 out:
@@ -7242,25 +7250,34 @@ out_err:
 	return 0;
 }
 
-static void btrfs_submit_direct(int rw, struct bio *bio, struct inode *inode,
-				loff_t file_offset)
+static void btrfs_submit_direct(int rw, struct bio *dio_bio,
+				struct inode *inode, loff_t file_offset)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_dio_private *dip;
-	struct bio_vec *bvec = bio->bi_io_vec;
+	struct bio_vec *bvec = dio_bio->bi_io_vec;
+	struct bio *io_bio;
 	int skip_sum;
 	int write = rw & REQ_WRITE;
 	int ret = 0;
 
 	skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
 
+	io_bio = btrfs_bio_clone(dio_bio, GFP_NOFS);
+
+	if (!io_bio) {
+		ret = -ENOMEM;
+		goto free_ordered;
+	}
+
 	dip = kmalloc(sizeof(*dip), GFP_NOFS);
 	if (!dip) {
 		ret = -ENOMEM;
-		goto free_ordered;
+		goto free_io_bio;
 	}
 
-	dip->private = bio->bi_private;
+	dip->private = dio_bio->bi_private;
+	io_bio->bi_private = dio_bio->bi_private;
 	dip->inode = inode;
 	dip->logical_offset = file_offset;
 
@@ -7268,22 +7285,27 @@ static void btrfs_submit_direct(int rw, struct bio *bio, struct inode *inode,
 	do {
 		dip->bytes += bvec->bv_len;
 		bvec++;
-	} while (bvec <= (bio->bi_io_vec + bio->bi_vcnt - 1));
+	} while (bvec <= (dio_bio->bi_io_vec + dio_bio->bi_vcnt - 1));
 
-	dip->disk_bytenr = (u64)bio->bi_sector << 9;
-	bio->bi_private = dip;
+	dip->disk_bytenr = (u64)dio_bio->bi_sector << 9;
+	io_bio->bi_private = dip;
 	dip->errors = 0;
-	dip->orig_bio = bio;
+	dip->orig_bio = io_bio;
+	dip->dio_bio = dio_bio;
 	atomic_set(&dip->pending_bios, 0);
 
 	if (write)
-		bio->bi_end_io = btrfs_endio_direct_write;
+		io_bio->bi_end_io = btrfs_endio_direct_write;
 	else
-		bio->bi_end_io = btrfs_endio_direct_read;
+		io_bio->bi_end_io = btrfs_endio_direct_read;
 
 	ret = btrfs_submit_direct_hook(rw, dip, skip_sum);
 	if (!ret)
 		return;
+
+free_io_bio:
+	bio_put(io_bio);
+
 free_ordered:
 	/*
 	 * If this is a write, we need to clean up the reserved space and kill
@@ -7299,7 +7321,7 @@ free_ordered:
 		btrfs_put_ordered_extent(ordered);
 		btrfs_put_ordered_extent(ordered);
 	}
-	bio_endio(bio, ret);
+	bio_endio(dio_bio, ret);
 }
 
 static ssize_t check_direct_IO(struct btrfs_root *root, int rw, struct kiocb *iocb,
diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c
index 0740621daf6c..0525e1389f5b 100644
--- a/fs/btrfs/raid56.c
+++ b/fs/btrfs/raid56.c
@@ -1050,7 +1050,7 @@ static int rbio_add_io_page(struct btrfs_raid_bio *rbio,
 	}
 
 	/* put a new bio on the list */
-	bio = bio_alloc(GFP_NOFS, bio_max_len >> PAGE_SHIFT?:1);
+	bio = btrfs_io_bio_alloc(GFP_NOFS, bio_max_len >> PAGE_SHIFT?:1);
 	if (!bio)
 		return -ENOMEM;
 
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index f489e24659a4..79bd479317cb 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -1296,7 +1296,7 @@ static void scrub_recheck_block(struct btrfs_fs_info *fs_info,
 		}
 
 		WARN_ON(!page->page);
-		bio = bio_alloc(GFP_NOFS, 1);
+		bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
 		if (!bio) {
 			page->io_error = 1;
 			sblock->no_io_error_seen = 0;
@@ -1431,7 +1431,7 @@ static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
 			return -EIO;
 		}
 
-		bio = bio_alloc(GFP_NOFS, 1);
+		bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
 		if (!bio)
 			return -EIO;
 		bio->bi_bdev = page_bad->dev->bdev;
@@ -1522,7 +1522,7 @@ again:
 		sbio->dev = wr_ctx->tgtdev;
 		bio = sbio->bio;
 		if (!bio) {
-			bio = bio_alloc(GFP_NOFS, wr_ctx->pages_per_wr_bio);
+			bio = btrfs_io_bio_alloc(GFP_NOFS, wr_ctx->pages_per_wr_bio);
 			if (!bio) {
 				mutex_unlock(&wr_ctx->wr_lock);
 				return -ENOMEM;
@@ -1930,7 +1930,7 @@ again:
 		sbio->dev = spage->dev;
 		bio = sbio->bio;
 		if (!bio) {
-			bio = bio_alloc(GFP_NOFS, sctx->pages_per_rd_bio);
+			bio = btrfs_io_bio_alloc(GFP_NOFS, sctx->pages_per_rd_bio);
 			if (!bio)
 				return -ENOMEM;
 			sbio->bio = bio;
@@ -3307,7 +3307,7 @@ static int write_page_nocow(struct scrub_ctx *sctx,
 			"btrfs: scrub write_page_nocow(bdev == NULL) is unexpected!\n");
 		return -EIO;
 	}
-	bio = bio_alloc(GFP_NOFS, 1);
+	bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
 	if (!bio) {
 		spin_lock(&sctx->stat_lock);
 		sctx->stat.malloc_errors++;
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index a191bac31d85..317afc7a2af4 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -5019,42 +5019,16 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
 	return 0;
 }
 
-static void *merge_stripe_index_into_bio_private(void *bi_private,
-						 unsigned int stripe_index)
-{
-	/*
-	 * with single, dup, RAID0, RAID1 and RAID10, stripe_index is
-	 * at most 1.
-	 * The alternative solution (instead of stealing bits from the
-	 * pointer) would be to allocate an intermediate structure
-	 * that contains the old private pointer plus the stripe_index.
-	 */
-	BUG_ON((((uintptr_t)bi_private) & 3) != 0);
-	BUG_ON(stripe_index > 3);
-	return (void *)(((uintptr_t)bi_private) | stripe_index);
-}
-
-static struct btrfs_bio *extract_bbio_from_bio_private(void *bi_private)
-{
-	return (struct btrfs_bio *)(((uintptr_t)bi_private) & ~((uintptr_t)3));
-}
-
-static unsigned int extract_stripe_index_from_bio_private(void *bi_private)
-{
-	return (unsigned int)((uintptr_t)bi_private) & 3;
-}
-
 static void btrfs_end_bio(struct bio *bio, int err)
 {
-	struct btrfs_bio *bbio = extract_bbio_from_bio_private(bio->bi_private);
+	struct btrfs_bio *bbio = bio->bi_private;
 	int is_orig_bio = 0;
 
 	if (err) {
 		atomic_inc(&bbio->error);
 		if (err == -EIO || err == -EREMOTEIO) {
 			unsigned int stripe_index =
-				extract_stripe_index_from_bio_private(
-					bio->bi_private);
+				btrfs_io_bio(bio)->stripe_index;
 			struct btrfs_device *dev;
 
 			BUG_ON(stripe_index >= bbio->num_stripes);
@@ -5084,8 +5058,7 @@ static void btrfs_end_bio(struct bio *bio, int err)
 		}
 		bio->bi_private = bbio->private;
 		bio->bi_end_io = bbio->end_io;
-		bio->bi_bdev = (struct block_device *)
-					(unsigned long)bbio->mirror_num;
+		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
 		/* only send an error to the higher layers if it is
 		 * beyond the tolerance of the btrfs bio
 		 */
@@ -5211,8 +5184,7 @@ static void submit_stripe_bio(struct btrfs_root *root, struct btrfs_bio *bbio,
 	struct btrfs_device *dev = bbio->stripes[dev_nr].dev;
 
 	bio->bi_private = bbio;
-	bio->bi_private = merge_stripe_index_into_bio_private(
-			bio->bi_private, (unsigned int)dev_nr);
+	btrfs_io_bio(bio)->stripe_index = dev_nr;
 	bio->bi_end_io = btrfs_end_bio;
 	bio->bi_sector = physical >> 9;
 #ifdef DEBUG
@@ -5273,8 +5245,7 @@ static void bbio_error(struct btrfs_bio *bbio, struct bio *bio, u64 logical)
 	if (atomic_dec_and_test(&bbio->stripes_pending)) {
 		bio->bi_private = bbio->private;
 		bio->bi_end_io = bbio->end_io;
-		bio->bi_bdev = (struct block_device *)
-			(unsigned long)bbio->mirror_num;
+		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
 		bio->bi_sector = logical >> 9;
 		kfree(bbio);
 		bio_endio(bio, -EIO);
@@ -5352,7 +5323,7 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
 		}
 
 		if (dev_nr < total_devs - 1) {
-			bio = bio_clone(first_bio, GFP_NOFS);
+			bio = btrfs_bio_clone(first_bio, GFP_NOFS);
 			BUG_ON(!bio); /* -ENOMEM */
 		} else {
 			bio = first_bio;
diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
index 845ccbb0d2e3..f6247e2a47f7 100644
--- a/fs/btrfs/volumes.h
+++ b/fs/btrfs/volumes.h
@@ -152,6 +152,26 @@ struct btrfs_fs_devices {
 	int rotating;
 };
 
+/*
+ * we need the mirror number and stripe index to be passed around
+ * the call chain while we are processing end_io (especially errors).
+ * Really, what we need is a btrfs_bio structure that has this info
+ * and is properly sized with its stripe array, but we're not there
+ * quite yet.  We have our own btrfs bioset, and all of the bios
+ * we allocate are actually btrfs_io_bios.  We'll cram as much of
+ * struct btrfs_bio as we can into this over time.
+ */
+struct btrfs_io_bio {
+	unsigned long mirror_num;
+	unsigned long stripe_index;
+	struct bio bio;
+};
+
+static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio)
+{
+	return container_of(bio, struct btrfs_io_bio, bio);
+}
+
 struct btrfs_bio_stripe {
 	struct btrfs_device *dev;
 	u64 physical;
-- 
cgit v1.2.3