SAFE public projects git trees. - safe/jmp/linux-2.6/blob - kernel/trace/trace_functions_graph.c

   1 /*
   2  *
   3  * Function graph tracer.
   4  * Copyright (c) 2008-2009 Frederic Weisbecker <fweisbec@gmail.com>
   5  * Mostly borrowed from function tracer which
   6  * is Copyright (c) Steven Rostedt <srostedt@redhat.com>
   7  *
   8  */
   9 #include <linux/debugfs.h>
  10 #include <linux/uaccess.h>
  11 #include <linux/ftrace.h>
  12 #include <linux/fs.h>
  13
  14 #include "trace.h"
  15 #include "trace_output.h"
  16
  17 struct fgraph_cpu_data {
  18         pid_t           last_pid;
  19         int             depth;
  20         int             ignore;
  21         unsigned long   enter_funcs[FTRACE_RETFUNC_DEPTH];
  22 };
  23
  24 struct fgraph_data {
  25         struct fgraph_cpu_data          *cpu_data;
  26
  27         /* Place to preserve last processed entry. */
  28         struct ftrace_graph_ent_entry   ent;
  29         struct ftrace_graph_ret_entry   ret;
  30         int                             failed;
  31         int                             cpu;
  32 };
  33
  34 #define TRACE_GRAPH_INDENT      2
  35
  36 /* Flag options */
  37 #define TRACE_GRAPH_PRINT_OVERRUN       0x1
  38 #define TRACE_GRAPH_PRINT_CPU           0x2
  39 #define TRACE_GRAPH_PRINT_OVERHEAD      0x4
  40 #define TRACE_GRAPH_PRINT_PROC          0x8
  41 #define TRACE_GRAPH_PRINT_DURATION      0x10
  42 #define TRACE_GRAPH_PRINT_ABS_TIME      0X20
  43
  44 static struct tracer_opt trace_opts[] = {
  45         /* Display overruns? (for self-debug purpose) */
  46         { TRACER_OPT(funcgraph-overrun, TRACE_GRAPH_PRINT_OVERRUN) },
  47         /* Display CPU ? */
  48         { TRACER_OPT(funcgraph-cpu, TRACE_GRAPH_PRINT_CPU) },
  49         /* Display Overhead ? */
  50         { TRACER_OPT(funcgraph-overhead, TRACE_GRAPH_PRINT_OVERHEAD) },
  51         /* Display proc name/pid */
  52         { TRACER_OPT(funcgraph-proc, TRACE_GRAPH_PRINT_PROC) },
  53         /* Display duration of execution */
  54         { TRACER_OPT(funcgraph-duration, TRACE_GRAPH_PRINT_DURATION) },
  55         /* Display absolute time of an entry */
  56         { TRACER_OPT(funcgraph-abstime, TRACE_GRAPH_PRINT_ABS_TIME) },
  57         { } /* Empty entry */
  58 };
  59
  60 static struct tracer_flags tracer_flags = {
  61         /* Don't display overruns and proc by default */
  62         .val = TRACE_GRAPH_PRINT_CPU | TRACE_GRAPH_PRINT_OVERHEAD |
  63                TRACE_GRAPH_PRINT_DURATION,
  64         .opts = trace_opts
  65 };
  66
  67 static struct trace_array *graph_array;
  68
  69
  70 /* Add a function return address to the trace stack on thread info.*/
  71 int
  72 ftrace_push_return_trace(unsigned long ret, unsigned long func, int *depth,
  73                          unsigned long frame_pointer)
  74 {
  75         unsigned long long calltime;
  76         int index;
  77
  78         if (!current->ret_stack)
  79                 return -EBUSY;
  80
  81         /*
  82          * We must make sure the ret_stack is tested before we read
  83          * anything else.
  84          */
  85         smp_rmb();
  86
  87         /* The return trace stack is full */
  88         if (current->curr_ret_stack == FTRACE_RETFUNC_DEPTH - 1) {
  89                 atomic_inc(&current->trace_overrun);
  90                 return -EBUSY;
  91         }
  92
  93         calltime = trace_clock_local();
  94
  95         index = ++current->curr_ret_stack;
  96         barrier();
  97         current->ret_stack[index].ret = ret;
  98         current->ret_stack[index].func = func;
  99         current->ret_stack[index].calltime = calltime;
 100         current->ret_stack[index].subtime = 0;
 101         current->ret_stack[index].fp = frame_pointer;
 102         *depth = index;
 103
 104         return 0;
 105 }
 106
 107 /* Retrieve a function return address to the trace stack on thread info.*/
 108 static void
 109 ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
 110                         unsigned long frame_pointer)
 111 {
 112         int index;
 113
 114         index = current->curr_ret_stack;
 115
 116         if (unlikely(index < 0)) {
 117                 ftrace_graph_stop();
 118                 WARN_ON(1);
 119                 /* Might as well panic, otherwise we have no where to go */
 120                 *ret = (unsigned long)panic;
 121                 return;
 122         }
 123
 124 #ifdef CONFIG_HAVE_FUNCTION_GRAPH_FP_TEST
 125         /*
 126          * The arch may choose to record the frame pointer used
 127          * and check it here to make sure that it is what we expect it
 128          * to be. If gcc does not set the place holder of the return
 129          * address in the frame pointer, and does a copy instead, then
 130          * the function graph trace will fail. This test detects this
 131          * case.
 132          *
 133          * Currently, x86_32 with optimize for size (-Os) makes the latest
 134          * gcc do the above.
 135          */
 136         if (unlikely(current->ret_stack[index].fp != frame_pointer)) {
 137                 ftrace_graph_stop();
 138                 WARN(1, "Bad frame pointer: expected %lx, received %lx\n"
 139                      "  from func %ps return to %lx\n",
 140                      current->ret_stack[index].fp,
 141                      frame_pointer,
 142                      (void *)current->ret_stack[index].func,
 143                      current->ret_stack[index].ret);
 144                 *ret = (unsigned long)panic;
 145                 return;
 146         }
 147 #endif
 148
 149         *ret = current->ret_stack[index].ret;
 150         trace->func = current->ret_stack[index].func;
 151         trace->calltime = current->ret_stack[index].calltime;
 152         trace->overrun = atomic_read(&current->trace_overrun);
 153         trace->depth = index;
 154 }
 155
 156 /*
 157  * Send the trace to the ring-buffer.
 158  * @return the original return address.
 159  */
 160 unsigned long ftrace_return_to_handler(unsigned long frame_pointer)
 161 {
 162         struct ftrace_graph_ret trace;
 163         unsigned long ret;
 164
 165         ftrace_pop_return_trace(&trace, &ret, frame_pointer);
 166         trace.rettime = trace_clock_local();
 167         ftrace_graph_return(&trace);
 168         barrier();
 169         current->curr_ret_stack--;
 170
 171         if (unlikely(!ret)) {
 172                 ftrace_graph_stop();
 173                 WARN_ON(1);
 174                 /* Might as well panic. What else to do? */
 175                 ret = (unsigned long)panic;
 176         }
 177
 178         return ret;
 179 }
 180
 181 static int __trace_graph_entry(struct trace_array *tr,
 182                                 struct ftrace_graph_ent *trace,
 183                                 unsigned long flags,
 184                                 int pc)
 185 {
 186         struct ftrace_event_call *call = &event_funcgraph_entry;
 187         struct ring_buffer_event *event;
 188         struct ring_buffer *buffer = tr->buffer;
 189         struct ftrace_graph_ent_entry *entry;
 190
 191         if (unlikely(__this_cpu_read(ftrace_cpu_disabled)))
 192                 return 0;
 193
 194         event = trace_buffer_lock_reserve(buffer, TRACE_GRAPH_ENT,
 195                                           sizeof(*entry), flags, pc);
 196         if (!event)
 197                 return 0;
 198         entry   = ring_buffer_event_data(event);
 199         entry->graph_ent                        = *trace;
 200         if (!filter_current_check_discard(buffer, call, entry, event))
 201                 ring_buffer_unlock_commit(buffer, event);
 202
 203         return 1;
 204 }
 205
 206 int trace_graph_entry(struct ftrace_graph_ent *trace)
 207 {
 208         struct trace_array *tr = graph_array;
 209         struct trace_array_cpu *data;
 210         unsigned long flags;
 211         long disabled;
 212         int ret;
 213         int cpu;
 214         int pc;
 215
 216         if (!ftrace_trace_task(current))
 217                 return 0;
 218
 219         /* trace it when it is-nested-in or is a function enabled. */
 220         if (!(trace->depth || ftrace_graph_addr(trace->func)))
 221                 return 0;
 222
 223         local_irq_save(flags);
 224         cpu = raw_smp_processor_id();
 225         data = tr->data[cpu];
 226         disabled = atomic_inc_return(&data->disabled);
 227         if (likely(disabled == 1)) {
 228                 pc = preempt_count();
 229                 ret = __trace_graph_entry(tr, trace, flags, pc);
 230         } else {
 231                 ret = 0;
 232         }
 233
 234         atomic_dec(&data->disabled);
 235         local_irq_restore(flags);
 236
 237         return ret;
 238 }
 239
 240 static void __trace_graph_return(struct trace_array *tr,
 241                                 struct ftrace_graph_ret *trace,
 242                                 unsigned long flags,
 243                                 int pc)
 244 {
 245         struct ftrace_event_call *call = &event_funcgraph_exit;
 246         struct ring_buffer_event *event;
 247         struct ring_buffer *buffer = tr->buffer;
 248         struct ftrace_graph_ret_entry *entry;
 249
 250         if (unlikely(__this_cpu_read(ftrace_cpu_disabled)))
 251                 return;
 252
 253         event = trace_buffer_lock_reserve(buffer, TRACE_GRAPH_RET,
 254                                           sizeof(*entry), flags, pc);
 255         if (!event)
 256                 return;
 257         entry   = ring_buffer_event_data(event);
 258         entry->ret                              = *trace;
 259         if (!filter_current_check_discard(buffer, call, entry, event))
 260                 ring_buffer_unlock_commit(buffer, event);
 261 }
 262
 263 void trace_graph_return(struct ftrace_graph_ret *trace)
 264 {
 265         struct trace_array *tr = graph_array;
 266         struct trace_array_cpu *data;
 267         unsigned long flags;
 268         long disabled;
 269         int cpu;
 270         int pc;
 271
 272         local_irq_save(flags);
 273         cpu = raw_smp_processor_id();
 274         data = tr->data[cpu];
 275         disabled = atomic_inc_return(&data->disabled);
 276         if (likely(disabled == 1)) {
 277                 pc = preempt_count();
 278                 __trace_graph_return(tr, trace, flags, pc);
 279         }
 280         atomic_dec(&data->disabled);
 281         local_irq_restore(flags);
 282 }
 283
 284 void set_graph_array(struct trace_array *tr)
 285 {
 286         graph_array = tr;
 287
 288         /* Make graph_array visible before we start tracing */
 289
 290         smp_mb();
 291 }
 292
 293 static int graph_trace_init(struct trace_array *tr)
 294 {
 295         int ret;
 296
 297         set_graph_array(tr);
 298         ret = register_ftrace_graph(&trace_graph_return,
 299                                     &trace_graph_entry);
 300         if (ret)
 301                 return ret;
 302         tracing_start_cmdline_record();
 303
 304         return 0;
 305 }
 306
 307 static void graph_trace_reset(struct trace_array *tr)
 308 {
 309         tracing_stop_cmdline_record();
 310         unregister_ftrace_graph();
 311 }
 312
 313 static int max_bytes_for_cpu;
 314
 315 static enum print_line_t
 316 print_graph_cpu(struct trace_seq *s, int cpu)
 317 {
 318         int ret;
 319
 320         /*
 321          * Start with a space character - to make it stand out
 322          * to the right a bit when trace output is pasted into
 323          * email:
 324          */
 325         ret = trace_seq_printf(s, " %*d) ", max_bytes_for_cpu, cpu);
 326         if (!ret)
 327                 return TRACE_TYPE_PARTIAL_LINE;
 328
 329         return TRACE_TYPE_HANDLED;
 330 }
 331
 332 #define TRACE_GRAPH_PROCINFO_LENGTH     14
 333
 334 static enum print_line_t
 335 print_graph_proc(struct trace_seq *s, pid_t pid)
 336 {
 337         char comm[TASK_COMM_LEN];
 338         /* sign + log10(MAX_INT) + '\0' */
 339         char pid_str[11];
 340         int spaces = 0;
 341         int ret;
 342         int len;
 343         int i;
 344
 345         trace_find_cmdline(pid, comm);
 346         comm[7] = '\0';
 347         sprintf(pid_str, "%d", pid);
 348
 349         /* 1 stands for the "-" character */
 350         len = strlen(comm) + strlen(pid_str) + 1;
 351
 352         if (len < TRACE_GRAPH_PROCINFO_LENGTH)
 353                 spaces = TRACE_GRAPH_PROCINFO_LENGTH - len;
 354
 355         /* First spaces to align center */
 356         for (i = 0; i < spaces / 2; i++) {
 357                 ret = trace_seq_printf(s, " ");
 358                 if (!ret)
 359                         return TRACE_TYPE_PARTIAL_LINE;
 360         }
 361
 362         ret = trace_seq_printf(s, "%s-%s", comm, pid_str);
 363         if (!ret)
 364                 return TRACE_TYPE_PARTIAL_LINE;
 365
 366         /* Last spaces to align center */
 367         for (i = 0; i < spaces - (spaces / 2); i++) {
 368                 ret = trace_seq_printf(s, " ");
 369                 if (!ret)
 370                         return TRACE_TYPE_PARTIAL_LINE;
 371         }
 372         return TRACE_TYPE_HANDLED;
 373 }
 374
 375
 376 static enum print_line_t
 377 print_graph_lat_fmt(struct trace_seq *s, struct trace_entry *entry)
 378 {
 379         if (!trace_seq_putc(s, ' '))
 380                 return 0;
 381
 382         return trace_print_lat_fmt(s, entry);
 383 }
 384
 385 /* If the pid changed since the last trace, output this event */
 386 static enum print_line_t
 387 verif_pid(struct trace_seq *s, pid_t pid, int cpu, struct fgraph_data *data)
 388 {
 389         pid_t prev_pid;
 390         pid_t *last_pid;
 391         int ret;
 392
 393         if (!data)
 394                 return TRACE_TYPE_HANDLED;
 395
 396         last_pid = &(per_cpu_ptr(data->cpu_data, cpu)->last_pid);
 397
 398         if (*last_pid == pid)
 399                 return TRACE_TYPE_HANDLED;
 400
 401         prev_pid = *last_pid;
 402         *last_pid = pid;
 403
 404         if (prev_pid == -1)
 405                 return TRACE_TYPE_HANDLED;
 406 /*
 407  * Context-switch trace line:
 408
 409  ------------------------------------------
 410  | 1)  migration/0--1  =>  sshd-1755
 411  ------------------------------------------
 412
 413  */
 414         ret = trace_seq_printf(s,
 415                 " ------------------------------------------\n");
 416         if (!ret)
 417                 return TRACE_TYPE_PARTIAL_LINE;
 418
 419         ret = print_graph_cpu(s, cpu);
 420         if (ret == TRACE_TYPE_PARTIAL_LINE)
 421                 return TRACE_TYPE_PARTIAL_LINE;
 422
 423         ret = print_graph_proc(s, prev_pid);
 424         if (ret == TRACE_TYPE_PARTIAL_LINE)
 425                 return TRACE_TYPE_PARTIAL_LINE;
 426
 427         ret = trace_seq_printf(s, " => ");
 428         if (!ret)
 429                 return TRACE_TYPE_PARTIAL_LINE;
 430
 431         ret = print_graph_proc(s, pid);
 432         if (ret == TRACE_TYPE_PARTIAL_LINE)
 433                 return TRACE_TYPE_PARTIAL_LINE;
 434
 435         ret = trace_seq_printf(s,
 436                 "\n ------------------------------------------\n\n");
 437         if (!ret)
 438                 return TRACE_TYPE_PARTIAL_LINE;
 439
 440         return TRACE_TYPE_HANDLED;
 441 }
 442
 443 static struct ftrace_graph_ret_entry *
 444 get_return_for_leaf(struct trace_iterator *iter,
 445                 struct ftrace_graph_ent_entry *curr)
 446 {
 447         struct fgraph_data *data = iter->private;
 448         struct ring_buffer_iter *ring_iter = NULL;
 449         struct ring_buffer_event *event;
 450         struct ftrace_graph_ret_entry *next;
 451
 452         /*
 453          * If the previous output failed to write to the seq buffer,
 454          * then we just reuse the data from before.
 455          */
 456         if (data && data->failed) {
 457                 curr = &data->ent;
 458                 next = &data->ret;
 459         } else {
 460
 461                 ring_iter = iter->buffer_iter[iter->cpu];
 462
 463                 /* First peek to compare current entry and the next one */
 464                 if (ring_iter)
 465                         event = ring_buffer_iter_peek(ring_iter, NULL);
 466                 else {
 467                         /*
 468                          * We need to consume the current entry to see
 469                          * the next one.
 470                          */
 471                         ring_buffer_consume(iter->tr->buffer, iter->cpu, NULL);
 472                         event = ring_buffer_peek(iter->tr->buffer, iter->cpu,
 473                                                  NULL);
 474                 }
 475
 476                 if (!event)
 477                         return NULL;
 478
 479                 next = ring_buffer_event_data(event);
 480
 481                 if (data) {
 482                         /*
 483                          * Save current and next entries for later reference
 484                          * if the output fails.
 485                          */
 486                         data->ent = *curr;
 487                         data->ret = *next;
 488                 }
 489         }
 490
 491         if (next->ent.type != TRACE_GRAPH_RET)
 492                 return NULL;
 493
 494         if (curr->ent.pid != next->ent.pid ||
 495                         curr->graph_ent.func != next->ret.func)
 496                 return NULL;
 497
 498         /* this is a leaf, now advance the iterator */
 499         if (ring_iter)
 500                 ring_buffer_read(ring_iter, NULL);
 501
 502         return next;
 503 }
 504
 505 /* Signal a overhead of time execution to the output */
 506 static int
 507 print_graph_overhead(unsigned long long duration, struct trace_seq *s)
 508 {
 509         /* If duration disappear, we don't need anything */
 510         if (!(tracer_flags.val & TRACE_GRAPH_PRINT_DURATION))
 511                 return 1;
 512
 513         /* Non nested entry or return */
 514         if (duration == -1)
 515                 return trace_seq_printf(s, "  ");
 516
 517         if (tracer_flags.val & TRACE_GRAPH_PRINT_OVERHEAD) {
 518                 /* Duration exceeded 100 msecs */
 519                 if (duration > 100000ULL)
 520                         return trace_seq_printf(s, "! ");
 521
 522                 /* Duration exceeded 10 msecs */
 523                 if (duration > 10000ULL)
 524                         return trace_seq_printf(s, "+ ");
 525         }
 526
 527         return trace_seq_printf(s, "  ");
 528 }
 529
 530 static int print_graph_abs_time(u64 t, struct trace_seq *s)
 531 {
 532         unsigned long usecs_rem;
 533
 534         usecs_rem = do_div(t, NSEC_PER_SEC);
 535         usecs_rem /= 1000;
 536
 537         return trace_seq_printf(s, "%5lu.%06lu |  ",
 538                         (unsigned long)t, usecs_rem);
 539 }
 540
 541 static enum print_line_t
 542 print_graph_irq(struct trace_iterator *iter, unsigned long addr,
 543                 enum trace_type type, int cpu, pid_t pid)
 544 {
 545         int ret;
 546         struct trace_seq *s = &iter->seq;
 547
 548         if (addr < (unsigned long)__irqentry_text_start ||
 549                 addr >= (unsigned long)__irqentry_text_end)
 550                 return TRACE_TYPE_UNHANDLED;
 551
 552         /* Absolute time */
 553         if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME) {
 554                 ret = print_graph_abs_time(iter->ts, s);
 555                 if (!ret)
 556                         return TRACE_TYPE_PARTIAL_LINE;
 557         }
 558
 559         /* Cpu */
 560         if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU) {
 561                 ret = print_graph_cpu(s, cpu);
 562                 if (ret == TRACE_TYPE_PARTIAL_LINE)
 563                         return TRACE_TYPE_PARTIAL_LINE;
 564         }
 565
 566         /* Proc */
 567         if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC) {
 568                 ret = print_graph_proc(s, pid);
 569                 if (ret == TRACE_TYPE_PARTIAL_LINE)
 570                         return TRACE_TYPE_PARTIAL_LINE;
 571                 ret = trace_seq_printf(s, " | ");
 572                 if (!ret)
 573                         return TRACE_TYPE_PARTIAL_LINE;
 574         }
 575
 576         /* No overhead */
 577         ret = print_graph_overhead(-1, s);
 578         if (!ret)
 579                 return TRACE_TYPE_PARTIAL_LINE;
 580
 581         if (type == TRACE_GRAPH_ENT)
 582                 ret = trace_seq_printf(s, "==========>");
 583         else
 584                 ret = trace_seq_printf(s, "<==========");
 585
 586         if (!ret)
 587                 return TRACE_TYPE_PARTIAL_LINE;
 588
 589         /* Don't close the duration column if haven't one */
 590         if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION)
 591                 trace_seq_printf(s, " |");
 592         ret = trace_seq_printf(s, "\n");
 593
 594         if (!ret)
 595                 return TRACE_TYPE_PARTIAL_LINE;
 596         return TRACE_TYPE_HANDLED;
 597 }
 598
 599 enum print_line_t
 600 trace_print_graph_duration(unsigned long long duration, struct trace_seq *s)
 601 {
 602         unsigned long nsecs_rem = do_div(duration, 1000);
 603         /* log10(ULONG_MAX) + '\0' */
 604         char msecs_str[21];
 605         char nsecs_str[5];
 606         int ret, len;
 607         int i;
 608
 609         sprintf(msecs_str, "%lu", (unsigned long) duration);
 610
 611         /* Print msecs */
 612         ret = trace_seq_printf(s, "%s", msecs_str);
 613         if (!ret)
 614                 return TRACE_TYPE_PARTIAL_LINE;
 615
 616         len = strlen(msecs_str);
 617
 618         /* Print nsecs (we don't want to exceed 7 numbers) */
 619         if (len < 7) {
 620                 snprintf(nsecs_str, 8 - len, "%03lu", nsecs_rem);
 621                 ret = trace_seq_printf(s, ".%s", nsecs_str);
 622                 if (!ret)
 623                         return TRACE_TYPE_PARTIAL_LINE;
 624                 len += strlen(nsecs_str);
 625         }
 626
 627         ret = trace_seq_printf(s, " us ");
 628         if (!ret)
 629                 return TRACE_TYPE_PARTIAL_LINE;
 630
 631         /* Print remaining spaces to fit the row's width */
 632         for (i = len; i < 7; i++) {
 633                 ret = trace_seq_printf(s, " ");
 634                 if (!ret)
 635                         return TRACE_TYPE_PARTIAL_LINE;
 636         }
 637         return TRACE_TYPE_HANDLED;
 638 }
 639
 640 static enum print_line_t
 641 print_graph_duration(unsigned long long duration, struct trace_seq *s)
 642 {
 643         int ret;
 644
 645         ret = trace_print_graph_duration(duration, s);
 646         if (ret != TRACE_TYPE_HANDLED)
 647                 return ret;
 648
 649         ret = trace_seq_printf(s, "|  ");
 650         if (!ret)
 651                 return TRACE_TYPE_PARTIAL_LINE;
 652
 653         return TRACE_TYPE_HANDLED;
 654 }
 655
 656 /* Case of a leaf function on its call entry */
 657 static enum print_line_t
 658 print_graph_entry_leaf(struct trace_iterator *iter,
 659                 struct ftrace_graph_ent_entry *entry,
 660                 struct ftrace_graph_ret_entry *ret_entry, struct trace_seq *s)
 661 {
 662         struct fgraph_data *data = iter->private;
 663         struct ftrace_graph_ret *graph_ret;
 664         struct ftrace_graph_ent *call;
 665         unsigned long long duration;
 666         int ret;
 667         int i;
 668
 669         graph_ret = &ret_entry->ret;
 670         call = &entry->graph_ent;
 671         duration = graph_ret->rettime - graph_ret->calltime;
 672
 673         if (data) {
 674                 struct fgraph_cpu_data *cpu_data;
 675                 int cpu = iter->cpu;
 676
 677                 cpu_data = per_cpu_ptr(data->cpu_data, cpu);
 678
 679                 /*
 680                  * Comments display at + 1 to depth. Since
 681                  * this is a leaf function, keep the comments
 682                  * equal to this depth.
 683                  */
 684                 cpu_data->depth = call->depth - 1;
 685
 686                 /* No need to keep this function around for this depth */
 687                 if (call->depth < FTRACE_RETFUNC_DEPTH)
 688                         cpu_data->enter_funcs[call->depth] = 0;
 689         }
 690
 691         /* Overhead */
 692         ret = print_graph_overhead(duration, s);
 693         if (!ret)
 694                 return TRACE_TYPE_PARTIAL_LINE;
 695
 696         /* Duration */
 697         if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION) {
 698                 ret = print_graph_duration(duration, s);
 699                 if (ret == TRACE_TYPE_PARTIAL_LINE)
 700                         return TRACE_TYPE_PARTIAL_LINE;
 701         }
 702
 703         /* Function */
 704         for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
 705                 ret = trace_seq_printf(s, " ");
 706                 if (!ret)
 707                         return TRACE_TYPE_PARTIAL_LINE;
 708         }
 709
 710         ret = trace_seq_printf(s, "%ps();\n", (void *)call->func);
 711         if (!ret)
 712                 return TRACE_TYPE_PARTIAL_LINE;
 713
 714         return TRACE_TYPE_HANDLED;
 715 }
 716
 717 static enum print_line_t
 718 print_graph_entry_nested(struct trace_iterator *iter,
 719                          struct ftrace_graph_ent_entry *entry,
 720                          struct trace_seq *s, int cpu)
 721 {
 722         struct ftrace_graph_ent *call = &entry->graph_ent;
 723         struct fgraph_data *data = iter->private;
 724         int ret;
 725         int i;
 726
 727         if (data) {
 728                 struct fgraph_cpu_data *cpu_data;
 729                 int cpu = iter->cpu;
 730
 731                 cpu_data = per_cpu_ptr(data->cpu_data, cpu);
 732                 cpu_data->depth = call->depth;
 733
 734                 /* Save this function pointer to see if the exit matches */
 735                 if (call->depth < FTRACE_RETFUNC_DEPTH)
 736                         cpu_data->enter_funcs[call->depth] = call->func;
 737         }
 738
 739         /* No overhead */
 740         ret = print_graph_overhead(-1, s);
 741         if (!ret)
 742                 return TRACE_TYPE_PARTIAL_LINE;
 743
 744         /* No time */
 745         if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION) {
 746                 ret = trace_seq_printf(s, "            |  ");
 747                 if (!ret)
 748                         return TRACE_TYPE_PARTIAL_LINE;
 749         }
 750
 751         /* Function */
 752         for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) {
 753                 ret = trace_seq_printf(s, " ");
 754                 if (!ret)
 755                         return TRACE_TYPE_PARTIAL_LINE;
 756         }
 757
 758         ret = trace_seq_printf(s, "%ps() {\n", (void *)call->func);
 759         if (!ret)
 760                 return TRACE_TYPE_PARTIAL_LINE;
 761
 762         /*
 763          * we already consumed the current entry to check the next one
 764          * and see if this is a leaf.
 765          */
 766         return TRACE_TYPE_NO_CONSUME;
 767 }
 768
 769 static enum print_line_t
 770 print_graph_prologue(struct trace_iterator *iter, struct trace_seq *s,
 771                      int type, unsigned long addr)
 772 {
 773         struct fgraph_data *data = iter->private;
 774         struct trace_entry *ent = iter->ent;
 775         int cpu = iter->cpu;
 776         int ret;
 777
 778         /* Pid */
 779         if (verif_pid(s, ent->pid, cpu, data) == TRACE_TYPE_PARTIAL_LINE)
 780                 return TRACE_TYPE_PARTIAL_LINE;
 781
 782         if (type) {
 783                 /* Interrupt */
 784                 ret = print_graph_irq(iter, addr, type, cpu, ent->pid);
 785                 if (ret == TRACE_TYPE_PARTIAL_LINE)
 786                         return TRACE_TYPE_PARTIAL_LINE;
 787         }
 788
 789         /* Absolute time */
 790         if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME) {
 791                 ret = print_graph_abs_time(iter->ts, s);
 792                 if (!ret)
 793                         return TRACE_TYPE_PARTIAL_LINE;
 794         }
 795
 796         /* Cpu */
 797         if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU) {
 798                 ret = print_graph_cpu(s, cpu);
 799                 if (ret == TRACE_TYPE_PARTIAL_LINE)
 800                         return TRACE_TYPE_PARTIAL_LINE;
 801         }
 802
 803         /* Proc */
 804         if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC) {
 805                 ret = print_graph_proc(s, ent->pid);
 806                 if (ret == TRACE_TYPE_PARTIAL_LINE)
 807                         return TRACE_TYPE_PARTIAL_LINE;
 808
 809                 ret = trace_seq_printf(s, " | ");
 810                 if (!ret)
 811                         return TRACE_TYPE_PARTIAL_LINE;
 812         }
 813
 814         /* Latency format */
 815         if (trace_flags & TRACE_ITER_LATENCY_FMT) {
 816                 ret = print_graph_lat_fmt(s, ent);
 817                 if (ret == TRACE_TYPE_PARTIAL_LINE)
 818                         return TRACE_TYPE_PARTIAL_LINE;
 819         }
 820
 821         return 0;
 822 }
 823
 824 static enum print_line_t
 825 print_graph_entry(struct ftrace_graph_ent_entry *field, struct trace_seq *s,
 826                         struct trace_iterator *iter)
 827 {
 828         struct fgraph_data *data = iter->private;
 829         struct ftrace_graph_ent *call = &field->graph_ent;
 830         struct ftrace_graph_ret_entry *leaf_ret;
 831         static enum print_line_t ret;
 832         int cpu = iter->cpu;
 833
 834         if (print_graph_prologue(iter, s, TRACE_GRAPH_ENT, call->func))
 835                 return TRACE_TYPE_PARTIAL_LINE;
 836
 837         leaf_ret = get_return_for_leaf(iter, field);
 838         if (leaf_ret)
 839                 ret = print_graph_entry_leaf(iter, field, leaf_ret, s);
 840         else
 841                 ret = print_graph_entry_nested(iter, field, s, cpu);
 842
 843         if (data) {
 844                 /*
 845                  * If we failed to write our output, then we need to make
 846                  * note of it. Because we already consumed our entry.
 847                  */
 848                 if (s->full) {
 849                         data->failed = 1;
 850                         data->cpu = cpu;
 851                 } else
 852                         data->failed = 0;
 853         }
 854
 855         return ret;
 856 }
 857
 858 static enum print_line_t
 859 print_graph_return(struct ftrace_graph_ret *trace, struct trace_seq *s,
 860                    struct trace_entry *ent, struct trace_iterator *iter)
 861 {
 862         unsigned long long duration = trace->rettime - trace->calltime;
 863         struct fgraph_data *data = iter->private;
 864         pid_t pid = ent->pid;
 865         int cpu = iter->cpu;
 866         int func_match = 1;
 867         int ret;
 868         int i;
 869
 870         if (data) {
 871                 struct fgraph_cpu_data *cpu_data;
 872                 int cpu = iter->cpu;
 873
 874                 cpu_data = per_cpu_ptr(data->cpu_data, cpu);
 875
 876                 /*
 877                  * Comments display at + 1 to depth. This is the
 878                  * return from a function, we now want the comments
 879                  * to display at the same level of the bracket.
 880                  */
 881                 cpu_data->depth = trace->depth - 1;
 882
 883                 if (trace->depth < FTRACE_RETFUNC_DEPTH) {
 884                         if (cpu_data->enter_funcs[trace->depth] != trace->func)
 885                                 func_match = 0;
 886                         cpu_data->enter_funcs[trace->depth] = 0;
 887                 }
 888         }
 889
 890         if (print_graph_prologue(iter, s, 0, 0))
 891                 return TRACE_TYPE_PARTIAL_LINE;
 892
 893         /* Overhead */
 894         ret = print_graph_overhead(duration, s);
 895         if (!ret)
 896                 return TRACE_TYPE_PARTIAL_LINE;
 897
 898         /* Duration */
 899         if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION) {
 900                 ret = print_graph_duration(duration, s);
 901                 if (ret == TRACE_TYPE_PARTIAL_LINE)
 902                         return TRACE_TYPE_PARTIAL_LINE;
 903         }
 904
 905         /* Closing brace */
 906         for (i = 0; i < trace->depth * TRACE_GRAPH_INDENT; i++) {
 907                 ret = trace_seq_printf(s, " ");
 908                 if (!ret)
 909                         return TRACE_TYPE_PARTIAL_LINE;
 910         }
 911
 912         /*
 913          * If the return function does not have a matching entry,
 914          * then the entry was lost. Instead of just printing
 915          * the '}' and letting the user guess what function this
 916          * belongs to, write out the function name.
 917          */
 918         if (func_match) {
 919                 ret = trace_seq_printf(s, "}\n");
 920                 if (!ret)
 921                         return TRACE_TYPE_PARTIAL_LINE;
 922         } else {
 923                 ret = trace_seq_printf(s, "} (%ps)\n", (void *)trace->func);
 924                 if (!ret)
 925                         return TRACE_TYPE_PARTIAL_LINE;
 926         }
 927
 928         /* Overrun */
 929         if (tracer_flags.val & TRACE_GRAPH_PRINT_OVERRUN) {
 930                 ret = trace_seq_printf(s, " (Overruns: %lu)\n",
 931                                         trace->overrun);
 932                 if (!ret)
 933                         return TRACE_TYPE_PARTIAL_LINE;
 934         }
 935
 936         ret = print_graph_irq(iter, trace->func, TRACE_GRAPH_RET, cpu, pid);
 937         if (ret == TRACE_TYPE_PARTIAL_LINE)
 938                 return TRACE_TYPE_PARTIAL_LINE;
 939
 940         return TRACE_TYPE_HANDLED;
 941 }
 942
 943 static enum print_line_t
 944 print_graph_comment(struct trace_seq *s,  struct trace_entry *ent,
 945                     struct trace_iterator *iter)
 946 {
 947         unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK);
 948         struct fgraph_data *data = iter->private;
 949         struct trace_event *event;
 950         int depth = 0;
 951         int ret;
 952         int i;
 953
 954         if (data)
 955                 depth = per_cpu_ptr(data->cpu_data, iter->cpu)->depth;
 956
 957         if (print_graph_prologue(iter, s, 0, 0))
 958                 return TRACE_TYPE_PARTIAL_LINE;
 959
 960         /* No overhead */
 961         ret = print_graph_overhead(-1, s);
 962         if (!ret)
 963                 return TRACE_TYPE_PARTIAL_LINE;
 964
 965         /* No time */
 966         if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION) {
 967                 ret = trace_seq_printf(s, "            |  ");
 968                 if (!ret)
 969                         return TRACE_TYPE_PARTIAL_LINE;
 970         }
 971
 972         /* Indentation */
 973         if (depth > 0)
 974                 for (i = 0; i < (depth + 1) * TRACE_GRAPH_INDENT; i++) {
 975                         ret = trace_seq_printf(s, " ");
 976                         if (!ret)
 977                                 return TRACE_TYPE_PARTIAL_LINE;
 978                 }
 979
 980         /* The comment */
 981         ret = trace_seq_printf(s, "/* ");
 982         if (!ret)
 983                 return TRACE_TYPE_PARTIAL_LINE;
 984
 985         switch (iter->ent->type) {
 986         case TRACE_BPRINT:
 987                 ret = trace_print_bprintk_msg_only(iter);
 988                 if (ret != TRACE_TYPE_HANDLED)
 989                         return ret;
 990                 break;
 991         case TRACE_PRINT:
 992                 ret = trace_print_printk_msg_only(iter);
 993                 if (ret != TRACE_TYPE_HANDLED)
 994                         return ret;
 995                 break;
 996         default:
 997                 event = ftrace_find_event(ent->type);
 998                 if (!event)
 999                         return TRACE_TYPE_UNHANDLED;
1000
1001                 ret = event->trace(iter, sym_flags);
1002                 if (ret != TRACE_TYPE_HANDLED)
1003                         return ret;
1004         }
1005
1006         /* Strip ending newline */
1007         if (s->buffer[s->len - 1] == '\n') {
1008                 s->buffer[s->len - 1] = '\0';
1009                 s->len--;
1010         }
1011
1012         ret = trace_seq_printf(s, " */\n");
1013         if (!ret)
1014                 return TRACE_TYPE_PARTIAL_LINE;
1015
1016         return TRACE_TYPE_HANDLED;
1017 }
1018
1019
1020 enum print_line_t
1021 print_graph_function(struct trace_iterator *iter)
1022 {
1023         struct ftrace_graph_ent_entry *field;
1024         struct fgraph_data *data = iter->private;
1025         struct trace_entry *entry = iter->ent;
1026         struct trace_seq *s = &iter->seq;
1027         int cpu = iter->cpu;
1028         int ret;
1029
1030         if (data && per_cpu_ptr(data->cpu_data, cpu)->ignore) {
1031                 per_cpu_ptr(data->cpu_data, cpu)->ignore = 0;
1032                 return TRACE_TYPE_HANDLED;
1033         }
1034
1035         /*
1036          * If the last output failed, there's a possibility we need
1037          * to print out the missing entry which would never go out.
1038          */
1039         if (data && data->failed) {
1040                 field = &data->ent;
1041                 iter->cpu = data->cpu;
1042                 ret = print_graph_entry(field, s, iter);
1043                 if (ret == TRACE_TYPE_HANDLED && iter->cpu != cpu) {
1044                         per_cpu_ptr(data->cpu_data, iter->cpu)->ignore = 1;
1045                         ret = TRACE_TYPE_NO_CONSUME;
1046                 }
1047                 iter->cpu = cpu;
1048                 return ret;
1049         }
1050
1051         switch (entry->type) {
1052         case TRACE_GRAPH_ENT: {
1053                 /*
1054                  * print_graph_entry() may consume the current event,
1055                  * thus @field may become invalid, so we need to save it.
1056                  * sizeof(struct ftrace_graph_ent_entry) is very small,
1057                  * it can be safely saved at the stack.
1058                  */
1059                 struct ftrace_graph_ent_entry saved;
1060                 trace_assign_type(field, entry);
1061                 saved = *field;
1062                 return print_graph_entry(&saved, s, iter);
1063         }
1064         case TRACE_GRAPH_RET: {
1065                 struct ftrace_graph_ret_entry *field;
1066                 trace_assign_type(field, entry);
1067                 return print_graph_return(&field->ret, s, entry, iter);
1068         }
1069         default:
1070                 return print_graph_comment(s, entry, iter);
1071         }
1072
1073         return TRACE_TYPE_HANDLED;
1074 }
1075
1076 static void print_lat_header(struct seq_file *s)
1077 {
1078         static const char spaces[] = "                " /* 16 spaces */
1079                 "    "                                  /* 4 spaces */
1080                 "                 ";                    /* 17 spaces */
1081         int size = 0;
1082
1083         if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME)
1084                 size += 16;
1085         if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
1086                 size += 4;
1087         if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
1088                 size += 17;
1089
1090         seq_printf(s, "#%.*s  _-----=> irqs-off        \n", size, spaces);
1091         seq_printf(s, "#%.*s / _----=> need-resched    \n", size, spaces);
1092         seq_printf(s, "#%.*s| / _---=> hardirq/softirq \n", size, spaces);
1093         seq_printf(s, "#%.*s|| / _--=> preempt-depth   \n", size, spaces);
1094         seq_printf(s, "#%.*s||| / _-=> lock-depth      \n", size, spaces);
1095         seq_printf(s, "#%.*s|||| /                     \n", size, spaces);
1096 }
1097
1098 static void print_graph_headers(struct seq_file *s)
1099 {
1100         int lat = trace_flags & TRACE_ITER_LATENCY_FMT;
1101
1102         if (lat)
1103                 print_lat_header(s);
1104
1105         /* 1st line */
1106         seq_printf(s, "#");
1107         if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME)
1108                 seq_printf(s, "     TIME       ");
1109         if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
1110                 seq_printf(s, " CPU");
1111         if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
1112                 seq_printf(s, "  TASK/PID       ");
1113         if (lat)
1114                 seq_printf(s, "|||||");
1115         if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION)
1116                 seq_printf(s, "  DURATION   ");
1117         seq_printf(s, "               FUNCTION CALLS\n");
1118
1119         /* 2nd line */
1120         seq_printf(s, "#");
1121         if (tracer_flags.val & TRACE_GRAPH_PRINT_ABS_TIME)
1122                 seq_printf(s, "      |         ");
1123         if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
1124                 seq_printf(s, " |  ");
1125         if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
1126                 seq_printf(s, "   |    |        ");
1127         if (lat)
1128                 seq_printf(s, "|||||");
1129         if (tracer_flags.val & TRACE_GRAPH_PRINT_DURATION)
1130                 seq_printf(s, "   |   |      ");
1131         seq_printf(s, "               |   |   |   |\n");
1132 }
1133
1134 static void graph_trace_open(struct trace_iterator *iter)
1135 {
1136         /* pid and depth on the last trace processed */
1137         struct fgraph_data *data;
1138         int cpu;
1139
1140         iter->private = NULL;
1141
1142         data = kzalloc(sizeof(*data), GFP_KERNEL);
1143         if (!data)
1144                 goto out_err;
1145
1146         data->cpu_data = alloc_percpu(struct fgraph_cpu_data);
1147         if (!data->cpu_data)
1148                 goto out_err_free;
1149
1150         for_each_possible_cpu(cpu) {
1151                 pid_t *pid = &(per_cpu_ptr(data->cpu_data, cpu)->last_pid);
1152                 int *depth = &(per_cpu_ptr(data->cpu_data, cpu)->depth);
1153                 int *ignore = &(per_cpu_ptr(data->cpu_data, cpu)->ignore);
1154                 *pid = -1;
1155                 *depth = 0;
1156                 *ignore = 0;
1157         }
1158
1159         iter->private = data;
1160
1161         return;
1162
1163  out_err_free:
1164         kfree(data);
1165  out_err:
1166         pr_warning("function graph tracer: not enough memory\n");
1167 }
1168
1169 static void graph_trace_close(struct trace_iterator *iter)
1170 {
1171         struct fgraph_data *data = iter->private;
1172
1173         if (data) {
1174                 free_percpu(data->cpu_data);
1175                 kfree(data);
1176         }
1177 }
1178
1179 static struct tracer graph_trace __read_mostly = {
1180         .name           = "function_graph",
1181         .open           = graph_trace_open,
1182         .pipe_open      = graph_trace_open,
1183         .close          = graph_trace_close,
1184         .pipe_close     = graph_trace_close,
1185         .wait_pipe      = poll_wait_pipe,
1186         .init           = graph_trace_init,
1187         .reset          = graph_trace_reset,
1188         .print_line     = print_graph_function,
1189         .print_header   = print_graph_headers,
1190         .flags          = &tracer_flags,
1191 #ifdef CONFIG_FTRACE_SELFTEST
1192         .selftest       = trace_selftest_startup_function_graph,
1193 #endif
1194 };
1195
1196 static __init int init_graph_trace(void)
1197 {
1198         max_bytes_for_cpu = snprintf(NULL, 0, "%d", nr_cpu_ids - 1);
1199
1200         return register_tracer(&graph_trace);
1201 }
1202
1203 device_initcall(init_graph_trace);