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