Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / kernel / msm / a49ebbabb084d345991b72818a119616431416f2 / . / Documentation / lockstat.txt
blob: 65f4c795015dead0cfde0f60381b7efcec67da9b [file] [log] [blame]
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]1
2LOCK STATISTICS
3
4- WHAT
5
6As the name suggests, it provides statistics on locks.
7
8- WHY
9
10Because things like lock contention can severely impact performance.
11
12- HOW
13
14Lockdep already has hooks in the lock functions and maps lock instances to
15lock classes. We build on that. The graph below shows the relation between
16the lock functions and the various hooks therein.
17
18 __acquire
19 |
20 lock _____
21 | \
22 | __contended
23 | |
24 | <wait>
25 | _______/
26 |/
27 |
28 __acquired
29 |
30 .
31 <hold>
32 .
33 |
34 __release
35 |
36 unlock
37
38lock, unlock - the regular lock functions
39__* - the hooks
40<> - states
41
42With these hooks we provide the following statistics:
43
44 con-bounces - number of lock contention that involved x-cpu data
45 contentions - number of lock acquisitions that had to wait
46 wait time min - shortest (non-0) time we ever had to wait for a lock
47 max - longest time we ever had to wait for a lock
48 total - total time we spend waiting on this lock
49 acq-bounces - number of lock acquisitions that involved x-cpu data
50 acquisitions - number of times we took the lock
51 hold time min - shortest (non-0) time we ever held the lock
52 max - longest time we ever held the lock
53 total - total time this lock was held
54
55From these number various other statistics can be derived, such as:
56
57 hold time average = hold time total / acquisitions
58
59These numbers are gathered per lock class, per read/write state (when
60applicable).
61
62It also tracks 4 contention points per class. A contention point is a call site
63that had to wait on lock acquisition.
64
Frank Rowandf5754bf2009-11-19 13:54:05 -0800[diff] [blame]65 - CONFIGURATION
66
67Lock statistics are enabled via CONFIG_LOCK_STATS.
68
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]69 - USAGE
70
Frank Rowandf5754bf2009-11-19 13:54:05 -0800[diff] [blame]71Enable collection of statistics:
72
73# echo 1 >/proc/sys/kernel/lock_stat
74
75Disable collection of statistics:
76
77# echo 0 >/proc/sys/kernel/lock_stat
78
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]79Look at the current lock statistics:
80
81( line numbers not part of actual output, done for clarity in the explanation
82 below )
83
84# less /proc/lock_stat
85
Peter Zijlstrac7e78cf2008-10-16 23:17:09 +0200[diff] [blame]8601 lock_stat version 0.3
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]8702 -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
8803 class name con-bounces contentions waittime-min waittime-max waittime-total acq-bounces acquisitions holdtime-min holdtime-max holdtime-total
8904 -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
9005
Peter Zijlstrac7e78cf2008-10-16 23:17:09 +0200[diff] [blame]9106 &mm->mmap_sem-W: 233 538 18446744073708 22924.27 607243.51 1342 45806 1.71 8595.89 1180582.34
9207 &mm->mmap_sem-R: 205 587 18446744073708 28403.36 731975.00 1940 412426 0.58 187825.45 6307502.88
9308 ---------------
9409 &mm->mmap_sem 487 [<ffffffff8053491f>] do_page_fault+0x466/0x928
9510 &mm->mmap_sem 179 [<ffffffff802a6200>] sys_mprotect+0xcd/0x21d
9611 &mm->mmap_sem 279 [<ffffffff80210a57>] sys_mmap+0x75/0xce
9712 &mm->mmap_sem 76 [<ffffffff802a490b>] sys_munmap+0x32/0x59
9813 ---------------
9914 &mm->mmap_sem 270 [<ffffffff80210a57>] sys_mmap+0x75/0xce
10015 &mm->mmap_sem 431 [<ffffffff8053491f>] do_page_fault+0x466/0x928
10116 &mm->mmap_sem 138 [<ffffffff802a490b>] sys_munmap+0x32/0x59
10217 &mm->mmap_sem 145 [<ffffffff802a6200>] sys_mprotect+0xcd/0x21d
10318
10419 ...............................................................................................................................................................................................
10520
10621 dcache_lock: 621 623 0.52 118.26 1053.02 6745 91930 0.29 316.29 118423.41
10722 -----------
10823 dcache_lock 179 [<ffffffff80378274>] _atomic_dec_and_lock+0x34/0x54
10924 dcache_lock 113 [<ffffffff802cc17b>] d_alloc+0x19a/0x1eb
11025 dcache_lock 99 [<ffffffff802ca0dc>] d_rehash+0x1b/0x44
11126 dcache_lock 104 [<ffffffff802cbca0>] d_instantiate+0x36/0x8a
11227 -----------
11328 dcache_lock 192 [<ffffffff80378274>] _atomic_dec_and_lock+0x34/0x54
11429 dcache_lock 98 [<ffffffff802ca0dc>] d_rehash+0x1b/0x44
11530 dcache_lock 72 [<ffffffff802cc17b>] d_alloc+0x19a/0x1eb
11631 dcache_lock 112 [<ffffffff802cbca0>] d_instantiate+0x36/0x8a
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]117
118This excerpt shows the first two lock class statistics. Line 01 shows the
119output version - each time the format changes this will be updated. Line 02-04
Peter Zijlstrac7e78cf2008-10-16 23:17:09 +0200[diff] [blame]120show the header with column descriptions. Lines 05-18 and 20-31 show the actual
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]121statistics. These statistics come in two parts; the actual stats separated by a
Peter Zijlstrac7e78cf2008-10-16 23:17:09 +0200[diff] [blame]122short separator (line 08, 13) from the contention points.
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]123
Peter Zijlstrac7e78cf2008-10-16 23:17:09 +0200[diff] [blame]124The first lock (05-18) is a read/write lock, and shows two lines above the
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]125short separator. The contention points don't match the column descriptors,
Peter Zijlstrac7e78cf2008-10-16 23:17:09 +0200[diff] [blame]126they have two: contentions and [<IP>] symbol. The second set of contention
127points are the points we're contending with.
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]128
Peter Zijlstracdad7222008-09-03 16:00:57 +0200[diff] [blame]129The integer part of the time values is in us.
Peter Zijlstraa560aa42007-10-07 00:24:33 -0700[diff] [blame]130
131View the top contending locks:
132
133# grep : /proc/lock_stat | head
134 &inode->i_data.tree_lock-W: 15 21657 0.18 1093295.30 11547131054.85 58 10415 0.16 87.51 6387.60
135 &inode->i_data.tree_lock-R: 0 0 0.00 0.00 0.00 23302 231198 0.25 8.45 98023.38
136 dcache_lock: 1037 1161 0.38 45.32 774.51 6611 243371 0.15 306.48 77387.24
137 &inode->i_mutex: 161 286 18446744073709 62882.54 1244614.55 3653 20598 18446744073709 62318.60 1693822.74
138 &zone->lru_lock: 94 94 0.53 7.33 92.10 4366 32690 0.29 59.81 16350.06
139 &inode->i_data.i_mmap_lock: 79 79 0.40 3.77 53.03 11779 87755 0.28 116.93 29898.44
140 &q->__queue_lock: 48 50 0.52 31.62 86.31 774 13131 0.17 113.08 12277.52
141 &rq->rq_lock_key: 43 47 0.74 68.50 170.63 3706 33929 0.22 107.99 17460.62
142 &rq->rq_lock_key#2: 39 46 0.75 6.68 49.03 2979 32292 0.17 125.17 17137.63
143 tasklist_lock-W: 15 15 1.45 10.87 32.70 1201 7390 0.58 62.55 13648.47
144
145Clear the statistics:
146
147# echo 0 > /proc/lock_stat