linux/fs/btrfs
Miao Xie 361048f586 Btrfs: fix full backref problem when inserting shared block reference
If we create several snapshots at the same time, the following BUG_ON() will be
triggered.

	kernel BUG at fs/btrfs/extent-tree.c:6047!

Steps to reproduce:
 # mkfs.btrfs <partition>
 # mount <partition> <mnt>
 # cd <mnt>
 # for ((i=0;i<2400;i++)); do touch long_name_to_make_tree_more_deep$i; done
 # for ((i=0; i<4; i++))
 > do
 > mkdir $i
 > for ((j=0; j<200; j++))
 > do
 > btrfs sub snap . $i/$j
 > done &
 > done

The reason is:
Before transaction commit, some operations changed the fs tree and new tree
blocks were allocated because of COW. We used the implicit non-shared back
reference for those newly allocated tree blocks because they were not shared by
two or more trees.

And then we created the first snapshot for the fs tree, according to the back
reference rules, we also used implicit back refs for the child tree blocks of
the root node of the fs tree, now those child nodes/leaves were shared by two
trees.

Then We didn't deal with the delayed references, and continued to change the fs
tree(created the second snapshot and inserted the dir item of the new snapshot
into the fs tree). According to the rules of the back reference, we added full
back refs for those tree blocks whose parents have be shared by two trees.
Now some newly allocated tree blocks had two types of the references.

As we know, the delayed reference system handles these delayed references from
back to front, and the full delayed reference is inserted after the implicit
ones. So when we dealt with the back references of those newly allocated tree
blocks, the full references was dealt with at first. And if the first reference
is a shared back reference and the tree block that the reference points to is
newly allocated, It would be considered as a tree block which is shared by two
or more trees when it is allocated and should be a full back reference not a
implicit one, the flag of its reference also should be set to FULL_BACKREF.
But in fact, it was a non-shared tree block with a implicit reference at
beginning, so it was not compulsory to set the flags to FULL_BACKREF. So BUG_ON
was triggered.

We have several methods to fix this bug:
1. deal with delayed references after the snapshot is created and before we
   change the source tree of the snapshot. This is the easiest and safest way.
2. modify the sort method of the delayed reference tree, make the full delayed
   references be inserted before the implicit ones. It is also very easy, but
   I don't know if it will introduce some problems or not.
3. modify select_delayed_ref() and make it select the implicit delayed reference
   at first. This way is not so good because it may wastes CPU time if we have
   lots of delayed references.
4. set the flags to FULL_BACKREF, this method is a little complex comparing with
   the 1st way.

I chose the 1st way to fix it.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2012-10-01 15:19:10 -04:00
..
acl.c Btrfs: cache no acl on new inodes 2012-05-30 10:23:27 -04:00
async-thread.c Btrfs: call the ordered free operation without any locks held 2012-07-25 16:15:07 -04:00
async-thread.h btrfs: return void in functions without error conditions 2012-03-22 01:45:34 +01:00
backref.c Btrfs: fix gcc warnings for 32bit compiles 2012-10-01 15:19:01 -04:00
backref.h Merge branch 'send-v2' of git://github.com/ablock84/linux-btrfs into for-linus 2012-07-25 19:19:10 -04:00
btrfs_inode.h Btrfs: fix a bug in checking whether a inode is already in log 2012-10-01 15:19:06 -04:00
check-integrity.c Btrfs: introduce subvol uuids and times 2012-07-25 23:28:38 +02:00
check-integrity.h Btrfs: add optional integrity check code 2011-12-21 19:14:09 +01:00
compat.h Btrfs: drop remaining LINUX_KERNEL_VERSION checks and compat code 2009-01-06 09:38:55 -05:00
compression.c Btrfs: barrier before waitqueue_active 2012-08-28 16:53:33 -04:00
compression.h btrfs: return void in functions without error conditions 2012-03-22 01:45:34 +01:00
ctree.c Btrfs: fix btrfs send for inline items and compression 2012-10-01 15:19:00 -04:00
ctree.h Btrfs: btrfs_drop_extent_cache should never fail 2012-10-01 15:19:09 -04:00
delayed-inode.c Btrfs: barrier before waitqueue_active 2012-08-28 16:53:33 -04:00
delayed-inode.h Btrfs: flush delayed inodes if we're short on space 2012-07-23 15:41:40 -04:00
delayed-ref.c Btrfs: allow delayed refs to be merged 2012-08-28 16:53:38 -04:00
delayed-ref.h Btrfs: allow delayed refs to be merged 2012-08-28 16:53:38 -04:00
dir-item.c btrfs: replace many BUG_ONs with proper error handling 2012-03-22 11:52:54 +01:00
disk-io.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs 2012-08-29 11:36:22 -07:00
disk-io.h Btrfs: remove superblock writing after fatal error 2012-08-28 16:53:38 -04:00
export.c ->encode_fh() API change 2012-05-29 23:28:33 -04:00
export.h
extent_io.c Btrfs: fix btrfs send for inline items and compression 2012-10-01 15:19:00 -04:00
extent_io.h Merge branch 'for-chris' of git://git.jan-o-sch.net/btrfs-unstable into for-linus 2012-05-31 16:49:53 -04:00
extent_map.c Btrfs: improve fsync by filtering extents that we want 2012-10-01 15:19:05 -04:00
extent_map.h Btrfs: improve fsync by filtering extents that we want 2012-10-01 15:19:05 -04:00
extent-tree.c Btrfs: add hole punching 2012-10-01 15:19:07 -04:00
file-item.c Btrfs: fix gcc warnings for 32bit compiles 2012-10-01 15:19:01 -04:00
file.c Btrfs: btrfs_drop_extent_cache should never fail 2012-10-01 15:19:09 -04:00
free-space-cache.c Btrfs: do not count in readonly bytes 2012-07-23 16:28:03 -04:00
free-space-cache.h btrfs: remove all unused functions 2011-05-06 12:34:03 +02:00
hash.h Btrfs: remove crc32c.h and use libcrc32c directly. 2009-06-10 11:29:53 -04:00
inode-item.c Merge branch 'error-handling' into for-linus 2012-03-28 20:31:37 -04:00
inode-map.c Btrfs: show useful info in space reservation tracepoint 2012-03-29 09:57:44 -04:00
inode-map.h Btrfs: Support reading/writing on disk free ino cache 2011-04-25 16:46:11 +08:00
inode.c Btrfs: do not take cleanup_work_sem in btrfs_run_delayed_iputs() 2012-10-01 15:19:08 -04:00
ioctl.c Btrfs: remove unused hint byte argument for btrfs_drop_extents 2012-10-01 15:19:06 -04:00
ioctl.h Merge branch 'send-v2' of git://github.com/ablock84/linux-btrfs into for-linus 2012-07-25 19:19:10 -04:00
Kconfig Btrfs: add config option to enable btrfs integrity check 2011-12-21 19:14:16 +01:00
locking.c Btrfs: fix a misplaced address operator in a condition 2012-08-28 16:53:23 -04:00
locking.h btrfs: return void in functions without error conditions 2012-03-22 01:45:34 +01:00
lzo.c btrfs: remove the second argument of k[un]map_atomic() 2012-03-20 21:48:21 +08:00
Makefile Merge branch 'send-v2' of git://github.com/ablock84/linux-btrfs into for-linus 2012-07-25 19:19:10 -04:00
ordered-data.c btrfs: nuke pdflush from comments 2012-08-04 12:15:35 +04:00
ordered-data.h Btrfs: finish ordered extents in their own thread 2012-05-30 10:23:33 -04:00
orphan.c btrfs: replace many BUG_ONs with proper error handling 2012-03-22 11:52:54 +01:00
print-tree.c Btrfs: read device stats on mount, write modified ones during commit 2012-05-30 10:23:41 -04:00
print-tree.h
qgroup.c Btrfs: fix gcc warnings for 32bit compiles 2012-10-01 15:19:01 -04:00
rcu-string.h Btrfs: use rcu to protect device->name 2012-06-14 21:29:16 -04:00
reada.c Btrfs: set ioprio of scrub readahead to idle 2012-05-30 10:23:43 -04:00
relocation.c btrfs: use printk_get_level and printk_skip_level, add __printf, fix fallout 2012-07-30 17:25:14 -07:00
root-tree.c Btrfs: fix some endian bugs handling the root times 2012-08-28 16:53:26 -04:00
scrub.c Btrfs: fix possible memory leak in scrub_setup_recheck_block() 2012-10-01 15:19:09 -04:00
send.c Btrfs: fix gcc warnings for 32bit compiles 2012-10-01 15:19:01 -04:00
send.h Btrfs: introduce BTRFS_IOC_SEND for btrfs send/receive 2012-07-25 23:30:19 +02:00
struct-funcs.c Btrfs: rewrite BTRFS_SETGET_FUNCS 2012-07-23 16:28:06 -04:00
super.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs 2012-08-29 11:36:22 -07:00
sysfs.c btrfs: Remove unused sysfs code 2011-06-17 14:54:18 -04:00
transaction.c Btrfs: fix full backref problem when inserting shared block reference 2012-10-01 15:19:10 -04:00
transaction.h Btrfs: fix a bug in checking whether a inode is already in log 2012-10-01 15:19:06 -04:00
tree-defrag.c btrfs: drop unused parameter from btrfs_release_path 2011-05-02 13:57:22 +02:00
tree-log.c Btrfs: add hole punching 2012-10-01 15:19:07 -04:00
tree-log.h btrfs: return void in functions without error conditions 2012-03-22 01:45:34 +01:00
ulist.c Btrfs: make aux field of ulist 64 bit 2012-10-01 15:18:53 -04:00
ulist.h Btrfs: make aux field of ulist 64 bit 2012-10-01 15:18:53 -04:00
version.h Update Btrfs files for in-kernel usage 2008-09-25 15:41:59 -04:00
volumes.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs 2012-08-29 11:36:22 -07:00
volumes.h Btrfs: revert checksum error statistic which can cause a BUG() 2012-08-28 16:53:39 -04:00
xattr.c Btrfs: use i_version instead of our own sequence 2012-05-30 10:23:27 -04:00
xattr.h fs/vfs/security: pass last path component to LSM on inode creation 2011-02-01 11:12:29 -05:00
zlib.c btrfs: remove the second argument of k[un]map_atomic() 2012-03-20 21:48:21 +08:00