kernel/trace/trace_selftest.c

   1 /* Include in trace.c */
   2
   3 #include <linux/stringify.h>
   4 #include <linux/kthread.h>
   5 #include <linux/delay.h>
   6
   7 static inline int trace_valid_entry(struct trace_entry *entry)
   8 {
   9         switch (entry->type) {
  10         case TRACE_FN:
  11         case TRACE_CTX:
  12         case TRACE_WAKE:
  13         case TRACE_STACK:
  14         case TRACE_PRINT:
  15         case TRACE_SPECIAL:
  16         case TRACE_BRANCH:
  17         case TRACE_GRAPH_ENT:
  18         case TRACE_GRAPH_RET:
  19         case TRACE_HW_BRANCHES:
  20                 return 1;
  21         }
  22         return 0;
  23 }
  24
  25 static int trace_test_buffer_cpu(struct trace_array *tr, int cpu)
  26 {
  27         struct ring_buffer_event *event;
  28         struct trace_entry *entry;
  29         unsigned int loops = 0;
  30
  31         while ((event = ring_buffer_consume(tr->buffer, cpu, NULL))) {
  32                 entry = ring_buffer_event_data(event);
  33
  34                 /*
  35                  * The ring buffer is a size of trace_buf_size, if
  36                  * we loop more than the size, there's something wrong
  37                  * with the ring buffer.
  38                  */
  39                 if (loops++ > trace_buf_size) {
  40                         printk(KERN_CONT ".. bad ring buffer ");
  41                         goto failed;
  42                 }
  43                 if (!trace_valid_entry(entry)) {
  44                         printk(KERN_CONT ".. invalid entry %d ",
  45                                 entry->type);
  46                         goto failed;
  47                 }
  48         }
  49         return 0;
  50
  51  failed:
  52         /* disable tracing */
  53         tracing_disabled = 1;
  54         printk(KERN_CONT ".. corrupted trace buffer .. ");
  55         return -1;
  56 }
  57
  58 /*
  59  * Test the trace buffer to see if all the elements
  60  * are still sane.
  61  */
  62 static int trace_test_buffer(struct trace_array *tr, unsigned long *count)
  63 {
  64         unsigned long flags, cnt = 0;
  65         int cpu, ret = 0;
  66
  67         /* Don't allow flipping of max traces now */
  68         local_irq_save(flags);
  69         __raw_spin_lock(&ftrace_max_lock);
  70
  71         cnt = ring_buffer_entries(tr->buffer);
  72
  73         /*
  74          * The trace_test_buffer_cpu runs a while loop to consume all data.
  75          * If the calling tracer is broken, and is constantly filling
  76          * the buffer, this will run forever, and hard lock the box.
  77          * We disable the ring buffer while we do this test to prevent
  78          * a hard lock up.
  79          */
  80         tracing_off();
  81         for_each_possible_cpu(cpu) {
  82                 ret = trace_test_buffer_cpu(tr, cpu);
  83                 if (ret)
  84                         break;
  85         }
  86         tracing_on();
  87         __raw_spin_unlock(&ftrace_max_lock);
  88         local_irq_restore(flags);
  89
  90         if (count)
  91                 *count = cnt;
  92
  93         return ret;
  94 }
  95
  96 static inline void warn_failed_init_tracer(struct tracer *trace, int init_ret)
  97 {
  98         printk(KERN_WARNING "Failed to init %s tracer, init returned %d\n",
  99                 trace->name, init_ret);
 100 }
 101 #ifdef CONFIG_FUNCTION_TRACER
 102
 103 #ifdef CONFIG_DYNAMIC_FTRACE
 104
 105 /* Test dynamic code modification and ftrace filters */
 106 int trace_selftest_startup_dynamic_tracing(struct tracer *trace,
 107                                            struct trace_array *tr,
 108                                            int (*func)(void))
 109 {
 110         int save_ftrace_enabled = ftrace_enabled;
 111         int save_tracer_enabled = tracer_enabled;
 112         unsigned long count;
 113         char *func_name;
 114         int ret;
 115
 116         /* The ftrace test PASSED */
 117         printk(KERN_CONT "PASSED\n");
 118         pr_info("Testing dynamic ftrace: ");
 119
 120         /* enable tracing, and record the filter function */
 121         ftrace_enabled = 1;
 122         tracer_enabled = 1;
 123
 124         /* passed in by parameter to fool gcc from optimizing */
 125         func();
 126
 127         /*
 128          * Some archs *cough*PowerPC*cough* add characters to the
 129          * start of the function names. We simply put a '*' to
 130          * accommodate them.
 131          */
 132         func_name = "*" __stringify(DYN_FTRACE_TEST_NAME);
 133
 134         /* filter only on our function */
 135         ftrace_set_filter(func_name, strlen(func_name), 1);
 136
 137         /* enable tracing */
 138         ret = tracer_init(trace, tr);
 139         if (ret) {
 140                 warn_failed_init_tracer(trace, ret);
 141                 goto out;
 142         }
 143
 144         /* Sleep for a 1/10 of a second */
 145         msleep(100);
 146
 147         /* we should have nothing in the buffer */
 148         ret = trace_test_buffer(tr, &count);
 149         if (ret)
 150                 goto out;
 151
 152         if (count) {
 153                 ret = -1;
 154                 printk(KERN_CONT ".. filter did not filter .. ");
 155                 goto out;
 156         }
 157
 158         /* call our function again */
 159         func();
 160
 161         /* sleep again */
 162         msleep(100);
 163
 164         /* stop the tracing. */
 165         tracing_stop();
 166         ftrace_enabled = 0;
 167
 168         /* check the trace buffer */
 169         ret = trace_test_buffer(tr, &count);
 170         trace->reset(tr);
 171         tracing_start();
 172
 173         /* we should only have one item */
 174         if (!ret && count != 1) {
 175                 printk(KERN_CONT ".. filter failed count=%ld ..", count);
 176                 ret = -1;
 177                 goto out;
 178         }
 179
 180  out:
 181         ftrace_enabled = save_ftrace_enabled;
 182         tracer_enabled = save_tracer_enabled;
 183
 184         /* Enable tracing on all functions again */
 185         ftrace_set_filter(NULL, 0, 1);
 186
 187         return ret;
 188 }
 189 #else
 190 # define trace_selftest_startup_dynamic_tracing(trace, tr, func) ({ 0; })
 191 #endif /* CONFIG_DYNAMIC_FTRACE */
 192
 193 /*
 194  * Simple verification test of ftrace function tracer.
 195  * Enable ftrace, sleep 1/10 second, and then read the trace
 196  * buffer to see if all is in order.
 197  */
 198 int
 199 trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr)
 200 {
 201         int save_ftrace_enabled = ftrace_enabled;
 202         int save_tracer_enabled = tracer_enabled;
 203         unsigned long count;
 204         int ret;
 205
 206         /* make sure msleep has been recorded */
 207         msleep(1);
 208
 209         /* start the tracing */
 210         ftrace_enabled = 1;
 211         tracer_enabled = 1;
 212
 213         ret = tracer_init(trace, tr);
 214         if (ret) {
 215                 warn_failed_init_tracer(trace, ret);
 216                 goto out;
 217         }
 218
 219         /* Sleep for a 1/10 of a second */
 220         msleep(100);
 221         /* stop the tracing. */
 222         tracing_stop();
 223         ftrace_enabled = 0;
 224
 225         /* check the trace buffer */
 226         ret = trace_test_buffer(tr, &count);
 227         trace->reset(tr);
 228         tracing_start();
 229
 230         if (!ret && !count) {
 231                 printk(KERN_CONT ".. no entries found ..");
 232                 ret = -1;
 233                 goto out;
 234         }
 235
 236         ret = trace_selftest_startup_dynamic_tracing(trace, tr,
 237                                                      DYN_FTRACE_TEST_NAME);
 238
 239  out:
 240         ftrace_enabled = save_ftrace_enabled;
 241         tracer_enabled = save_tracer_enabled;
 242
 243         /* kill ftrace totally if we failed */
 244         if (ret)
 245                 ftrace_kill();
 246
 247         return ret;
 248 }
 249 #endif /* CONFIG_FUNCTION_TRACER */
 250
 251
 252 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
 253
 254 /* Maximum number of functions to trace before diagnosing a hang */
 255 #define GRAPH_MAX_FUNC_TEST     100000000
 256
 257 static void __ftrace_dump(bool disable_tracing);
 258 static unsigned int graph_hang_thresh;
 259
 260 /* Wrap the real function entry probe to avoid possible hanging */
 261 static int trace_graph_entry_watchdog(struct ftrace_graph_ent *trace)
 262 {
 263         /* This is harmlessly racy, we want to approximately detect a hang */
 264         if (unlikely(++graph_hang_thresh > GRAPH_MAX_FUNC_TEST)) {
 265                 ftrace_graph_stop();
 266                 printk(KERN_WARNING "BUG: Function graph tracer hang!\n");
 267                 if (ftrace_dump_on_oops)
 268                         __ftrace_dump(false);
 269                 return 0;
 270         }
 271
 272         return trace_graph_entry(trace);
 273 }
 274
 275 /*
 276  * Pretty much the same than for the function tracer from which the selftest
 277  * has been borrowed.
 278  */
 279 int
 280 trace_selftest_startup_function_graph(struct tracer *trace,
 281                                         struct trace_array *tr)
 282 {
 283         int ret;
 284         unsigned long count;
 285
 286         /*
 287          * Simulate the init() callback but we attach a watchdog callback
 288          * to detect and recover from possible hangs
 289          */
 290         tracing_reset_online_cpus(tr);
 291         ret = register_ftrace_graph(&trace_graph_return,
 292                                     &trace_graph_entry_watchdog);
 293         if (ret) {
 294                 warn_failed_init_tracer(trace, ret);
 295                 goto out;
 296         }
 297         tracing_start_cmdline_record();
 298
 299         /* Sleep for a 1/10 of a second */
 300         msleep(100);
 301
 302         /* Have we just recovered from a hang? */
 303         if (graph_hang_thresh > GRAPH_MAX_FUNC_TEST) {
 304                 tracing_selftest_disabled = true;
 305                 ret = -1;
 306                 goto out;
 307         }
 308
 309         tracing_stop();
 310
 311         /* check the trace buffer */
 312         ret = trace_test_buffer(tr, &count);
 313
 314         trace->reset(tr);
 315         tracing_start();
 316
 317         if (!ret && !count) {
 318                 printk(KERN_CONT ".. no entries found ..");
 319                 ret = -1;
 320                 goto out;
 321         }
 322
 323         /* Don't test dynamic tracing, the function tracer already did */
 324
 325 out:
 326         /* Stop it if we failed */
 327         if (ret)
 328                 ftrace_graph_stop();
 329
 330         return ret;
 331 }
 332 #endif /* CONFIG_FUNCTION_GRAPH_TRACER */
 333
 334
 335 #ifdef CONFIG_IRQSOFF_TRACER
 336 int
 337 trace_selftest_startup_irqsoff(struct tracer *trace, struct trace_array *tr)
 338 {
 339         unsigned long save_max = tracing_max_latency;
 340         unsigned long count;
 341         int ret;
 342
 343         /* start the tracing */
 344         ret = tracer_init(trace, tr);
 345         if (ret) {
 346                 warn_failed_init_tracer(trace, ret);
 347                 return ret;
 348         }
 349
 350         /* reset the max latency */
 351         tracing_max_latency = 0;
 352         /* disable interrupts for a bit */
 353         local_irq_disable();
 354         udelay(100);
 355         local_irq_enable();
 356
 357         /*
 358          * Stop the tracer to avoid a warning subsequent
 359          * to buffer flipping failure because tracing_stop()
 360          * disables the tr and max buffers, making flipping impossible
 361          * in case of parallels max irqs off latencies.
 362          */
 363         trace->stop(tr);
 364         /* stop the tracing. */
 365         tracing_stop();
 366         /* check both trace buffers */
 367         ret = trace_test_buffer(tr, NULL);
 368         if (!ret)
 369                 ret = trace_test_buffer(&max_tr, &count);
 370         trace->reset(tr);
 371         tracing_start();
 372
 373         if (!ret && !count) {
 374                 printk(KERN_CONT ".. no entries found ..");
 375                 ret = -1;
 376         }
 377
 378         tracing_max_latency = save_max;
 379
 380         return ret;
 381 }
 382 #endif /* CONFIG_IRQSOFF_TRACER */
 383
 384 #ifdef CONFIG_PREEMPT_TRACER
 385 int
 386 trace_selftest_startup_preemptoff(struct tracer *trace, struct trace_array *tr)
 387 {
 388         unsigned long save_max = tracing_max_latency;
 389         unsigned long count;
 390         int ret;
 391
 392         /*
 393          * Now that the big kernel lock is no longer preemptable,
 394          * and this is called with the BKL held, it will always
 395          * fail. If preemption is already disabled, simply
 396          * pass the test. When the BKL is removed, or becomes
 397          * preemptible again, we will once again test this,
 398          * so keep it in.
 399          */
 400         if (preempt_count()) {
 401                 printk(KERN_CONT "can not test ... force ");
 402                 return 0;
 403         }
 404
 405         /* start the tracing */
 406         ret = tracer_init(trace, tr);
 407         if (ret) {
 408                 warn_failed_init_tracer(trace, ret);
 409                 return ret;
 410         }
 411
 412         /* reset the max latency */
 413         tracing_max_latency = 0;
 414         /* disable preemption for a bit */
 415         preempt_disable();
 416         udelay(100);
 417         preempt_enable();
 418
 419         /*
 420          * Stop the tracer to avoid a warning subsequent
 421          * to buffer flipping failure because tracing_stop()
 422          * disables the tr and max buffers, making flipping impossible
 423          * in case of parallels max preempt off latencies.
 424          */
 425         trace->stop(tr);
 426         /* stop the tracing. */
 427         tracing_stop();
 428         /* check both trace buffers */
 429         ret = trace_test_buffer(tr, NULL);
 430         if (!ret)
 431                 ret = trace_test_buffer(&max_tr, &count);
 432         trace->reset(tr);
 433         tracing_start();
 434
 435         if (!ret && !count) {
 436                 printk(KERN_CONT ".. no entries found ..");
 437                 ret = -1;
 438         }
 439
 440         tracing_max_latency = save_max;
 441
 442         return ret;
 443 }
 444 #endif /* CONFIG_PREEMPT_TRACER */
 445
 446 #if defined(CONFIG_IRQSOFF_TRACER) && defined(CONFIG_PREEMPT_TRACER)
 447 int
 448 trace_selftest_startup_preemptirqsoff(struct tracer *trace, struct trace_array *tr)
 449 {
 450         unsigned long save_max = tracing_max_latency;
 451         unsigned long count;
 452         int ret;
 453
 454         /*
 455          * Now that the big kernel lock is no longer preemptable,
 456          * and this is called with the BKL held, it will always
 457          * fail. If preemption is already disabled, simply
 458          * pass the test. When the BKL is removed, or becomes
 459          * preemptible again, we will once again test this,
 460          * so keep it in.
 461          */
 462         if (preempt_count()) {
 463                 printk(KERN_CONT "can not test ... force ");
 464                 return 0;
 465         }
 466
 467         /* start the tracing */
 468         ret = tracer_init(trace, tr);
 469         if (ret) {
 470                 warn_failed_init_tracer(trace, ret);
 471                 goto out_no_start;
 472         }
 473
 474         /* reset the max latency */
 475         tracing_max_latency = 0;
 476
 477         /* disable preemption and interrupts for a bit */
 478         preempt_disable();
 479         local_irq_disable();
 480         udelay(100);
 481         preempt_enable();
 482         /* reverse the order of preempt vs irqs */
 483         local_irq_enable();
 484
 485         /*
 486          * Stop the tracer to avoid a warning subsequent
 487          * to buffer flipping failure because tracing_stop()
 488          * disables the tr and max buffers, making flipping impossible
 489          * in case of parallels max irqs/preempt off latencies.
 490          */
 491         trace->stop(tr);
 492         /* stop the tracing. */
 493         tracing_stop();
 494         /* check both trace buffers */
 495         ret = trace_test_buffer(tr, NULL);
 496         if (ret)
 497                 goto out;
 498
 499         ret = trace_test_buffer(&max_tr, &count);
 500         if (ret)
 501                 goto out;
 502
 503         if (!ret && !count) {
 504                 printk(KERN_CONT ".. no entries found ..");
 505                 ret = -1;
 506                 goto out;
 507         }
 508
 509         /* do the test by disabling interrupts first this time */
 510         tracing_max_latency = 0;
 511         tracing_start();
 512         trace->start(tr);
 513
 514         preempt_disable();
 515         local_irq_disable();
 516         udelay(100);
 517         preempt_enable();
 518         /* reverse the order of preempt vs irqs */
 519         local_irq_enable();
 520
 521         trace->stop(tr);
 522         /* stop the tracing. */
 523         tracing_stop();
 524         /* check both trace buffers */
 525         ret = trace_test_buffer(tr, NULL);
 526         if (ret)
 527                 goto out;
 528
 529         ret = trace_test_buffer(&max_tr, &count);
 530
 531         if (!ret && !count) {
 532                 printk(KERN_CONT ".. no entries found ..");
 533                 ret = -1;
 534                 goto out;
 535         }
 536
 537 out:
 538         tracing_start();
 539 out_no_start:
 540         trace->reset(tr);
 541         tracing_max_latency = save_max;
 542
 543         return ret;
 544 }
 545 #endif /* CONFIG_IRQSOFF_TRACER && CONFIG_PREEMPT_TRACER */
 546
 547 #ifdef CONFIG_NOP_TRACER
 548 int
 549 trace_selftest_startup_nop(struct tracer *trace, struct trace_array *tr)
 550 {
 551         /* What could possibly go wrong? */
 552         return 0;
 553 }
 554 #endif
 555
 556 #ifdef CONFIG_SCHED_TRACER
 557 static int trace_wakeup_test_thread(void *data)
 558 {
 559         /* Make this a RT thread, doesn't need to be too high */
 560         struct sched_param param = { .sched_priority = 5 };
 561         struct completion *x = data;
 562
 563         sched_setscheduler(current, SCHED_FIFO, &param);
 564
 565         /* Make it know we have a new prio */
 566         complete(x);
 567
 568         /* now go to sleep and let the test wake us up */
 569         set_current_state(TASK_INTERRUPTIBLE);
 570         schedule();
 571
 572         /* we are awake, now wait to disappear */
 573         while (!kthread_should_stop()) {
 574                 /*
 575                  * This is an RT task, do short sleeps to let
 576                  * others run.
 577                  */
 578                 msleep(100);
 579         }
 580
 581         return 0;
 582 }
 583
 584 int
 585 trace_selftest_startup_wakeup(struct tracer *trace, struct trace_array *tr)
 586 {
 587         unsigned long save_max = tracing_max_latency;
 588         struct task_struct *p;
 589         struct completion isrt;
 590         unsigned long count;
 591         int ret;
 592
 593         init_completion(&isrt);
 594
 595         /* create a high prio thread */
 596         p = kthread_run(trace_wakeup_test_thread, &isrt, "ftrace-test");
 597         if (IS_ERR(p)) {
 598                 printk(KERN_CONT "Failed to create ftrace wakeup test thread ");
 599                 return -1;
 600         }
 601
 602         /* make sure the thread is running at an RT prio */
 603         wait_for_completion(&isrt);
 604
 605         /* start the tracing */
 606         ret = tracer_init(trace, tr);
 607         if (ret) {
 608                 warn_failed_init_tracer(trace, ret);
 609                 return ret;
 610         }
 611
 612         /* reset the max latency */
 613         tracing_max_latency = 0;
 614
 615         /* sleep to let the RT thread sleep too */
 616         msleep(100);
 617
 618         /*
 619          * Yes this is slightly racy. It is possible that for some
 620          * strange reason that the RT thread we created, did not
 621          * call schedule for 100ms after doing the completion,
 622          * and we do a wakeup on a task that already is awake.
 623          * But that is extremely unlikely, and the worst thing that
 624          * happens in such a case, is that we disable tracing.
 625          * Honestly, if this race does happen something is horrible
 626          * wrong with the system.
 627          */
 628
 629         wake_up_process(p);
 630
 631         /* give a little time to let the thread wake up */
 632         msleep(100);
 633
 634         /* stop the tracing. */
 635         tracing_stop();
 636         /* check both trace buffers */
 637         ret = trace_test_buffer(tr, NULL);
 638         if (!ret)
 639                 ret = trace_test_buffer(&max_tr, &count);
 640
 641
 642         trace->reset(tr);
 643         tracing_start();
 644
 645         tracing_max_latency = save_max;
 646
 647         /* kill the thread */
 648         kthread_stop(p);
 649
 650         if (!ret && !count) {
 651                 printk(KERN_CONT ".. no entries found ..");
 652                 ret = -1;
 653         }
 654
 655         return ret;
 656 }
 657 #endif /* CONFIG_SCHED_TRACER */
 658
 659 #ifdef CONFIG_CONTEXT_SWITCH_TRACER
 660 int
 661 trace_selftest_startup_sched_switch(struct tracer *trace, struct trace_array *tr)
 662 {
 663         unsigned long count;
 664         int ret;
 665
 666         /* start the tracing */
 667         ret = tracer_init(trace, tr);
 668         if (ret) {
 669                 warn_failed_init_tracer(trace, ret);
 670                 return ret;
 671         }
 672
 673         /* Sleep for a 1/10 of a second */
 674         msleep(100);
 675         /* stop the tracing. */
 676         tracing_stop();
 677         /* check the trace buffer */
 678         ret = trace_test_buffer(tr, &count);
 679         trace->reset(tr);
 680         tracing_start();
 681
 682         if (!ret && !count) {
 683                 printk(KERN_CONT ".. no entries found ..");
 684                 ret = -1;
 685         }
 686
 687         return ret;
 688 }
 689 #endif /* CONFIG_CONTEXT_SWITCH_TRACER */
 690
 691 #ifdef CONFIG_SYSPROF_TRACER
 692 int
 693 trace_selftest_startup_sysprof(struct tracer *trace, struct trace_array *tr)
 694 {
 695         unsigned long count;
 696         int ret;
 697
 698         /* start the tracing */
 699         ret = tracer_init(trace, tr);
 700         if (ret) {
 701                 warn_failed_init_tracer(trace, ret);
 702                 return ret;
 703         }
 704
 705         /* Sleep for a 1/10 of a second */
 706         msleep(100);
 707         /* stop the tracing. */
 708         tracing_stop();
 709         /* check the trace buffer */
 710         ret = trace_test_buffer(tr, &count);
 711         trace->reset(tr);
 712         tracing_start();
 713
 714         if (!ret && !count) {
 715                 printk(KERN_CONT ".. no entries found ..");
 716                 ret = -1;
 717         }
 718
 719         return ret;
 720 }
 721 #endif /* CONFIG_SYSPROF_TRACER */
 722
 723 #ifdef CONFIG_BRANCH_TRACER
 724 int
 725 trace_selftest_startup_branch(struct tracer *trace, struct trace_array *tr)
 726 {
 727         unsigned long count;
 728         int ret;
 729
 730         /* start the tracing */
 731         ret = tracer_init(trace, tr);
 732         if (ret) {
 733                 warn_failed_init_tracer(trace, ret);
 734                 return ret;
 735         }
 736
 737         /* Sleep for a 1/10 of a second */
 738         msleep(100);
 739         /* stop the tracing. */
 740         tracing_stop();
 741         /* check the trace buffer */
 742         ret = trace_test_buffer(tr, &count);
 743         trace->reset(tr);
 744         tracing_start();
 745
 746         if (!ret && !count) {
 747                 printk(KERN_CONT ".. no entries found ..");
 748                 ret = -1;
 749         }
 750
 751         return ret;
 752 }
 753 #endif /* CONFIG_BRANCH_TRACER */
 754
 755 #ifdef CONFIG_HW_BRANCH_TRACER
 756 int
 757 trace_selftest_startup_hw_branches(struct tracer *trace,
 758                                    struct trace_array *tr)
 759 {
 760         struct trace_iterator *iter;
 761         struct tracer tracer;
 762         unsigned long count;
 763         int ret;
 764
 765         if (!trace->open) {
 766                 printk(KERN_CONT "missing open function...");
 767                 return -1;
 768         }
 769
 770         ret = tracer_init(trace, tr);
 771         if (ret) {
 772                 warn_failed_init_tracer(trace, ret);
 773                 return ret;
 774         }
 775
 776         /*
 777          * The hw-branch tracer needs to collect the trace from the various
 778          * cpu trace buffers - before tracing is stopped.
 779          */
 780         iter = kzalloc(sizeof(*iter), GFP_KERNEL);
 781         if (!iter)
 782                 return -ENOMEM;
 783
 784         memcpy(&tracer, trace, sizeof(tracer));
 785
 786         iter->trace = &tracer;
 787         iter->tr = tr;
 788         iter->pos = -1;
 789         mutex_init(&iter->mutex);
 790
 791         trace->open(iter);
 792
 793         mutex_destroy(&iter->mutex);
 794         kfree(iter);
 795
 796         tracing_stop();
 797
 798         ret = trace_test_buffer(tr, &count);
 799         trace->reset(tr);
 800         tracing_start();
 801
 802         if (!ret && !count) {
 803                 printk(KERN_CONT "no entries found..");
 804                 ret = -1;
 805         }
 806
 807         return ret;
 808 }
 809 #endif /* CONFIG_HW_BRANCH_TRACER */