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 (see Documentation/locking/lockdep-design.txt).
16The graph below shows the relation between the lock functions and the various
17hooks therein.
18
19        __acquire
20            |
21           lock _____
22            |        \
23            |    __contended
24            |         |
25            |       <wait>
26            | _______/
27            |/
28            |
29       __acquired
30            |
31            .
32          <hold>
33            .
34            |
35       __release
36            |
37         unlock
38
39lock, unlock	- the regular lock functions
40__*		- the hooks
41<> 		- states
42
43With these hooks we provide the following statistics:
44
45 con-bounces       - number of lock contention that involved x-cpu data
46 contentions       - number of lock acquisitions that had to wait
47 wait time min     - shortest (non-0) time we ever had to wait for a lock
48           max     - longest time we ever had to wait for a lock
49	   total   - total time we spend waiting on this lock
50	   avg     - average time spent waiting on this lock
51 acq-bounces       - number of lock acquisitions that involved x-cpu data
52 acquisitions      - number of times we took the lock
53 hold time min     - shortest (non-0) time we ever held the lock
54	   max     - longest time we ever held the lock
55	   total   - total time this lock was held
56	   avg     - average time this lock was held
57
58These numbers are gathered per lock class, per read/write state (when
59applicable).
60
61It also tracks 4 contention points per class. A contention point is a call site
62that had to wait on lock acquisition.
63
64 - CONFIGURATION
65
66Lock statistics are enabled via CONFIG_LOCK_STAT.
67
68 - USAGE
69
70Enable collection of statistics:
71
72# echo 1 >/proc/sys/kernel/lock_stat
73
74Disable collection of statistics:
75
76# echo 0 >/proc/sys/kernel/lock_stat
77
78Look at the current lock statistics:
79
80( line numbers not part of actual output, done for clarity in the explanation
81  below )
82
83# less /proc/lock_stat
84
8501 lock_stat version 0.4
8602-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
8703                              class name    con-bounces    contentions   waittime-min   waittime-max waittime-total   waittime-avg    acq-bounces   acquisitions   holdtime-min   holdtime-max holdtime-total   holdtime-avg
8804-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
8905
9006                         &mm->mmap_sem-W:            46             84           0.26         939.10       16371.53         194.90          47291        2922365           0.16     2220301.69 17464026916.32        5975.99
9107                         &mm->mmap_sem-R:            37            100           1.31      299502.61      325629.52        3256.30         212344       34316685           0.10        7744.91    95016910.20           2.77
9208                         ---------------
9309                           &mm->mmap_sem              1          [<ffffffff811502a7>] khugepaged_scan_mm_slot+0x57/0x280
9419                           &mm->mmap_sem             96          [<ffffffff815351c4>] __do_page_fault+0x1d4/0x510
9511                           &mm->mmap_sem             34          [<ffffffff81113d77>] vm_mmap_pgoff+0x87/0xd0
9612                           &mm->mmap_sem             17          [<ffffffff81127e71>] vm_munmap+0x41/0x80
9713                         ---------------
9814                           &mm->mmap_sem              1          [<ffffffff81046fda>] dup_mmap+0x2a/0x3f0
9915                           &mm->mmap_sem             60          [<ffffffff81129e29>] SyS_mprotect+0xe9/0x250
10016                           &mm->mmap_sem             41          [<ffffffff815351c4>] __do_page_fault+0x1d4/0x510
10117                           &mm->mmap_sem             68          [<ffffffff81113d77>] vm_mmap_pgoff+0x87/0xd0
10218
10319.............................................................................................................................................................................................................................
10420
10521                         unix_table_lock:           110            112           0.21          49.24         163.91           1.46          21094          66312           0.12         624.42       31589.81           0.48
10622                         ---------------
10723                         unix_table_lock             45          [<ffffffff8150ad8e>] unix_create1+0x16e/0x1b0
10824                         unix_table_lock             47          [<ffffffff8150b111>] unix_release_sock+0x31/0x250
10925                         unix_table_lock             15          [<ffffffff8150ca37>] unix_find_other+0x117/0x230
11026                         unix_table_lock              5          [<ffffffff8150a09f>] unix_autobind+0x11f/0x1b0
11127                         ---------------
11228                         unix_table_lock             39          [<ffffffff8150b111>] unix_release_sock+0x31/0x250
11329                         unix_table_lock             49          [<ffffffff8150ad8e>] unix_create1+0x16e/0x1b0
11430                         unix_table_lock             20          [<ffffffff8150ca37>] unix_find_other+0x117/0x230
11531                         unix_table_lock              4          [<ffffffff8150a09f>] unix_autobind+0x11f/0x1b0
116
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
120show the header with column descriptions. Lines 05-18 and 20-31 show the actual
121statistics. These statistics come in two parts; the actual stats separated by a
122short separator (line 08, 13) from the contention points.
123
124Lines 09-12 show the first 4 recorded contention points (the code
125which tries to get the lock) and lines 14-17 show the first 4 recorded
126contended points (the lock holder). It is possible that the max
127con-bounces point is missing in the statistics.
128
129The first lock (05-18) is a read/write lock, and shows two lines above the
130short separator. The contention points don't match the column descriptors,
131they have two: contentions and [<IP>] symbol. The second set of contention
132points are the points we're contending with.
133
134The integer part of the time values is in us.
135
136Dealing with nested locks, subclasses may appear:
137
13832...........................................................................................................................................................................................................................
13933
14034                               &rq->lock:       13128          13128           0.43         190.53      103881.26           7.91          97454        3453404           0.00         401.11    13224683.11           3.82
14135                               ---------
14236                               &rq->lock          645          [<ffffffff8103bfc4>] task_rq_lock+0x43/0x75
14337                               &rq->lock          297          [<ffffffff8104ba65>] try_to_wake_up+0x127/0x25a
14438                               &rq->lock          360          [<ffffffff8103c4c5>] select_task_rq_fair+0x1f0/0x74a
14539                               &rq->lock          428          [<ffffffff81045f98>] scheduler_tick+0x46/0x1fb
14640                               ---------
14741                               &rq->lock           77          [<ffffffff8103bfc4>] task_rq_lock+0x43/0x75
14842                               &rq->lock          174          [<ffffffff8104ba65>] try_to_wake_up+0x127/0x25a
14943                               &rq->lock         4715          [<ffffffff8103ed4b>] double_rq_lock+0x42/0x54
15044                               &rq->lock          893          [<ffffffff81340524>] schedule+0x157/0x7b8
15145
15246...........................................................................................................................................................................................................................
15347
15448                             &rq->lock/1:        1526          11488           0.33         388.73      136294.31          11.86          21461          38404           0.00          37.93      109388.53           2.84
15549                             -----------
15650                             &rq->lock/1        11526          [<ffffffff8103ed58>] double_rq_lock+0x4f/0x54
15751                             -----------
15852                             &rq->lock/1         5645          [<ffffffff8103ed4b>] double_rq_lock+0x42/0x54
15953                             &rq->lock/1         1224          [<ffffffff81340524>] schedule+0x157/0x7b8
16054                             &rq->lock/1         4336          [<ffffffff8103ed58>] double_rq_lock+0x4f/0x54
16155                             &rq->lock/1          181          [<ffffffff8104ba65>] try_to_wake_up+0x127/0x25a
162
163Line 48 shows statistics for the second subclass (/1) of &rq->lock class
164(subclass starts from 0), since in this case, as line 50 suggests,
165double_rq_lock actually acquires a nested lock of two spinlocks.
166
167View the top contending locks:
168
169# grep : /proc/lock_stat | head
170			clockevents_lock:       2926159        2947636           0.15       46882.81  1784540466.34         605.41        3381345        3879161           0.00        2260.97    53178395.68          13.71
171		     tick_broadcast_lock:        346460         346717           0.18        2257.43    39364622.71         113.54        3642919        4242696           0.00        2263.79    49173646.60          11.59
172		  &mapping->i_mmap_mutex:        203896         203899           3.36      645530.05 31767507988.39      155800.21        3361776        8893984           0.17        2254.15    14110121.02           1.59
173			       &rq->lock:        135014         136909           0.18         606.09      842160.68           6.15        1540728       10436146           0.00         728.72    17606683.41           1.69
174	       &(&zone->lru_lock)->rlock:         93000          94934           0.16          59.18      188253.78           1.98        1199912        3809894           0.15         391.40     3559518.81           0.93
175			 tasklist_lock-W:         40667          41130           0.23        1189.42      428980.51          10.43         270278         510106           0.16         653.51     3939674.91           7.72
176			 tasklist_lock-R:         21298          21305           0.20        1310.05      215511.12          10.12         186204         241258           0.14        1162.33     1179779.23           4.89
177			      rcu_node_1:         47656          49022           0.16         635.41      193616.41           3.95         844888        1865423           0.00         764.26     1656226.96           0.89
178       &(&dentry->d_lockref.lock)->rlock:         39791          40179           0.15        1302.08       88851.96           2.21        2790851       12527025           0.10        1910.75     3379714.27           0.27
179			      rcu_node_0:         29203          30064           0.16         786.55     1555573.00          51.74          88963         244254           0.00         398.87      428872.51           1.76
180
181Clear the statistics:
182
183# echo 0 > /proc/lock_stat
184