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 <<
17 f) direct compact <<
18 g) write-protect copy <<
19 h) IRQ/SOFTIRQ <<
20 16
21 and makes these statistics available to usersp 17 and makes these statistics available to userspace through
22 the taskstats interface. 18 the taskstats interface.
23 19
24 Such delays provide feedback for setting a tas 20 Such delays provide feedback for setting a task's cpu priority,
25 io priority and rss limit values appropriately 21 io priority and rss limit values appropriately. Long delays for
26 important tasks could be a trigger for raising 22 important tasks could be a trigger for raising its corresponding priority.
27 23
28 The functionality, through its use of the task 24 The functionality, through its use of the taskstats interface, also provides
29 delay statistics aggregated for all tasks (or 25 delay statistics aggregated for all tasks (or threads) belonging to a
30 thread group (corresponding to a traditional U 26 thread group (corresponding to a traditional Unix process). This is a commonly
31 needed aggregation that is more efficiently do 27 needed aggregation that is more efficiently done by the kernel.
32 28
33 Userspace utilities, particularly resource man 29 Userspace utilities, particularly resource management applications, can also
34 aggregate delay statistics into arbitrary grou 30 aggregate delay statistics into arbitrary groups. To enable this, delay
35 statistics of a task are available both during 31 statistics of a task are available both during its lifetime as well as on its
36 exit, ensuring continuous and complete monitor 32 exit, ensuring continuous and complete monitoring can be done.
37 33
38 34
39 Interface 35 Interface
40 --------- 36 ---------
41 37
42 Delay accounting uses the taskstats interface 38 Delay accounting uses the taskstats interface which is described
43 in detail in a separate document in this direc 39 in detail in a separate document in this directory. Taskstats returns a
44 generic data structure to userspace correspond 40 generic data structure to userspace corresponding to per-pid and per-tgid
45 statistics. The delay accounting functionality 41 statistics. The delay accounting functionality populates specific fields of
46 this structure. See 42 this structure. See
47 43
48 include/uapi/linux/taskstats.h !! 44 include/linux/taskstats.h
49 45
50 for a description of the fields pertaining to 46 for a description of the fields pertaining to delay accounting.
51 It will generally be in the form of counters r 47 It will generally be in the form of counters returning the cumulative
52 delay seen for cpu, sync block I/O, swapin, me !! 48 delay seen for cpu, sync block I/O, swapin, memory reclaim etc.
53 cache, direct compact, write-protect copy, IRQ <<
54 49
55 Taking the difference of two successive readin 50 Taking the difference of two successive readings of a given
56 counter (say cpu_delay_total) for a task will 51 counter (say cpu_delay_total) for a task will give the delay
57 experienced by the task waiting for the corres 52 experienced by the task waiting for the corresponding resource
58 in that interval. 53 in that interval.
59 54
60 When a task exits, records containing the per- 55 When a task exits, records containing the per-task statistics
61 are sent to userspace without requiring a comm 56 are sent to userspace without requiring a command. If it is the last exiting
62 task of a thread group, the per-tgid statistic 57 task of a thread group, the per-tgid statistics are also sent. More details
63 are given in the taskstats interface descripti 58 are given in the taskstats interface description.
64 59
65 The getdelays.c userspace utility in tools/acc 60 The getdelays.c userspace utility in tools/accounting directory allows simple
66 commands to be run and the corresponding delay 61 commands to be run and the corresponding delay statistics to be displayed. It
67 also serves as an example of using the tasksta 62 also serves as an example of using the taskstats interface.
68 63
69 Usage 64 Usage
70 ----- 65 -----
71 66
72 Compile the kernel with:: 67 Compile the kernel with::
73 68
74 CONFIG_TASK_DELAY_ACCT=y 69 CONFIG_TASK_DELAY_ACCT=y
75 CONFIG_TASKSTATS=y 70 CONFIG_TASKSTATS=y
76 71
77 Delay accounting is disabled by default at boo 72 Delay accounting is disabled by default at boot up.
78 To enable, add:: 73 To enable, add::
79 74
80 delayacct 75 delayacct
81 76
82 to the kernel boot options. The rest of the in 77 to the kernel boot options. The rest of the instructions below assume this has
83 been done. Alternatively, use sysctl kernel.ta 78 been done. Alternatively, use sysctl kernel.task_delayacct to switch the state
84 at runtime. Note however that only tasks start 79 at runtime. Note however that only tasks started after enabling it will have
85 delayacct information. 80 delayacct information.
86 81
87 After the system has booted up, use a utility 82 After the system has booted up, use a utility
88 similar to getdelays.c to access the delays 83 similar to getdelays.c to access the delays
89 seen by a given task or a task group (tgid). 84 seen by a given task or a task group (tgid).
90 The utility also allows a given command to be 85 The utility also allows a given command to be
91 executed and the corresponding delays to be 86 executed and the corresponding delays to be
92 seen. 87 seen.
93 88
94 General format of the getdelays command:: 89 General format of the getdelays command::
95 90
96 getdelays [-dilv] [-t tgid] [-p pid] !! 91 getdelays [-t tgid] [-p pid] [-c cmd...]
>> 92
97 93
98 Get delays, since system boot, for pid 10:: 94 Get delays, since system boot, for pid 10::
99 95
100 # ./getdelays -d -p 10 !! 96 # ./getdelays -p 10
101 (output similar to next case) 97 (output similar to next case)
102 98
103 Get sum of delays, since system boot, for all 99 Get sum of delays, since system boot, for all pids with tgid 5::
104 100
105 # ./getdelays -d -t 5 !! 101 # ./getdelays -t 5
106 print delayacct stats ON !! 102
107 TGID 5 <<
108 <<
109 <<
110 CPU count real total <<
111 8 7000000 <<
112 IO count delay total <<
113 0 0 0 <<
114 SWAP count delay total <<
115 0 0 <<
116 RECLAIM count delay total <<
117 0 0 0 <<
118 THRASHING count delay total <<
119 0 0 <<
120 COMPACT count delay total <<
121 0 0 <<
122 WPCOPY count delay total <<
123 0 0 <<
124 IRQ count delay total <<
125 0 0 <<
126 <<
127 Get IO accounting for pid 1, it works only wit <<
128 <<
129 # ./getdelays -i -p 1 <<
130 printing IO accounting <<
131 linuxrc: read=65536, write=0, cancelle <<
132 103
133 The above command can be used with -v to get m !! 104 CPU count real total virtual total delay total
>> 105 7876 92005750 100000000 24001500
>> 106 IO count delay total
>> 107 0 0
>> 108 SWAP count delay total
>> 109 0 0
>> 110 RECLAIM count delay total
>> 111 0 0
>> 112
>> 113 Get delays seen in executing a given simple command::
>> 114
>> 115 # ./getdelays -c ls /
>> 116
>> 117 bin data1 data3 data5 dev home media opt root srv sys usr
>> 118 boot data2 data4 data6 etc lib mnt proc sbin subdomain tmp var
>> 119
>> 120
>> 121 CPU count real total virtual total delay total
>> 122 6 4000250 4000000 0
>> 123 IO count delay total
>> 124 0 0
>> 125 SWAP count delay total
>> 126 0 0
>> 127 RECLAIM count delay total
>> 128 0 0
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.