git.oblomov.eu Git - linux-2.6/blob - 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         if (!runqueue_is_locked()) {
 169                 if (tracing_thresh) {
 170                         printk(KERN_INFO "(%16s-%-5d|#%d): %lu us critical"
 171                                " section violates %lu us threshold.\n",
 172                                current->comm, current->pid,
 173                                raw_smp_processor_id(),
 174                                latency, nsecs_to_usecs(tracing_thresh));
 175                 } else {
 176                         printk(KERN_INFO "(%16s-%-5d|#%d): new %lu us"
 177                                " maximum-latency critical section.\n",
 178                                current->comm, current->pid,
 179                                raw_smp_processor_id(),
 180                                latency);
 181                 }
 182         }
 183
 184         max_sequence++;
 185
 186 out_unlock:
 187         spin_unlock_irqrestore(&max_trace_lock, flags);
 188
 189 out:
 190         data->critical_sequence = max_sequence;
 191         data->preempt_timestamp = ftrace_now(cpu);
 192         tracing_reset(data);
 193         trace_function(tr, data, CALLER_ADDR0, parent_ip, flags);
 194 }
 195
 196 static inline void
 197 start_critical_timing(unsigned long ip, unsigned long parent_ip)
 198 {
 199         int cpu;
 200         struct trace_array *tr = irqsoff_trace;
 201         struct trace_array_cpu *data;
 202         unsigned long flags;
 203
 204         if (likely(!tracer_enabled))
 205                 return;
 206
 207         cpu = raw_smp_processor_id();
 208
 209         if (per_cpu(tracing_cpu, cpu))
 210                 return;
 211
 212         data = tr->data[cpu];
 213
 214         if (unlikely(!data) || atomic_read(&data->disabled))
 215                 return;
 216
 217         atomic_inc(&data->disabled);
 218
 219         data->critical_sequence = max_sequence;
 220         data->preempt_timestamp = ftrace_now(cpu);
 221         data->critical_start = parent_ip ? : ip;
 222         tracing_reset(data);
 223
 224         local_save_flags(flags);
 225
 226         trace_function(tr, data, ip, parent_ip, flags);
 227
 228         per_cpu(tracing_cpu, cpu) = 1;
 229
 230         atomic_dec(&data->disabled);
 231 }
 232
 233 static inline void
 234 stop_critical_timing(unsigned long ip, unsigned long parent_ip)
 235 {
 236         int cpu;
 237         struct trace_array *tr = irqsoff_trace;
 238         struct trace_array_cpu *data;
 239         unsigned long flags;
 240
 241         cpu = raw_smp_processor_id();
 242         /* Always clear the tracing cpu on stopping the trace */
 243         if (unlikely(per_cpu(tracing_cpu, cpu)))
 244                 per_cpu(tracing_cpu, cpu) = 0;
 245         else
 246                 return;
 247
 248         if (!tracer_enabled)
 249                 return;
 250
 251         data = tr->data[cpu];
 252
 253         if (unlikely(!data) || unlikely(!head_page(data)) ||
 254             !data->critical_start || atomic_read(&data->disabled))
 255                 return;
 256
 257         atomic_inc(&data->disabled);
 258
 259         local_save_flags(flags);
 260         trace_function(tr, data, ip, parent_ip, flags);
 261         check_critical_timing(tr, data, parent_ip ? : ip, cpu);
 262         data->critical_start = 0;
 263         atomic_dec(&data->disabled);
 264 }
 265
 266 /* start and stop critical timings used to for stoppage (in idle) */
 267 void start_critical_timings(void)
 268 {
 269         if (preempt_trace() || irq_trace())
 270                 start_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
 271 }
 272
 273 void stop_critical_timings(void)
 274 {
 275         if (preempt_trace() || irq_trace())
 276                 stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
 277 }
 278
 279 #ifdef CONFIG_IRQSOFF_TRACER
 280 #ifdef CONFIG_PROVE_LOCKING
 281 void time_hardirqs_on(unsigned long a0, unsigned long a1)
 282 {
 283         if (!preempt_trace() && irq_trace())
 284                 stop_critical_timing(a0, a1);
 285 }
 286
 287 void time_hardirqs_off(unsigned long a0, unsigned long a1)
 288 {
 289         if (!preempt_trace() && irq_trace())
 290                 start_critical_timing(a0, a1);
 291 }
 292
 293 #else /* !CONFIG_PROVE_LOCKING */
 294
 295 /*
 296  * Stubs:
 297  */
 298
 299 void early_boot_irqs_off(void)
 300 {
 301 }
 302
 303 void early_boot_irqs_on(void)
 304 {
 305 }
 306
 307 void trace_softirqs_on(unsigned long ip)
 308 {
 309 }
 310
 311 void trace_softirqs_off(unsigned long ip)
 312 {
 313 }
 314
 315 inline void print_irqtrace_events(struct task_struct *curr)
 316 {
 317 }
 318
 319 /*
 320  * We are only interested in hardirq on/off events:
 321  */
 322 void trace_hardirqs_on(void)
 323 {
 324         if (!preempt_trace() && irq_trace())
 325                 stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
 326 }
 327 EXPORT_SYMBOL(trace_hardirqs_on);
 328
 329 void trace_hardirqs_off(void)
 330 {
 331         if (!preempt_trace() && irq_trace())
 332                 start_critical_timing(CALLER_ADDR0, CALLER_ADDR1);
 333 }
 334 EXPORT_SYMBOL(trace_hardirqs_off);
 335
 336 void trace_hardirqs_on_caller(unsigned long caller_addr)
 337 {
 338         if (!preempt_trace() && irq_trace())
 339                 stop_critical_timing(CALLER_ADDR0, caller_addr);
 340 }
 341 EXPORT_SYMBOL(trace_hardirqs_on_caller);
 342
 343 void trace_hardirqs_off_caller(unsigned long caller_addr)
 344 {
 345         if (!preempt_trace() && irq_trace())
 346                 start_critical_timing(CALLER_ADDR0, caller_addr);
 347 }
 348 EXPORT_SYMBOL(trace_hardirqs_off_caller);
 349
 350 #endif /* CONFIG_PROVE_LOCKING */
 351 #endif /*  CONFIG_IRQSOFF_TRACER */
 352
 353 #ifdef CONFIG_PREEMPT_TRACER
 354 void trace_preempt_on(unsigned long a0, unsigned long a1)
 355 {
 356         stop_critical_timing(a0, a1);
 357 }
 358
 359 void trace_preempt_off(unsigned long a0, unsigned long a1)
 360 {
 361         start_critical_timing(a0, a1);
 362 }
 363 #endif /* CONFIG_PREEMPT_TRACER */
 364
 365 static void start_irqsoff_tracer(struct trace_array *tr)
 366 {
 367         register_ftrace_function(&trace_ops);
 368         tracer_enabled = 1;
 369 }
 370
 371 static void stop_irqsoff_tracer(struct trace_array *tr)
 372 {
 373         tracer_enabled = 0;
 374         unregister_ftrace_function(&trace_ops);
 375 }
 376
 377 static void __irqsoff_tracer_init(struct trace_array *tr)
 378 {
 379         irqsoff_trace = tr;
 380         /* make sure that the tracer is visible */
 381         smp_wmb();
 382
 383         if (tr->ctrl)
 384                 start_irqsoff_tracer(tr);
 385 }
 386
 387 static void irqsoff_tracer_reset(struct trace_array *tr)
 388 {
 389         if (tr->ctrl)
 390                 stop_irqsoff_tracer(tr);
 391 }
 392
 393 static void irqsoff_tracer_ctrl_update(struct trace_array *tr)
 394 {
 395         if (tr->ctrl)
 396                 start_irqsoff_tracer(tr);
 397         else
 398                 stop_irqsoff_tracer(tr);
 399 }
 400
 401 static void irqsoff_tracer_open(struct trace_iterator *iter)
 402 {
 403         /* stop the trace while dumping */
 404         if (iter->tr->ctrl)
 405                 stop_irqsoff_tracer(iter->tr);
 406 }
 407
 408 static void irqsoff_tracer_close(struct trace_iterator *iter)
 409 {
 410         if (iter->tr->ctrl)
 411                 start_irqsoff_tracer(iter->tr);
 412 }
 413
 414 #ifdef CONFIG_IRQSOFF_TRACER
 415 static void irqsoff_tracer_init(struct trace_array *tr)
 416 {
 417         trace_type = TRACER_IRQS_OFF;
 418
 419         __irqsoff_tracer_init(tr);
 420 }
 421 static struct tracer irqsoff_tracer __read_mostly =
 422 {
 423         .name           = "irqsoff",
 424         .init           = irqsoff_tracer_init,
 425         .reset          = irqsoff_tracer_reset,
 426         .open           = irqsoff_tracer_open,
 427         .close          = irqsoff_tracer_close,
 428         .ctrl_update    = irqsoff_tracer_ctrl_update,
 429         .print_max      = 1,
 430 #ifdef CONFIG_FTRACE_SELFTEST
 431         .selftest    = trace_selftest_startup_irqsoff,
 432 #endif
 433 };
 434 # define register_irqsoff(trace) register_tracer(&trace)
 435 #else
 436 # define register_irqsoff(trace) do { } while (0)
 437 #endif
 438
 439 #ifdef CONFIG_PREEMPT_TRACER
 440 static void preemptoff_tracer_init(struct trace_array *tr)
 441 {
 442         trace_type = TRACER_PREEMPT_OFF;
 443
 444         __irqsoff_tracer_init(tr);
 445 }
 446
 447 static struct tracer preemptoff_tracer __read_mostly =
 448 {
 449         .name           = "preemptoff",
 450         .init           = preemptoff_tracer_init,
 451         .reset          = irqsoff_tracer_reset,
 452         .open           = irqsoff_tracer_open,
 453         .close          = irqsoff_tracer_close,
 454         .ctrl_update    = irqsoff_tracer_ctrl_update,
 455         .print_max      = 1,
 456 #ifdef CONFIG_FTRACE_SELFTEST
 457         .selftest    = trace_selftest_startup_preemptoff,
 458 #endif
 459 };
 460 # define register_preemptoff(trace) register_tracer(&trace)
 461 #else
 462 # define register_preemptoff(trace) do { } while (0)
 463 #endif
 464
 465 #if defined(CONFIG_IRQSOFF_TRACER) && \
 466         defined(CONFIG_PREEMPT_TRACER)
 467
 468 static void preemptirqsoff_tracer_init(struct trace_array *tr)
 469 {
 470         trace_type = TRACER_IRQS_OFF | TRACER_PREEMPT_OFF;
 471
 472         __irqsoff_tracer_init(tr);
 473 }
 474
 475 static struct tracer preemptirqsoff_tracer __read_mostly =
 476 {
 477         .name           = "preemptirqsoff",
 478         .init           = preemptirqsoff_tracer_init,
 479         .reset          = irqsoff_tracer_reset,
 480         .open           = irqsoff_tracer_open,
 481         .close          = irqsoff_tracer_close,
 482         .ctrl_update    = irqsoff_tracer_ctrl_update,
 483         .print_max      = 1,
 484 #ifdef CONFIG_FTRACE_SELFTEST
 485         .selftest    = trace_selftest_startup_preemptirqsoff,
 486 #endif
 487 };
 488
 489 # define register_preemptirqsoff(trace) register_tracer(&trace)
 490 #else
 491 # define register_preemptirqsoff(trace) do { } while (0)
 492 #endif
 493
 494 __init static int init_irqsoff_tracer(void)
 495 {
 496         register_irqsoff(irqsoff_tracer);
 497         register_preemptoff(preemptoff_tracer);
 498         register_preemptirqsoff(preemptirqsoff_tracer);
 499
 500         return 0;
 501 }
 502 device_initcall(init_irqsoff_tracer);