xref: /openbmc/linux/kernel/trace/trace_irqsoff.c (revision baa7eb025ab14f3cba2e35c0a8648f9c9f01d24f) 
1 /*
2  * trace irqs off critical timings
3  *
4  * Copyright (C) 2007-2008 Steven Rostedt <srostedt@redhat.com>
5  * Copyright (C) 2008 Ingo Molnar <mingo@redhat.com>
6  *
7  * From code in the latency_tracer, that is:
8  *
9  *  Copyright (C) 2004-2006 Ingo Molnar
10  *  Copyright (C) 2004 William Lee Irwin III
11  */
12 #include <linux/kallsyms.h>
13 #include <linux/debugfs.h>
14 #include <linux/uaccess.h>
15 #include <linux/module.h>
16 #include <linux/ftrace.h>
17 #include <linux/fs.h>
18 
19 #include "trace.h"
20 
21 static struct trace_array		*irqsoff_trace __read_mostly;
22 static int				tracer_enabled __read_mostly;
23 
24 static DEFINE_PER_CPU(int, tracing_cpu);
25 
26 static DEFINE_SPINLOCK(max_trace_lock);
27 
28 enum {
29 	TRACER_IRQS_OFF		= (1 << 1),
30 	TRACER_PREEMPT_OFF	= (1 << 2),
31 };
32 
33 static int trace_type __read_mostly;
34 
35 static int save_lat_flag;
36 
37 static void stop_irqsoff_tracer(struct trace_array *tr, int graph);
38 static int start_irqsoff_tracer(struct trace_array *tr, int graph);
39 
40 #ifdef CONFIG_PREEMPT_TRACER
41 static inline int
42 preempt_trace(void)
43 {
44 	return ((trace_type & TRACER_PREEMPT_OFF) && preempt_count());
45 }
46 #else
47 # define preempt_trace() (0)
48 #endif
49 
50 #ifdef CONFIG_IRQSOFF_TRACER
51 static inline int
52 irq_trace(void)
53 {
54 	return ((trace_type & TRACER_IRQS_OFF) &&
55 		irqs_disabled());
56 }
57 #else
58 # define irq_trace() (0)
59 #endif
60 
61 #define TRACE_DISPLAY_GRAPH	1
62 
63 static struct tracer_opt trace_opts[] = {
64 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
65 	/* display latency trace as call graph */
66 	{ TRACER_OPT(display-graph, TRACE_DISPLAY_GRAPH) },
67 #endif
68 	{ } /* Empty entry */
69 };
70 
71 static struct tracer_flags tracer_flags = {
72 	.val  = 0,
73 	.opts = trace_opts,
74 };
75 
76 #define is_graph() (tracer_flags.val & TRACE_DISPLAY_GRAPH)
77 
78 /*
79  * Sequence count - we record it when starting a measurement and
80  * skip the latency if the sequence has changed - some other section
81  * did a maximum and could disturb our measurement with serial console
82  * printouts, etc. Truly coinciding maximum latencies should be rare
83  * and what happens together happens separately as well, so this doesnt
84  * decrease the validity of the maximum found:
85  */
86 static __cacheline_aligned_in_smp	unsigned long max_sequence;
87 
88 #ifdef CONFIG_FUNCTION_TRACER
89 /*
90  * Prologue for the preempt and irqs off function tracers.
91  *
92  * Returns 1 if it is OK to continue, and data->disabled is
93  *            incremented.
94  *         0 if the trace is to be ignored, and data->disabled
95  *            is kept the same.
96  *
97  * Note, this function is also used outside this ifdef but
98  *  inside the #ifdef of the function graph tracer below.
99  *  This is OK, since the function graph tracer is
100  *  dependent on the function tracer.
101  */
102 static int func_prolog_dec(struct trace_array *tr,
103 			   struct trace_array_cpu **data,
104 			   unsigned long *flags)
105 {
106 	long disabled;
107 	int cpu;
108 
109 	/*
110 	 * Does not matter if we preempt. We test the flags
111 	 * afterward, to see if irqs are disabled or not.
112 	 * If we preempt and get a false positive, the flags
113 	 * test will fail.
114 	 */
115 	cpu = raw_smp_processor_id();
116 	if (likely(!per_cpu(tracing_cpu, cpu)))
117 		return 0;
118 
119 	local_save_flags(*flags);
120 	/* slight chance to get a false positive on tracing_cpu */
121 	if (!irqs_disabled_flags(*flags))
122 		return 0;
123 
124 	*data = tr->data[cpu];
125 	disabled = atomic_inc_return(&(*data)->disabled);
126 
127 	if (likely(disabled == 1))
128 		return 1;
129 
130 	atomic_dec(&(*data)->disabled);
131 
132 	return 0;
133 }
134 
135 /*
136  * irqsoff uses its own tracer function to keep the overhead down:
137  */
138 static void
139 irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip)
140 {
141 	struct trace_array *tr = irqsoff_trace;
142 	struct trace_array_cpu *data;
143 	unsigned long flags;
144 
145 	if (!func_prolog_dec(tr, &data, &flags))
146 		return;
147 
148 	trace_function(tr, ip, parent_ip, flags, preempt_count());
149 
150 	atomic_dec(&data->disabled);
151 }
152 
153 static struct ftrace_ops trace_ops __read_mostly =
154 {
155 	.func = irqsoff_tracer_call,
156 };
157 #endif /* CONFIG_FUNCTION_TRACER */
158 
159 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
160 static int irqsoff_set_flag(u32 old_flags, u32 bit, int set)
161 {
162 	int cpu;
163 
164 	if (!(bit & TRACE_DISPLAY_GRAPH))
165 		return -EINVAL;
166 
167 	if (!(is_graph() ^ set))
168 		return 0;
169 
170 	stop_irqsoff_tracer(irqsoff_trace, !set);
171 
172 	for_each_possible_cpu(cpu)
173 		per_cpu(tracing_cpu, cpu) = 0;
174 
175 	tracing_max_latency = 0;
176 	tracing_reset_online_cpus(irqsoff_trace);
177 
178 	return start_irqsoff_tracer(irqsoff_trace, set);
179 }
180 
181 static int irqsoff_graph_entry(struct ftrace_graph_ent *trace)
182 {
183 	struct trace_array *tr = irqsoff_trace;
184 	struct trace_array_cpu *data;
185 	unsigned long flags;
186 	int ret;
187 	int pc;
188 
189 	if (!func_prolog_dec(tr, &data, &flags))
190 		return 0;
191 
192 	pc = preempt_count();
193 	ret = __trace_graph_entry(tr, trace, flags, pc);
194 	atomic_dec(&data->disabled);
195 
196 	return ret;
197 }
198 
199 static void irqsoff_graph_return(struct ftrace_graph_ret *trace)
200 {
201 	struct trace_array *tr = irqsoff_trace;
202 	struct trace_array_cpu *data;
203 	unsigned long flags;
204 	int pc;
205 
206 	if (!func_prolog_dec(tr, &data, &flags))
207 		return;
208 
209 	pc = preempt_count();
210 	__trace_graph_return(tr, trace, flags, pc);
211 	atomic_dec(&data->disabled);
212 }
213 
214 static void irqsoff_trace_open(struct trace_iterator *iter)
215 {
216 	if (is_graph())
217 		graph_trace_open(iter);
218 
219 }
220 
221 static void irqsoff_trace_close(struct trace_iterator *iter)
222 {
223 	if (iter->private)
224 		graph_trace_close(iter);
225 }
226 
227 #define GRAPH_TRACER_FLAGS (TRACE_GRAPH_PRINT_CPU | \
228 			    TRACE_GRAPH_PRINT_PROC)
229 
230 static enum print_line_t irqsoff_print_line(struct trace_iterator *iter)
231 {
232 	/*
233 	 * In graph mode call the graph tracer output function,
234 	 * otherwise go with the TRACE_FN event handler
235 	 */
236 	if (is_graph())
237 		return print_graph_function_flags(iter, GRAPH_TRACER_FLAGS);
238 
239 	return TRACE_TYPE_UNHANDLED;
240 }
241 
242 static void irqsoff_print_header(struct seq_file *s)
243 {
244 	if (is_graph())
245 		print_graph_headers_flags(s, GRAPH_TRACER_FLAGS);
246 	else
247 		trace_default_header(s);
248 }
249 
250 static void
251 __trace_function(struct trace_array *tr,
252 		 unsigned long ip, unsigned long parent_ip,
253 		 unsigned long flags, int pc)
254 {
255 	if (is_graph())
256 		trace_graph_function(tr, ip, parent_ip, flags, pc);
257 	else
258 		trace_function(tr, ip, parent_ip, flags, pc);
259 }
260 
261 #else
262 #define __trace_function trace_function
263 
264 static int irqsoff_set_flag(u32 old_flags, u32 bit, int set)
265 {
266 	return -EINVAL;
267 }
268 
269 static int irqsoff_graph_entry(struct ftrace_graph_ent *trace)
270 {
271 	return -1;
272 }
273 
274 static enum print_line_t irqsoff_print_line(struct trace_iterator *iter)
275 {
276 	return TRACE_TYPE_UNHANDLED;
277 }
278 
279 static void irqsoff_graph_return(struct ftrace_graph_ret *trace) { }
280 static void irqsoff_print_header(struct seq_file *s) { }
281 static void irqsoff_trace_open(struct trace_iterator *iter) { }
282 static void irqsoff_trace_close(struct trace_iterator *iter) { }
283 #endif /* CONFIG_FUNCTION_GRAPH_TRACER */
284 
285 /*
286  * Should this new latency be reported/recorded?
287  */
288 static int report_latency(cycle_t delta)
289 {
290 	if (tracing_thresh) {
291 		if (delta < tracing_thresh)
292 			return 0;
293 	} else {
294 		if (delta <= tracing_max_latency)
295 			return 0;
296 	}
297 	return 1;
298 }
299 
300 static void
301 check_critical_timing(struct trace_array *tr,
302 		      struct trace_array_cpu *data,
303 		      unsigned long parent_ip,
304 		      int cpu)
305 {
306 	cycle_t T0, T1, delta;
307 	unsigned long flags;
308 	int pc;
309 
310 	T0 = data->preempt_timestamp;
311 	T1 = ftrace_now(cpu);
312 	delta = T1-T0;
313 
314 	local_save_flags(flags);
315 
316 	pc = preempt_count();
317 
318 	if (!report_latency(delta))
319 		goto out;
320 
321 	spin_lock_irqsave(&max_trace_lock, flags);
322 
323 	/* check if we are still the max latency */
324 	if (!report_latency(delta))
325 		goto out_unlock;
326 
327 	__trace_function(tr, CALLER_ADDR0, parent_ip, flags, pc);
328 	/* Skip 5 functions to get to the irq/preempt enable function */
329 	__trace_stack(tr, flags, 5, pc);
330 
331 	if (data->critical_sequence != max_sequence)
332 		goto out_unlock;
333 
334 	data->critical_end = parent_ip;
335 
336 	if (likely(!is_tracing_stopped())) {
337 		tracing_max_latency = delta;
338 		update_max_tr_single(tr, current, cpu);
339 	}
340 
341 	max_sequence++;
342 
343 out_unlock:
344 	spin_unlock_irqrestore(&max_trace_lock, flags);
345 
346 out:
347 	data->critical_sequence = max_sequence;
348 	data->preempt_timestamp = ftrace_now(cpu);
349 	__trace_function(tr, CALLER_ADDR0, parent_ip, flags, pc);
350 }
351 
352 static inline void
353 start_critical_timing(unsigned long ip, unsigned long parent_ip)
354 {
355 	int cpu;
356 	struct trace_array *tr = irqsoff_trace;
357 	struct trace_array_cpu *data;
358 	unsigned long flags;
359 
360 	if (likely(!tracer_enabled))
361 		return;
362 
363 	cpu = raw_smp_processor_id();
364 
365 	if (per_cpu(tracing_cpu, cpu))
366 		return;
367 
368 	data = tr->data[cpu];
369 
370 	if (unlikely(!data) || atomic_read(&data->disabled))
371 		return;
372 
373 	atomic_inc(&data->disabled);
374 
375 	data->critical_sequence = max_sequence;
376 	data->preempt_timestamp = ftrace_now(cpu);
377 	data->critical_start = parent_ip ? : ip;
378 
379 	local_save_flags(flags);
380 
381 	__trace_function(tr, ip, parent_ip, flags, preempt_count());
382 
383 	per_cpu(tracing_cpu, cpu) = 1;
384 
385 	atomic_dec(&data->disabled);
386 }
387 
388 static inline void
389 stop_critical_timing(unsigned long ip, unsigned long parent_ip)
390 {
391 	int cpu;
392 	struct trace_array *tr = irqsoff_trace;
393 	struct trace_array_cpu *data;
394 	unsigned long flags;
395 
396 	cpu = raw_smp_processor_id();
397 	/* Always clear the tracing cpu on stopping the trace */
398 	if (unlikely(per_cpu(tracing_cpu, cpu)))
399 		per_cpu(tracing_cpu, cpu) = 0;
400 	else
401 		return;
402 
403 	if (!tracer_enabled)
404 		return;
405 
406 	data = tr->data[cpu];
407 
408 	if (unlikely(!data) ||
409 	    !data->critical_start || atomic_read(&data->disabled))
410 		return;
411 
412 	atomic_inc(&data->disabled);
413 
414 	local_save_flags(flags);
415 	__trace_function(tr, ip, parent_ip, flags, preempt_count());
416 	check_critical_timing(tr, data, parent_ip ? : ip, cpu);
417 	data->critical_start = 0;
418 	atomic_dec(&data->disabled);
419 }
420 
421 /* start and stop critical timings used to for stoppage (in idle) */
422 void start_critical_timings(void)
423 {
424 	if (preempt_trace() || irq_trace())
425 		start_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
426 }
427 EXPORT_SYMBOL_GPL(start_critical_timings);
428 
429 void stop_critical_timings(void)
430 {
431 	if (preempt_trace() || irq_trace())
432 		stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
433 }
434 EXPORT_SYMBOL_GPL(stop_critical_timings);
435 
436 #ifdef CONFIG_IRQSOFF_TRACER
437 #ifdef CONFIG_PROVE_LOCKING
438 void time_hardirqs_on(unsigned long a0, unsigned long a1)
439 {
440 	if (!preempt_trace() && irq_trace())
441 		stop_critical_timing(a0, a1);
442 }
443 
444 void time_hardirqs_off(unsigned long a0, unsigned long a1)
445 {
446 	if (!preempt_trace() && irq_trace())
447 		start_critical_timing(a0, a1);
448 }
449 
450 #else /* !CONFIG_PROVE_LOCKING */
451 
452 /*
453  * Stubs:
454  */
455 
456 void early_boot_irqs_off(void)
457 {
458 }
459 
460 void early_boot_irqs_on(void)
461 {
462 }
463 
464 void trace_softirqs_on(unsigned long ip)
465 {
466 }
467 
468 void trace_softirqs_off(unsigned long ip)
469 {
470 }
471 
472 inline void print_irqtrace_events(struct task_struct *curr)
473 {
474 }
475 
476 /*
477  * We are only interested in hardirq on/off events:
478  */
479 void trace_hardirqs_on(void)
480 {
481 	if (!preempt_trace() && irq_trace())
482 		stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
483 }
484 EXPORT_SYMBOL(trace_hardirqs_on);
485 
486 void trace_hardirqs_off(void)
487 {
488 	if (!preempt_trace() && irq_trace())
489 		start_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
490 }
491 EXPORT_SYMBOL(trace_hardirqs_off);
492 
493 void trace_hardirqs_on_caller(unsigned long caller_addr)
494 {
495 	if (!preempt_trace() && irq_trace())
496 		stop_critical_timing(CALLER_ADDR0, caller_addr);
497 }
498 EXPORT_SYMBOL(trace_hardirqs_on_caller);
499 
500 void trace_hardirqs_off_caller(unsigned long caller_addr)
501 {
502 	if (!preempt_trace() && irq_trace())
503 		start_critical_timing(CALLER_ADDR0, caller_addr);
504 }
505 EXPORT_SYMBOL(trace_hardirqs_off_caller);
506 
507 #endif /* CONFIG_PROVE_LOCKING */
508 #endif /*  CONFIG_IRQSOFF_TRACER */
509 
510 #ifdef CONFIG_PREEMPT_TRACER
511 void trace_preempt_on(unsigned long a0, unsigned long a1)
512 {
513 	if (preempt_trace())
514 		stop_critical_timing(a0, a1);
515 }
516 
517 void trace_preempt_off(unsigned long a0, unsigned long a1)
518 {
519 	if (preempt_trace())
520 		start_critical_timing(a0, a1);
521 }
522 #endif /* CONFIG_PREEMPT_TRACER */
523 
524 static int start_irqsoff_tracer(struct trace_array *tr, int graph)
525 {
526 	int ret = 0;
527 
528 	if (!graph)
529 		ret = register_ftrace_function(&trace_ops);
530 	else
531 		ret = register_ftrace_graph(&irqsoff_graph_return,
532 					    &irqsoff_graph_entry);
533 
534 	if (!ret && tracing_is_enabled())
535 		tracer_enabled = 1;
536 	else
537 		tracer_enabled = 0;
538 
539 	return ret;
540 }
541 
542 static void stop_irqsoff_tracer(struct trace_array *tr, int graph)
543 {
544 	tracer_enabled = 0;
545 
546 	if (!graph)
547 		unregister_ftrace_function(&trace_ops);
548 	else
549 		unregister_ftrace_graph();
550 }
551 
552 static void __irqsoff_tracer_init(struct trace_array *tr)
553 {
554 	save_lat_flag = trace_flags & TRACE_ITER_LATENCY_FMT;
555 	trace_flags |= TRACE_ITER_LATENCY_FMT;
556 
557 	tracing_max_latency = 0;
558 	irqsoff_trace = tr;
559 	/* make sure that the tracer is visible */
560 	smp_wmb();
561 	tracing_reset_online_cpus(tr);
562 
563 	if (start_irqsoff_tracer(tr, is_graph()))
564 		printk(KERN_ERR "failed to start irqsoff tracer\n");
565 }
566 
567 static void irqsoff_tracer_reset(struct trace_array *tr)
568 {
569 	stop_irqsoff_tracer(tr, is_graph());
570 
571 	if (!save_lat_flag)
572 		trace_flags &= ~TRACE_ITER_LATENCY_FMT;
573 }
574 
575 static void irqsoff_tracer_start(struct trace_array *tr)
576 {
577 	tracer_enabled = 1;
578 }
579 
580 static void irqsoff_tracer_stop(struct trace_array *tr)
581 {
582 	tracer_enabled = 0;
583 }
584 
585 #ifdef CONFIG_IRQSOFF_TRACER
586 static int irqsoff_tracer_init(struct trace_array *tr)
587 {
588 	trace_type = TRACER_IRQS_OFF;
589 
590 	__irqsoff_tracer_init(tr);
591 	return 0;
592 }
593 static struct tracer irqsoff_tracer __read_mostly =
594 {
595 	.name		= "irqsoff",
596 	.init		= irqsoff_tracer_init,
597 	.reset		= irqsoff_tracer_reset,
598 	.start		= irqsoff_tracer_start,
599 	.stop		= irqsoff_tracer_stop,
600 	.print_max	= 1,
601 	.print_header   = irqsoff_print_header,
602 	.print_line     = irqsoff_print_line,
603 	.flags		= &tracer_flags,
604 	.set_flag	= irqsoff_set_flag,
605 #ifdef CONFIG_FTRACE_SELFTEST
606 	.selftest    = trace_selftest_startup_irqsoff,
607 #endif
608 	.open           = irqsoff_trace_open,
609 	.close          = irqsoff_trace_close,
610 	.use_max_tr	= 1,
611 };
612 # define register_irqsoff(trace) register_tracer(&trace)
613 #else
614 # define register_irqsoff(trace) do { } while (0)
615 #endif
616 
617 #ifdef CONFIG_PREEMPT_TRACER
618 static int preemptoff_tracer_init(struct trace_array *tr)
619 {
620 	trace_type = TRACER_PREEMPT_OFF;
621 
622 	__irqsoff_tracer_init(tr);
623 	return 0;
624 }
625 
626 static struct tracer preemptoff_tracer __read_mostly =
627 {
628 	.name		= "preemptoff",
629 	.init		= preemptoff_tracer_init,
630 	.reset		= irqsoff_tracer_reset,
631 	.start		= irqsoff_tracer_start,
632 	.stop		= irqsoff_tracer_stop,
633 	.print_max	= 1,
634 	.print_header   = irqsoff_print_header,
635 	.print_line     = irqsoff_print_line,
636 	.flags		= &tracer_flags,
637 	.set_flag	= irqsoff_set_flag,
638 #ifdef CONFIG_FTRACE_SELFTEST
639 	.selftest    = trace_selftest_startup_preemptoff,
640 #endif
641 	.open		= irqsoff_trace_open,
642 	.close		= irqsoff_trace_close,
643 	.use_max_tr	= 1,
644 };
645 # define register_preemptoff(trace) register_tracer(&trace)
646 #else
647 # define register_preemptoff(trace) do { } while (0)
648 #endif
649 
650 #if defined(CONFIG_IRQSOFF_TRACER) && \
651 	defined(CONFIG_PREEMPT_TRACER)
652 
653 static int preemptirqsoff_tracer_init(struct trace_array *tr)
654 {
655 	trace_type = TRACER_IRQS_OFF | TRACER_PREEMPT_OFF;
656 
657 	__irqsoff_tracer_init(tr);
658 	return 0;
659 }
660 
661 static struct tracer preemptirqsoff_tracer __read_mostly =
662 {
663 	.name		= "preemptirqsoff",
664 	.init		= preemptirqsoff_tracer_init,
665 	.reset		= irqsoff_tracer_reset,
666 	.start		= irqsoff_tracer_start,
667 	.stop		= irqsoff_tracer_stop,
668 	.print_max	= 1,
669 	.print_header   = irqsoff_print_header,
670 	.print_line     = irqsoff_print_line,
671 	.flags		= &tracer_flags,
672 	.set_flag	= irqsoff_set_flag,
673 #ifdef CONFIG_FTRACE_SELFTEST
674 	.selftest    = trace_selftest_startup_preemptirqsoff,
675 #endif
676 	.open		= irqsoff_trace_open,
677 	.close		= irqsoff_trace_close,
678 	.use_max_tr	= 1,
679 };
680 
681 # define register_preemptirqsoff(trace) register_tracer(&trace)
682 #else
683 # define register_preemptirqsoff(trace) do { } while (0)
684 #endif
685 
686 __init static int init_irqsoff_tracer(void)
687 {
688 	register_irqsoff(irqsoff_tracer);
689 	register_preemptoff(preemptoff_tracer);
690 	register_preemptirqsoff(preemptirqsoff_tracer);
691 
692 	return 0;
693 }
694 device_initcall(init_irqsoff_tracer);
695