Fix race/oops in tty layer after BKL pushdown
[linux-2.6] / kernel / trace / trace_irqsoff.c
1 /*
2  * trace irqs off criticall 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 #ifdef CONFIG_PREEMPT_TRACER
36 static inline int
37 preempt_trace(void)
38 {
39         return ((trace_type & TRACER_PREEMPT_OFF) && preempt_count());
40 }
41 #else
42 # define preempt_trace() (0)
43 #endif
44
45 #ifdef CONFIG_IRQSOFF_TRACER
46 static inline int
47 irq_trace(void)
48 {
49         return ((trace_type & TRACER_IRQS_OFF) &&
50                 irqs_disabled());
51 }
52 #else
53 # define irq_trace() (0)
54 #endif
55
56 /*
57  * Sequence count - we record it when starting a measurement and
58  * skip the latency if the sequence has changed - some other section
59  * did a maximum and could disturb our measurement with serial console
60  * printouts, etc. Truly coinciding maximum latencies should be rare
61  * and what happens together happens separately as well, so this doesnt
62  * decrease the validity of the maximum found:
63  */
64 static __cacheline_aligned_in_smp       unsigned long max_sequence;
65
66 #ifdef CONFIG_FTRACE
67 /*
68  * irqsoff uses its own tracer function to keep the overhead down:
69  */
70 static void
71 irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip)
72 {
73         struct trace_array *tr = irqsoff_trace;
74         struct trace_array_cpu *data;
75         unsigned long flags;
76         long disabled;
77         int cpu;
78
79         /*
80          * Does not matter if we preempt. We test the flags
81          * afterward, to see if irqs are disabled or not.
82          * If we preempt and get a false positive, the flags
83          * test will fail.
84          */
85         cpu = raw_smp_processor_id();
86         if (likely(!per_cpu(tracing_cpu, cpu)))
87                 return;
88
89         local_save_flags(flags);
90         /* slight chance to get a false positive on tracing_cpu */
91         if (!irqs_disabled_flags(flags))
92                 return;
93
94         data = tr->data[cpu];
95         disabled = atomic_inc_return(&data->disabled);
96
97         if (likely(disabled == 1))
98                 trace_function(tr, data, ip, parent_ip, flags);
99
100         atomic_dec(&data->disabled);
101 }
102
103 static struct ftrace_ops trace_ops __read_mostly =
104 {
105         .func = irqsoff_tracer_call,
106 };
107 #endif /* CONFIG_FTRACE */
108
109 /*
110  * Should this new latency be reported/recorded?
111  */
112 static int report_latency(cycle_t delta)
113 {
114         if (tracing_thresh) {
115                 if (delta < tracing_thresh)
116                         return 0;
117         } else {
118                 if (delta <= tracing_max_latency)
119                         return 0;
120         }
121         return 1;
122 }
123
124 static void
125 check_critical_timing(struct trace_array *tr,
126                       struct trace_array_cpu *data,
127                       unsigned long parent_ip,
128                       int cpu)
129 {
130         unsigned long latency, t0, t1;
131         cycle_t T0, T1, delta;
132         unsigned long flags;
133
134         /*
135          * usecs conversion is slow so we try to delay the conversion
136          * as long as possible:
137          */
138         T0 = data->preempt_timestamp;
139         T1 = ftrace_now(cpu);
140         delta = T1-T0;
141
142         local_save_flags(flags);
143
144         if (!report_latency(delta))
145                 goto out;
146
147         spin_lock_irqsave(&max_trace_lock, flags);
148
149         /* check if we are still the max latency */
150         if (!report_latency(delta))
151                 goto out_unlock;
152
153         trace_function(tr, data, CALLER_ADDR0, parent_ip, flags);
154
155         latency = nsecs_to_usecs(delta);
156
157         if (data->critical_sequence != max_sequence)
158                 goto out_unlock;
159
160         tracing_max_latency = delta;
161         t0 = nsecs_to_usecs(T0);
162         t1 = nsecs_to_usecs(T1);
163
164         data->critical_end = parent_ip;
165
166         update_max_tr_single(tr, current, cpu);
167
168         max_sequence++;
169
170 out_unlock:
171         spin_unlock_irqrestore(&max_trace_lock, flags);
172
173 out:
174         data->critical_sequence = max_sequence;
175         data->preempt_timestamp = ftrace_now(cpu);
176         tracing_reset(data);
177         trace_function(tr, data, CALLER_ADDR0, parent_ip, flags);
178 }
179
180 static inline void
181 start_critical_timing(unsigned long ip, unsigned long parent_ip)
182 {
183         int cpu;
184         struct trace_array *tr = irqsoff_trace;
185         struct trace_array_cpu *data;
186         unsigned long flags;
187
188         if (likely(!tracer_enabled))
189                 return;
190
191         cpu = raw_smp_processor_id();
192
193         if (per_cpu(tracing_cpu, cpu))
194                 return;
195
196         data = tr->data[cpu];
197
198         if (unlikely(!data) || atomic_read(&data->disabled))
199                 return;
200
201         atomic_inc(&data->disabled);
202
203         data->critical_sequence = max_sequence;
204         data->preempt_timestamp = ftrace_now(cpu);
205         data->critical_start = parent_ip ? : ip;
206         tracing_reset(data);
207
208         local_save_flags(flags);
209
210         trace_function(tr, data, ip, parent_ip, flags);
211
212         per_cpu(tracing_cpu, cpu) = 1;
213
214         atomic_dec(&data->disabled);
215 }
216
217 static inline void
218 stop_critical_timing(unsigned long ip, unsigned long parent_ip)
219 {
220         int cpu;
221         struct trace_array *tr = irqsoff_trace;
222         struct trace_array_cpu *data;
223         unsigned long flags;
224
225         cpu = raw_smp_processor_id();
226         /* Always clear the tracing cpu on stopping the trace */
227         if (unlikely(per_cpu(tracing_cpu, cpu)))
228                 per_cpu(tracing_cpu, cpu) = 0;
229         else
230                 return;
231
232         if (!tracer_enabled)
233                 return;
234
235         data = tr->data[cpu];
236
237         if (unlikely(!data) || unlikely(!head_page(data)) ||
238             !data->critical_start || atomic_read(&data->disabled))
239                 return;
240
241         atomic_inc(&data->disabled);
242
243         local_save_flags(flags);
244         trace_function(tr, data, ip, parent_ip, flags);
245         check_critical_timing(tr, data, parent_ip ? : ip, cpu);
246         data->critical_start = 0;
247         atomic_dec(&data->disabled);
248 }
249
250 /* start and stop critical timings used to for stoppage (in idle) */
251 void start_critical_timings(void)
252 {
253         if (preempt_trace() || irq_trace())
254                 start_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
255 }
256 EXPORT_SYMBOL_GPL(start_critical_timings);
257
258 void stop_critical_timings(void)
259 {
260         if (preempt_trace() || irq_trace())
261                 stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
262 }
263 EXPORT_SYMBOL_GPL(stop_critical_timings);
264
265 #ifdef CONFIG_IRQSOFF_TRACER
266 #ifdef CONFIG_PROVE_LOCKING
267 void time_hardirqs_on(unsigned long a0, unsigned long a1)
268 {
269         if (!preempt_trace() && irq_trace())
270                 stop_critical_timing(a0, a1);
271 }
272
273 void time_hardirqs_off(unsigned long a0, unsigned long a1)
274 {
275         if (!preempt_trace() && irq_trace())
276                 start_critical_timing(a0, a1);
277 }
278
279 #else /* !CONFIG_PROVE_LOCKING */
280
281 /*
282  * Stubs:
283  */
284
285 void early_boot_irqs_off(void)
286 {
287 }
288
289 void early_boot_irqs_on(void)
290 {
291 }
292
293 void trace_softirqs_on(unsigned long ip)
294 {
295 }
296
297 void trace_softirqs_off(unsigned long ip)
298 {
299 }
300
301 inline void print_irqtrace_events(struct task_struct *curr)
302 {
303 }
304
305 /*
306  * We are only interested in hardirq on/off events:
307  */
308 void trace_hardirqs_on(void)
309 {
310         if (!preempt_trace() && irq_trace())
311                 stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
312 }
313 EXPORT_SYMBOL(trace_hardirqs_on);
314
315 void trace_hardirqs_off(void)
316 {
317         if (!preempt_trace() && irq_trace())
318                 start_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
319 }
320 EXPORT_SYMBOL(trace_hardirqs_off);
321
322 void trace_hardirqs_on_caller(unsigned long caller_addr)
323 {
324         if (!preempt_trace() && irq_trace())
325                 stop_critical_timing(CALLER_ADDR0, caller_addr);
326 }
327 EXPORT_SYMBOL(trace_hardirqs_on_caller);
328
329 void trace_hardirqs_off_caller(unsigned long caller_addr)
330 {
331         if (!preempt_trace() && irq_trace())
332                 start_critical_timing(CALLER_ADDR0, caller_addr);
333 }
334 EXPORT_SYMBOL(trace_hardirqs_off_caller);
335
336 #endif /* CONFIG_PROVE_LOCKING */
337 #endif /*  CONFIG_IRQSOFF_TRACER */
338
339 #ifdef CONFIG_PREEMPT_TRACER
340 void trace_preempt_on(unsigned long a0, unsigned long a1)
341 {
342         if (preempt_trace())
343                 stop_critical_timing(a0, a1);
344 }
345
346 void trace_preempt_off(unsigned long a0, unsigned long a1)
347 {
348         if (preempt_trace())
349                 start_critical_timing(a0, a1);
350 }
351 #endif /* CONFIG_PREEMPT_TRACER */
352
353 static void start_irqsoff_tracer(struct trace_array *tr)
354 {
355         register_ftrace_function(&trace_ops);
356         tracer_enabled = 1;
357 }
358
359 static void stop_irqsoff_tracer(struct trace_array *tr)
360 {
361         tracer_enabled = 0;
362         unregister_ftrace_function(&trace_ops);
363 }
364
365 static void __irqsoff_tracer_init(struct trace_array *tr)
366 {
367         irqsoff_trace = tr;
368         /* make sure that the tracer is visible */
369         smp_wmb();
370
371         if (tr->ctrl)
372                 start_irqsoff_tracer(tr);
373 }
374
375 static void irqsoff_tracer_reset(struct trace_array *tr)
376 {
377         if (tr->ctrl)
378                 stop_irqsoff_tracer(tr);
379 }
380
381 static void irqsoff_tracer_ctrl_update(struct trace_array *tr)
382 {
383         if (tr->ctrl)
384                 start_irqsoff_tracer(tr);
385         else
386                 stop_irqsoff_tracer(tr);
387 }
388
389 static void irqsoff_tracer_open(struct trace_iterator *iter)
390 {
391         /* stop the trace while dumping */
392         if (iter->tr->ctrl)
393                 stop_irqsoff_tracer(iter->tr);
394 }
395
396 static void irqsoff_tracer_close(struct trace_iterator *iter)
397 {
398         if (iter->tr->ctrl)
399                 start_irqsoff_tracer(iter->tr);
400 }
401
402 #ifdef CONFIG_IRQSOFF_TRACER
403 static void irqsoff_tracer_init(struct trace_array *tr)
404 {
405         trace_type = TRACER_IRQS_OFF;
406
407         __irqsoff_tracer_init(tr);
408 }
409 static struct tracer irqsoff_tracer __read_mostly =
410 {
411         .name           = "irqsoff",
412         .init           = irqsoff_tracer_init,
413         .reset          = irqsoff_tracer_reset,
414         .open           = irqsoff_tracer_open,
415         .close          = irqsoff_tracer_close,
416         .ctrl_update    = irqsoff_tracer_ctrl_update,
417         .print_max      = 1,
418 #ifdef CONFIG_FTRACE_SELFTEST
419         .selftest    = trace_selftest_startup_irqsoff,
420 #endif
421 };
422 # define register_irqsoff(trace) register_tracer(&trace)
423 #else
424 # define register_irqsoff(trace) do { } while (0)
425 #endif
426
427 #ifdef CONFIG_PREEMPT_TRACER
428 static void preemptoff_tracer_init(struct trace_array *tr)
429 {
430         trace_type = TRACER_PREEMPT_OFF;
431
432         __irqsoff_tracer_init(tr);
433 }
434
435 static struct tracer preemptoff_tracer __read_mostly =
436 {
437         .name           = "preemptoff",
438         .init           = preemptoff_tracer_init,
439         .reset          = irqsoff_tracer_reset,
440         .open           = irqsoff_tracer_open,
441         .close          = irqsoff_tracer_close,
442         .ctrl_update    = irqsoff_tracer_ctrl_update,
443         .print_max      = 1,
444 #ifdef CONFIG_FTRACE_SELFTEST
445         .selftest    = trace_selftest_startup_preemptoff,
446 #endif
447 };
448 # define register_preemptoff(trace) register_tracer(&trace)
449 #else
450 # define register_preemptoff(trace) do { } while (0)
451 #endif
452
453 #if defined(CONFIG_IRQSOFF_TRACER) && \
454         defined(CONFIG_PREEMPT_TRACER)
455
456 static void preemptirqsoff_tracer_init(struct trace_array *tr)
457 {
458         trace_type = TRACER_IRQS_OFF | TRACER_PREEMPT_OFF;
459
460         __irqsoff_tracer_init(tr);
461 }
462
463 static struct tracer preemptirqsoff_tracer __read_mostly =
464 {
465         .name           = "preemptirqsoff",
466         .init           = preemptirqsoff_tracer_init,
467         .reset          = irqsoff_tracer_reset,
468         .open           = irqsoff_tracer_open,
469         .close          = irqsoff_tracer_close,
470         .ctrl_update    = irqsoff_tracer_ctrl_update,
471         .print_max      = 1,
472 #ifdef CONFIG_FTRACE_SELFTEST
473         .selftest    = trace_selftest_startup_preemptirqsoff,
474 #endif
475 };
476
477 # define register_preemptirqsoff(trace) register_tracer(&trace)
478 #else
479 # define register_preemptirqsoff(trace) do { } while (0)
480 #endif
481
482 __init static int init_irqsoff_tracer(void)
483 {
484         register_irqsoff(irqsoff_tracer);
485         register_preemptoff(preemptoff_tracer);
486         register_preemptirqsoff(preemptirqsoff_tracer);
487
488         return 0;
489 }
490 device_initcall(init_irqsoff_tracer);