1 ================ 1 ================
2 Delay accounting 2 Delay accounting
3 ================ 3 ================
4 4
5 Tasks encounter delays in execution when they 5 Tasks encounter delays in execution when they wait
6 for some kernel resource to become available e 6 for some kernel resource to become available e.g. a
7 runnable task may wait for a free CPU to run o 7 runnable task may wait for a free CPU to run on.
8 8
9 The per-task delay accounting functionality me 9 The per-task delay accounting functionality measures
10 the delays experienced by a task while 10 the delays experienced by a task while
11 11
12 a) waiting for a CPU (while being runnable) 12 a) waiting for a CPU (while being runnable)
13 b) completion of synchronous block I/O initiat 13 b) completion of synchronous block I/O initiated by the task
14 c) swapping in pages 14 c) swapping in pages
15 d) memory reclaim 15 d) memory reclaim
16 e) thrashing 16 e) thrashing
17 f) direct compact 17 f) direct compact
18 g) write-protect copy 18 g) write-protect copy
19 h) IRQ/SOFTIRQ <<
20 19
21 and makes these statistics available to usersp 20 and makes these statistics available to userspace through
22 the taskstats interface. 21 the taskstats interface.
23 22
24 Such delays provide feedback for setting a tas 23 Such delays provide feedback for setting a task's cpu priority,
25 io priority and rss limit values appropriately 24 io priority and rss limit values appropriately. Long delays for
26 important tasks could be a trigger for raising 25 important tasks could be a trigger for raising its corresponding priority.
27 26
28 The functionality, through its use of the task 27 The functionality, through its use of the taskstats interface, also provides
29 delay statistics aggregated for all tasks (or 28 delay statistics aggregated for all tasks (or threads) belonging to a
30 thread group (corresponding to a traditional U 29 thread group (corresponding to a traditional Unix process). This is a commonly
31 needed aggregation that is more efficiently do 30 needed aggregation that is more efficiently done by the kernel.
32 31
33 Userspace utilities, particularly resource man 32 Userspace utilities, particularly resource management applications, can also
34 aggregate delay statistics into arbitrary grou 33 aggregate delay statistics into arbitrary groups. To enable this, delay
35 statistics of a task are available both during 34 statistics of a task are available both during its lifetime as well as on its
36 exit, ensuring continuous and complete monitor 35 exit, ensuring continuous and complete monitoring can be done.
37 36
38 37
39 Interface 38 Interface
40 --------- 39 ---------
41 40
42 Delay accounting uses the taskstats interface 41 Delay accounting uses the taskstats interface which is described
43 in detail in a separate document in this direc 42 in detail in a separate document in this directory. Taskstats returns a
44 generic data structure to userspace correspond 43 generic data structure to userspace corresponding to per-pid and per-tgid
45 statistics. The delay accounting functionality 44 statistics. The delay accounting functionality populates specific fields of
46 this structure. See 45 this structure. See
47 46
48 include/uapi/linux/taskstats.h 47 include/uapi/linux/taskstats.h
49 48
50 for a description of the fields pertaining to 49 for a description of the fields pertaining to delay accounting.
51 It will generally be in the form of counters r 50 It will generally be in the form of counters returning the cumulative
52 delay seen for cpu, sync block I/O, swapin, me 51 delay seen for cpu, sync block I/O, swapin, memory reclaim, thrash page
53 cache, direct compact, write-protect copy, IRQ !! 52 cache, direct compact, write-protect copy etc.
54 53
55 Taking the difference of two successive readin 54 Taking the difference of two successive readings of a given
56 counter (say cpu_delay_total) for a task will 55 counter (say cpu_delay_total) for a task will give the delay
57 experienced by the task waiting for the corres 56 experienced by the task waiting for the corresponding resource
58 in that interval. 57 in that interval.
59 58
60 When a task exits, records containing the per- 59 When a task exits, records containing the per-task statistics
61 are sent to userspace without requiring a comm 60 are sent to userspace without requiring a command. If it is the last exiting
62 task of a thread group, the per-tgid statistic 61 task of a thread group, the per-tgid statistics are also sent. More details
63 are given in the taskstats interface descripti 62 are given in the taskstats interface description.
64 63
65 The getdelays.c userspace utility in tools/acc 64 The getdelays.c userspace utility in tools/accounting directory allows simple
66 commands to be run and the corresponding delay 65 commands to be run and the corresponding delay statistics to be displayed. It
67 also serves as an example of using the tasksta 66 also serves as an example of using the taskstats interface.
68 67
69 Usage 68 Usage
70 ----- 69 -----
71 70
72 Compile the kernel with:: 71 Compile the kernel with::
73 72
74 CONFIG_TASK_DELAY_ACCT=y 73 CONFIG_TASK_DELAY_ACCT=y
75 CONFIG_TASKSTATS=y 74 CONFIG_TASKSTATS=y
76 75
77 Delay accounting is disabled by default at boo 76 Delay accounting is disabled by default at boot up.
78 To enable, add:: 77 To enable, add::
79 78
80 delayacct 79 delayacct
81 80
82 to the kernel boot options. The rest of the in 81 to the kernel boot options. The rest of the instructions below assume this has
83 been done. Alternatively, use sysctl kernel.ta 82 been done. Alternatively, use sysctl kernel.task_delayacct to switch the state
84 at runtime. Note however that only tasks start 83 at runtime. Note however that only tasks started after enabling it will have
85 delayacct information. 84 delayacct information.
86 85
87 After the system has booted up, use a utility 86 After the system has booted up, use a utility
88 similar to getdelays.c to access the delays 87 similar to getdelays.c to access the delays
89 seen by a given task or a task group (tgid). 88 seen by a given task or a task group (tgid).
90 The utility also allows a given command to be 89 The utility also allows a given command to be
91 executed and the corresponding delays to be 90 executed and the corresponding delays to be
92 seen. 91 seen.
93 92
94 General format of the getdelays command:: 93 General format of the getdelays command::
95 94
96 getdelays [-dilv] [-t tgid] [-p pid] 95 getdelays [-dilv] [-t tgid] [-p pid]
97 96
98 Get delays, since system boot, for pid 10:: 97 Get delays, since system boot, for pid 10::
99 98
100 # ./getdelays -d -p 10 99 # ./getdelays -d -p 10
101 (output similar to next case) 100 (output similar to next case)
102 101
103 Get sum of delays, since system boot, for all 102 Get sum of delays, since system boot, for all pids with tgid 5::
104 103
105 # ./getdelays -d -t 5 104 # ./getdelays -d -t 5
106 print delayacct stats ON 105 print delayacct stats ON
107 TGID 5 106 TGID 5
108 107
109 108
110 CPU count real total 109 CPU count real total virtual total delay total delay average
111 8 7000000 110 8 7000000 6872122 3382277 0.423ms
112 IO count delay total 111 IO count delay total delay average
113 0 0 0 112 0 0 0.000ms
114 SWAP count delay total 113 SWAP count delay total delay average
115 0 0 114 0 0 0.000ms
116 RECLAIM count delay total 115 RECLAIM count delay total delay average
117 0 0 0 116 0 0 0.000ms
118 THRASHING count delay total 117 THRASHING count delay total delay average
119 0 0 118 0 0 0.000ms
120 COMPACT count delay total 119 COMPACT count delay total delay average
121 0 0 120 0 0 0.000ms
122 WPCOPY count delay total !! 121 WPCOPY count delay total delay average
123 0 0 <<
124 IRQ count delay total <<
125 0 0 122 0 0 0.000ms
126 123
127 Get IO accounting for pid 1, it works only wit 124 Get IO accounting for pid 1, it works only with -p::
128 125
129 # ./getdelays -i -p 1 126 # ./getdelays -i -p 1
130 printing IO accounting 127 printing IO accounting
131 linuxrc: read=65536, write=0, cancelle 128 linuxrc: read=65536, write=0, cancelled_write=0
132 129
133 The above command can be used with -v to get m 130 The above command can be used with -v to get more debug information.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.