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 using simgrid::mc::remote;
  38
  39 extern "C" {
  40
  41 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(mc_process, mc,
  42                                 "MC process information");
  43
  44 // ***** Helper stuff
  45
  46 #define SO_RE "\\.so[\\.0-9]*$"
  47 #define VERSION_RE "-[\\.0-9]*$"
  48
  49 static const char *const FILTERED_LIBS[] = {
  50   "libstdc++",
  51   "libboost_context",
  52   "libc++",
  53   "libm",
  54   "libgcc_s",
  55   "libpthread",
  56   "libunwind",
  57   "libunwind-x86_64",
  58   "libunwind-x86",
  59   "libunwind-ptrace",
  60   "libdw",
  61   "libdl",
  62   "librt",
  63   "liblzma",
  64   "libelf",
  65   "libbz2",
  66   "libz",
  67   "libelf",
  68   "libc",
  69   "ld"
  70 };
  71
  72 static bool MC_is_simgrid_lib(const char* libname)
  73 {
  74   return !strcmp(libname, "libsimgrid");
  75 }
  76
  77 static bool MC_is_filtered_lib(const char* libname)
  78 {
  79   for (const char* filtered_lib : FILTERED_LIBS)
  80     if (strcmp(libname, filtered_lib)==0)
  81       return true;
  82   return false;
  83 }
  84
  85 struct s_mc_memory_map_re {
  86   regex_t so_re;
  87   regex_t version_re;
  88 };
  89
  90 static char* MC_get_lib_name(const char* pathname, struct s_mc_memory_map_re* res)
  91 {
  92   const char* map_basename = basename((char*) pathname);
  93
  94   regmatch_t match;
  95   if(regexec(&res->so_re, map_basename, 1, &match, 0))
  96     return NULL;
  97
  98   char* libname = strndup(map_basename, match.rm_so);
  99
 100   // Strip the version suffix:
 101   if(libname && !regexec(&res->version_re, libname, 1, &match, 0)) {
 102     char* temp = libname;
 103     libname = strndup(temp, match.rm_so);
 104     free(temp);
 105   }
 106
 107   return libname;
 108 }
 109
 110 static ssize_t pread_whole(int fd, void *buf, size_t count, off_t offset)
 111 {
 112   char* buffer = (char*) buf;
 113   ssize_t real_count = count;
 114   while (count) {
 115     ssize_t res = pread(fd, buffer, count, offset);
 116     if (res > 0) {
 117       count  -= res;
 118       buffer += res;
 119       offset += res;
 120     } else if (res==0) {
 121       return -1;
 122     } else if (errno != EINTR) {
 123       return -1;
 124     }
 125   }
 126   return real_count;
 127 }
 128
 129 static ssize_t pwrite_whole(int fd, const void *buf, size_t count, off_t offset)
 130 {
 131   const char* buffer = (const char*) buf;
 132   ssize_t real_count = count;
 133   while (count) {
 134     ssize_t res = pwrite(fd, buffer, count, offset);
 135     if (res > 0) {
 136       count  -= res;
 137       buffer += res;
 138       offset += res;
 139     } else if (res==0) {
 140       return -1;
 141     } else if (errno != EINTR) {
 142       return -1;
 143     }
 144   }
 145   return real_count;
 146 }
 147
 148 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 149 static const void* zero_buffer;
 150 static const size_t zero_buffer_size = 10 * 4096;
 151
 152 static void MC_zero_buffer_init(void)
 153 {
 154   int fd = open("/dev/zero", O_RDONLY);
 155   if (fd<0)
 156     xbt_die("Could not open /dev/zero");
 157   zero_buffer = mmap(NULL, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 158   if (zero_buffer == MAP_FAILED)
 159     xbt_die("Could not map the zero buffer");
 160   close(fd);
 161 }
 162
 163 }
 164
 165 namespace simgrid {
 166 namespace mc {
 167
 168 int open_vm(pid_t pid, int flags)
 169 {
 170   const size_t buffer_size = 30;
 171   char buffer[buffer_size];
 172   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long) pid);
 173   if (res < 0 || (size_t) res >= buffer_size) {
 174     errno = ENAMETOOLONG;
 175     return -1;
 176   }
 177   return open(buffer, flags);
 178 }
 179
 180
 181 }
 182 }
 183
 184 // ***** Process
 185
 186 namespace simgrid {
 187 namespace mc {
 188
 189 Process::Process(pid_t pid, int sockfd)
 190 {
 191   Process* process = this;
 192
 193   process->process_flags = MC_PROCESS_NO_FLAG;
 194   process->socket_ = sockfd;
 195   process->pid_ = pid;
 196   if (pid==getpid())
 197     process->process_flags |= MC_PROCESS_SELF_FLAG;
 198   process->running_ = true;
 199   process->status_ = 0;
 200   process->memory_map_ = get_memory_map(pid);
 201   process->cache_flags = MC_PROCESS_CACHE_FLAG_NONE;
 202   process->heap = NULL;
 203   process->heap_info = NULL;
 204   process->init_memory_map_info();
 205
 206   // Open the memory file
 207   if (process->is_self())
 208     process->memory_file = -1;
 209   else {
 210     int fd = open_vm(process->pid_, O_RDWR);
 211     if (fd<0)
 212       xbt_die("Could not open file for process virtual address space");
 213     process->memory_file = fd;
 214   }
 215
 216   // Read std_heap (is a struct mdesc*):
 217   dw_variable_t std_heap_var = process->find_variable("__mmalloc_default_mdp");
 218   if (!std_heap_var)
 219     xbt_die("No heap information in the target process");
 220   if(!std_heap_var->address)
 221     xbt_die("No constant address for this variable");
 222   process->read_bytes(&process->heap_address, sizeof(struct mdesc*),
 223     remote(std_heap_var->address),
 224     simgrid::mc::ProcessIndexDisabled);
 225
 226   process->smx_process_infos = MC_smx_process_info_list_new();
 227   process->smx_old_process_infos = MC_smx_process_info_list_new();
 228
 229   process->unw_addr_space = unw_create_addr_space(&mc_unw_accessors  , __BYTE_ORDER);
 230   if (process->process_flags & MC_PROCESS_SELF_FLAG) {
 231     process->unw_underlying_addr_space = unw_local_addr_space;
 232     process->unw_underlying_context = NULL;
 233   } else {
 234     process->unw_underlying_addr_space = unw_create_addr_space(&mc_unw_vmread_accessors, __BYTE_ORDER);
 235     process->unw_underlying_context = _UPT_create(pid);
 236   }
 237 }
 238
 239 Process::~Process()
 240 {
 241   Process* process = this;
 242
 243   if (this->socket_ >= 0 && close(this->socket_) < 0)
 244     xbt_die("Could not close communication socket");
 245
 246   process->process_flags = MC_PROCESS_NO_FLAG;
 247   process->pid_ = 0;
 248
 249   process->maestro_stack_start_ = nullptr;
 250   process->maestro_stack_end_ = nullptr;
 251
 252   xbt_dynar_free(&process->smx_process_infos);
 253   xbt_dynar_free(&process->smx_old_process_infos);
 254
 255   size_t i;
 256   for (i=0; i!=process->object_infos_size; ++i) {
 257     delete process->object_infos[i];
 258     process->object_infos[i] = nullptr;
 259   }
 260   free(process->object_infos);
 261   process->object_infos = NULL;
 262   process->object_infos_size = 0;
 263   if (process->memory_file >= 0) {
 264     close(process->memory_file);
 265   }
 266
 267   if (process->unw_underlying_addr_space != unw_local_addr_space) {
 268     unw_destroy_addr_space(process->unw_underlying_addr_space);
 269     _UPT_destroy(process->unw_underlying_context);
 270   }
 271   process->unw_underlying_context = NULL;
 272   process->unw_underlying_addr_space = NULL;
 273
 274   unw_destroy_addr_space(process->unw_addr_space);
 275   process->unw_addr_space = NULL;
 276
 277   process->cache_flags = MC_PROCESS_CACHE_FLAG_NONE;
 278
 279   free(process->heap);
 280   process->heap = NULL;
 281
 282   free(process->heap_info);
 283   process->heap_info = NULL;
 284 }
 285
 286 /** Refresh the information about the process
 287  *
 288  *  Do not use direclty, this is used by the getters when appropriate
 289  *  in order to have fresh data.
 290  */
 291 void Process::refresh_heap()
 292 {
 293   xbt_assert(mc_mode == MC_MODE_SERVER);
 294   xbt_assert(!this->is_self());
 295   // Read/dereference/refresh the std_heap pointer:
 296   if (!this->heap) {
 297     this->heap = (struct mdesc*) malloc(sizeof(struct mdesc));
 298   }
 299   this->read_bytes(this->heap, sizeof(struct mdesc), remote(this->heap_address),
 300     simgrid::mc::ProcessIndexDisabled);
 301   this->cache_flags |= MC_PROCESS_CACHE_FLAG_HEAP;
 302 }
 303
 304 /** Refresh the information about the process
 305  *
 306  *  Do not use direclty, this is used by the getters when appropriate
 307  *  in order to have fresh data.
 308  * */
 309 void Process::refresh_malloc_info()
 310 {
 311   xbt_assert(mc_mode == MC_MODE_SERVER);
 312   xbt_assert(!this->is_self());
 313   if (!(this->cache_flags & MC_PROCESS_CACHE_FLAG_HEAP))
 314     this->refresh_heap();
 315   // Refresh process->heapinfo:
 316   size_t malloc_info_bytesize =
 317     (this->heap->heaplimit + 1) * sizeof(malloc_info);
 318   this->heap_info = (malloc_info*) realloc(this->heap_info, malloc_info_bytesize);
 319   this->read_bytes(this->heap_info, malloc_info_bytesize,
 320     remote(this->heap->heapinfo), simgrid::mc::ProcessIndexDisabled);
 321   this->cache_flags |= MC_PROCESS_CACHE_FLAG_MALLOC_INFO;
 322 }
 323
 324 /** @brief Finds the range of the different memory segments and binary paths */
 325 void Process::init_memory_map_info()
 326 {
 327   XBT_DEBUG("Get debug information ...");
 328   this->maestro_stack_start_ = nullptr;
 329   this->maestro_stack_end_ = nullptr;
 330   this->object_infos = NULL;
 331   this->object_infos_size = 0;
 332   this->binary_info = NULL;
 333   this->libsimgrid_info = NULL;
 334
 335   struct s_mc_memory_map_re res;
 336
 337   if(regcomp(&res.so_re, SO_RE, 0) || regcomp(&res.version_re, VERSION_RE, 0))
 338     xbt_die(".so regexp did not compile");
 339
 340   std::vector<simgrid::mc::VmMap> const& maps = this->memory_map_;
 341
 342   const char* current_name = NULL;
 343
 344   for (size_t i=0; i < maps.size(); i++) {
 345     simgrid::mc::VmMap const& reg = maps[i];
 346     const char* pathname = maps[i].pathname.c_str();
 347
 348     // Nothing to do
 349     if (maps[i].pathname.empty()) {
 350       current_name = NULL;
 351       continue;
 352     }
 353
 354     // [stack], [vvar], [vsyscall], [vdso] ...
 355     if (pathname[0] == '[') {
 356       if ((reg.prot & PROT_WRITE) && !memcmp(pathname, "[stack]", 7)) {
 357         this->maestro_stack_start_ = remote(reg.start_addr);
 358         this->maestro_stack_end_ = remote(reg.end_addr);
 359       }
 360       current_name = NULL;
 361       continue;
 362     }
 363
 364     if (current_name && strcmp(current_name, pathname)==0)
 365       continue;
 366
 367     current_name = pathname;
 368     if (!(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 369       continue;
 370
 371     const bool is_executable = !i;
 372     char* libname = NULL;
 373     if (!is_executable) {
 374       libname = MC_get_lib_name(pathname, &res);
 375       if(!libname)
 376         continue;
 377       if (MC_is_filtered_lib(libname)) {
 378         free(libname);
 379         continue;
 380       }
 381     }
 382
 383     mc_object_info_t info =
 384       MC_find_object_info(this->memory_map_, pathname, is_executable);
 385     this->object_infos = (mc_object_info_t*) realloc(this->object_infos,
 386       (this->object_infos_size+1) * sizeof(mc_object_info_t));
 387     this->object_infos[this->object_infos_size] = info;
 388     this->object_infos_size++;
 389     if (is_executable)
 390       this->binary_info = info;
 391     else if (libname && MC_is_simgrid_lib(libname))
 392       this->libsimgrid_info = info;
 393     free(libname);
 394   }
 395
 396   regfree(&res.so_re);
 397   regfree(&res.version_re);
 398
 399   // Resolve time (including accross differents objects):
 400   for (size_t i=0; i!=this->object_infos_size; ++i)
 401     MC_post_process_object_info(this, this->object_infos[i]);
 402
 403   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 404   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 405
 406   XBT_DEBUG("Get debug information done !");
 407 }
 408
 409 mc_object_info_t Process::find_object_info(remote_ptr<void> addr) const
 410 {
 411   size_t i;
 412   for (i = 0; i != this->object_infos_size; ++i) {
 413     if (addr.address() >= (std::uint64_t)this->object_infos[i]->start
 414         && addr.address() <= (std::uint64_t)this->object_infos[i]->end) {
 415       return this->object_infos[i];
 416     }
 417   }
 418   return NULL;
 419 }
 420
 421 mc_object_info_t Process::find_object_info_exec(remote_ptr<void> addr) const
 422 {
 423   size_t i;
 424   for (i = 0; i != this->object_infos_size; ++i) {
 425     if (addr.address() >= (std::uint64_t)this->object_infos[i]->start_exec
 426         && addr.address() <= (std::uint64_t)this->object_infos[i]->end_exec) {
 427       return this->object_infos[i];
 428     }
 429   }
 430   return NULL;
 431 }
 432
 433 mc_object_info_t Process::find_object_info_rw(remote_ptr<void> addr) const
 434 {
 435   size_t i;
 436   for (i = 0; i != this->object_infos_size; ++i) {
 437     if (addr.address() >= (std::uint64_t)this->object_infos[i]->start_rw
 438         && addr.address() <= (std::uint64_t)this->object_infos[i]->end_rw) {
 439       return this->object_infos[i];
 440     }
 441   }
 442   return NULL;
 443 }
 444
 445 dw_frame_t Process::find_function(remote_ptr<void> ip) const
 446 {
 447   mc_object_info_t info = this->find_object_info_exec(ip);
 448   if (info == NULL)
 449     return NULL;
 450   else
 451     return MC_file_object_info_find_function(info, (void*) ip.address());
 452 }
 453
 454 /** Find (one occurence of) the named variable definition
 455  */
 456 dw_variable_t Process::find_variable(const char* name) const
 457 {
 458   const size_t n = this->object_infos_size;
 459   size_t i;
 460
 461   // First lookup the variable in the executable shared object.
 462   // A global variable used directly by the executable code from a library
 463   // is reinstanciated in the executable memory .data/.bss.
 464   // We need to look up the variable in the execvutable first.
 465   if (this->binary_info) {
 466     mc_object_info_t info = this->binary_info;
 467     dw_variable_t var = MC_file_object_info_find_variable_by_name(info, name);
 468     if (var)
 469       return var;
 470   }
 471
 472   for (i=0; i!=n; ++i) {
 473     mc_object_info_t info = this->object_infos[i];
 474     dw_variable_t var = MC_file_object_info_find_variable_by_name(info, name);
 475     if (var)
 476       return var;
 477   }
 478
 479   return NULL;
 480 }
 481
 482 void Process::read_variable(const char* name, void* target, size_t size) const
 483 {
 484   dw_variable_t var = this->find_variable(name);
 485   if (!var->address)
 486     xbt_die("No simple location for this variable");
 487   if (!var->type->full_type)
 488     xbt_die("Partial type for %s, cannot check size", name);
 489   if ((size_t) var->type->full_type->byte_size != size)
 490     xbt_die("Unexpected size for %s (expected %zi, was %zi)",
 491       name, size, (size_t) var->type->full_type->byte_size);
 492   this->read_bytes(target, size, remote(var->address));
 493 }
 494
 495 char* Process::read_string(remote_ptr<void> address) const
 496 {
 497   if (!address)
 498     return NULL;
 499   if (this->is_self())
 500     return strdup((char*) address.address());
 501
 502   off_t len = 128;
 503   char* res = (char*) malloc(len);
 504   off_t off = 0;
 505
 506   while (1) {
 507     ssize_t c = pread(this->memory_file, res + off, len - off, (off_t) address.address() + off);
 508     if (c == -1) {
 509       if (errno == EINTR)
 510         continue;
 511       else
 512         xbt_die("Could not read from from remote process");
 513     }
 514     if (c==0)
 515       xbt_die("Could not read string from remote process");
 516
 517     void* p = memchr(res + off, '\0', c);
 518     if (p)
 519       return res;
 520
 521     off += c;
 522     if (off == len) {
 523       len *= 2;
 524       res = (char*) realloc(res, len);
 525     }
 526   }
 527 }
 528
 529 const void *Process::read_bytes(void* buffer, std::size_t size,
 530   remote_ptr<void> address, int process_index,
 531   AddressSpace::ReadMode mode) const
 532 {
 533   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 534     mc_object_info_t info = this->find_object_info_rw((void*)address.address());
 535     // Segment overlap is not handled.
 536     if (MC_object_info_is_privatized(info)) {
 537       if (process_index < 0)
 538         xbt_die("Missing process index");
 539       if (process_index >= (int) MC_smpi_process_count())
 540         xbt_die("Invalid process index");
 541
 542       // Read smpi_privatisation_regions from MCed:
 543       smpi_privatisation_region_t remote_smpi_privatisation_regions =
 544         mc_model_checker->process().read_variable<smpi_privatisation_region_t>(
 545           "smpi_privatisation_regions");
 546
 547       s_smpi_privatisation_region_t privatisation_region =
 548         mc_model_checker->process().read<s_smpi_privatisation_region_t>(
 549           remote(remote_smpi_privatisation_regions + process_index));
 550
 551       // Address translation in the privaization segment:
 552       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 553       address = remote((char*)privatisation_region.address + offset);
 554     }
 555   }
 556
 557   if (this->is_self()) {
 558     if (mode == simgrid::mc::AddressSpace::Lazy)
 559       return (void*)address.address();
 560     else {
 561       memcpy(buffer, (void*)address.address(), size);
 562       return buffer;
 563     }
 564   } else {
 565     if (pread_whole(this->memory_file, buffer, size, (off_t) address.address()) < 0)
 566       xbt_die("Read from process %lli failed", (long long) this->pid_);
 567     return buffer;
 568   }
 569 }
 570
 571 /** Write data to a process memory
 572  *
 573  *  @param process the process
 574  *  @param local   local memory address (source)
 575  *  @param remote  target process memory address (target)
 576  *  @param len     data size
 577  */
 578 void Process::write_bytes(const void* buffer, size_t len, remote_ptr<void> address)
 579 {
 580   if (this->is_self()) {
 581     memcpy((void*)address.address(), buffer, len);
 582   } else {
 583     if (pwrite_whole(this->memory_file, buffer, len, address.address()) < 0)
 584       xbt_die("Write to process %lli failed", (long long) this->pid_);
 585   }
 586 }
 587
 588 void Process::clear_bytes(remote_ptr<void> address, size_t len)
 589 {
 590   if (this->is_self()) {
 591     memset((void*)address.address(), 0, len);
 592   } else {
 593     pthread_once(&zero_buffer_flag, MC_zero_buffer_init);
 594     while (len) {
 595       size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 596       this->write_bytes(zero_buffer, s, address);
 597       address = remote((char*) address.address() + s);
 598       len -= s;
 599     }
 600   }
 601 }
 602
 603 void Process::ignore_region(std::uint64_t addr, std::size_t size)
 604 {
 605   IgnoredRegion region;
 606   region.addr = addr;
 607   region.size = size;
 608
 609   if (ignored_regions_.empty()) {
 610     ignored_regions_.push_back(region);
 611     return;
 612   }
 613
 614   unsigned int cursor = 0;
 615   IgnoredRegion* current_region = nullptr;
 616
 617   int start = 0;
 618   int end = ignored_regions_.size() - 1;
 619   while (start <= end) {
 620     cursor = (start + end) / 2;
 621     current_region = &ignored_regions_[cursor];
 622     if (current_region->addr == addr) {
 623       if (current_region->size == size)
 624         return;
 625       else if (current_region->size < size)
 626         start = cursor + 1;
 627       else
 628         end = cursor - 1;
 629     } else if (current_region->addr < addr)
 630       start = cursor + 1;
 631     else
 632       end = cursor - 1;
 633   }
 634
 635   std::size_t position;
 636   if (current_region->addr == addr) {
 637     if (current_region->size < size) {
 638       position = cursor + 1;
 639     } else {
 640       position = cursor;
 641     }
 642   } else if (current_region->addr < addr) {
 643     position = cursor + 1;
 644   } else {
 645     position = cursor;
 646   }
 647   ignored_regions_.insert(
 648     ignored_regions_.begin() + position, region);
 649 }
 650
 651 }
 652 }