blob: 0e50bc2aa1e2e4f4682f5b5540a49987bfb00884 [file] [log] [blame]
Paul E. McKenneya241ec62005-10-30 15:03:12 -08001RCU Torture Test Operation
2
3
4CONFIG_RCU_TORTURE_TEST
5
6The CONFIG_RCU_TORTURE_TEST config option is available for all RCU
7implementations. It creates an rcutorture kernel module that can
8be loaded to run a torture test. The test periodically outputs
9status messages via printk(), which can be examined via the dmesg
Paul E. McKenney72e9bb52006-06-27 02:54:03 -070010command (perhaps grepping for "torture"). The test is started
Paul E. McKenneya241ec62005-10-30 15:03:12 -080011when the module is loaded, and stops when the module is unloaded.
12
Paul E. McKenney31a72bc2008-06-18 09:26:49 -070013CONFIG_RCU_TORTURE_TEST_RUNNABLE
14
15It is also possible to specify CONFIG_RCU_TORTURE_TEST=y, which will
16result in the tests being loaded into the base kernel. In this case,
17the CONFIG_RCU_TORTURE_TEST_RUNNABLE config option is used to specify
18whether the RCU torture tests are to be started immediately during
19boot or whether the /proc/sys/kernel/rcutorture_runnable file is used
20to enable them. This /proc file can be used to repeatedly pause and
21restart the tests, regardless of the initial state specified by the
22CONFIG_RCU_TORTURE_TEST_RUNNABLE config option.
23
24You will normally -not- want to start the RCU torture tests during boot
25(and thus the default is CONFIG_RCU_TORTURE_TEST_RUNNABLE=n), but doing
26this can sometimes be useful in finding boot-time bugs.
Paul E. McKenneya241ec62005-10-30 15:03:12 -080027
28
29MODULE PARAMETERS
30
31This module has the following parameters:
32
Paul E. McKenney4c540052010-01-14 16:10:57 -080033fqs_duration Duration (in microseconds) of artificially induced bursts
34 of force_quiescent_state() invocations. In RCU
35 implementations having force_quiescent_state(), these
36 bursts help force races between forcing a given grace
37 period and that grace period ending on its own.
38
39fqs_holdoff Holdoff time (in microseconds) between consecutive calls
40 to force_quiescent_state() within a burst.
41
42fqs_stutter Wait time (in seconds) between consecutive bursts
43 of calls to force_quiescent_state().
44
Paul E. McKenney0729fbf2008-06-25 12:24:52 -070045irqreaders Says to invoke RCU readers from irq level. This is currently
46 done via timers. Defaults to "1" for variants of RCU that
47 permit this. (Or, more accurately, variants of RCU that do
48 -not- permit this know to ignore this variable.)
Paul E. McKenneya241ec62005-10-30 15:03:12 -080049
Josh Triplettb772e1d2006-10-04 02:17:13 -070050nfakewriters This is the number of RCU fake writer threads to run. Fake
51 writer threads repeatedly use the synchronous "wait for
52 current readers" function of the interface selected by
53 torture_type, with a delay between calls to allow for various
54 different numbers of writers running in parallel.
55 nfakewriters defaults to 4, which provides enough parallelism
56 to trigger special cases caused by multiple writers, such as
57 the synchronize_srcu() early return optimization.
58
Paul E. McKenney0729fbf2008-06-25 12:24:52 -070059nreaders This is the number of RCU reading threads supported.
60 The default is twice the number of CPUs. Why twice?
61 To properly exercise RCU implementations with preemptible
62 read-side critical sections.
63
64shuffle_interval
65 The number of seconds to keep the test threads affinitied
66 to a particular subset of the CPUs, defaults to 3 seconds.
67 Used in conjunction with test_no_idle_hz.
68
Paul E. McKenneya241ec62005-10-30 15:03:12 -080069stat_interval The number of seconds between output of torture
70 statistics (via printk()). Regardless of the interval,
71 statistics are printed when the module is unloaded.
72 Setting the interval to zero causes the statistics to
73 be printed -only- when the module is unloaded, and this
74 is the default.
75
Paul E. McKenneyd120f652008-06-18 05:21:44 -070076stutter The length of time to run the test before pausing for this
77 same period of time. Defaults to "stutter=5", so as
78 to run and pause for (roughly) five-second intervals.
79 Specifying "stutter=0" causes the test to run continuously
80 without pausing, which is the old default behavior.
81
Paul E. McKenney29766f12006-06-27 02:54:02 -070082test_no_idle_hz Whether or not to test the ability of RCU to operate in
83 a kernel that disables the scheduling-clock interrupt to
84 idle CPUs. Boolean parameter, "1" to test, "0" otherwise.
Paul E. McKenneyf85d6c72008-01-25 21:08:25 +010085 Defaults to omitting this test.
Paul E. McKenney29766f12006-06-27 02:54:02 -070086
Josh Triplett20d2e422006-10-04 02:17:15 -070087torture_type The type of RCU to test: "rcu" for the rcu_read_lock() API,
88 "rcu_sync" for rcu_read_lock() with synchronous reclamation,
Josh Triplett11a14702006-10-04 02:17:16 -070089 "rcu_bh" for the rcu_read_lock_bh() API, "rcu_bh_sync" for
Josh Triplett4b6c2cc2006-10-04 02:17:16 -070090 rcu_read_lock_bh() with synchronous reclamation, "srcu" for
Paul E. McKenney240ebbf2009-06-25 09:08:18 -070091 the "srcu_read_lock()" API, "sched" for the use of
92 preempt_disable() together with synchronize_sched(),
93 and "sched_expedited" for the use of preempt_disable()
94 with synchronize_sched_expedited().
Paul E. McKenney72e9bb52006-06-27 02:54:03 -070095
Paul E. McKenneya241ec62005-10-30 15:03:12 -080096verbose Enable debug printk()s. Default is disabled.
97
98
99OUTPUT
100
101The statistics output is as follows:
102
Paul E. McKenney72e9bb52006-06-27 02:54:03 -0700103 rcu-torture: --- Start of test: nreaders=16 stat_interval=0 verbose=0
104 rcu-torture: rtc: 0000000000000000 ver: 1916 tfle: 0 rta: 1916 rtaf: 0 rtf: 1915
105 rcu-torture: Reader Pipe: 1466408 9747 0 0 0 0 0 0 0 0 0
106 rcu-torture: Reader Batch: 1464477 11678 0 0 0 0 0 0 0 0
107 rcu-torture: Free-Block Circulation: 1915 1915 1915 1915 1915 1915 1915 1915 1915 1915 0
108 rcu-torture: --- End of test
Paul E. McKenneya241ec62005-10-30 15:03:12 -0800109
Paul E. McKenney72e9bb52006-06-27 02:54:03 -0700110The command "dmesg | grep torture:" will extract this information on
Paul E. McKenneya241ec62005-10-30 15:03:12 -0800111most systems. On more esoteric configurations, it may be necessary to
112use other commands to access the output of the printk()s used by
113the RCU torture test. The printk()s use KERN_ALERT, so they should
114be evident. ;-)
115
116The entries are as follows:
117
Paul E. McKenneya241ec62005-10-30 15:03:12 -0800118o "rtc": The hexadecimal address of the structure currently visible
119 to readers.
120
121o "ver": The number of times since boot that the rcutw writer task
122 has changed the structure visible to readers.
123
124o "tfle": If non-zero, indicates that the "torture freelist"
125 containing structure to be placed into the "rtc" area is empty.
126 This condition is important, since it can fool you into thinking
127 that RCU is working when it is not. :-/
128
129o "rta": Number of structures allocated from the torture freelist.
130
131o "rtaf": Number of allocations from the torture freelist that have
132 failed due to the list being empty.
133
134o "rtf": Number of frees into the torture freelist.
135
136o "Reader Pipe": Histogram of "ages" of structures seen by readers.
137 If any entries past the first two are non-zero, RCU is broken.
138 And rcutorture prints the error flag string "!!!" to make sure
139 you notice. The age of a newly allocated structure is zero,
140 it becomes one when removed from reader visibility, and is
141 incremented once per grace period subsequently -- and is freed
142 after passing through (RCU_TORTURE_PIPE_LEN-2) grace periods.
143
144 The output displayed above was taken from a correctly working
145 RCU. If you want to see what it looks like when broken, break
146 it yourself. ;-)
147
148o "Reader Batch": Another histogram of "ages" of structures seen
149 by readers, but in terms of counter flips (or batches) rather
150 than in terms of grace periods. The legal number of non-zero
Paul E. McKenneyf85d6c72008-01-25 21:08:25 +0100151 entries is again two. The reason for this separate view is that
152 it is sometimes easier to get the third entry to show up in the
Paul E. McKenneya241ec62005-10-30 15:03:12 -0800153 "Reader Batch" list than in the "Reader Pipe" list.
154
155o "Free-Block Circulation": Shows the number of torture structures
156 that have reached a given point in the pipeline. The first element
157 should closely correspond to the number of structures allocated,
158 the second to the number that have been removed from reader view,
159 and all but the last remaining to the corresponding number of
160 passes through a grace period. The last entry should be zero,
161 as it is only incremented if a torture structure's counter
162 somehow gets incremented farther than it should.
163
Paul E. McKenneyb2896d22006-10-04 02:17:03 -0700164Different implementations of RCU can provide implementation-specific
165additional information. For example, SRCU provides the following:
166
167 srcu-torture: rtc: f8cf46a8 ver: 355 tfle: 0 rta: 356 rtaf: 0 rtf: 346 rtmbe: 0
168 srcu-torture: Reader Pipe: 559738 939 0 0 0 0 0 0 0 0 0
169 srcu-torture: Reader Batch: 560434 243 0 0 0 0 0 0 0 0
170 srcu-torture: Free-Block Circulation: 355 354 353 352 351 350 349 348 347 346 0
171 srcu-torture: per-CPU(idx=1): 0(0,1) 1(0,1) 2(0,0) 3(0,1)
172
173The first four lines are similar to those for RCU. The last line shows
174the per-CPU counter state. The numbers in parentheses are the values
175of the "old" and "current" counters for the corresponding CPU. The
176"idx" value maps the "old" and "current" values to the underlying array,
177and is useful for debugging.
178
Paul E. McKenney240ebbf2009-06-25 09:08:18 -0700179Similarly, sched_expedited RCU provides the following:
180
181 sched_expedited-torture: rtc: d0000000016c1880 ver: 1090796 tfle: 0 rta: 1090796 rtaf: 0 rtf: 1090787 rtmbe: 0 nt: 27713319
182 sched_expedited-torture: Reader Pipe: 12660320201 95875 0 0 0 0 0 0 0 0 0
183 sched_expedited-torture: Reader Batch: 12660424885 0 0 0 0 0 0 0 0 0 0
184 sched_expedited-torture: Free-Block Circulation: 1090795 1090795 1090794 1090793 1090792 1090791 1090790 1090789 1090788 1090787 0
185 state: -1 / 0:0 3:0 4:0
186
187As before, the first four lines are similar to those for RCU.
188The last line shows the task-migration state. The first number is
189-1 if synchronize_sched_expedited() is idle, -2 if in the process of
190posting wakeups to the migration kthreads, and N when waiting on CPU N.
191Each of the colon-separated fields following the "/" is a CPU:state pair.
192Valid states are "0" for idle, "1" for waiting for quiescent state,
193"2" for passed through quiescent state, and "3" when a race with a
194CPU-hotplug event forces use of the synchronize_sched() primitive.
195
Paul E. McKenneya241ec62005-10-30 15:03:12 -0800196
197USAGE
198
199The following script may be used to torture RCU:
200
201 #!/bin/sh
202
203 modprobe rcutorture
204 sleep 100
205 rmmod rcutorture
Paul E. McKenney72e9bb52006-06-27 02:54:03 -0700206 dmesg | grep torture:
Paul E. McKenneya241ec62005-10-30 15:03:12 -0800207
208The output can be manually inspected for the error flag of "!!!".
209One could of course create a more elaborate script that automatically
Paul E. McKenney29766f12006-06-27 02:54:02 -0700210checked for such errors. The "rmmod" command forces a "SUCCESS" or
211"FAILURE" indication to be printk()ed.