Wang Shilong 901ed070df ext4: reduce lock contention in __ext4_new_inode ... While running number of creating file threads concurrently, we found heavy lock contention on group spinlock: FUNC TOTAL_TIME(us) COUNT AVG(us) ext4_create 1707443399 1440000 1185.72 _raw_spin_lock 1317641501 180899929 7.28 jbd2__journal_start 287821030 1453950 197.96 jbd2_journal_get_write_access 33441470 73077185 0.46 ext4_add_nondir 29435963 1440000 20.44 ext4_add_entry 26015166 1440049 18.07 ext4_dx_add_entry 25729337 1432814 17.96 ext4_mark_inode_dirty 12302433 5774407 2.13 most of cpu time blames to _raw_spin_lock, here is some testing numbers with/without patch. Test environment: Server : SuperMicro Sever (2 x E5-2690 v3@2.60GHz, 128GB 2133MHz DDR4 Memory, 8GbFC) Storage : 2 x RAID1 (DDN SFA7700X, 4 x Toshiba PX02SMU020 200GB Read Intensive SSD) format command: mkfs.ext4 -J size=4096 test command: mpirun -np 48 mdtest -n 30000 -d /ext4/mdtest.out -F -C \ -r -i 1 -v -p 10 -u #first run to load inode mpirun -np 48 mdtest -n 30000 -d /ext4/mdtest.out -F -C \ -r -i 3 -v -p 10 -u Kernel version: 4.13.0-rc3 Test 1,440,000 files with 48 directories by 48 processes: Without patch: File Creation File removal 79,033 289,569 ops/per second 81,463 285,359 79,875 288,475 With patch: File Creation File removal 810669 301694 812805 302711 813965 297670 Creation performance is improved more than 10X with large journal size. The main problem here is we test bitmap and do some check and journal operations which could be slept, then we test and set with lock hold, this could be racy, and make 'inode' steal by other process. However, after first try, we could confirm handle has been started and inode bitmap journaled too, then we could find and set bit with lock hold directly, this will mostly gurateee success with second try. Tested-by: Shuichi Ihara <sihara@ddn.com> Signed-off-by: Wang Shilong <wshilong@ddn.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz>		2017-08-24 12:56:35 -04:00
..
9p	9p: Implement show_options	2017-07-11 06:08:58 -04:00
adfs
affs	affs: Implement show_options	2017-07-11 06:06:17 -04:00
afs	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
autofs4	Fix up over-eager 'wait_queue_t' renaming	2017-07-10 11:40:19 -07:00
befs	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
bfs	bfs: fix sanity checks for empty files	2017-07-12 16:26:00 -07:00
btrfs	Merge branch 'for-4.13-part3' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux	2017-07-28 12:26:59 -07:00
cachefiles	sched/wait: Disambiguate wq_entry->task_list and wq_head->task_list naming	2017-06-20 12:19:14 +02:00
ceph	ceph: fix race in concurrent readdir	2017-07-17 14:54:59 +02:00
cifs	Add wait_for_random_bytes() and get_random_*_wait() functions so that	2017-07-15 12:44:02 -07:00
coda	fs: implement vfs_iter_write using do_iter_write	2017-06-29 17:49:23 -04:00
configfs	configfs: Introduce config_item_get_unless_zero()	2017-06-12 13:20:20 +02:00
cramfs
crypto	The first major feature for ext4 this merge window is the largedir	2017-07-09 09:31:22 -07:00
debugfs	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
devpts
dlm
ecryptfs	ecryptfs: Convert to separately allocated bdi	2017-04-20 12:09:55 -06:00
efivarfs	VFS: Kill off s_options and helpers	2017-07-11 06:09:21 -04:00
efs
exofs	mm: drop "wait" parameter from write_one_page()	2017-07-05 18:44:22 -04:00
exportfs
ext2	ext2: preserve i_mode if ext2_set_acl() fails	2017-07-18 11:23:56 +02:00
ext4	ext4: reduce lock contention in __ext4_new_inode	2017-08-24 12:56:35 -04:00
f2fs	f2fs: avoid cpu lockup	2017-07-17 19:23:18 -07:00
fat
freevxfs
fscache
fuse	Merge branch 'for-linus' of git://git.kernel.dk/linux-block	2017-05-20 16:12:30 -07:00
gfs2	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
hfs	fs: semove set but not checked AOP_FLAG_UNINTERRUPTIBLE flag	2017-05-08 17:15:14 -07:00
hfsplus	hfsplus: Don't clear SGID when inheriting ACLs	2017-07-18 18:23:39 +02:00
hostfs
hpfs
hugetlbfs	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
isofs	isofs: Fix off-by-one in 'session' mount option parsing	2017-07-18 12:33:16 +02:00
jbd2	Writeback error handling fixes (pile #2)	2017-07-07 19:38:17 -07:00
jffs2	jffs2: fix spelling mistake: "requestied" -> "requested"	2017-04-19 11:35:55 -07:00
jfs	JFS fixes for 4.13	2017-07-25 08:51:57 -07:00
kernfs
lockd	sunrpc: mark all struct svc_version instances as const	2017-07-13 15:58:03 -04:00
minix	Merge branch 'work.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-08 10:50:54 -07:00
ncpfs	mm: per-cgroup memory reclaim stats	2017-07-06 16:24:35 -07:00
nfs	Two more NFS client bugfixes for 4.13	2017-08-02 20:56:44 -07:00
nfs_common
nfsd	nfsd: Fix a memory scribble in the callback channel	2017-07-17 13:15:06 -04:00
nilfs2	Merge branch 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip	2017-07-03 13:08:04 -07:00
nls
notify	dentry name snapshots	2017-07-07 20:09:10 -04:00
ntfs	ntfs: Use ERR_CAST() to avoid cross-structure cast	2017-05-28 10:11:48 -07:00
ocfs2	ocfs2: don't clear SGID when inheriting ACLs	2017-08-02 17:16:13 -07:00
omfs	omfs: Implement show_options	2017-07-06 03:31:46 -04:00
openpromfs
orangefs	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
overlayfs	ovl: check for bad and whiteout index on lookup	2017-07-20 11:08:21 +02:00
proc	Now that IPC and other changes have landed, enable manual markings for	2017-07-19 08:55:18 -07:00
pstore	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
qnx4
qnx6
quota	quota: add get_inode_usage callback to transfer multi-inode charges	2017-06-22 11:46:48 -04:00
ramfs	ramfs: Implement show_options	2017-07-06 03:31:46 -04:00
reiserfs	reiserfs: preserve i_mode if __reiserfs_set_acl() fails	2017-07-18 11:24:08 +02:00
romfs
squashfs
sysfs	sysfs: be careful of error returns from ops->show()	2017-04-08 17:33:32 +02:00
sysv	mm: drop "wait" parameter from write_one_page()	2017-07-05 18:44:22 -04:00
tracefs	VFS: Don't use save/replace_mount_options if not using generic_show_options	2017-07-06 03:31:46 -04:00
ubifs	ubifs: Set double hash cookie also for RENAME_EXCHANGE	2017-07-14 22:50:57 +02:00
udf	udf: Convert udf_disk_stamp_to_time() to use mktime64()	2017-06-14 11:21:02 +02:00
ufs	Writeback error handling fixes (pile #1)	2017-07-07 18:39:15 -07:00
xfs	xfs: fix multi-AG deadlock in xfs_bunmapi	2017-07-26 08:20:03 -07:00
aio.c	fs: add O_DIRECT and aio support for sending down write life time hints	2017-06-27 12:05:36 -06:00
anon_inodes.c
attr.c
bad_inode.c
binfmt_aout.c
binfmt_elf.c	binfmt_elf: safely increment argv pointers	2017-07-10 16:32:36 -07:00
binfmt_elf_fdpic.c
binfmt_em86.c
binfmt_flat.c	binfmt_flat: Use %u to format u32	2017-07-16 09:24:05 -07:00
binfmt_misc.c	fs: constify tree_descr arrays passed to simple_fill_super()	2017-04-26 23:54:06 -04:00
binfmt_script.c
block_dev.c	Writeback error handling fixes (pile #2)	2017-07-07 19:38:17 -07:00
buffer.c	fs/buffer.c: make bh_lru_install() more efficient	2017-07-10 16:32:30 -07:00
char_dev.c
compat.c	fs/compat.c: trim unused includes	2017-04-17 12:52:27 -04:00
compat_binfmt_elf.c
compat_ioctl.c	Merge branch 'work.__copy_in_user' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-08 10:15:02 -07:00
coredump.c
dax.c	Writeback error handling fixes (pile #2)	2017-07-07 19:38:17 -07:00
dcache.c	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
dcookies.c
direct-io.c	fs: add O_DIRECT and aio support for sending down write life time hints	2017-06-27 12:05:36 -06:00
drop_caches.c
eventfd.c	There has been a fair amount of activity in the docs tree this time	2017-07-03 21:13:25 -07:00
eventpoll.c	kcmp: fs/epoll: wrap kcmp code with CONFIG_CHECKPOINT_RESTORE	2017-07-12 16:26:01 -07:00
exec.c	exec: Limit arg stack to at most 75% of _STK_LIM	2017-07-07 20:05:08 -07:00
fcntl.c	vfs: fix flock compat thinko	2017-07-07 13:48:18 -07:00
fhandle.c	fhandle: move compat syscalls from compat.c	2017-04-17 12:52:26 -04:00
file.c	fs/file.c: replace alloc_fdmem() with kvmalloc() alternative	2017-07-06 16:24:30 -07:00
file_table.c	fs: new infrastructure for writeback error handling and reporting	2017-07-06 07:02:25 -04:00
filesystems.c	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
fs-writeback.c	writeback: rework wb_[dec|inc]_stat family of functions	2017-07-12 16:26:05 -07:00
fs_pin.c	sched/wait: Disambiguate wq_entry->task_list and wq_head->task_list naming	2017-06-20 12:19:14 +02:00
fs_struct.c
inode.c	Merge branch 'work.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-08 10:50:54 -07:00
internal.h	Merge branch 'work.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-05-09 09:12:53 -07:00
ioctl.c
iomap.c	Changes since last update:	2017-07-14 22:57:32 -07:00
Kconfig	fs/Kconfig: kill CONFIG_PERCPU_RWSEM some more	2017-07-12 16:26:00 -07:00
Kconfig.binfmt
libfs.c	fs: convert __generic_file_fsync to use errseq_t based reporting	2017-07-06 07:02:29 -04:00
locks.c	fs/locks: pass kernel struct flock to fcntl_getlk/setlk	2017-05-27 06:07:19 -04:00
Makefile
mbcache.c	ext4: xattr inode deduplication	2017-06-22 11:44:55 -04:00
mount.h	Now that IPC and other changes have landed, enable manual markings for	2017-07-19 08:55:18 -07:00
mpage.c	There has been a fair amount of activity in the docs tree this time	2017-07-03 21:13:25 -07:00
namei.c	Now that IPC and other changes have landed, enable manual markings for	2017-07-19 08:55:18 -07:00
namespace.c	Merge branch 'work.mount' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-15 12:00:42 -07:00
no-block.c
nsfs.c	VFS: Provide empty name qstr	2017-07-06 03:27:09 -04:00
open.c	Writeback error handling fixes (pile #2)	2017-07-07 19:38:17 -07:00
pipe.c	VFS: Provide empty name qstr	2017-07-06 03:27:09 -04:00
pnode.c	mnt: Make propagate_umount less slow for overlapping mount propagation trees	2017-05-23 08:41:17 -05:00
pnode.h
posix_acl.c
proc_namespace.c
read_write.c	Merge branch 'work.read_write' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-07 21:48:15 -07:00
readdir.c	readdir: move compat syscalls from compat.c	2017-04-17 12:52:24 -04:00
select.c	Merge branch 'misc.compat' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-06 20:57:13 -07:00
seq_file.c	mm: introduce kv[mz]alloc helpers	2017-05-08 17:15:12 -07:00
signalfd.c	sched/wait: Rename wait_queue_t => wait_queue_entry_t	2017-06-20 12:18:27 +02:00
splice.c	fs: implement vfs_iter_write using do_iter_write	2017-06-29 17:49:23 -04:00
stack.c
stat.c	ufs: restore maintaining ->i_blocks	2017-06-09 16:28:01 -04:00
statfs.c	Merge branch 'work.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs	2017-07-08 10:50:54 -07:00
super.c	VFS: Kill off s_options and helpers	2017-07-11 06:09:21 -04:00
sync.c	fs: remove call_fsync helper function	2017-07-05 18:44:23 -04:00
timerfd.c	timerfd: Use get_itimerspec64() and put_itimerspec64()	2017-06-30 04:14:38 -04:00
userfaultfd.c	userfaultfd: non-cooperative: flush event_wqh at release time	2017-08-02 17:16:13 -07:00
utimes.c	utimes: move compat syscalls from compat.c	2017-04-17 12:52:23 -04:00
xattr.c	treewide: use kv[mz]alloc* rather than opencoded variants	2017-05-08 17:15:13 -07:00