linux/include
Dave Chinner 7747bd4bce sync: don't block the flusher thread waiting on IO
When sync does it's WB_SYNC_ALL writeback, it issues data Io and
then immediately waits for IO completion. This is done in the
context of the flusher thread, and hence completely ties up the
flusher thread for the backing device until all the dirty inodes
have been synced. On filesystems that are dirtying inodes constantly
and quickly, this means the flusher thread can be tied up for
minutes per sync call and hence badly affect system level write IO
performance as the page cache cannot be cleaned quickly.

We already have a wait loop for IO completion for sync(2), so cut
this out of the flusher thread and delegate it to wait_sb_inodes().
Hence we can do rapid IO submission, and then wait for it all to
complete.

Effect of sync on fsmark before the patch:

FSUse%        Count         Size    Files/sec     App Overhead
.....
     0       640000         4096      35154.6          1026984
     0       720000         4096      36740.3          1023844
     0       800000         4096      36184.6           916599
     0       880000         4096       1282.7          1054367
     0       960000         4096       3951.3           918773
     0      1040000         4096      40646.2           996448
     0      1120000         4096      43610.1           895647
     0      1200000         4096      40333.1           921048

And a single sync pass took:

  real    0m52.407s
  user    0m0.000s
  sys     0m0.090s

After the patch, there is no impact on fsmark results, and each
individual sync(2) operation run concurrently with the same fsmark
workload takes roughly 7s:

  real    0m6.930s
  user    0m0.000s
  sys     0m0.039s

IOWs, sync is 7-8x faster on a busy filesystem and does not have an
adverse impact on ongoing async data write operations.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-02 09:16:42 -07:00
..
acpi ACPI / dock / PCI: Synchronous handling of dock events for PCI devices 2013-06-24 11:22:53 +02:00
asm-generic Merge branch 'fixes' of git://git.kernel.org/pub/scm/virt/kvm/kvm 2013-06-11 11:16:43 -07:00
clocksource ARM: late Exynos multiplatform changes 2013-05-07 11:28:42 -07:00
crypto crypto: sha512 - Expose generic sha512 routine to be callable from other modules 2013-04-25 21:00:57 +08:00
drm Merge branch 'drm-radeon-sun-hainan' of git://people.freedesktop.org/~airlied/linux 2013-05-21 08:50:57 -07:00
dt-bindings
keys
linux sync: don't block the flusher thread waiting on IO 2013-07-02 09:16:42 -07:00
math-emu
media Merge branch 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media 2013-06-18 06:25:08 -10:00
memory
misc
net ip_tunnel: remove __net_init/exit from exported functions 2013-06-13 03:00:59 -07:00
pcmcia
ras
rdma
rxrpc
scsi Merge branch 'postmerge' into for-linus 2013-05-10 07:54:01 -07:00
sound ASoC: dapm: Treat DAI widgets like AIF widgets for power 2013-06-07 15:54:50 +01:00
target target: Propigate up ->cmd_kref put return via transport_generic_free_cmd 2013-05-31 01:21:23 -07:00
trace Fixed regressions (two stability regressions and a performance 2013-05-14 09:30:54 -07:00
uapi netlink: export netlink_diag.h header 2013-06-19 18:30:27 -07:00
video Merge branch 'fbdev-3.10-fixes' of git://gitorious.org/linux-omap-dss2/linux into linux-fbdev/for-3.10-fixes 2013-05-29 17:00:34 +08:00
xen xenbus: delay xenbus frontend resume if xenstored is not running 2013-05-29 09:04:19 -04:00
Kbuild UAPI: remove empty Kbuild files 2013-04-30 17:04:09 -07:00