1.. SPDX-License-Identifier: GPL-2.0
2
3==============================
4Using RCU's CPU Stall Detector
5==============================
6
7This document first discusses what sorts of issues RCU's CPU stall
8detector can locate, and then discusses kernel parameters and Kconfig
9options that can be used to fine-tune the detector's operation.  Finally,
10this document explains the stall detector's "splat" format.
11
12
13What Causes RCU CPU Stall Warnings?
14===================================
15
16So your kernel printed an RCU CPU stall warning.  The next question is
17"What caused it?"  The following problems can result in RCU CPU stall
18warnings:
19
20-	A CPU looping in an RCU read-side critical section.
21
22-	A CPU looping with interrupts disabled.
23
24-	A CPU looping with preemption disabled.
25
26-	A CPU looping with bottom halves disabled.
27
28-	For !CONFIG_PREEMPTION kernels, a CPU looping anywhere in the kernel
29	without invoking schedule().  If the looping in the kernel is
30	really expected and desirable behavior, you might need to add
31	some calls to cond_resched().
32
33-	Booting Linux using a console connection that is too slow to
34	keep up with the boot-time console-message rate.  For example,
35	a 115Kbaud serial console can be *way* too slow to keep up
36	with boot-time message rates, and will frequently result in
37	RCU CPU stall warning messages.  Especially if you have added
38	debug printk()s.
39
40-	Anything that prevents RCU's grace-period kthreads from running.
41	This can result in the "All QSes seen" console-log message.
42	This message will include information on when the kthread last
43	ran and how often it should be expected to run.  It can also
44	result in the ``rcu_.*kthread starved for`` console-log message,
45	which will include additional debugging information.
46
47-	A CPU-bound real-time task in a CONFIG_PREEMPTION kernel, which might
48	happen to preempt a low-priority task in the middle of an RCU
49	read-side critical section.   This is especially damaging if
50	that low-priority task is not permitted to run on any other CPU,
51	in which case the next RCU grace period can never complete, which
52	will eventually cause the system to run out of memory and hang.
53	While the system is in the process of running itself out of
54	memory, you might see stall-warning messages.
55
56-	A CPU-bound real-time task in a CONFIG_PREEMPT_RT kernel that
57	is running at a higher priority than the RCU softirq threads.
58	This will prevent RCU callbacks from ever being invoked,
59	and in a CONFIG_PREEMPT_RCU kernel will further prevent
60	RCU grace periods from ever completing.  Either way, the
61	system will eventually run out of memory and hang.  In the
62	CONFIG_PREEMPT_RCU case, you might see stall-warning
63	messages.
64
65	You can use the rcutree.kthread_prio kernel boot parameter to
66	increase the scheduling priority of RCU's kthreads, which can
67	help avoid this problem.  However, please note that doing this
68	can increase your system's context-switch rate and thus degrade
69	performance.
70
71-	A periodic interrupt whose handler takes longer than the time
72	interval between successive pairs of interrupts.  This can
73	prevent RCU's kthreads and softirq handlers from running.
74	Note that certain high-overhead debugging options, for example
75	the function_graph tracer, can result in interrupt handler taking
76	considerably longer than normal, which can in turn result in
77	RCU CPU stall warnings.
78
79-	Testing a workload on a fast system, tuning the stall-warning
80	timeout down to just barely avoid RCU CPU stall warnings, and then
81	running the same workload with the same stall-warning timeout on a
82	slow system.  Note that thermal throttling and on-demand governors
83	can cause a single system to be sometimes fast and sometimes slow!
84
85-	A hardware or software issue shuts off the scheduler-clock
86	interrupt on a CPU that is not in dyntick-idle mode.  This
87	problem really has happened, and seems to be most likely to
88	result in RCU CPU stall warnings for CONFIG_NO_HZ_COMMON=n kernels.
89
90-	A hardware or software issue that prevents time-based wakeups
91	from occurring.  These issues can range from misconfigured or
92	buggy timer hardware through bugs in the interrupt or exception
93	path (whether hardware, firmware, or software) through bugs
94	in Linux's timer subsystem through bugs in the scheduler, and,
95	yes, even including bugs in RCU itself.  It can also result in
96	the ``rcu_.*timer wakeup didn't happen for`` console-log message,
97	which will include additional debugging information.
98
99-	A bug in the RCU implementation.
100
101-	A hardware failure.  This is quite unlikely, but has occurred
102	at least once in real life.  A CPU failed in a running system,
103	becoming unresponsive, but not causing an immediate crash.
104	This resulted in a series of RCU CPU stall warnings, eventually
105	leading the realization that the CPU had failed.
106
107The RCU, RCU-sched, and RCU-tasks implementations have CPU stall warning.
108Note that SRCU does *not* have CPU stall warnings.  Please note that
109RCU only detects CPU stalls when there is a grace period in progress.
110No grace period, no CPU stall warnings.
111
112To diagnose the cause of the stall, inspect the stack traces.
113The offending function will usually be near the top of the stack.
114If you have a series of stall warnings from a single extended stall,
115comparing the stack traces can often help determine where the stall
116is occurring, which will usually be in the function nearest the top of
117that portion of the stack which remains the same from trace to trace.
118If you can reliably trigger the stall, ftrace can be quite helpful.
119
120RCU bugs can often be debugged with the help of CONFIG_RCU_TRACE
121and with RCU's event tracing.  For information on RCU's event tracing,
122see include/trace/events/rcu.h.
123
124
125Fine-Tuning the RCU CPU Stall Detector
126======================================
127
128The rcuupdate.rcu_cpu_stall_suppress module parameter disables RCU's
129CPU stall detector, which detects conditions that unduly delay RCU grace
130periods.  This module parameter enables CPU stall detection by default,
131but may be overridden via boot-time parameter or at runtime via sysfs.
132The stall detector's idea of what constitutes "unduly delayed" is
133controlled by a set of kernel configuration variables and cpp macros:
134
135CONFIG_RCU_CPU_STALL_TIMEOUT
136----------------------------
137
138	This kernel configuration parameter defines the period of time
139	that RCU will wait from the beginning of a grace period until it
140	issues an RCU CPU stall warning.  This time period is normally
141	21 seconds.
142
143	This configuration parameter may be changed at runtime via the
144	/sys/module/rcupdate/parameters/rcu_cpu_stall_timeout, however
145	this parameter is checked only at the beginning of a cycle.
146	So if you are 10 seconds into a 40-second stall, setting this
147	sysfs parameter to (say) five will shorten the timeout for the
148	*next* stall, or the following warning for the current stall
149	(assuming the stall lasts long enough).  It will not affect the
150	timing of the next warning for the current stall.
151
152	Stall-warning messages may be enabled and disabled completely via
153	/sys/module/rcupdate/parameters/rcu_cpu_stall_suppress.
154
155RCU_STALL_DELAY_DELTA
156---------------------
157
158	Although the lockdep facility is extremely useful, it does add
159	some overhead.  Therefore, under CONFIG_PROVE_RCU, the
160	RCU_STALL_DELAY_DELTA macro allows five extra seconds before
161	giving an RCU CPU stall warning message.  (This is a cpp
162	macro, not a kernel configuration parameter.)
163
164RCU_STALL_RAT_DELAY
165-------------------
166
167	The CPU stall detector tries to make the offending CPU print its
168	own warnings, as this often gives better-quality stack traces.
169	However, if the offending CPU does not detect its own stall in
170	the number of jiffies specified by RCU_STALL_RAT_DELAY, then
171	some other CPU will complain.  This delay is normally set to
172	two jiffies.  (This is a cpp macro, not a kernel configuration
173	parameter.)
174
175rcupdate.rcu_task_stall_timeout
176-------------------------------
177
178	This boot/sysfs parameter controls the RCU-tasks stall warning
179	interval.  A value of zero or less suppresses RCU-tasks stall
180	warnings.  A positive value sets the stall-warning interval
181	in seconds.  An RCU-tasks stall warning starts with the line:
182
183		INFO: rcu_tasks detected stalls on tasks:
184
185	And continues with the output of sched_show_task() for each
186	task stalling the current RCU-tasks grace period.
187
188
189Interpreting RCU's CPU Stall-Detector "Splats"
190==============================================
191
192For non-RCU-tasks flavors of RCU, when a CPU detects that some other
193CPU is stalling, it will print a message similar to the following::
194
195	INFO: rcu_sched detected stalls on CPUs/tasks:
196	2-...: (3 GPs behind) idle=06c/0/0 softirq=1453/1455 fqs=0
197	16-...: (0 ticks this GP) idle=81c/0/0 softirq=764/764 fqs=0
198	(detected by 32, t=2603 jiffies, g=7075, q=625)
199
200This message indicates that CPU 32 detected that CPUs 2 and 16 were both
201causing stalls, and that the stall was affecting RCU-sched.  This message
202will normally be followed by stack dumps for each CPU.  Please note that
203PREEMPT_RCU builds can be stalled by tasks as well as by CPUs, and that
204the tasks will be indicated by PID, for example, "P3421".  It is even
205possible for an rcu_state stall to be caused by both CPUs *and* tasks,
206in which case the offending CPUs and tasks will all be called out in the list.
207In some cases, CPUs will detect themselves stalling, which will result
208in a self-detected stall.
209
210CPU 2's "(3 GPs behind)" indicates that this CPU has not interacted with
211the RCU core for the past three grace periods.  In contrast, CPU 16's "(0
212ticks this GP)" indicates that this CPU has not taken any scheduling-clock
213interrupts during the current stalled grace period.
214
215The "idle=" portion of the message prints the dyntick-idle state.
216The hex number before the first "/" is the low-order 12 bits of the
217dynticks counter, which will have an even-numbered value if the CPU
218is in dyntick-idle mode and an odd-numbered value otherwise.  The hex
219number between the two "/"s is the value of the nesting, which will be
220a small non-negative number if in the idle loop (as shown above) and a
221very large positive number otherwise.
222
223The "softirq=" portion of the message tracks the number of RCU softirq
224handlers that the stalled CPU has executed.  The number before the "/"
225is the number that had executed since boot at the time that this CPU
226last noted the beginning of a grace period, which might be the current
227(stalled) grace period, or it might be some earlier grace period (for
228example, if the CPU might have been in dyntick-idle mode for an extended
229time period).  The number after the "/" is the number that have executed
230since boot until the current time.  If this latter number stays constant
231across repeated stall-warning messages, it is possible that RCU's softirq
232handlers are no longer able to execute on this CPU.  This can happen if
233the stalled CPU is spinning with interrupts are disabled, or, in -rt
234kernels, if a high-priority process is starving RCU's softirq handler.
235
236The "fqs=" shows the number of force-quiescent-state idle/offline
237detection passes that the grace-period kthread has made across this
238CPU since the last time that this CPU noted the beginning of a grace
239period.
240
241The "detected by" line indicates which CPU detected the stall (in this
242case, CPU 32), how many jiffies have elapsed since the start of the grace
243period (in this case 2603), the grace-period sequence number (7075), and
244an estimate of the total number of RCU callbacks queued across all CPUs
245(625 in this case).
246
247In kernels with CONFIG_RCU_FAST_NO_HZ, more information is printed
248for each CPU::
249
250	0: (64628 ticks this GP) idle=dd5/3fffffffffffffff/0 softirq=82/543 last_accelerate: a345/d342 dyntick_enabled: 1
251
252The "last_accelerate:" prints the low-order 16 bits (in hex) of the
253jiffies counter when this CPU last invoked rcu_try_advance_all_cbs()
254from rcu_needs_cpu() or last invoked rcu_accelerate_cbs() from
255rcu_prepare_for_idle(). "dyntick_enabled: 1" indicates that dyntick-idle
256processing is enabled.
257
258If the grace period ends just as the stall warning starts printing,
259there will be a spurious stall-warning message, which will include
260the following::
261
262	INFO: Stall ended before state dump start
263
264This is rare, but does happen from time to time in real life.  It is also
265possible for a zero-jiffy stall to be flagged in this case, depending
266on how the stall warning and the grace-period initialization happen to
267interact.  Please note that it is not possible to entirely eliminate this
268sort of false positive without resorting to things like stop_machine(),
269which is overkill for this sort of problem.
270
271If all CPUs and tasks have passed through quiescent states, but the
272grace period has nevertheless failed to end, the stall-warning splat
273will include something like the following::
274
275	All QSes seen, last rcu_preempt kthread activity 23807 (4297905177-4297881370), jiffies_till_next_fqs=3, root ->qsmask 0x0
276
277The "23807" indicates that it has been more than 23 thousand jiffies
278since the grace-period kthread ran.  The "jiffies_till_next_fqs"
279indicates how frequently that kthread should run, giving the number
280of jiffies between force-quiescent-state scans, in this case three,
281which is way less than 23807.  Finally, the root rcu_node structure's
282->qsmask field is printed, which will normally be zero.
283
284If the relevant grace-period kthread has been unable to run prior to
285the stall warning, as was the case in the "All QSes seen" line above,
286the following additional line is printed::
287
288	rcu_sched kthread starved for 23807 jiffies! g7075 f0x0 RCU_GP_WAIT_FQS(3) ->state=0x1 ->cpu=5
289	Unless rcu_sched kthread gets sufficient CPU time, OOM is now expected behavior.
290
291Starving the grace-period kthreads of CPU time can of course result
292in RCU CPU stall warnings even when all CPUs and tasks have passed
293through the required quiescent states.  The "g" number shows the current
294grace-period sequence number, the "f" precedes the ->gp_flags command
295to the grace-period kthread, the "RCU_GP_WAIT_FQS" indicates that the
296kthread is waiting for a short timeout, the "state" precedes value of the
297task_struct ->state field, and the "cpu" indicates that the grace-period
298kthread last ran on CPU 5.
299
300If the relevant grace-period kthread does not wake from FQS wait in a
301reasonable time, then the following additional line is printed::
302
303	kthread timer wakeup didn't happen for 23804 jiffies! g7076 f0x0 RCU_GP_WAIT_FQS(5) ->state=0x402
304
305The "23804" indicates that kthread's timer expired more than 23 thousand
306jiffies ago.  The rest of the line has meaning similar to the kthread
307starvation case.
308
309Additionally, the following line is printed::
310
311	Possible timer handling issue on cpu=4 timer-softirq=11142
312
313Here "cpu" indicates that the grace-period kthread last ran on CPU 4,
314where it queued the fqs timer.  The number following the "timer-softirq"
315is the current ``TIMER_SOFTIRQ`` count on cpu 4.  If this value does not
316change on successive RCU CPU stall warnings, there is further reason to
317suspect a timer problem.
318
319These messages are usually followed by stack dumps of the CPUs and tasks
320involved in the stall.  These stack traces can help you locate the cause
321of the stall, keeping in mind that the CPU detecting the stall will have
322an interrupt frame that is mainly devoted to detecting the stall.
323
324
325Multiple Warnings From One Stall
326================================
327
328If a stall lasts long enough, multiple stall-warning messages will
329be printed for it.  The second and subsequent messages are printed at
330longer intervals, so that the time between (say) the first and second
331message will be about three times the interval between the beginning
332of the stall and the first message.  It can be helpful to compare the
333stack dumps for the different messages for the same stalled grace period.
334
335
336Stall Warnings for Expedited Grace Periods
337==========================================
338
339If an expedited grace period detects a stall, it will place a message
340like the following in dmesg::
341
342	INFO: rcu_sched detected expedited stalls on CPUs/tasks: { 7-... } 21119 jiffies s: 73 root: 0x2/.
343
344This indicates that CPU 7 has failed to respond to a reschedule IPI.
345The three periods (".") following the CPU number indicate that the CPU
346is online (otherwise the first period would instead have been "O"),
347that the CPU was online at the beginning of the expedited grace period
348(otherwise the second period would have instead been "o"), and that
349the CPU has been online at least once since boot (otherwise, the third
350period would instead have been "N").  The number before the "jiffies"
351indicates that the expedited grace period has been going on for 21,119
352jiffies.  The number following the "s:" indicates that the expedited
353grace-period sequence counter is 73.  The fact that this last value is
354odd indicates that an expedited grace period is in flight.  The number
355following "root:" is a bitmask that indicates which children of the root
356rcu_node structure correspond to CPUs and/or tasks that are blocking the
357current expedited grace period.  If the tree had more than one level,
358additional hex numbers would be printed for the states of the other
359rcu_node structures in the tree.
360
361As with normal grace periods, PREEMPT_RCU builds can be stalled by
362tasks as well as by CPUs, and that the tasks will be indicated by PID,
363for example, "P3421".
364
365It is entirely possible to see stall warnings from normal and from
366expedited grace periods at about the same time during the same run.
367