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 Module parameters are prefixed by "rcutorture." in
  14 Documentation/admin-guide/kernel-parameters.txt.
  15
  16 OUTPUT
  17
  18 The statistics output is as follows:
  19
  20         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
  21         rcu-torture: rtc:           (null) ver: 155441 tfle: 0 rta: 155441 rtaf: 8884 rtf: 155440 rtmbe: 0 rtbe: 0 rtbke: 0 rtbre: 0 rtbf: 0 rtb: 0 nt: 3055767
  22         rcu-torture: Reader Pipe:  727860534 34213 0 0 0 0 0 0 0 0 0
  23         rcu-torture: Reader Batch:  727877838 17003 0 0 0 0 0 0 0 0 0
  24         rcu-torture: Free-Block Circulation:  155440 155440 155440 155440 155440 155440 155440 155440 155440 155440 0
  25         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
  26
  27 The command "dmesg | grep torture:" will extract this information on
  28 most systems.  On more esoteric configurations, it may be necessary to
  29 use other commands to access the output of the printk()s used by
  30 the RCU torture test.  The printk()s use KERN_ALERT, so they should
  31 be evident.  ;-)
  32
  33 The first and last lines show the rcutorture module parameters, and the
  34 last line shows either "SUCCESS" or "FAILURE", based on rcutorture's
  35 automatic determination as to whether RCU operated correctly.
  36
  37 The entries are as follows:
  38
  39 o       "rtc": The hexadecimal address of the structure currently visible
  40         to readers.
  41
  42 o       "ver": The number of times since boot that the RCU writer task
  43         has changed the structure visible to readers.
  44
  45 o       "tfle": If non-zero, indicates that the "torture freelist"
  46         containing structures to be placed into the "rtc" area is empty.
  47         This condition is important, since it can fool you into thinking
  48         that RCU is working when it is not.  :-/
  49
  50 o       "rta": Number of structures allocated from the torture freelist.
  51
  52 o       "rtaf": Number of allocations from the torture freelist that have
  53         failed due to the list being empty.  It is not unusual for this
  54         to be non-zero, but it is bad for it to be a large fraction of
  55         the value indicated by "rta".
  56
  57 o       "rtf": Number of frees into the torture freelist.
  58
  59 o       "rtmbe": A non-zero value indicates that rcutorture believes that
  60         rcu_assign_pointer() and rcu_dereference() are not working
  61         correctly.  This value should be zero.
  62
  63 o       "rtbe": A non-zero value indicates that one of the rcu_barrier()
  64         family of functions is not working correctly.
  65
  66 o       "rtbke": rcutorture was unable to create the real-time kthreads
  67         used to force RCU priority inversion.  This value should be zero.
  68
  69 o       "rtbre": Although rcutorture successfully created the kthreads
  70         used to force RCU priority inversion, it was unable to set them
  71         to the real-time priority level of 1.  This value should be zero.
  72
  73 o       "rtbf": The number of times that RCU priority boosting failed
  74         to resolve RCU priority inversion.
  75
  76 o       "rtb": The number of times that rcutorture attempted to force
  77         an RCU priority inversion condition.  If you are testing RCU
  78         priority boosting via the "test_boost" module parameter, this
  79         value should be non-zero.
  80
  81 o       "nt": The number of times rcutorture ran RCU read-side code from
  82         within a timer handler.  This value should be non-zero only
  83         if you specified the "irqreader" module parameter.
  84
  85 o       "Reader Pipe": Histogram of "ages" of structures seen by readers.
  86         If any entries past the first two are non-zero, RCU is broken.
  87         And rcutorture prints the error flag string "!!!" to make sure
  88         you notice.  The age of a newly allocated structure is zero,
  89         it becomes one when removed from reader visibility, and is
  90         incremented once per grace period subsequently -- and is freed
  91         after passing through (RCU_TORTURE_PIPE_LEN-2) grace periods.
  92
  93         The output displayed above was taken from a correctly working
  94         RCU.  If you want to see what it looks like when broken, break
  95         it yourself.  ;-)
  96
  97 o       "Reader Batch": Another histogram of "ages" of structures seen
  98         by readers, but in terms of counter flips (or batches) rather
  99         than in terms of grace periods.  The legal number of non-zero
 100         entries is again two.  The reason for this separate view is that
 101         it is sometimes easier to get the third entry to show up in the
 102         "Reader Batch" list than in the "Reader Pipe" list.
 103
 104 o       "Free-Block Circulation": Shows the number of torture structures
 105         that have reached a given point in the pipeline.  The first element
 106         should closely correspond to the number of structures allocated,
 107         the second to the number that have been removed from reader view,
 108         and all but the last remaining to the corresponding number of
 109         passes through a grace period.  The last entry should be zero,
 110         as it is only incremented if a torture structure's counter
 111         somehow gets incremented farther than it should.
 112
 113 Different implementations of RCU can provide implementation-specific
 114 additional information.  For example, Tree SRCU provides the following
 115 additional line:
 116
 117         srcud-torture: Tree SRCU per-CPU(idx=0): 0(35,-21) 1(-4,24) 2(1,1) 3(-26,20) 4(28,-47) 5(-9,4) 6(-10,14) 7(-14,11) T(1,6)
 118
 119 This line shows the per-CPU counter state, in this case for Tree SRCU
 120 using a dynamically allocated srcu_struct (hence "srcud-" rather than
 121 "srcu-").  The numbers in parentheses are the values of the "old" and
 122 "current" counters for the corresponding CPU.  The "idx" value maps the
 123 "old" and "current" values to the underlying array, and is useful for
 124 debugging.  The final "T" entry contains the totals of the counters.
 125
 126
 127 USAGE
 128
 129 The following script may be used to torture RCU:
 130
 131         #!/bin/sh
 132
 133         modprobe rcutorture
 134         sleep 3600
 135         rmmod rcutorture
 136         dmesg | grep torture:
 137
 138 The output can be manually inspected for the error flag of "!!!".
 139 One could of course create a more elaborate script that automatically
 140 checked for such errors.  The "rmmod" command forces a "SUCCESS",
 141 "FAILURE", or "RCU_HOTPLUG" indication to be printk()ed.  The first
 142 two are self-explanatory, while the last indicates that while there
 143 were no RCU failures, CPU-hotplug problems were detected.
 144
 145 However, the tools/testing/selftests/rcutorture/bin/kvm.sh script
 146 provides better automation, including automatic failure analysis.
 147 It assumes a qemu/kvm-enabled platform, and runs guest OSes out of initrd.
 148 See tools/testing/selftests/rcutorture/doc/initrd.txt for instructions
 149 on setting up such an initrd.