1 RCU Torture Test Operation
4 CONFIG_RCU_TORTURE_TEST
6 The CONFIG_RCU_TORTURE_TEST config option is available for all RCU
7 implementations. It creates an rcutorture kernel module that can
8 be loaded to run a torture test. The test periodically outputs
9 status messages via printk(), which can be examined via the dmesg
10 command (perhaps grepping for "torture"). The test is started
11 when the module is loaded, and stops when the module is unloaded.
13 CONFIG_RCU_TORTURE_TEST_RUNNABLE
15 It is also possible to specify CONFIG_RCU_TORTURE_TEST=y, which will
16 result in the tests being loaded into the base kernel. In this case,
17 the CONFIG_RCU_TORTURE_TEST_RUNNABLE config option is used to specify
18 whether the RCU torture tests are to be started immediately during
19 boot or whether the /proc/sys/kernel/rcutorture_runnable file is used
20 to enable them. This /proc file can be used to repeatedly pause and
21 restart the tests, regardless of the initial state specified by the
22 CONFIG_RCU_TORTURE_TEST_RUNNABLE config option.
24 You 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
26 this can sometimes be useful in finding boot-time bugs.
31 This module has the following parameters:
33 fqs_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.
39 fqs_holdoff Holdoff time (in microseconds) between consecutive calls
40 to force_quiescent_state() within a burst.
42 fqs_stutter Wait time (in seconds) between consecutive bursts
43 of calls to force_quiescent_state().
45 irqreader 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.)
50 nfakewriters 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.
59 nreaders 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.
65 The number of seconds between each attempt to execute a
66 randomly selected CPU-hotplug operation. Defaults to
67 zero, which disables CPU hotplugging. In HOTPLUG_CPU=n
68 kernels, rcutorture will silently refuse to do any
69 CPU-hotplug operations regardless of what value is
70 specified for onoff_interval.
73 The number of seconds to keep the test threads affinitied
74 to a particular subset of the CPUs, defaults to 3 seconds.
75 Used in conjunction with test_no_idle_hz.
77 shutdown_secs The number of seconds to run the test before terminating
78 the test and powering off the system. The default is
79 zero, which disables test termination and system shutdown.
80 This capability is useful for automated testing.
82 stat_interval The number of seconds between output of torture
83 statistics (via printk()). Regardless of the interval,
84 statistics are printed when the module is unloaded.
85 Setting the interval to zero causes the statistics to
86 be printed -only- when the module is unloaded, and this
89 stutter The length of time to run the test before pausing for this
90 same period of time. Defaults to "stutter=5", so as
91 to run and pause for (roughly) five-second intervals.
92 Specifying "stutter=0" causes the test to run continuously
93 without pausing, which is the old default behavior.
95 test_boost Whether or not to test the ability of RCU to do priority
96 boosting. Defaults to "test_boost=1", which performs
97 RCU priority-inversion testing only if the selected
98 RCU implementation supports priority boosting. Specifying
99 "test_boost=0" never performs RCU priority-inversion
100 testing. Specifying "test_boost=2" performs RCU
101 priority-inversion testing even if the selected RCU
102 implementation does not support RCU priority boosting,
103 which can be used to test rcutorture's ability to
104 carry out RCU priority-inversion testing.
107 The number of seconds in an RCU priority-inversion test
108 cycle. Defaults to "test_boost_interval=7". It is
109 usually wise for this value to be relatively prime to
110 the value selected for "stutter".
113 The number of seconds to do RCU priority-inversion testing
114 within any given "test_boost_interval". Defaults to
115 "test_boost_duration=4".
117 test_no_idle_hz Whether or not to test the ability of RCU to operate in
118 a kernel that disables the scheduling-clock interrupt to
119 idle CPUs. Boolean parameter, "1" to test, "0" otherwise.
120 Defaults to omitting this test.
122 torture_type The type of RCU to test, with string values as follows:
124 "rcu": rcu_read_lock(), rcu_read_unlock() and call_rcu().
126 "rcu_sync": rcu_read_lock(), rcu_read_unlock(), and
129 "rcu_expedited": rcu_read_lock(), rcu_read_unlock(), and
130 synchronize_rcu_expedited().
132 "rcu_bh": rcu_read_lock_bh(), rcu_read_unlock_bh(), and
135 "rcu_bh_sync": rcu_read_lock_bh(), rcu_read_unlock_bh(),
136 and synchronize_rcu_bh().
138 "rcu_bh_expedited": rcu_read_lock_bh(), rcu_read_unlock_bh(),
139 and synchronize_rcu_bh_expedited().
141 "srcu": srcu_read_lock(), srcu_read_unlock() and
144 "srcu_expedited": srcu_read_lock(), srcu_read_unlock() and
145 synchronize_srcu_expedited().
147 "sched": preempt_disable(), preempt_enable(), and
150 "sched_sync": preempt_disable(), preempt_enable(), and
153 "sched_expedited": preempt_disable(), preempt_enable(), and
154 synchronize_sched_expedited().
158 verbose Enable debug printk()s. Default is disabled.
163 The statistics output is as follows:
165 rcu-torture:--- Start of test: nreaders=16 nfakewriters=4 stat_interval=30 verbose=0 test_no_idle_hz=1 shuffle_interval=3 stutter=5 irqreader=1 fqs_duration=0 fqs_holdoff=0 fqs_stutter=3 test_boost=1/0 test_boost_interval=7 test_boost_duration=4
166 rcu-torture: rtc: (null) ver: 155441 tfle: 0 rta: 155441 rtaf: 8884 rtf: 155440 rtmbe: 0 rtbke: 0 rtbre: 0 rtbf: 0 rtb: 0 nt: 3055767
167 rcu-torture: Reader Pipe: 727860534 34213 0 0 0 0 0 0 0 0 0
168 rcu-torture: Reader Batch: 727877838 17003 0 0 0 0 0 0 0 0 0
169 rcu-torture: Free-Block Circulation: 155440 155440 155440 155440 155440 155440 155440 155440 155440 155440 0
170 rcu-torture:--- End of test: SUCCESS: nreaders=16 nfakewriters=4 stat_interval=30 verbose=0 test_no_idle_hz=1 shuffle_interval=3 stutter=5 irqreader=1 fqs_duration=0 fqs_holdoff=0 fqs_stutter=3 test_boost=1/0 test_boost_interval=7 test_boost_duration=4
172 The command "dmesg | grep torture:" will extract this information on
173 most systems. On more esoteric configurations, it may be necessary to
174 use other commands to access the output of the printk()s used by
175 the RCU torture test. The printk()s use KERN_ALERT, so they should
178 The first and last lines show the rcutorture module parameters, and the
179 last line shows either "SUCCESS" or "FAILURE", based on rcutorture's
180 automatic determination as to whether RCU operated correctly.
182 The entries are as follows:
184 o "rtc": The hexadecimal address of the structure currently visible
187 o "ver": The number of times since boot that the RCU writer task
188 has changed the structure visible to readers.
190 o "tfle": If non-zero, indicates that the "torture freelist"
191 containing structures to be placed into the "rtc" area is empty.
192 This condition is important, since it can fool you into thinking
193 that RCU is working when it is not. :-/
195 o "rta": Number of structures allocated from the torture freelist.
197 o "rtaf": Number of allocations from the torture freelist that have
198 failed due to the list being empty. It is not unusual for this
199 to be non-zero, but it is bad for it to be a large fraction of
200 the value indicated by "rta".
202 o "rtf": Number of frees into the torture freelist.
204 o "rtmbe": A non-zero value indicates that rcutorture believes that
205 rcu_assign_pointer() and rcu_dereference() are not working
206 correctly. This value should be zero.
208 o "rtbke": rcutorture was unable to create the real-time kthreads
209 used to force RCU priority inversion. This value should be zero.
211 o "rtbre": Although rcutorture successfully created the kthreads
212 used to force RCU priority inversion, it was unable to set them
213 to the real-time priority level of 1. This value should be zero.
215 o "rtbf": The number of times that RCU priority boosting failed
216 to resolve RCU priority inversion.
218 o "rtb": The number of times that rcutorture attempted to force
219 an RCU priority inversion condition. If you are testing RCU
220 priority boosting via the "test_boost" module parameter, this
221 value should be non-zero.
223 o "nt": The number of times rcutorture ran RCU read-side code from
224 within a timer handler. This value should be non-zero only
225 if you specified the "irqreader" module parameter.
227 o "Reader Pipe": Histogram of "ages" of structures seen by readers.
228 If any entries past the first two are non-zero, RCU is broken.
229 And rcutorture prints the error flag string "!!!" to make sure
230 you notice. The age of a newly allocated structure is zero,
231 it becomes one when removed from reader visibility, and is
232 incremented once per grace period subsequently -- and is freed
233 after passing through (RCU_TORTURE_PIPE_LEN-2) grace periods.
235 The output displayed above was taken from a correctly working
236 RCU. If you want to see what it looks like when broken, break
239 o "Reader Batch": Another histogram of "ages" of structures seen
240 by readers, but in terms of counter flips (or batches) rather
241 than in terms of grace periods. The legal number of non-zero
242 entries is again two. The reason for this separate view is that
243 it is sometimes easier to get the third entry to show up in the
244 "Reader Batch" list than in the "Reader Pipe" list.
246 o "Free-Block Circulation": Shows the number of torture structures
247 that have reached a given point in the pipeline. The first element
248 should closely correspond to the number of structures allocated,
249 the second to the number that have been removed from reader view,
250 and all but the last remaining to the corresponding number of
251 passes through a grace period. The last entry should be zero,
252 as it is only incremented if a torture structure's counter
253 somehow gets incremented farther than it should.
255 Different implementations of RCU can provide implementation-specific
256 additional information. For example, SRCU provides the following
259 srcu-torture: per-CPU(idx=1): 0(0,1) 1(0,1) 2(0,0) 3(0,1)
261 This line shows the per-CPU counter state. The numbers in parentheses are
262 the values of the "old" and "current" counters for the corresponding CPU.
263 The "idx" value maps the "old" and "current" values to the underlying
264 array, and is useful for debugging.
269 The following script may be used to torture RCU:
276 dmesg | grep torture:
278 The output can be manually inspected for the error flag of "!!!".
279 One could of course create a more elaborate script that automatically
280 checked for such errors. The "rmmod" command forces a "SUCCESS" or
281 "FAILURE" indication to be printk()ed.