1================
2Delay accounting
3================
4
5Tasks encounter delays in execution when they wait
6for some kernel resource to become available e.g. a
7runnable task may wait for a free CPU to run on.
8
9The per-task delay accounting functionality measures
10the delays experienced by a task while
11
12a) waiting for a CPU (while being runnable)
13b) completion of synchronous block I/O initiated by the task
14c) swapping in pages
15d) memory reclaim
16e) thrashing
17f) direct compact
18g) write-protect copy
19h) IRQ/SOFTIRQ
20
21and makes these statistics available to userspace through
22the taskstats interface.
23
24Such delays provide feedback for setting a task's cpu priority,
25io priority and rss limit values appropriately. Long delays for
26important tasks could be a trigger for raising its corresponding priority.
27
28The functionality, through its use of the taskstats interface, also provides
29delay statistics aggregated for all tasks (or threads) belonging to a
30thread group (corresponding to a traditional Unix process). This is a commonly
31needed aggregation that is more efficiently done by the kernel.
32
33Userspace utilities, particularly resource management applications, can also
34aggregate delay statistics into arbitrary groups. To enable this, delay
35statistics of a task are available both during its lifetime as well as on its
36exit, ensuring continuous and complete monitoring can be done.
37
38
39Interface
40---------
41
42Delay accounting uses the taskstats interface which is described
43in detail in a separate document in this directory. Taskstats returns a
44generic data structure to userspace corresponding to per-pid and per-tgid
45statistics. The delay accounting functionality populates specific fields of
46this structure. See
47
48     include/uapi/linux/taskstats.h
49
50for a description of the fields pertaining to delay accounting.
51It will generally be in the form of counters returning the cumulative
52delay seen for cpu, sync block I/O, swapin, memory reclaim, thrash page
53cache, direct compact, write-protect copy, IRQ/SOFTIRQ etc.
54
55Taking the difference of two successive readings of a given
56counter (say cpu_delay_total) for a task will give the delay
57experienced by the task waiting for the corresponding resource
58in that interval.
59
60When a task exits, records containing the per-task statistics
61are sent to userspace without requiring a command. If it is the last exiting
62task of a thread group, the per-tgid statistics are also sent. More details
63are given in the taskstats interface description.
64
65The getdelays.c userspace utility in tools/accounting directory allows simple
66commands to be run and the corresponding delay statistics to be displayed. It
67also serves as an example of using the taskstats interface.
68
69Usage
70-----
71
72Compile the kernel with::
73
74	CONFIG_TASK_DELAY_ACCT=y
75	CONFIG_TASKSTATS=y
76
77Delay accounting is disabled by default at boot up.
78To enable, add::
79
80   delayacct
81
82to the kernel boot options. The rest of the instructions below assume this has
83been done. Alternatively, use sysctl kernel.task_delayacct to switch the state
84at runtime. Note however that only tasks started after enabling it will have
85delayacct information.
86
87After the system has booted up, use a utility
88similar to  getdelays.c to access the delays
89seen by a given task or a task group (tgid).
90The utility also allows a given command to be
91executed and the corresponding delays to be
92seen.
93
94General format of the getdelays command::
95
96	getdelays [-dilv] [-t tgid] [-p pid]
97
98Get delays, since system boot, for pid 10::
99
100	# ./getdelays -d -p 10
101	(output similar to next case)
102
103Get sum of delays, since system boot, for all pids with tgid 5::
104
105	# ./getdelays -d -t 5
106	print delayacct stats ON
107	TGID	5
108
109
110	CPU             count     real total  virtual total    delay total  delay average
111	                    8        7000000        6872122        3382277          0.423ms
112	IO              count    delay total  delay average
113                   0              0          0.000ms
114	SWAP            count    delay total  delay average
115                       0              0          0.000ms
116	RECLAIM         count    delay total  delay average
117                   0              0          0.000ms
118	THRASHING       count    delay total  delay average
119                       0              0          0.000ms
120	COMPACT         count    delay total  delay average
121                       0              0          0.000ms
122	WPCOPY          count    delay total  delay average
123                       0              0          0.000ms
124	IRQ             count    delay total  delay average
125                       0              0          0.000ms
126
127Get IO accounting for pid 1, it works only with -p::
128
129	# ./getdelays -i -p 1
130	printing IO accounting
131	linuxrc: read=65536, write=0, cancelled_write=0
132
133The above command can be used with -v to get more debug information.
134