1 CONFIG_RCU_TRACE debugfs Files and Formats
4 The rcutree implementation of RCU provides debugfs trace output that
5 summarizes counters and state. This information is useful for debugging
6 RCU itself, and can sometimes also help to debug abuses of RCU.
7 The following sections describe the debugfs files and formats.
10 Hierarchical RCU debugfs Files and Formats
12 This implementation of RCU provides three debugfs files under the
13 top-level directory RCU: rcu/rcudata (which displays fields in struct
14 rcu_data), rcu/rcugp (which displays grace-period counters), and
15 rcu/rcuhier (which displays the struct rcu_node hierarchy).
17 The output of "cat rcu/rcudata" looks as follows:
20 0 c=17829 g=17829 pq=1 pqc=17829 qp=0 dt=10951/1 dn=0 df=1101 of=0 ri=36 ql=0 b=10
21 1 c=17829 g=17829 pq=1 pqc=17829 qp=0 dt=16117/1 dn=0 df=1015 of=0 ri=0 ql=0 b=10
22 2 c=17829 g=17829 pq=1 pqc=17829 qp=0 dt=1445/1 dn=0 df=1839 of=0 ri=0 ql=0 b=10
23 3 c=17829 g=17829 pq=1 pqc=17829 qp=0 dt=6681/1 dn=0 df=1545 of=0 ri=0 ql=0 b=10
24 4 c=17829 g=17829 pq=1 pqc=17829 qp=0 dt=1003/1 dn=0 df=1992 of=0 ri=0 ql=0 b=10
25 5 c=17829 g=17830 pq=1 pqc=17829 qp=1 dt=3887/1 dn=0 df=3331 of=0 ri=4 ql=2 b=10
26 6 c=17829 g=17829 pq=1 pqc=17829 qp=0 dt=859/1 dn=0 df=3224 of=0 ri=0 ql=0 b=10
27 7 c=17829 g=17830 pq=0 pqc=17829 qp=1 dt=3761/1 dn=0 df=1818 of=0 ri=0 ql=2 b=10
29 0 c=-275 g=-275 pq=1 pqc=-275 qp=0 dt=10951/1 dn=0 df=0 of=0 ri=0 ql=0 b=10
30 1 c=-275 g=-275 pq=1 pqc=-275 qp=0 dt=16117/1 dn=0 df=13 of=0 ri=0 ql=0 b=10
31 2 c=-275 g=-275 pq=1 pqc=-275 qp=0 dt=1445/1 dn=0 df=15 of=0 ri=0 ql=0 b=10
32 3 c=-275 g=-275 pq=1 pqc=-275 qp=0 dt=6681/1 dn=0 df=9 of=0 ri=0 ql=0 b=10
33 4 c=-275 g=-275 pq=1 pqc=-275 qp=0 dt=1003/1 dn=0 df=15 of=0 ri=0 ql=0 b=10
34 5 c=-275 g=-275 pq=1 pqc=-275 qp=0 dt=3887/1 dn=0 df=15 of=0 ri=0 ql=0 b=10
35 6 c=-275 g=-275 pq=1 pqc=-275 qp=0 dt=859/1 dn=0 df=15 of=0 ri=0 ql=0 b=10
36 7 c=-275 g=-275 pq=1 pqc=-275 qp=0 dt=3761/1 dn=0 df=15 of=0 ri=0 ql=0 b=10
38 The first section lists the rcu_data structures for rcu_sched, the second
39 for rcu_bh. Note that CONFIG_TREE_PREEMPT_RCU kernels will have an
40 additional section for rcu_preempt. Each section has one line per CPU,
41 or eight for this 8-CPU system. The fields are as follows:
43 o The number at the beginning of each line is the CPU number.
44 CPUs numbers followed by an exclamation mark are offline,
45 but have been online at least once since boot. There will be
46 no output for CPUs that have never been online, which can be
47 a good thing in the surprisingly common case where NR_CPUS is
48 substantially larger than the number of actual CPUs.
50 o "c" is the count of grace periods that this CPU believes have
51 completed. CPUs in dynticks idle mode may lag quite a ways
52 behind, for example, CPU 4 under "rcu_sched" above, which has
53 slept through the past 25 RCU grace periods. It is not unusual
54 to see CPUs lagging by thousands of grace periods.
56 o "g" is the count of grace periods that this CPU believes have
57 started. Again, CPUs in dynticks idle mode may lag behind.
58 If the "c" and "g" values are equal, this CPU has already
59 reported a quiescent state for the last RCU grace period that
60 it is aware of, otherwise, the CPU believes that it owes RCU a
63 o "pq" indicates that this CPU has passed through a quiescent state
64 for the current grace period. It is possible for "pq" to be
65 "1" and "c" different than "g", which indicates that although
66 the CPU has passed through a quiescent state, either (1) this
67 CPU has not yet reported that fact, (2) some other CPU has not
68 yet reported for this grace period, or (3) both.
70 o "pqc" indicates which grace period the last-observed quiescent
71 state for this CPU corresponds to. This is important for handling
72 the race between CPU 0 reporting an extended dynticks-idle
73 quiescent state for CPU 1 and CPU 1 suddenly waking up and
74 reporting its own quiescent state. If CPU 1 was the last CPU
75 for the current grace period, then the CPU that loses this race
76 will attempt to incorrectly mark CPU 1 as having checked in for
77 the next grace period!
79 o "qp" indicates that RCU still expects a quiescent state from
82 o "dt" is the current value of the dyntick counter that is incremented
83 when entering or leaving dynticks idle state, either by the
84 scheduler or by irq. The number after the "/" is the interrupt
85 nesting depth when in dyntick-idle state, or one greater than
86 the interrupt-nesting depth otherwise.
88 This field is displayed only for CONFIG_NO_HZ kernels.
90 o "dn" is the current value of the dyntick counter that is incremented
91 when entering or leaving dynticks idle state via NMI. If both
92 the "dt" and "dn" values are even, then this CPU is in dynticks
93 idle mode and may be ignored by RCU. If either of these two
94 counters is odd, then RCU must be alert to the possibility of
95 an RCU read-side critical section running on this CPU.
97 This field is displayed only for CONFIG_NO_HZ kernels.
99 o "df" is the number of times that some other CPU has forced a
100 quiescent state on behalf of this CPU due to this CPU being in
103 This field is displayed only for CONFIG_NO_HZ kernels.
105 o "of" is the number of times that some other CPU has forced a
106 quiescent state on behalf of this CPU due to this CPU being
107 offline. In a perfect world, this might neve happen, but it
108 turns out that offlining and onlining a CPU can take several grace
109 periods, and so there is likely to be an extended period of time
110 when RCU believes that the CPU is online when it really is not.
111 Please note that erring in the other direction (RCU believing a
112 CPU is offline when it is really alive and kicking) is a fatal
113 error, so it makes sense to err conservatively.
115 o "ri" is the number of times that RCU has seen fit to send a
116 reschedule IPI to this CPU in order to get it to report a
119 o "ql" is the number of RCU callbacks currently residing on
120 this CPU. This is the total number of callbacks, regardless
121 of what state they are in (new, waiting for grace period to
122 start, waiting for grace period to end, ready to invoke).
124 o "b" is the batch limit for this CPU. If more than this number
125 of RCU callbacks is ready to invoke, then the remainder will
128 There is also an rcu/rcudata.csv file with the same information in
129 comma-separated-variable spreadsheet format.
132 The output of "cat rcu/rcugp" looks as follows:
134 rcu_sched: completed=33062 gpnum=33063
135 rcu_bh: completed=464 gpnum=464
137 Again, this output is for both "rcu_sched" and "rcu_bh". Note that
138 kernels built with CONFIG_TREE_PREEMPT_RCU will have an additional
139 "rcu_preempt" line. The fields are taken from the rcu_state structure,
142 o "completed" is the number of grace periods that have completed.
143 It is comparable to the "c" field from rcu/rcudata in that a
144 CPU whose "c" field matches the value of "completed" is aware
145 that the corresponding RCU grace period has completed.
147 o "gpnum" is the number of grace periods that have started. It is
148 comparable to the "g" field from rcu/rcudata in that a CPU
149 whose "g" field matches the value of "gpnum" is aware that the
150 corresponding RCU grace period has started.
152 If these two fields are equal (as they are for "rcu_bh" above),
153 then there is no grace period in progress, in other words, RCU
154 is idle. On the other hand, if the two fields differ (as they
155 do for "rcu_sched" above), then an RCU grace period is in progress.
158 The output of "cat rcu/rcuhier" looks as follows, with very long lines:
160 c=6902 g=6903 s=2 jfq=3 j=72c7 nfqs=13142/nfqsng=0(13142) fqlh=6 oqlen=0
162 3/3 .>. 0:35 ^0 0/0 .>. 36:71 ^1 0/0 .>. 72:107 ^2 0/0 .>. 108:127 ^3
163 3/3f .>. 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
165 c=-226 g=-226 s=1 jfq=-5701 j=72c7 nfqs=88/nfqsng=0(88) fqlh=0 oqlen=0
167 0/3 .>. 0:35 ^0 0/0 .>. 36:71 ^1 0/0 .>. 72:107 ^2 0/0 .>. 108:127 ^3
168 0/3f .>. 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
170 This is once again split into "rcu_sched" and "rcu_bh" portions,
171 and CONFIG_TREE_PREEMPT_RCU kernels will again have an additional
172 "rcu_preempt" section. The fields are as follows:
174 o "c" is exactly the same as "completed" under rcu/rcugp.
176 o "g" is exactly the same as "gpnum" under rcu/rcugp.
178 o "s" is the "signaled" state that drives force_quiescent_state()'s
181 o "jfq" is the number of jiffies remaining for this grace period
182 before force_quiescent_state() is invoked to help push things
183 along. Note that CPUs in dyntick-idle mode thoughout the grace
184 period will not report on their own, but rather must be check by
185 some other CPU via force_quiescent_state().
187 o "j" is the low-order four hex digits of the jiffies counter.
188 Yes, Paul did run into a number of problems that turned out to
189 be due to the jiffies counter no longer counting. Why do you ask?
191 o "nfqs" is the number of calls to force_quiescent_state() since
194 o "nfqsng" is the number of useless calls to force_quiescent_state(),
195 where there wasn't actually a grace period active. This can
196 happen due to races. The number in parentheses is the difference
197 between "nfqs" and "nfqsng", or the number of times that
198 force_quiescent_state() actually did some real work.
200 o "fqlh" is the number of calls to force_quiescent_state() that
201 exited immediately (without even being counted in nfqs above)
202 due to contention on ->fqslock.
204 o "oqlen" is the number of callbacks on the "orphan" callback
205 list. RCU callbacks are placed on this list by CPUs going
206 offline, and are "adopted" either by the CPU helping the outgoing
207 CPU or by the next rcu_barrier*() call, whichever comes first.
209 o Each element of the form "1/1 0:127 ^0" represents one struct
210 rcu_node. Each line represents one level of the hierarchy, from
211 root to leaves. It is best to think of the rcu_data structures
212 as forming yet another level after the leaves. Note that there
213 might be either one, two, or three levels of rcu_node structures,
214 depending on the relationship between CONFIG_RCU_FANOUT and
217 o The numbers separated by the "/" are the qsmask followed
218 by the qsmaskinit. The qsmask will have one bit
219 set for each entity in the next lower level that
220 has not yet checked in for the current grace period.
221 The qsmaskinit will have one bit for each entity that is
222 currently expected to check in during each grace period.
223 The value of qsmaskinit is assigned to that of qsmask
224 at the beginning of each grace period.
226 For example, for "rcu_sched", the qsmask of the first
227 entry of the lowest level is 0x14, meaning that we
228 are still waiting for CPUs 2 and 4 to check in for the
229 current grace period.
231 o The characters separated by the ">" indicate the state
232 of the blocked-tasks lists. A "T" preceding the ">"
233 indicates that at least one task blocked in an RCU
234 read-side critical section blocks the current grace
235 period, while a "." preceding the ">" indicates otherwise.
236 The character following the ">" indicates similarly for
237 the next grace period. A "T" should appear in this
238 field only for rcu-preempt.
240 o The numbers separated by the ":" are the range of CPUs
241 served by this struct rcu_node. This can be helpful
242 in working out how the hierarchy is wired together.
244 For example, the first entry at the lowest level shows
245 "0:5", indicating that it covers CPUs 0 through 5.
247 o The number after the "^" indicates the bit in the
248 next higher level rcu_node structure that this
249 rcu_node structure corresponds to.
251 For example, the first entry at the lowest level shows
252 "^0", indicating that it corresponds to bit zero in
253 the first entry at the middle level.
256 The output of "cat rcu/rcu_pending" looks as follows:
259 0 np=255892 qsp=53936 cbr=0 cng=14417 gpc=10033 gps=24320 nf=6445 nn=146741
260 1 np=261224 qsp=54638 cbr=0 cng=25723 gpc=16310 gps=2849 nf=5912 nn=155792
261 2 np=237496 qsp=49664 cbr=0 cng=2762 gpc=45478 gps=1762 nf=1201 nn=136629
262 3 np=236249 qsp=48766 cbr=0 cng=286 gpc=48049 gps=1218 nf=207 nn=137723
263 4 np=221310 qsp=46850 cbr=0 cng=26 gpc=43161 gps=4634 nf=3529 nn=123110
264 5 np=237332 qsp=48449 cbr=0 cng=54 gpc=47920 gps=3252 nf=201 nn=137456
265 6 np=219995 qsp=46718 cbr=0 cng=50 gpc=42098 gps=6093 nf=4202 nn=120834
266 7 np=249893 qsp=49390 cbr=0 cng=72 gpc=38400 gps=17102 nf=41 nn=144888
268 0 np=146741 qsp=1419 cbr=0 cng=6 gpc=0 gps=0 nf=2 nn=145314
269 1 np=155792 qsp=12597 cbr=0 cng=0 gpc=4 gps=8 nf=3 nn=143180
270 2 np=136629 qsp=18680 cbr=0 cng=0 gpc=7 gps=6 nf=0 nn=117936
271 3 np=137723 qsp=2843 cbr=0 cng=0 gpc=10 gps=7 nf=0 nn=134863
272 4 np=123110 qsp=12433 cbr=0 cng=0 gpc=4 gps=2 nf=0 nn=110671
273 5 np=137456 qsp=4210 cbr=0 cng=0 gpc=6 gps=5 nf=0 nn=133235
274 6 np=120834 qsp=9902 cbr=0 cng=0 gpc=6 gps=3 nf=2 nn=110921
275 7 np=144888 qsp=26336 cbr=0 cng=0 gpc=8 gps=2 nf=0 nn=118542
277 As always, this is once again split into "rcu_sched" and "rcu_bh"
278 portions, with CONFIG_TREE_PREEMPT_RCU kernels having an additional
279 "rcu_preempt" section. The fields are as follows:
281 o "np" is the number of times that __rcu_pending() has been invoked
282 for the corresponding flavor of RCU.
284 o "qsp" is the number of times that the RCU was waiting for a
285 quiescent state from this CPU.
287 o "cbr" is the number of times that this CPU had RCU callbacks
288 that had passed through a grace period, and were thus ready
291 o "cng" is the number of times that this CPU needed another
292 grace period while RCU was idle.
294 o "gpc" is the number of times that an old grace period had
295 completed, but this CPU was not yet aware of it.
297 o "gps" is the number of times that a new grace period had started,
298 but this CPU was not yet aware of it.
300 o "nf" is the number of times that this CPU suspected that the
301 current grace period had run for too long, and thus needed to
304 Please note that "forcing" consists of sending resched IPIs
305 to holdout CPUs. If that CPU really still is in an old RCU
306 read-side critical section, then we really do have to wait for it.
307 The assumption behing "forcing" is that the CPU is not still in
308 an old RCU read-side critical section, but has not yet responded
309 for some other reason.
311 o "nn" is the number of times that this CPU needed nothing. Alert
312 readers will note that the rcu "nn" number for a given CPU very
313 closely matches the rcu_bh "np" number for that same CPU. This
314 is due to short-circuit evaluation in rcu_pending().