src/mc/mc_process.cpp

   1 /* Copyright (c) 2014-2015. The SimGrid Team.
   2  * All rights reserved.                                                     */
   3
   4 /* This program is free software; you can redistribute it and/or modify it
   5  * under the terms of the license (GNU LGPL) which comes with this package. */
   6
   7 #include <assert.h>
   8 #include <stddef.h>
   9 #include <stdbool.h>
  10 #include <stdint.h>
  11 #include <errno.h>
  12
  13 #include <sys/types.h>
  14 #include <fcntl.h>
  15 #include <unistd.h>
  16 #include <regex.h>
  17 #include <sys/mman.h> // PROT_*
  18
  19 #include <pthread.h>
  20
  21 #include <libgen.h>
  22
  23 #include <libunwind.h>
  24 #include <libunwind-ptrace.h>
  25
  26 #include <xbt/mmalloc.h>
  27
  28 #include "mc_process.h"
  29 #include "mc_object_info.h"
  30 #include "AddressSpace.hpp"
  31 #include "mc_unw.h"
  32 #include "mc_snapshot.h"
  33 #include "mc_ignore.h"
  34 #include "mc_smx.h"
  35 #include "mc_server.h"
  36
  37 extern "C" {
  38
  39 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(mc_process, mc,
  40                                 "MC process information");
  41
  42 static void MC_process_init_memory_map_info(mc_process_t process);
  43 static void MC_process_open_memory_file(mc_process_t process);
  44
  45 // ***** Destructor callbacks
  46
  47 // ***** mc_address_space methods for mc_process
  48
  49 static mc_process_t MC_process_get_process(mc_process_t p) {
  50   return p;
  51 }
  52
  53 }
  54
  55 // ***** mc_process
  56
  57 namespace simgrid {
  58 namespace mc {
  59
  60 Process::Process(pid_t pid, int sockfd)
  61 {
  62   Process* process = this;
  63
  64   process->process_flags = MC_PROCESS_NO_FLAG;
  65   process->socket = sockfd;
  66   process->pid = pid;
  67   if (pid==getpid())
  68     process->process_flags |= MC_PROCESS_SELF_FLAG;
  69   process->running = true;
  70   process->status = 0;
  71   process->memory_map = MC_get_memory_map(pid);
  72   process->memory_file = -1;
  73   process->cache_flags = MC_PROCESS_CACHE_FLAG_NONE;
  74   process->heap = NULL;
  75   process->heap_info = NULL;
  76   MC_process_init_memory_map_info(process);
  77   MC_process_open_memory_file(process);
  78
  79   // Read std_heap (is a struct mdesc*):
  80   dw_variable_t std_heap_var = MC_process_find_variable_by_name(process,
  81     "__mmalloc_default_mdp");
  82   if (!std_heap_var)
  83     xbt_die("No heap information in the target process");
  84   if(!std_heap_var->address)
  85     xbt_die("No constant address for this variable");
  86   MC_process_read(process, simgrid::mc::AddressSpace::Normal,
  87     &process->heap_address, std_heap_var->address, sizeof(struct mdesc*),
  88     simgrid::mc::ProcessIndexDisabled);
  89
  90   process->smx_process_infos = MC_smx_process_info_list_new();
  91   process->smx_old_process_infos = MC_smx_process_info_list_new();
  92
  93   process->checkpoint_ignore = MC_checkpoint_ignore_new();
  94
  95   process->unw_addr_space = unw_create_addr_space(&mc_unw_accessors  , __BYTE_ORDER);
  96   if (process->process_flags & MC_PROCESS_SELF_FLAG) {
  97     process->unw_underlying_addr_space = unw_local_addr_space;
  98     process->unw_underlying_context = NULL;
  99   } else {
 100     process->unw_underlying_addr_space = unw_create_addr_space(&mc_unw_vmread_accessors, __BYTE_ORDER);
 101     process->unw_underlying_context = _UPT_create(pid);
 102   }
 103 }
 104
 105 Process::~Process()
 106 {
 107   Process* process = this;
 108
 109   process->process_flags = MC_PROCESS_NO_FLAG;
 110   process->pid = 0;
 111
 112   MC_free_memory_map(process->memory_map);
 113   process->memory_map = NULL;
 114
 115   process->maestro_stack_start = NULL;
 116   process->maestro_stack_end = NULL;
 117
 118   xbt_dynar_free(&process->checkpoint_ignore);
 119
 120   xbt_dynar_free(&process->smx_process_infos);
 121   xbt_dynar_free(&process->smx_old_process_infos);
 122
 123   size_t i;
 124   for (i=0; i!=process->object_infos_size; ++i) {
 125     MC_free_object_info(&process->object_infos[i]);
 126   }
 127   free(process->object_infos);
 128   process->object_infos = NULL;
 129   process->object_infos_size = 0;
 130   if (process->memory_file >= 0) {
 131     close(process->memory_file);
 132   }
 133
 134   if (process->unw_underlying_addr_space != unw_local_addr_space) {
 135     unw_destroy_addr_space(process->unw_underlying_addr_space);
 136     _UPT_destroy(process->unw_underlying_context);
 137   }
 138   process->unw_underlying_context = NULL;
 139   process->unw_underlying_addr_space = NULL;
 140
 141   unw_destroy_addr_space(process->unw_addr_space);
 142   process->unw_addr_space = NULL;
 143
 144   process->cache_flags = MC_PROCESS_CACHE_FLAG_NONE;
 145
 146   free(process->heap);
 147   process->heap = NULL;
 148
 149   free(process->heap_info);
 150   process->heap_info = NULL;
 151 }
 152
 153 /** Refresh the information about the process
 154  *
 155  *  Do not use direclty, this is used by the getters when appropriate
 156  *  in order to have fresh data.
 157  */
 158 void Process::refresh_heap()
 159 {
 160   xbt_assert(mc_mode == MC_MODE_SERVER);
 161   xbt_assert(!this->is_self());
 162   // Read/dereference/refresh the std_heap pointer:
 163   if (!this->heap) {
 164     this->heap = (struct mdesc*) malloc(sizeof(struct mdesc));
 165   }
 166   MC_process_read(this, simgrid::mc::AddressSpace::Normal,
 167     this->heap, this->heap_address, sizeof(struct mdesc),
 168     simgrid::mc::ProcessIndexDisabled
 169     );
 170   this->cache_flags |= MC_PROCESS_CACHE_FLAG_HEAP;
 171 }
 172
 173 /** Refresh the information about the process
 174  *
 175  *  Do not use direclty, this is used by the getters when appropriate
 176  *  in order to have fresh data.
 177  * */
 178 void Process::refresh_malloc_info()
 179 {
 180   xbt_assert(mc_mode == MC_MODE_SERVER);
 181   xbt_assert(!this->is_self());
 182   if (!(this->cache_flags & MC_PROCESS_CACHE_FLAG_HEAP))
 183     this->refresh_heap();
 184   // Refresh process->heapinfo:
 185   size_t malloc_info_bytesize =
 186     (this->heap->heaplimit + 1) * sizeof(malloc_info);
 187   this->heap_info = (malloc_info*) realloc(this->heap_info, malloc_info_bytesize);
 188   MC_process_read(this, simgrid::mc::AddressSpace::Normal,
 189     this->heap_info,
 190     this->heap->heapinfo, malloc_info_bytesize,
 191     simgrid::mc::ProcessIndexDisabled);
 192   this->cache_flags |= MC_PROCESS_CACHE_FLAG_MALLOC_INFO;
 193 }
 194
 195 }
 196 }
 197
 198 extern "C" {
 199
 200 #define SO_RE "\\.so[\\.0-9]*$"
 201 #define VERSION_RE "-[\\.0-9]*$"
 202
 203 const char* FILTERED_LIBS[] = {
 204   "libstdc++",
 205   "libc++",
 206   "libm",
 207   "libgcc_s",
 208   "libpthread",
 209   "libunwind",
 210   "libunwind-x86_64",
 211   "libunwind-x86",
 212   "libunwind-ptrace",
 213   "libdw",
 214   "libdl",
 215   "librt",
 216   "liblzma",
 217   "libelf",
 218   "libbz2",
 219   "libz",
 220   "libelf",
 221   "libc",
 222   "ld"
 223 };
 224
 225 static bool MC_is_simgrid_lib(const char* libname)
 226 {
 227   return !strcmp(libname, "libsimgrid");
 228 }
 229
 230 static bool MC_is_filtered_lib(const char* libname)
 231 {
 232   const size_t n = sizeof(FILTERED_LIBS) / sizeof(const char*);
 233   size_t i;
 234   for (i=0; i!=n; ++i)
 235     if (strcmp(libname, FILTERED_LIBS[i])==0)
 236       return true;
 237   return false;
 238 }
 239
 240 struct s_mc_memory_map_re {
 241   regex_t so_re;
 242   regex_t version_re;
 243 };
 244
 245 static char* MC_get_lib_name(const char* pathname, struct s_mc_memory_map_re* res) {
 246   const char* map_basename = basename((char*) pathname);
 247
 248   regmatch_t match;
 249   if(regexec(&res->so_re, map_basename, 1, &match, 0))
 250     return NULL;
 251
 252   char* libname = strndup(map_basename, match.rm_so);
 253
 254   // Strip the version suffix:
 255   if(libname && !regexec(&res->version_re, libname, 1, &match, 0)) {
 256     char* temp = libname;
 257     libname = strndup(temp, match.rm_so);
 258     free(temp);
 259   }
 260
 261   return libname;
 262 }
 263
 264 /** @brief Finds the range of the different memory segments and binary paths */
 265 static void MC_process_init_memory_map_info(mc_process_t process)
 266 {
 267   XBT_DEBUG("Get debug information ...");
 268   process->maestro_stack_start = NULL;
 269   process->maestro_stack_end = NULL;
 270   process->object_infos = NULL;
 271   process->object_infos_size = 0;
 272   process->binary_info = NULL;
 273   process->libsimgrid_info = NULL;
 274
 275   struct s_mc_memory_map_re res;
 276
 277   if(regcomp(&res.so_re, SO_RE, 0) || regcomp(&res.version_re, VERSION_RE, 0))
 278     xbt_die(".so regexp did not compile");
 279
 280   memory_map_t maps = process->memory_map;
 281
 282   const char* current_name = NULL;
 283
 284   for (ssize_t i=0; i < maps->mapsize; i++) {
 285     map_region_t reg = &(maps->regions[i]);
 286     const char* pathname = maps->regions[i].pathname;
 287
 288     // Nothing to do
 289     if (maps->regions[i].pathname == NULL) {
 290       current_name = NULL;
 291       continue;
 292     }
 293
 294     // [stack], [vvar], [vsyscall], [vdso] ...
 295     if (pathname[0] == '[') {
 296       if ((reg->prot & PROT_WRITE) && !memcmp(pathname, "[stack]", 7)) {
 297         process->maestro_stack_start = reg->start_addr;
 298         process->maestro_stack_end = reg->end_addr;
 299       }
 300       current_name = NULL;
 301       continue;
 302     }
 303
 304     if (current_name && strcmp(current_name, pathname)==0)
 305       continue;
 306
 307     current_name = pathname;
 308     if (!(reg->prot & PROT_READ) && (reg->prot & PROT_EXEC))
 309       continue;
 310
 311     const bool is_executable = !i;
 312     char* libname = NULL;
 313     if (!is_executable) {
 314       libname = MC_get_lib_name(pathname, &res);
 315       if(!libname)
 316         continue;
 317       if (MC_is_filtered_lib(libname)) {
 318         free(libname);
 319         continue;
 320       }
 321     }
 322
 323     mc_object_info_t info =
 324       MC_find_object_info(process->memory_map, pathname, is_executable);
 325     process->object_infos = (mc_object_info_t*) realloc(process->object_infos,
 326       (process->object_infos_size+1) * sizeof(mc_object_info_t*));
 327     process->object_infos[process->object_infos_size] = info;
 328     process->object_infos_size++;
 329     if (is_executable)
 330       process->binary_info = info;
 331     else if (libname && MC_is_simgrid_lib(libname))
 332       process->libsimgrid_info = info;
 333     free(libname);
 334   }
 335
 336   regfree(&res.so_re);
 337   regfree(&res.version_re);
 338
 339   // Resolve time (including accress differents objects):
 340   for (size_t i=0; i!=process->object_infos_size; ++i)
 341     MC_post_process_object_info(process, process->object_infos[i]);
 342
 343   xbt_assert(process->maestro_stack_start, "Did not find maestro_stack_start");
 344   xbt_assert(process->maestro_stack_end, "Did not find maestro_stack_end");
 345
 346   XBT_DEBUG("Get debug information done !");
 347 }
 348
 349 mc_object_info_t MC_process_find_object_info(mc_process_t process, const void *addr)
 350 {
 351   size_t i;
 352   for (i = 0; i != process->object_infos_size; ++i) {
 353     if (addr >= (void *) process->object_infos[i]->start
 354         && addr <= (void *) process->object_infos[i]->end) {
 355       return process->object_infos[i];
 356     }
 357   }
 358   return NULL;
 359 }
 360
 361 mc_object_info_t MC_process_find_object_info_exec(mc_process_t process, const void *addr)
 362 {
 363   size_t i;
 364   for (i = 0; i != process->object_infos_size; ++i) {
 365     if (addr >= (void *) process->object_infos[i]->start_exec
 366         && addr <= (void *) process->object_infos[i]->end_exec) {
 367       return process->object_infos[i];
 368     }
 369   }
 370   return NULL;
 371 }
 372
 373 mc_object_info_t MC_process_find_object_info_rw(mc_process_t process, const void *addr)
 374 {
 375   size_t i;
 376   for (i = 0; i != process->object_infos_size; ++i) {
 377     if (addr >= (void *) process->object_infos[i]->start_rw
 378         && addr <= (void *) process->object_infos[i]->end_rw) {
 379       return process->object_infos[i];
 380     }
 381   }
 382   return NULL;
 383 }
 384
 385 // Functions, variables…
 386
 387 dw_frame_t MC_process_find_function(mc_process_t process, const void *ip)
 388 {
 389   mc_object_info_t info = MC_process_find_object_info_exec(process, ip);
 390   if (info == NULL)
 391     return NULL;
 392   else
 393     return MC_file_object_info_find_function(info, ip);
 394 }
 395
 396 dw_variable_t MC_process_find_variable_by_name(mc_process_t process, const char* name)
 397 {
 398   const size_t n = process->object_infos_size;
 399   size_t i;
 400
 401   // First lookup the variable in the executable shared object.
 402   // A global variable used directly by the executable code from a library
 403   // is reinstanciated in the executable memory .data/.bss.
 404   // We need to look up the variable in the execvutable first.
 405   if (process->binary_info) {
 406     mc_object_info_t info = process->binary_info;
 407     dw_variable_t var = MC_file_object_info_find_variable_by_name(info, name);
 408     if (var)
 409       return var;
 410   }
 411
 412   for (i=0; i!=n; ++i) {
 413     mc_object_info_t info =process->object_infos[i];
 414     dw_variable_t var = MC_file_object_info_find_variable_by_name(info, name);
 415     if (var)
 416       return var;
 417   }
 418
 419   return NULL;
 420 }
 421
 422 void MC_process_read_variable(mc_process_t process, const char* name, void* target, size_t size)
 423 {
 424   dw_variable_t var = MC_process_find_variable_by_name(process, name);
 425   if (!var->address)
 426     xbt_die("No simple location for this variable");
 427   if (!var->type->full_type)
 428     xbt_die("Partial type for %s, cannot check size", name);
 429   if ((size_t) var->type->full_type->byte_size != size)
 430     xbt_die("Unexpected size for %s (expected %zi, was %zi)",
 431       name, size, (size_t) var->type->full_type->byte_size);
 432   MC_process_read(process, simgrid::mc::AddressSpace::Normal, target, var->address, size,
 433     simgrid::mc::ProcessIndexAny);
 434 }
 435
 436 char* MC_process_read_string(mc_process_t process, void* address)
 437 {
 438   if (!address)
 439     return NULL;
 440   if (process->is_self())
 441     return strdup((char*) address);
 442
 443   off_t len = 128;
 444   char* res = (char*) malloc(len);
 445   off_t off = 0;
 446
 447   while (1) {
 448     ssize_t c = pread(process->memory_file, res + off, len - off, (off_t) address + off);
 449     if (c == -1) {
 450       if (errno == EINTR)
 451         continue;
 452       else
 453         xbt_die("Could not read from from remote process");
 454     }
 455     if (c==0)
 456       xbt_die("Could not read string from remote process");
 457
 458     void* p = memchr(res + off, '\0', c);
 459     if (p)
 460       return res;
 461
 462     off += c;
 463     if (off == len) {
 464       len *= 2;
 465       res = (char*) realloc(res, len);
 466     }
 467   }
 468 }
 469
 470 // ***** Memory access
 471
 472 int MC_process_vm_open(pid_t pid, int flags)
 473 {
 474   const size_t buffer_size = 30;
 475   char buffer[buffer_size];
 476   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long) pid);
 477   if (res < 0 || (size_t) res >= buffer_size) {
 478     errno = ENAMETOOLONG;
 479     return -1;
 480   }
 481   return open(buffer, flags);
 482 }
 483
 484 static void MC_process_open_memory_file(mc_process_t process)
 485 {
 486   if (process->is_self() || process->memory_file >= 0)
 487     return;
 488
 489   int fd = MC_process_vm_open(process->pid, O_RDWR);
 490   if (fd<0)
 491     xbt_die("Could not open file for process virtual address space");
 492   process->memory_file = fd;
 493 }
 494
 495 static ssize_t pread_whole(int fd, void *buf, size_t count, off_t offset)
 496 {
 497   char* buffer = (char*) buf;
 498   ssize_t real_count = count;
 499   while (count) {
 500     ssize_t res = pread(fd, buffer, count, offset);
 501     if (res > 0) {
 502       count  -= res;
 503       buffer += res;
 504       offset += res;
 505     } else if (res==0) {
 506       return -1;
 507     } else if (errno != EINTR) {
 508       return -1;
 509     }
 510   }
 511   return real_count;
 512 }
 513
 514 static ssize_t pwrite_whole(int fd, const void *buf, size_t count, off_t offset)
 515 {
 516   const char* buffer = (const char*) buf;
 517   ssize_t real_count = count;
 518   while (count) {
 519     ssize_t res = pwrite(fd, buffer, count, offset);
 520     if (res > 0) {
 521       count  -= res;
 522       buffer += res;
 523       offset += res;
 524     } else if (res==0) {
 525       return -1;
 526     } else if (errno != EINTR) {
 527       return -1;
 528     }
 529   }
 530   return real_count;
 531 }
 532
 533 }
 534
 535 namespace simgrid {
 536 namespace mc {
 537
 538 const void *Process::read_bytes(void* buffer, std::size_t size,
 539   remote_ptr<void> address, int process_index,
 540   AddressSpace::ReadMode mode)
 541 {
 542   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 543     mc_object_info_t info = MC_process_find_object_info_rw(this, (void*)address.address());
 544     // Segment overlap is not handled.
 545     if (MC_object_info_is_privatized(info)) {
 546       if (process_index < 0)
 547         xbt_die("Missing process index");
 548       // Address translation in the privaization segment:
 549       // TODO, fix me (broken)
 550       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 551       address = remote(address.address() - offset);
 552     }
 553   }
 554
 555   if (this->is_self()) {
 556     if (mode == simgrid::mc::AddressSpace::Lazy)
 557       return (void*)address.address();
 558     else {
 559       memcpy(buffer, (void*)address.address(), size);
 560       return buffer;
 561     }
 562   } else {
 563     if (pread_whole(this->memory_file, buffer, size, (off_t) address.address()) < 0)
 564       xbt_die("Read from process %lli failed", (long long) this->pid);
 565     return buffer;
 566   }
 567 }
 568
 569 }
 570 }
 571
 572 extern "C" {
 573
 574 const void* MC_process_read_dynar_element(mc_process_t process,
 575   void* local, const void* remote_dynar, size_t i, size_t len)
 576 {
 577   s_xbt_dynar_t d;
 578   MC_process_read_simple(process, &d, remote_dynar, sizeof(d));
 579   if (i >= d.used)
 580     xbt_die("Out of bound index %zi/%lu", i, d.used);
 581   if (len != d.elmsize)
 582     xbt_die("Bad size in MC_process_read_dynar_element");
 583   MC_process_read_simple(process, local, xbt_dynar_get_ptr(&d, i), len);
 584   return local;
 585 }
 586
 587 void MC_process_write(mc_process_t process, const void* local, void* remote, size_t len)
 588 {
 589   if (process->is_self()) {
 590     memcpy(remote, local, len);
 591   } else {
 592     if (pwrite_whole(process->memory_file, local, len, (off_t) remote) < 0)
 593       xbt_die("Write to process %lli failed", (long long) process->pid);
 594   }
 595 }
 596
 597 unsigned long MC_process_read_dynar_length(mc_process_t process, const void* remote_dynar)
 598 {
 599   if (!remote_dynar)
 600     return 0;
 601   unsigned long res;
 602   MC_process_read_simple(process, &res,
 603     &((xbt_dynar_t)remote_dynar)->used, sizeof(res));
 604   return res;
 605 }
 606
 607 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 608 static const void* zero_buffer;
 609 static const size_t zero_buffer_size = 10 * 4096;
 610
 611 static void MC_zero_buffer_init(void)
 612 {
 613   int fd = open("/dev/zero", O_RDONLY);
 614   if (fd<0)
 615     xbt_die("Could not open /dev/zero");
 616   zero_buffer = mmap(NULL, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 617   if (zero_buffer == MAP_FAILED)
 618     xbt_die("Could not map the zero buffer");
 619   close(fd);
 620 }
 621
 622 void MC_process_clear_memory(mc_process_t process, void* remote, size_t len)
 623 {
 624   if (process->is_self()) {
 625     memset(remote, 0, len);
 626   } else {
 627     pthread_once(&zero_buffer_flag, MC_zero_buffer_init);
 628     while (len) {
 629       size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 630       MC_process_write(process, zero_buffer, remote, s);
 631       remote = (char*) remote + s;
 632       len -= s;
 633     }
 634   }
 635 }
 636
 637 }