linux/Documentation/RCU/trace.txt
<<
>>
Prefs
   1CONFIG_RCU_TRACE debugfs Files and Formats
   2
   3
   4The rcutree and rcutiny implementations of RCU provide debugfs trace
   5output that summarizes counters and state.  This information is useful for
   6debugging RCU itself, and can sometimes also help to debug abuses of RCU.
   7The following sections describe the debugfs files and formats, first
   8for rcutree and next for rcutiny.
   9
  10
  11CONFIG_TREE_RCU and CONFIG_TREE_PREEMPT_RCU debugfs Files and Formats
  12
  13These implementations of RCU provides several debugfs files under the
  14top-level directory "rcu":
  15
  16rcu/rcudata:
  17        Displays fields in struct rcu_data.
  18rcu/rcudata.csv:
  19        Comma-separated values spreadsheet version of rcudata.
  20rcu/rcugp:
  21        Displays grace-period counters.
  22rcu/rcuhier:
  23        Displays the struct rcu_node hierarchy.
  24rcu/rcu_pending:
  25        Displays counts of the reasons rcu_pending() decided that RCU had
  26        work to do.
  27rcu/rcutorture:
  28        Displays rcutorture test progress.
  29rcu/rcuboost:
  30        Displays RCU boosting statistics.  Only present if
  31        CONFIG_RCU_BOOST=y.
  32
  33The output of "cat rcu/rcudata" looks as follows:
  34
  35rcu_sched:
  36  0 c=20972 g=20973 pq=1 pgp=20973 qp=0 dt=545/1/0 df=50 of=0 ql=163 qs=NRW. kt=0/W/0 ktl=ebc3 b=10 ci=153737 co=0 ca=0
  37  1 c=20972 g=20973 pq=1 pgp=20973 qp=0 dt=967/1/0 df=58 of=0 ql=634 qs=NRW. kt=0/W/1 ktl=58c b=10 ci=191037 co=0 ca=0
  38  2 c=20972 g=20973 pq=1 pgp=20973 qp=0 dt=1081/1/0 df=175 of=0 ql=74 qs=N.W. kt=0/W/2 ktl=da94 b=10 ci=75991 co=0 ca=0
  39  3 c=20942 g=20943 pq=1 pgp=20942 qp=1 dt=1846/0/0 df=404 of=0 ql=0 qs=.... kt=0/W/3 ktl=d1cd b=10 ci=72261 co=0 ca=0
  40  4 c=20972 g=20973 pq=1 pgp=20973 qp=0 dt=369/1/0 df=83 of=0 ql=48 qs=N.W. kt=0/W/4 ktl=e0e7 b=10 ci=128365 co=0 ca=0
  41  5 c=20972 g=20973 pq=1 pgp=20973 qp=0 dt=381/1/0 df=64 of=0 ql=169 qs=NRW. kt=0/W/5 ktl=fb2f b=10 ci=164360 co=0 ca=0
  42  6 c=20972 g=20973 pq=1 pgp=20973 qp=0 dt=1037/1/0 df=183 of=0 ql=62 qs=N.W. kt=0/W/6 ktl=d2ad b=10 ci=65663 co=0 ca=0
  43  7 c=20897 g=20897 pq=1 pgp=20896 qp=0 dt=1572/0/0 df=382 of=0 ql=0 qs=.... kt=0/W/7 ktl=cf15 b=10 ci=75006 co=0 ca=0
  44rcu_bh:
  45  0 c=1480 g=1480 pq=1 pgp=1480 qp=0 dt=545/1/0 df=6 of=0 ql=0 qs=.... kt=0/W/0 ktl=ebc3 b=10 ci=0 co=0 ca=0
  46  1 c=1480 g=1480 pq=1 pgp=1480 qp=0 dt=967/1/0 df=3 of=0 ql=0 qs=.... kt=0/W/1 ktl=58c b=10 ci=151 co=0 ca=0
  47  2 c=1480 g=1480 pq=1 pgp=1480 qp=0 dt=1081/1/0 df=6 of=0 ql=0 qs=.... kt=0/W/2 ktl=da94 b=10 ci=0 co=0 ca=0
  48  3 c=1480 g=1480 pq=1 pgp=1480 qp=0 dt=1846/0/0 df=8 of=0 ql=0 qs=.... kt=0/W/3 ktl=d1cd b=10 ci=0 co=0 ca=0
  49  4 c=1480 g=1480 pq=1 pgp=1480 qp=0 dt=369/1/0 df=6 of=0 ql=0 qs=.... kt=0/W/4 ktl=e0e7 b=10 ci=0 co=0 ca=0
  50  5 c=1480 g=1480 pq=1 pgp=1480 qp=0 dt=381/1/0 df=4 of=0 ql=0 qs=.... kt=0/W/5 ktl=fb2f b=10 ci=0 co=0 ca=0
  51  6 c=1480 g=1480 pq=1 pgp=1480 qp=0 dt=1037/1/0 df=6 of=0 ql=0 qs=.... kt=0/W/6 ktl=d2ad b=10 ci=0 co=0 ca=0
  52  7 c=1474 g=1474 pq=1 pgp=1473 qp=0 dt=1572/0/0 df=8 of=0 ql=0 qs=.... kt=0/W/7 ktl=cf15 b=10 ci=0 co=0 ca=0
  53
  54The first section lists the rcu_data structures for rcu_sched, the second
  55for rcu_bh.  Note that CONFIG_TREE_PREEMPT_RCU kernels will have an
  56additional section for rcu_preempt.  Each section has one line per CPU,
  57or eight for this 8-CPU system.  The fields are as follows:
  58
  59o       The number at the beginning of each line is the CPU number.
  60        CPUs numbers followed by an exclamation mark are offline,
  61        but have been online at least once since boot.  There will be
  62        no output for CPUs that have never been online, which can be
  63        a good thing in the surprisingly common case where NR_CPUS is
  64        substantially larger than the number of actual CPUs.
  65
  66o       "c" is the count of grace periods that this CPU believes have
  67        completed.  Offlined CPUs and CPUs in dynticks idle mode may
  68        lag quite a ways behind, for example, CPU 6 under "rcu_sched"
  69        above, which has been offline through not quite 40,000 RCU grace
  70        periods.  It is not unusual to see CPUs lagging by thousands of
  71        grace periods.
  72
  73o       "g" is the count of grace periods that this CPU believes have
  74        started.  Again, offlined CPUs and CPUs in dynticks idle mode
  75        may lag behind.  If the "c" and "g" values are equal, this CPU
  76        has already reported a quiescent state for the last RCU grace
  77        period that it is aware of, otherwise, the CPU believes that it
  78        owes RCU a quiescent state.
  79
  80o       "pq" indicates that this CPU has passed through a quiescent state
  81        for the current grace period.  It is possible for "pq" to be
  82        "1" and "c" different than "g", which indicates that although
  83        the CPU has passed through a quiescent state, either (1) this
  84        CPU has not yet reported that fact, (2) some other CPU has not
  85        yet reported for this grace period, or (3) both.
  86
  87o       "pgp" indicates which grace period the last-observed quiescent
  88        state for this CPU corresponds to.  This is important for handling
  89        the race between CPU 0 reporting an extended dynticks-idle
  90        quiescent state for CPU 1 and CPU 1 suddenly waking up and
  91        reporting its own quiescent state.  If CPU 1 was the last CPU
  92        for the current grace period, then the CPU that loses this race
  93        will attempt to incorrectly mark CPU 1 as having checked in for
  94        the next grace period!
  95
  96o       "qp" indicates that RCU still expects a quiescent state from
  97        this CPU.  Offlined CPUs and CPUs in dyntick idle mode might
  98        well have qp=1, which is OK: RCU is still ignoring them.
  99
 100o       "dt" is the current value of the dyntick counter that is incremented
 101        when entering or leaving dynticks idle state, either by the
 102        scheduler or by irq.  This number is even if the CPU is in
 103        dyntick idle mode and odd otherwise.  The number after the first
 104        "/" is the interrupt nesting depth when in dyntick-idle state,
 105        or one greater than the interrupt-nesting depth otherwise.
 106        The number after the second "/" is the NMI nesting depth.
 107
 108o       "df" is the number of times that some other CPU has forced a
 109        quiescent state on behalf of this CPU due to this CPU being in
 110        dynticks-idle state.
 111
 112o       "of" is the number of times that some other CPU has forced a
 113        quiescent state on behalf of this CPU due to this CPU being
 114        offline.  In a perfect world, this might never happen, but it
 115        turns out that offlining and onlining a CPU can take several grace
 116        periods, and so there is likely to be an extended period of time
 117        when RCU believes that the CPU is online when it really is not.
 118        Please note that erring in the other direction (RCU believing a
 119        CPU is offline when it is really alive and kicking) is a fatal
 120        error, so it makes sense to err conservatively.
 121
 122o       "ql" is the number of RCU callbacks currently residing on
 123        this CPU.  This is the total number of callbacks, regardless
 124        of what state they are in (new, waiting for grace period to
 125        start, waiting for grace period to end, ready to invoke).
 126
 127o       "qs" gives an indication of the state of the callback queue
 128        with four characters:
 129
 130        "N"     Indicates that there are callbacks queued that are not
 131                ready to be handled by the next grace period, and thus
 132                will be handled by the grace period following the next
 133                one.
 134
 135        "R"     Indicates that there are callbacks queued that are
 136                ready to be handled by the next grace period.
 137
 138        "W"     Indicates that there are callbacks queued that are
 139                waiting on the current grace period.
 140
 141        "D"     Indicates that there are callbacks queued that have
 142                already been handled by a prior grace period, and are
 143                thus waiting to be invoked.  Note that callbacks in
 144                the process of being invoked are not counted here.
 145                Callbacks in the process of being invoked are those
 146                that have been removed from the rcu_data structures
 147                queues by rcu_do_batch(), but which have not yet been
 148                invoked.
 149
 150        If there are no callbacks in a given one of the above states,
 151        the corresponding character is replaced by ".".
 152
 153o       "kt" is the per-CPU kernel-thread state.  The digit preceding
 154        the first slash is zero if there is no work pending and 1
 155        otherwise.  The character between the first pair of slashes is
 156        as follows:
 157
 158        "S"     The kernel thread is stopped, in other words, all
 159                CPUs corresponding to this rcu_node structure are
 160                offline.
 161
 162        "R"     The kernel thread is running.
 163
 164        "W"     The kernel thread is waiting because there is no work
 165                for it to do.
 166
 167        "O"     The kernel thread is waiting because it has been
 168                forced off of its designated CPU or because its
 169                ->cpus_allowed mask permits it to run on other than
 170                its designated CPU.
 171
 172        "Y"     The kernel thread is yielding to avoid hogging CPU.
 173
 174        "?"     Unknown value, indicates a bug.
 175
 176        The number after the final slash is the CPU that the kthread
 177        is actually running on.
 178
 179        This field is displayed only for CONFIG_RCU_BOOST kernels.
 180
 181o       "ktl" is the low-order 16 bits (in hexadecimal) of the count of
 182        the number of times that this CPU's per-CPU kthread has gone
 183        through its loop servicing invoke_rcu_cpu_kthread() requests.
 184
 185        This field is displayed only for CONFIG_RCU_BOOST kernels.
 186
 187o       "b" is the batch limit for this CPU.  If more than this number
 188        of RCU callbacks is ready to invoke, then the remainder will
 189        be deferred.
 190
 191o       "ci" is the number of RCU callbacks that have been invoked for
 192        this CPU.  Note that ci+ql is the number of callbacks that have
 193        been registered in absence of CPU-hotplug activity.
 194
 195o       "co" is the number of RCU callbacks that have been orphaned due to
 196        this CPU going offline.  These orphaned callbacks have been moved
 197        to an arbitrarily chosen online CPU.
 198
 199o       "ca" is the number of RCU callbacks that have been adopted due to
 200        other CPUs going offline.  Note that ci+co-ca+ql is the number of
 201        RCU callbacks registered on this CPU.
 202
 203There is also an rcu/rcudata.csv file with the same information in
 204comma-separated-variable spreadsheet format.
 205
 206
 207The output of "cat rcu/rcugp" looks as follows:
 208
 209rcu_sched: completed=33062  gpnum=33063
 210rcu_bh: completed=464  gpnum=464
 211
 212Again, this output is for both "rcu_sched" and "rcu_bh".  Note that
 213kernels built with CONFIG_TREE_PREEMPT_RCU will have an additional
 214"rcu_preempt" line.  The fields are taken from the rcu_state structure,
 215and are as follows:
 216
 217o       "completed" is the number of grace periods that have completed.
 218        It is comparable to the "c" field from rcu/rcudata in that a
 219        CPU whose "c" field matches the value of "completed" is aware
 220        that the corresponding RCU grace period has completed.
 221
 222o       "gpnum" is the number of grace periods that have started.  It is
 223        comparable to the "g" field from rcu/rcudata in that a CPU
 224        whose "g" field matches the value of "gpnum" is aware that the
 225        corresponding RCU grace period has started.
 226
 227        If these two fields are equal (as they are for "rcu_bh" above),
 228        then there is no grace period in progress, in other words, RCU
 229        is idle.  On the other hand, if the two fields differ (as they
 230        do for "rcu_sched" above), then an RCU grace period is in progress.
 231
 232
 233The output of "cat rcu/rcuhier" looks as follows, with very long lines:
 234
 235c=6902 g=6903 s=2 jfq=3 j=72c7 nfqs=13142/nfqsng=0(13142) fqlh=6
 2361/1 ..>. 0:127 ^0
 2373/3 ..>. 0:35 ^0    0/0 ..>. 36:71 ^1    0/0 ..>. 72:107 ^2    0/0 ..>. 108:127 ^3
 2383/3f ..>. 0:5 ^0    2/3 ..>. 6:11 ^1    0/0 ..>. 12:17 ^2    0/0 ..>. 18:23 ^3    0/0 ..>. 24:29 ^4    0/0 ..>. 30:35 ^5    0/0 ..>. 36:41 ^0    0/0 ..>. 42:47 ^1    0/0 ..>. 48:53 ^2    0/0 ..>. 54:59 ^3    0/0 ..>. 60:65 ^4    0/0 ..>. 66:71 ^5    0/0 ..>. 72:77 ^0    0/0 ..>. 78:83 ^1    0/0 ..>. 84:89 ^2    0/0 ..>. 90:95 ^3    0/0 ..>. 96:101 ^4    0/0 ..>. 102:107 ^5    0/0 ..>. 108:113 ^0    0/0 ..>. 114:119 ^1    0/0 ..>. 120:125 ^2    0/0 ..>. 126:127 ^3
 239rcu_bh:
 240c=-226 g=-226 s=1 jfq=-5701 j=72c7 nfqs=88/nfqsng=0(88) fqlh=0
 2410/1 ..>. 0:127 ^0
 2420/3 ..>. 0:35 ^0    0/0 ..>. 36:71 ^1    0/0 ..>. 72:107 ^2    0/0 ..>. 108:127 ^3
 2430/3f ..>. 0:5 ^0    0/3 ..>. 6:11 ^1    0/0 ..>. 12:17 ^2    0/0 ..>. 18:23 ^3    0/0 ..>. 24:29 ^4    0/0 ..>. 30:35 ^5    0/0 ..>. 36:41 ^0    0/0 ..>. 42:47 ^1    0/0 ..>. 48:53 ^2    0/0 ..>. 54:59 ^3    0/0 ..>. 60:65 ^4    0/0 ..>. 66:71 ^5    0/0 ..>. 72:77 ^0    0/0 ..>. 78:83 ^1    0/0 ..>. 84:89 ^2    0/0 ..>. 90:95 ^3    0/0 ..>. 96:101 ^4    0/0 ..>. 102:107 ^5    0/0 ..>. 108:113 ^0    0/0 ..>. 114:119 ^1    0/0 ..>. 120:125 ^2    0/0 ..>. 126:127 ^3
 244
 245This is once again split into "rcu_sched" and "rcu_bh" portions,
 246and CONFIG_TREE_PREEMPT_RCU kernels will again have an additional
 247"rcu_preempt" section.  The fields are as follows:
 248
 249o       "c" is exactly the same as "completed" under rcu/rcugp.
 250
 251o       "g" is exactly the same as "gpnum" under rcu/rcugp.
 252
 253o       "s" is the "signaled" state that drives force_quiescent_state()'s
 254        state machine.
 255
 256o       "jfq" is the number of jiffies remaining for this grace period
 257        before force_quiescent_state() is invoked to help push things
 258        along.  Note that CPUs in dyntick-idle mode throughout the grace
 259        period will not report on their own, but rather must be check by
 260        some other CPU via force_quiescent_state().
 261
 262o       "j" is the low-order four hex digits of the jiffies counter.
 263        Yes, Paul did run into a number of problems that turned out to
 264        be due to the jiffies counter no longer counting.  Why do you ask?
 265
 266o       "nfqs" is the number of calls to force_quiescent_state() since
 267        boot.
 268
 269o       "nfqsng" is the number of useless calls to force_quiescent_state(),
 270        where there wasn't actually a grace period active.  This can
 271        happen due to races.  The number in parentheses is the difference
 272        between "nfqs" and "nfqsng", or the number of times that
 273        force_quiescent_state() actually did some real work.
 274
 275o       "fqlh" is the number of calls to force_quiescent_state() that
 276        exited immediately (without even being counted in nfqs above)
 277        due to contention on ->fqslock.
 278
 279o       Each element of the form "1/1 0:127 ^0" represents one struct
 280        rcu_node.  Each line represents one level of the hierarchy, from
 281        root to leaves.  It is best to think of the rcu_data structures
 282        as forming yet another level after the leaves.  Note that there
 283        might be either one, two, or three levels of rcu_node structures,
 284        depending on the relationship between CONFIG_RCU_FANOUT and
 285        CONFIG_NR_CPUS.
 286
 287        o       The numbers separated by the "/" are the qsmask followed
 288                by the qsmaskinit.  The qsmask will have one bit
 289                set for each entity in the next lower level that
 290                has not yet checked in for the current grace period.
 291                The qsmaskinit will have one bit for each entity that is
 292                currently expected to check in during each grace period.
 293                The value of qsmaskinit is assigned to that of qsmask
 294                at the beginning of each grace period.
 295
 296                For example, for "rcu_sched", the qsmask of the first
 297                entry of the lowest level is 0x14, meaning that we
 298                are still waiting for CPUs 2 and 4 to check in for the
 299                current grace period.
 300
 301        o       The characters separated by the ">" indicate the state
 302                of the blocked-tasks lists.  A "G" preceding the ">"
 303                indicates that at least one task blocked in an RCU
 304                read-side critical section blocks the current grace
 305                period, while a "E" preceding the ">" indicates that
 306                at least one task blocked in an RCU read-side critical
 307                section blocks the current expedited grace period.
 308                A "T" character following the ">" indicates that at
 309                least one task is blocked within an RCU read-side
 310                critical section, regardless of whether any current
 311                grace period (expedited or normal) is inconvenienced.
 312                A "." character appears if the corresponding condition
 313                does not hold, so that "..>." indicates that no tasks
 314                are blocked.  In contrast, "GE>T" indicates maximal
 315                inconvenience from blocked tasks.
 316
 317        o       The numbers separated by the ":" are the range of CPUs
 318                served by this struct rcu_node.  This can be helpful
 319                in working out how the hierarchy is wired together.
 320
 321                For example, the first entry at the lowest level shows
 322                "0:5", indicating that it covers CPUs 0 through 5.
 323
 324        o       The number after the "^" indicates the bit in the
 325                next higher level rcu_node structure that this
 326                rcu_node structure corresponds to.
 327
 328                For example, the first entry at the lowest level shows
 329                "^0", indicating that it corresponds to bit zero in
 330                the first entry at the middle level.
 331
 332
 333The output of "cat rcu/rcu_pending" looks as follows:
 334
 335rcu_sched:
 336  0 np=255892 qsp=53936 rpq=85 cbr=0 cng=14417 gpc=10033 gps=24320 nf=6445 nn=146741
 337  1 np=261224 qsp=54638 rpq=33 cbr=0 cng=25723 gpc=16310 gps=2849 nf=5912 nn=155792
 338  2 np=237496 qsp=49664 rpq=23 cbr=0 cng=2762 gpc=45478 gps=1762 nf=1201 nn=136629
 339  3 np=236249 qsp=48766 rpq=98 cbr=0 cng=286 gpc=48049 gps=1218 nf=207 nn=137723
 340  4 np=221310 qsp=46850 rpq=7 cbr=0 cng=26 gpc=43161 gps=4634 nf=3529 nn=123110
 341  5 np=237332 qsp=48449 rpq=9 cbr=0 cng=54 gpc=47920 gps=3252 nf=201 nn=137456
 342  6 np=219995 qsp=46718 rpq=12 cbr=0 cng=50 gpc=42098 gps=6093 nf=4202 nn=120834
 343  7 np=249893 qsp=49390 rpq=42 cbr=0 cng=72 gpc=38400 gps=17102 nf=41 nn=144888
 344rcu_bh:
 345  0 np=146741 qsp=1419 rpq=6 cbr=0 cng=6 gpc=0 gps=0 nf=2 nn=145314
 346  1 np=155792 qsp=12597 rpq=3 cbr=0 cng=0 gpc=4 gps=8 nf=3 nn=143180
 347  2 np=136629 qsp=18680 rpq=1 cbr=0 cng=0 gpc=7 gps=6 nf=0 nn=117936
 348  3 np=137723 qsp=2843 rpq=0 cbr=0 cng=0 gpc=10 gps=7 nf=0 nn=134863
 349  4 np=123110 qsp=12433 rpq=0 cbr=0 cng=0 gpc=4 gps=2 nf=0 nn=110671
 350  5 np=137456 qsp=4210 rpq=1 cbr=0 cng=0 gpc=6 gps=5 nf=0 nn=133235
 351  6 np=120834 qsp=9902 rpq=2 cbr=0 cng=0 gpc=6 gps=3 nf=2 nn=110921
 352  7 np=144888 qsp=26336 rpq=0 cbr=0 cng=0 gpc=8 gps=2 nf=0 nn=118542
 353
 354As always, this is once again split into "rcu_sched" and "rcu_bh"
 355portions, with CONFIG_TREE_PREEMPT_RCU kernels having an additional
 356"rcu_preempt" section.  The fields are as follows:
 357
 358o       "np" is the number of times that __rcu_pending() has been invoked
 359        for the corresponding flavor of RCU.
 360
 361o       "qsp" is the number of times that the RCU was waiting for a
 362        quiescent state from this CPU.
 363
 364o       "rpq" is the number of times that the CPU had passed through
 365        a quiescent state, but not yet reported it to RCU.
 366
 367o       "cbr" is the number of times that this CPU had RCU callbacks
 368        that had passed through a grace period, and were thus ready
 369        to be invoked.
 370
 371o       "cng" is the number of times that this CPU needed another
 372        grace period while RCU was idle.
 373
 374o       "gpc" is the number of times that an old grace period had
 375        completed, but this CPU was not yet aware of it.
 376
 377o       "gps" is the number of times that a new grace period had started,
 378        but this CPU was not yet aware of it.
 379
 380o       "nf" is the number of times that this CPU suspected that the
 381        current grace period had run for too long, and thus needed to
 382        be forced.
 383
 384        Please note that "forcing" consists of sending resched IPIs
 385        to holdout CPUs.  If that CPU really still is in an old RCU
 386        read-side critical section, then we really do have to wait for it.
 387        The assumption behing "forcing" is that the CPU is not still in
 388        an old RCU read-side critical section, but has not yet responded
 389        for some other reason.
 390
 391o       "nn" is the number of times that this CPU needed nothing.  Alert
 392        readers will note that the rcu "nn" number for a given CPU very
 393        closely matches the rcu_bh "np" number for that same CPU.  This
 394        is due to short-circuit evaluation in rcu_pending().
 395
 396
 397The output of "cat rcu/rcutorture" looks as follows:
 398
 399rcutorture test sequence: 0 (test in progress)
 400rcutorture update version number: 615
 401
 402The first line shows the number of rcutorture tests that have completed
 403since boot.  If a test is currently running, the "(test in progress)"
 404string will appear as shown above.  The second line shows the number of
 405update cycles that the current test has started, or zero if there is
 406no test in progress.
 407
 408
 409The output of "cat rcu/rcuboost" looks as follows:
 410
 4110:5 tasks=.... kt=W ntb=0 neb=0 nnb=0 j=2f95 bt=300f
 412     balk: nt=0 egt=989 bt=0 nb=0 ny=0 nos=16
 4136:7 tasks=.... kt=W ntb=0 neb=0 nnb=0 j=2f95 bt=300f
 414     balk: nt=0 egt=225 bt=0 nb=0 ny=0 nos=6
 415
 416This information is output only for rcu_preempt.  Each two-line entry
 417corresponds to a leaf rcu_node strcuture.  The fields are as follows:
 418
 419o       "n:m" is the CPU-number range for the corresponding two-line
 420        entry.  In the sample output above, the first entry covers
 421        CPUs zero through five and the second entry covers CPUs 6
 422        and 7.
 423
 424o       "tasks=TNEB" gives the state of the various segments of the
 425        rnp->blocked_tasks list:
 426
 427        "T"     This indicates that there are some tasks that blocked
 428                while running on one of the corresponding CPUs while
 429                in an RCU read-side critical section.
 430
 431        "N"     This indicates that some of the blocked tasks are preventing
 432                the current normal (non-expedited) grace period from
 433                completing.
 434
 435        "E"     This indicates that some of the blocked tasks are preventing
 436                the current expedited grace period from completing.
 437
 438        "B"     This indicates that some of the blocked tasks are in
 439                need of RCU priority boosting.
 440
 441        Each character is replaced with "." if the corresponding
 442        condition does not hold.
 443
 444o       "kt" is the state of the RCU priority-boosting kernel
 445        thread associated with the corresponding rcu_node structure.
 446        The state can be one of the following:
 447
 448        "S"     The kernel thread is stopped, in other words, all
 449                CPUs corresponding to this rcu_node structure are
 450                offline.
 451
 452        "R"     The kernel thread is running.
 453
 454        "W"     The kernel thread is waiting because there is no work
 455                for it to do.
 456
 457        "Y"     The kernel thread is yielding to avoid hogging CPU.
 458
 459        "?"     Unknown value, indicates a bug.
 460
 461o       "ntb" is the number of tasks boosted.
 462
 463o       "neb" is the number of tasks boosted in order to complete an
 464        expedited grace period.
 465
 466o       "nnb" is the number of tasks boosted in order to complete a
 467        normal (non-expedited) grace period.  When boosting a task
 468        that was blocking both an expedited and a normal grace period,
 469        it is counted against the expedited total above.
 470
 471o       "j" is the low-order 16 bits of the jiffies counter in
 472        hexadecimal.
 473
 474o       "bt" is the low-order 16 bits of the value that the jiffies
 475        counter will have when we next start boosting, assuming that
 476        the current grace period does not end beforehand.  This is
 477        also in hexadecimal.
 478
 479o       "balk: nt" counts the number of times we didn't boost (in
 480        other words, we balked) even though it was time to boost because
 481        there were no blocked tasks to boost.  This situation occurs
 482        when there is one blocked task on one rcu_node structure and
 483        none on some other rcu_node structure.
 484
 485o       "egt" counts the number of times we balked because although
 486        there were blocked tasks, none of them were blocking the
 487        current grace period, whether expedited or otherwise.
 488
 489o       "bt" counts the number of times we balked because boosting
 490        had already been initiated for the current grace period.
 491
 492o       "nb" counts the number of times we balked because there
 493        was at least one task blocking the current non-expedited grace
 494        period that never had blocked.  If it is already running, it
 495        just won't help to boost its priority!
 496
 497o       "ny" counts the number of times we balked because it was
 498        not yet time to start boosting.
 499
 500o       "nos" counts the number of times we balked for other
 501        reasons, e.g., the grace period ended first.
 502
 503
 504CONFIG_TINY_RCU and CONFIG_TINY_PREEMPT_RCU debugfs Files and Formats
 505
 506These implementations of RCU provides a single debugfs file under the
 507top-level directory RCU, namely rcu/rcudata, which displays fields in
 508rcu_bh_ctrlblk, rcu_sched_ctrlblk and, for CONFIG_TINY_PREEMPT_RCU,
 509rcu_preempt_ctrlblk.
 510
 511The output of "cat rcu/rcudata" is as follows:
 512
 513rcu_preempt: qlen=24 gp=1097669 g197/p197/c197 tasks=...
 514             ttb=. btg=no ntb=184 neb=0 nnb=183 j=01f7 bt=0274
 515             normal balk: nt=1097669 gt=0 bt=371 b=0 ny=25073378 nos=0
 516             exp balk: bt=0 nos=0
 517rcu_sched: qlen: 0
 518rcu_bh: qlen: 0
 519
 520This is split into rcu_preempt, rcu_sched, and rcu_bh sections, with the
 521rcu_preempt section appearing only in CONFIG_TINY_PREEMPT_RCU builds.
 522The last three lines of the rcu_preempt section appear only in
 523CONFIG_RCU_BOOST kernel builds.  The fields are as follows:
 524
 525o       "qlen" is the number of RCU callbacks currently waiting either
 526        for an RCU grace period or waiting to be invoked.  This is the
 527        only field present for rcu_sched and rcu_bh, due to the
 528        short-circuiting of grace period in those two cases.
 529
 530o       "gp" is the number of grace periods that have completed.
 531
 532o       "g197/p197/c197" displays the grace-period state, with the
 533        "g" number being the number of grace periods that have started
 534        (mod 256), the "p" number being the number of grace periods
 535        that the CPU has responded to (also mod 256), and the "c"
 536        number being the number of grace periods that have completed
 537        (once again mode 256).
 538
 539        Why have both "gp" and "g"?  Because the data flowing into
 540        "gp" is only present in a CONFIG_RCU_TRACE kernel.
 541
 542o       "tasks" is a set of bits.  The first bit is "T" if there are
 543        currently tasks that have recently blocked within an RCU
 544        read-side critical section, the second bit is "N" if any of the
 545        aforementioned tasks are blocking the current RCU grace period,
 546        and the third bit is "E" if any of the aforementioned tasks are
 547        blocking the current expedited grace period.  Each bit is "."
 548        if the corresponding condition does not hold.
 549
 550o       "ttb" is a single bit.  It is "B" if any of the blocked tasks
 551        need to be priority boosted and "." otherwise.
 552
 553o       "btg" indicates whether boosting has been carried out during
 554        the current grace period, with "exp" indicating that boosting
 555        is in progress for an expedited grace period, "no" indicating
 556        that boosting has not yet started for a normal grace period,
 557        "begun" indicating that boosting has bebug for a normal grace
 558        period, and "done" indicating that boosting has completed for
 559        a normal grace period.
 560
 561o       "ntb" is the total number of tasks subjected to RCU priority boosting
 562        periods since boot.
 563
 564o       "neb" is the number of expedited grace periods that have had
 565        to resort to RCU priority boosting since boot.
 566
 567o       "nnb" is the number of normal grace periods that have had
 568        to resort to RCU priority boosting since boot.
 569
 570o       "j" is the low-order 16 bits of the jiffies counter in hexadecimal.
 571
 572o       "bt" is the low-order 16 bits of the value that the jiffies counter
 573        will have at the next time that boosting is scheduled to begin.
 574
 575o       In the line beginning with "normal balk", the fields are as follows:
 576
 577        o       "nt" is the number of times that the system balked from
 578                boosting because there were no blocked tasks to boost.
 579                Note that the system will balk from boosting even if the
 580                grace period is overdue when the currently running task
 581                is looping within an RCU read-side critical section.
 582                There is no point in boosting in this case, because
 583                boosting a running task won't make it run any faster.
 584
 585        o       "gt" is the number of times that the system balked
 586                from boosting because, although there were blocked tasks,
 587                none of them were preventing the current grace period
 588                from completing.
 589
 590        o       "bt" is the number of times that the system balked
 591                from boosting because boosting was already in progress.
 592
 593        o       "b" is the number of times that the system balked from
 594                boosting because boosting had already completed for
 595                the grace period in question.
 596
 597        o       "ny" is the number of times that the system balked from
 598                boosting because it was not yet time to start boosting
 599                the grace period in question.
 600
 601        o       "nos" is the number of times that the system balked from
 602                boosting for inexplicable ("not otherwise specified")
 603                reasons.  This can actually happen due to races involving
 604                increments of the jiffies counter.
 605
 606o       In the line beginning with "exp balk", the fields are as follows:
 607
 608        o       "bt" is the number of times that the system balked from
 609                boosting because there were no blocked tasks to boost.
 610
 611        o       "nos" is the number of times that the system balked from
 612                 boosting for inexplicable ("not otherwise specified")
 613                 reasons.
 614
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.