205 lines · plain
1===============2Lock Statistics3===============4 5What6====7 8As the name suggests, it provides statistics on locks.9 10 11Why12===13 14Because things like lock contention can severely impact performance.15 16How17===18 19Lockdep already has hooks in the lock functions and maps lock instances to20lock classes. We build on that (see Documentation/locking/lockdep-design.rst).21The graph below shows the relation between the lock functions and the various22hooks therein::23 24 __acquire25 |26 lock _____27 | \28 | __contended29 | |30 | <wait>31 | _______/32 |/33 |34 __acquired35 |36 .37 <hold>38 .39 |40 __release41 |42 unlock43 44 lock, unlock - the regular lock functions45 __* - the hooks46 <> - states47 48With these hooks we provide the following statistics:49 50 con-bounces51 - number of lock contention that involved x-cpu data52 contentions53 - number of lock acquisitions that had to wait54 wait time55 min56 - shortest (non-0) time we ever had to wait for a lock57 max58 - longest time we ever had to wait for a lock59 total60 - total time we spend waiting on this lock61 avg62 - average time spent waiting on this lock63 acq-bounces64 - number of lock acquisitions that involved x-cpu data65 acquisitions66 - number of times we took the lock67 hold time68 min69 - shortest (non-0) time we ever held the lock70 max71 - longest time we ever held the lock72 total73 - total time this lock was held74 avg75 - average time this lock was held76 77These numbers are gathered per lock class, per read/write state (when78applicable).79 80It also tracks 4 contention points per class. A contention point is a call site81that had to wait on lock acquisition.82 83Configuration84-------------85 86Lock statistics are enabled via CONFIG_LOCK_STAT.87 88Usage89-----90 91Enable collection of statistics::92 93 # echo 1 >/proc/sys/kernel/lock_stat94 95Disable collection of statistics::96 97 # echo 0 >/proc/sys/kernel/lock_stat98 99Look at the current lock statistics::100 101 ( line numbers not part of actual output, done for clarity in the explanation102 below )103 104 # less /proc/lock_stat105 106 01 lock_stat version 0.4107 02-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------108 03 class name con-bounces contentions waittime-min waittime-max waittime-total waittime-avg acq-bounces acquisitions holdtime-min holdtime-max holdtime-total holdtime-avg109 04-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------110 05111 06 &mm->mmap_sem-W: 46 84 0.26 939.10 16371.53 194.90 47291 2922365 0.16 2220301.69 17464026916.32 5975.99112 07 &mm->mmap_sem-R: 37 100 1.31 299502.61 325629.52 3256.30 212344 34316685 0.10 7744.91 95016910.20 2.77113 08 ---------------114 09 &mm->mmap_sem 1 [<ffffffff811502a7>] khugepaged_scan_mm_slot+0x57/0x280115 10 &mm->mmap_sem 96 [<ffffffff815351c4>] __do_page_fault+0x1d4/0x510116 11 &mm->mmap_sem 34 [<ffffffff81113d77>] vm_mmap_pgoff+0x87/0xd0117 12 &mm->mmap_sem 17 [<ffffffff81127e71>] vm_munmap+0x41/0x80118 13 ---------------119 14 &mm->mmap_sem 1 [<ffffffff81046fda>] dup_mmap+0x2a/0x3f0120 15 &mm->mmap_sem 60 [<ffffffff81129e29>] SyS_mprotect+0xe9/0x250121 16 &mm->mmap_sem 41 [<ffffffff815351c4>] __do_page_fault+0x1d4/0x510122 17 &mm->mmap_sem 68 [<ffffffff81113d77>] vm_mmap_pgoff+0x87/0xd0123 18124 19.............................................................................................................................................................................................................................125 20126 21 unix_table_lock: 110 112 0.21 49.24 163.91 1.46 21094 66312 0.12 624.42 31589.81 0.48127 22 ---------------128 23 unix_table_lock 45 [<ffffffff8150ad8e>] unix_create1+0x16e/0x1b0129 24 unix_table_lock 47 [<ffffffff8150b111>] unix_release_sock+0x31/0x250130 25 unix_table_lock 15 [<ffffffff8150ca37>] unix_find_other+0x117/0x230131 26 unix_table_lock 5 [<ffffffff8150a09f>] unix_autobind+0x11f/0x1b0132 27 ---------------133 28 unix_table_lock 39 [<ffffffff8150b111>] unix_release_sock+0x31/0x250134 29 unix_table_lock 49 [<ffffffff8150ad8e>] unix_create1+0x16e/0x1b0135 30 unix_table_lock 20 [<ffffffff8150ca37>] unix_find_other+0x117/0x230136 31 unix_table_lock 4 [<ffffffff8150a09f>] unix_autobind+0x11f/0x1b0137 138 139This excerpt shows the first two lock class statistics. Line 01 shows the140output version - each time the format changes this will be updated. Line 02-04141show the header with column descriptions. Lines 05-18 and 20-31 show the actual142statistics. These statistics come in two parts; the actual stats separated by a143short separator (line 08, 13) from the contention points.144 145Lines 09-12 show the first 4 recorded contention points (the code146which tries to get the lock) and lines 14-17 show the first 4 recorded147contended points (the lock holder). It is possible that the max148con-bounces point is missing in the statistics.149 150The first lock (05-18) is a read/write lock, and shows two lines above the151short separator. The contention points don't match the column descriptors,152they have two: contentions and [<IP>] symbol. The second set of contention153points are the points we're contending with.154 155The integer part of the time values is in us.156 157Dealing with nested locks, subclasses may appear::158 159 32...........................................................................................................................................................................................................................160 33161 34 &rq->lock: 13128 13128 0.43 190.53 103881.26 7.91 97454 3453404 0.00 401.11 13224683.11 3.82162 35 ---------163 36 &rq->lock 645 [<ffffffff8103bfc4>] task_rq_lock+0x43/0x75164 37 &rq->lock 297 [<ffffffff8104ba65>] try_to_wake_up+0x127/0x25a165 38 &rq->lock 360 [<ffffffff8103c4c5>] select_task_rq_fair+0x1f0/0x74a166 39 &rq->lock 428 [<ffffffff81045f98>] scheduler_tick+0x46/0x1fb167 40 ---------168 41 &rq->lock 77 [<ffffffff8103bfc4>] task_rq_lock+0x43/0x75169 42 &rq->lock 174 [<ffffffff8104ba65>] try_to_wake_up+0x127/0x25a170 43 &rq->lock 4715 [<ffffffff8103ed4b>] double_rq_lock+0x42/0x54171 44 &rq->lock 893 [<ffffffff81340524>] schedule+0x157/0x7b8172 45173 46...........................................................................................................................................................................................................................174 47175 48 &rq->lock/1: 1526 11488 0.33 388.73 136294.31 11.86 21461 38404 0.00 37.93 109388.53 2.84176 49 -----------177 50 &rq->lock/1 11526 [<ffffffff8103ed58>] double_rq_lock+0x4f/0x54178 51 -----------179 52 &rq->lock/1 5645 [<ffffffff8103ed4b>] double_rq_lock+0x42/0x54180 53 &rq->lock/1 1224 [<ffffffff81340524>] schedule+0x157/0x7b8181 54 &rq->lock/1 4336 [<ffffffff8103ed58>] double_rq_lock+0x4f/0x54182 55 &rq->lock/1 181 [<ffffffff8104ba65>] try_to_wake_up+0x127/0x25a183 184Line 48 shows statistics for the second subclass (/1) of &rq->lock class185(subclass starts from 0), since in this case, as line 50 suggests,186double_rq_lock actually acquires a nested lock of two spinlocks.187 188View the top contending locks::189 190 # grep : /proc/lock_stat | head191 clockevents_lock: 2926159 2947636 0.15 46882.81 1784540466.34 605.41 3381345 3879161 0.00 2260.97 53178395.68 13.71192 tick_broadcast_lock: 346460 346717 0.18 2257.43 39364622.71 113.54 3642919 4242696 0.00 2263.79 49173646.60 11.59193 &mapping->i_mmap_mutex: 203896 203899 3.36 645530.05 31767507988.39 155800.21 3361776 8893984 0.17 2254.15 14110121.02 1.59194 &rq->lock: 135014 136909 0.18 606.09 842160.68 6.15 1540728 10436146 0.00 728.72 17606683.41 1.69195 &(&zone->lru_lock)->rlock: 93000 94934 0.16 59.18 188253.78 1.98 1199912 3809894 0.15 391.40 3559518.81 0.93196 tasklist_lock-W: 40667 41130 0.23 1189.42 428980.51 10.43 270278 510106 0.16 653.51 3939674.91 7.72197 tasklist_lock-R: 21298 21305 0.20 1310.05 215511.12 10.12 186204 241258 0.14 1162.33 1179779.23 4.89198 rcu_node_1: 47656 49022 0.16 635.41 193616.41 3.95 844888 1865423 0.00 764.26 1656226.96 0.89199 &(&dentry->d_lockref.lock)->rlock: 39791 40179 0.15 1302.08 88851.96 2.21 2790851 12527025 0.10 1910.75 3379714.27 0.27200 rcu_node_0: 29203 30064 0.16 786.55 1555573.00 51.74 88963 244254 0.00 398.87 428872.51 1.76201 202Clear the statistics::203 204 # echo 0 > /proc/lock_stat205