Documentation/RCU/torture.txt

   1 RCU Torture Test Operation
   2
   3
   4 CONFIG_RCU_TORTURE_TEST
   5
   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.
  12
  13 CONFIG_RCU_TORTURE_TEST_RUNNABLE
  14
  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.
  23
  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.
  27
  28
  29 MODULE PARAMETERS
  30
  31 This module has the following parameters:
  32
  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.
  38
  39 fqs_holdoff     Holdoff time (in microseconds) between consecutive calls
  40                 to force_quiescent_state() within a burst.
  41
  42 fqs_stutter     Wait time (in seconds) between consecutive bursts
  43                 of calls to force_quiescent_state().
  44
  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.)
  49
  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.
  58
  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.
  63
  64 shuffle_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
  69 shutdown_secs   The number of seconds to run the test before terminating
  70                 the test and powering off the system.  The default is
  71                 zero, which disables test termination and system shutdown.
  72                 This capability is useful for automated testing.
  73
  74 stat_interval   The number of seconds between output of torture
  75                 statistics (via printk()).  Regardless of the interval,
  76                 statistics are printed when the module is unloaded.
  77                 Setting the interval to zero causes the statistics to
  78                 be printed -only- when the module is unloaded, and this
  79                 is the default.
  80
  81 stutter         The length of time to run the test before pausing for this
  82                 same period of time.  Defaults to "stutter=5", so as
  83                 to run and pause for (roughly) five-second intervals.
  84                 Specifying "stutter=0" causes the test to run continuously
  85                 without pausing, which is the old default behavior.
  86
  87 test_boost      Whether or not to test the ability of RCU to do priority
  88                 boosting.  Defaults to "test_boost=1", which performs
  89                 RCU priority-inversion testing only if the selected
  90                 RCU implementation supports priority boosting.  Specifying
  91                 "test_boost=0" never performs RCU priority-inversion
  92                 testing.  Specifying "test_boost=2" performs RCU
  93                 priority-inversion testing even if the selected RCU
  94                 implementation does not support RCU priority boosting,
  95                 which can be used to test rcutorture's ability to
  96                 carry out RCU priority-inversion testing.
  97
  98 test_boost_interval
  99                 The number of seconds in an RCU priority-inversion test
 100                 cycle.  Defaults to "test_boost_interval=7".  It is
 101                 usually wise for this value to be relatively prime to
 102                 the value selected for "stutter".
 103
 104 test_boost_duration
 105                 The number of seconds to do RCU priority-inversion testing
 106                 within any given "test_boost_interval".  Defaults to
 107                 "test_boost_duration=4".
 108
 109 test_no_idle_hz Whether or not to test the ability of RCU to operate in
 110                 a kernel that disables the scheduling-clock interrupt to
 111                 idle CPUs.  Boolean parameter, "1" to test, "0" otherwise.
 112                 Defaults to omitting this test.
 113
 114 torture_type    The type of RCU to test, with string values as follows:
 115
 116                 "rcu":  rcu_read_lock(), rcu_read_unlock() and call_rcu().
 117
 118                 "rcu_sync":  rcu_read_lock(), rcu_read_unlock(), and
 119                         synchronize_rcu().
 120
 121                 "rcu_expedited": rcu_read_lock(), rcu_read_unlock(), and
 122                         synchronize_rcu_expedited().
 123
 124                 "rcu_bh": rcu_read_lock_bh(), rcu_read_unlock_bh(), and
 125                         call_rcu_bh().
 126
 127                 "rcu_bh_sync": rcu_read_lock_bh(), rcu_read_unlock_bh(),
 128                         and synchronize_rcu_bh().
 129
 130                 "rcu_bh_expedited": rcu_read_lock_bh(), rcu_read_unlock_bh(),
 131                         and synchronize_rcu_bh_expedited().
 132
 133                 "srcu": srcu_read_lock(), srcu_read_unlock() and
 134                         synchronize_srcu().
 135
 136                 "srcu_expedited": srcu_read_lock(), srcu_read_unlock() and
 137                         synchronize_srcu_expedited().
 138
 139                 "sched": preempt_disable(), preempt_enable(), and
 140                         call_rcu_sched().
 141
 142                 "sched_sync": preempt_disable(), preempt_enable(), and
 143                         synchronize_sched().
 144
 145                 "sched_expedited": preempt_disable(), preempt_enable(), and
 146                         synchronize_sched_expedited().
 147
 148                 Defaults to "rcu".
 149
 150 verbose         Enable debug printk()s.  Default is disabled.
 151
 152
 153 OUTPUT
 154
 155 The statistics output is as follows:
 156
 157         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
 158         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
 159         rcu-torture: Reader Pipe:  727860534 34213 0 0 0 0 0 0 0 0 0
 160         rcu-torture: Reader Batch:  727877838 17003 0 0 0 0 0 0 0 0 0
 161         rcu-torture: Free-Block Circulation:  155440 155440 155440 155440 155440 155440 155440 155440 155440 155440 0
 162         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
 163
 164 The command "dmesg | grep torture:" will extract this information on
 165 most systems.  On more esoteric configurations, it may be necessary to
 166 use other commands to access the output of the printk()s used by
 167 the RCU torture test.  The printk()s use KERN_ALERT, so they should
 168 be evident.  ;-)
 169
 170 The first and last lines show the rcutorture module parameters, and the
 171 last line shows either "SUCCESS" or "FAILURE", based on rcutorture's
 172 automatic determination as to whether RCU operated correctly.
 173
 174 The entries are as follows:
 175
 176 o       "rtc": The hexadecimal address of the structure currently visible
 177         to readers.
 178
 179 o       "ver": The number of times since boot that the RCU writer task
 180         has changed the structure visible to readers.
 181
 182 o       "tfle": If non-zero, indicates that the "torture freelist"
 183         containing structures to be placed into the "rtc" area is empty.
 184         This condition is important, since it can fool you into thinking
 185         that RCU is working when it is not.  :-/
 186
 187 o       "rta": Number of structures allocated from the torture freelist.
 188
 189 o       "rtaf": Number of allocations from the torture freelist that have
 190         failed due to the list being empty.  It is not unusual for this
 191         to be non-zero, but it is bad for it to be a large fraction of
 192         the value indicated by "rta".
 193
 194 o       "rtf": Number of frees into the torture freelist.
 195
 196 o       "rtmbe": A non-zero value indicates that rcutorture believes that
 197         rcu_assign_pointer() and rcu_dereference() are not working
 198         correctly.  This value should be zero.
 199
 200 o       "rtbke": rcutorture was unable to create the real-time kthreads
 201         used to force RCU priority inversion.  This value should be zero.
 202
 203 o       "rtbre": Although rcutorture successfully created the kthreads
 204         used to force RCU priority inversion, it was unable to set them
 205         to the real-time priority level of 1.  This value should be zero.
 206
 207 o       "rtbf": The number of times that RCU priority boosting failed
 208         to resolve RCU priority inversion.
 209
 210 o       "rtb": The number of times that rcutorture attempted to force
 211         an RCU priority inversion condition.  If you are testing RCU
 212         priority boosting via the "test_boost" module parameter, this
 213         value should be non-zero.
 214
 215 o       "nt": The number of times rcutorture ran RCU read-side code from
 216         within a timer handler.  This value should be non-zero only
 217         if you specified the "irqreader" module parameter.
 218
 219 o       "Reader Pipe": Histogram of "ages" of structures seen by readers.
 220         If any entries past the first two are non-zero, RCU is broken.
 221         And rcutorture prints the error flag string "!!!" to make sure
 222         you notice.  The age of a newly allocated structure is zero,
 223         it becomes one when removed from reader visibility, and is
 224         incremented once per grace period subsequently -- and is freed
 225         after passing through (RCU_TORTURE_PIPE_LEN-2) grace periods.
 226
 227         The output displayed above was taken from a correctly working
 228         RCU.  If you want to see what it looks like when broken, break
 229         it yourself.  ;-)
 230
 231 o       "Reader Batch": Another histogram of "ages" of structures seen
 232         by readers, but in terms of counter flips (or batches) rather
 233         than in terms of grace periods.  The legal number of non-zero
 234         entries is again two.  The reason for this separate view is that
 235         it is sometimes easier to get the third entry to show up in the
 236         "Reader Batch" list than in the "Reader Pipe" list.
 237
 238 o       "Free-Block Circulation": Shows the number of torture structures
 239         that have reached a given point in the pipeline.  The first element
 240         should closely correspond to the number of structures allocated,
 241         the second to the number that have been removed from reader view,
 242         and all but the last remaining to the corresponding number of
 243         passes through a grace period.  The last entry should be zero,
 244         as it is only incremented if a torture structure's counter
 245         somehow gets incremented farther than it should.
 246
 247 Different implementations of RCU can provide implementation-specific
 248 additional information.  For example, SRCU provides the following
 249 additional line:
 250
 251         srcu-torture: per-CPU(idx=1): 0(0,1) 1(0,1) 2(0,0) 3(0,1)
 252
 253 This line shows the per-CPU counter state.  The numbers in parentheses are
 254 the values of the "old" and "current" counters for the corresponding CPU.
 255 The "idx" value maps the "old" and "current" values to the underlying
 256 array, and is useful for debugging.
 257
 258
 259 USAGE
 260
 261 The following script may be used to torture RCU:
 262
 263         #!/bin/sh
 264
 265         modprobe rcutorture
 266         sleep 100
 267         rmmod rcutorture
 268         dmesg | grep torture:
 269
 270 The output can be manually inspected for the error flag of "!!!".
 271 One could of course create a more elaborate script that automatically
 272 checked for such errors.  The "rmmod" command forces a "SUCCESS" or
 273 "FAILURE" indication to be printk()ed.