linux/fs/btrfs
Filipe Manana 00d80e342c Btrfs: fix quick exhaustion of the system array in the superblock
Omar reported that after commit 4fbcdf6694 ("Btrfs: fix -ENOSPC when
finishing block group creation"), introduced in 4.2-rc1, the following
test was failing due to exhaustion of the system array in the superblock:

  #!/bin/bash

  truncate -s 100T big.img
  mkfs.btrfs big.img
  mount -o loop big.img /mnt/loop

  num=5
  sz=10T
  for ((i = 0; i < $num; i++)); do
      echo fallocate $i $sz
      fallocate -l $sz /mnt/loop/testfile$i
  done
  btrfs filesystem sync /mnt/loop

  for ((i = 0; i < $num; i++)); do
        echo rm $i
        rm /mnt/loop/testfile$i
        btrfs filesystem sync /mnt/loop
  done
  umount /mnt/loop

This made btrfs_add_system_chunk() fail with -EFBIG due to excessive
allocation of system block groups. This happened because the test creates
a large number of data block groups per transaction and when committing
the transaction we start the writeout of the block group caches for all
the new new (dirty) block groups, which results in pre-allocating space
for each block group's free space cache using the same transaction handle.
That in turn often leads to creation of more block groups, and all get
attached to the new_bgs list of the same transaction handle to the point
of getting a list with over 1500 elements, and creation of new block groups
leads to the need of reserving space in the chunk block reserve and often
creating a new system block group too.

So that made us quickly exhaust the chunk block reserve/system space info,
because as of the commit mentioned before, we do reserve space for each
new block group in the chunk block reserve, unlike before where we would
not and would at most allocate one new system block group and therefore
would only ensure that there was enough space in the system space info to
allocate 1 new block group even if we ended up allocating thousands of
new block groups using the same transaction handle. That worked most of
the time because the computed required space at check_system_chunk() is
very pessimistic (assumes a chunk tree height of BTRFS_MAX_LEVEL/8 and
that all nodes/leafs in a path will be COWed and split) and since the
updates to the chunk tree all happen at btrfs_create_pending_block_groups
it is unlikely that a path needs to be COWed more than once (unless
writepages() for the btree inode is called by mm in between) and that
compensated for the need of creating any new nodes/leads in the chunk
tree.

So fix this by ensuring we don't accumulate a too large list of new block
groups in a transaction's handles new_bgs list, inserting/updating the
chunk tree for all accumulated new block groups and releasing the unused
space from the chunk block reserve whenever the list becomes sufficiently
large. This is a generic solution even though the problem currently can
only happen when starting the writeout of the free space caches for all
dirty block groups (btrfs_start_dirty_block_groups()).

Reported-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Tested-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
2015-07-22 18:20:54 -07:00
..
tests btrfs: qgroup: Switch self test to extent-oriented qgroup mechanism. 2015-06-10 09:26:05 -07:00
acl.c btrfs: remove useless ACL check 2014-06-09 17:20:42 -07:00
async-thread.c btrfs: Fix lockdep warning of wr_ctx->wr_lock in scrub_free_wr_ctx() 2015-06-10 07:04:52 -07:00
async-thread.h btrfs: Fix lockdep warning of wr_ctx->wr_lock in scrub_free_wr_ctx() 2015-06-10 07:04:52 -07:00
backref.c btrfs: backref: Add special time_seq == (u64)-1 case for 2015-06-10 09:25:45 -07:00
backref.h btrfs: cleanup, remove inode_item_info helper 2015-01-14 19:23:47 +01:00
btrfs_inode.h Btrfs: fix warning of bytes_may_use 2015-07-01 17:17:21 -07:00
check-integrity.c Merge branch 'cleanups-post-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux into for-linus-4.1 2015-03-25 10:52:48 -07:00
check-integrity.h block: submit_bio_wait() conversions 2013-11-24 16:33:41 -07:00
compression.c Merge branch 'cleanups-post-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux into for-linus-4.1 2015-03-25 10:52:48 -07:00
compression.h btrfs: constify structs with op functions or static definitions 2015-02-16 18:48:44 +01:00
ctree.c Btrfs: fix up read_tree_block to return proper error 2015-06-03 04:03:08 -07:00
ctree.h Btrfs: fix race between balance and unused block group deletion 2015-06-30 14:36:46 -07:00
delayed-inode.c Btrfs: fill ->last_trans for delayed inode in btrfs_fill_inode. 2015-04-26 06:27:03 -07:00
delayed-inode.h Btrfs: introduce the delayed inode ref deletion for the single link inode 2014-01-28 13:20:09 -08:00
delayed-ref.c btrfs: delayed-ref: double free in btrfs_add_delayed_tree_ref() 2015-06-24 12:28:03 -07:00
delayed-ref.h btrfs: qgroup: Add the ability to skip given qgroup for old/new_roots. 2015-06-10 09:26:23 -07:00
dev-replace.c btrfs: its btrfs_err() instead of btrfs_error() 2015-07-22 18:20:53 -07:00
dev-replace.h
dir-item.c Btrfs: make xattr replace operations atomic 2014-11-20 17:20:07 -08:00
disk-io.c btrfs: Avoid NULL pointer dereference of free_extent_buffer when read_tree_block() fail 2015-07-22 18:20:52 -07:00
disk-io.h Btrfs: disk-io: replace root args iff only fs_info used 2015-02-16 18:48:43 +01:00
export.c VFS: normal filesystems (and lustre): d_inode() annotations 2015-04-15 15:06:57 -04:00
export.h
extent_io.c Btrfs: set UNWRITTEN for prealloc'ed extents in fiemap 2015-06-03 04:03:03 -07:00
extent_io.h btrfs: constify structs with op functions or static definitions 2015-02-16 18:48:44 +01:00
extent_map.c Btrfs: do not move em to modified list when unpinning 2014-11-21 11:59:54 -08:00
extent_map.h Btrfs: fix NULL pointer crash when running balance and scrub concurrently 2014-06-19 14:20:55 -07:00
extent-tree.c Btrfs: fix quick exhaustion of the system array in the superblock 2015-07-22 18:20:54 -07:00
extent-tree.h btrfs: qgroup: Add new qgroup calculation function 2015-06-10 09:25:49 -07:00
file-item.c Merge branch 'cleanups-post-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux into for-linus-4.1 2015-03-25 10:52:48 -07:00
file.c Btrfs: avoid syncing log in the fast fsync path when not necessary 2015-06-10 07:02:43 -07:00
free-space-cache.c Btrfs: fix mutex unlock without prior lock on space cache truncation 2015-06-02 19:34:34 -07:00
free-space-cache.h Btrfs: allow block group cache writeout outside critical section in commit 2015-04-10 14:07:22 -07:00
hash.c btrfs: LLVMLinux: Remove VLAIS 2014-10-14 10:51:22 +02:00
hash.h Btrfs: fix btrfs boot when compiled as built-in 2014-01-28 13:20:31 -08:00
inode-item.c Btrfs: fix fsync log replay for inodes with a mix of regular refs and extrefs 2015-01-21 18:02:05 -08:00
inode-map.c Btrfs: fix race between caching kthread and returning inode to inode cache 2015-06-30 14:36:46 -07:00
inode-map.h
inode.c Btrfs: fix shrinking truncate when the no_holes feature is enabled 2015-07-11 22:33:14 +01:00
ioctl.c Btrfs: fix file corruption after cloning inline extents 2015-07-14 16:09:39 +01:00
Kconfig rcu: Make SRCU optional by using CONFIG_SRCU 2015-01-06 11:04:29 -08:00
locking.c btrfs: fix lockups from btrfs_clear_path_blocking 2014-11-19 10:34:35 -08:00
locking.h btrfs: fix lockups from btrfs_clear_path_blocking 2014-11-19 10:34:35 -08:00
lzo.c btrfs: constify structs with op functions or static definitions 2015-02-16 18:48:44 +01:00
Makefile Btrfs: add sanity tests for new qgroup accounting code 2014-06-09 17:20:49 -07:00
math.h btrfs: cleanup 64bit/32bit divs, compile time constants 2015-03-03 17:23:57 +01:00
ordered-data.c Btrfs: fix memory corruption on failure to submit bio for direct IO 2015-07-01 17:17:18 -07:00
ordered-data.h Btrfs: avoid syncing log in the fast fsync path when not necessary 2015-06-10 07:02:43 -07:00
orphan.c btrfs: kill the key type accessor helpers 2014-09-17 13:37:12 -07:00
print-tree.c btrfs: remove parameter blocksize from read_tree_block 2014-10-02 17:14:50 +02:00
print-tree.h
props.c btrfs: constify structs with op functions or static definitions 2015-02-16 18:48:44 +01:00
props.h Btrfs: add support for inode properties 2014-01-28 13:20:24 -08:00
qgroup.c btrfs: qgroup: allow user to clear the limitation on qgroup 2015-06-30 13:20:00 -07:00
qgroup.h btrfs: qgroup: Cleanup the old ref_node-oriented mechanism. 2015-06-10 09:26:11 -07:00
raid56.c Merge branch 'cleanups-post-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux into for-linus-4.1 2015-03-25 10:52:48 -07:00
raid56.h Btrfs: Make raid_map array be inlined in btrfs_bio structure 2015-01-21 18:06:47 -08:00
rcu-string.h
reada.c Btrfs: add ref_count and free function for btrfs_bio 2015-01-21 18:06:48 -08:00
relocation.c Btrfs: fix wrong check for btrfs_force_chunk_alloc() 2015-07-01 17:17:22 -07:00
root-tree.c Btrfs: use bitfield instead of integer data type for the some variants in btrfs_root 2014-06-09 17:20:40 -07:00
scrub.c btrfs: add error handling for scrub_workers_get() 2015-06-30 13:20:03 -07:00
send.c Btrfs: use received_uuid of parent during send 2015-06-12 13:20:38 -07:00
send.h
struct-funcs.c
super.c Btrfs: show subvol= and subvolid= in /proc/mounts 2015-06-03 04:03:02 -07:00
sysfs.c Btrfs: Check if kobject is initialized before put 2015-06-22 14:43:31 +02:00
sysfs.h Btrfs: sysfs: btrfs_sysfs_remove_fsid() make it non static 2015-05-27 12:27:22 +02:00
transaction.c btrfs: Fix lockdep warning of btrfs_run_delayed_iputs() 2015-07-22 18:20:51 -07:00
transaction.h btrfs: qgroup: Add the ability to skip given qgroup for old/new_roots. 2015-06-10 09:26:23 -07:00
tree-defrag.c btrfs: let tree defrag work in SSD mode 2015-06-02 19:34:33 -07:00
tree-log.c Btrfs: fix fsync after truncate when no_holes feature is enabled 2015-07-01 17:17:12 -07:00
tree-log.h Btrfs: fix metadata inconsistencies after directory fsync 2015-03-26 17:56:23 -07:00
ulist.c btrfs: ulist: Add ulist_del() function. 2015-06-10 09:26:17 -07:00
ulist.h btrfs: ulist: Add ulist_del() function. 2015-06-10 09:26:17 -07:00
uuid-tree.c Btrfs: make btrfs_search_forward return with nodes unlocked 2014-09-17 13:38:02 -07:00
volumes.c Btrfs: fix race between balance and unused block group deletion 2015-06-30 14:36:46 -07:00
volumes.h Btrfs: sysfs: add pointer to access fs_info from fs_devices 2015-05-27 12:27:21 +02:00
xattr.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2015-04-26 17:22:07 -07:00
xattr.h btrfs: use generic posix ACL infrastructure 2014-01-25 23:58:18 -05:00
zlib.c btrfs: constify structs with op functions or static definitions 2015-02-16 18:48:44 +01:00