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     MC_free_object_info(&process->object_infos[i]);
 258   }
 259   free(process->object_infos);
 260   process->object_infos = NULL;
 261   process->object_infos_size = 0;
 262   if (process->memory_file >= 0) {
 263     close(process->memory_file);
 264   }
 265
 266   if (process->unw_underlying_addr_space != unw_local_addr_space) {
 267     unw_destroy_addr_space(process->unw_underlying_addr_space);
 268     _UPT_destroy(process->unw_underlying_context);
 269   }
 270   process->unw_underlying_context = NULL;
 271   process->unw_underlying_addr_space = NULL;
 272
 273   unw_destroy_addr_space(process->unw_addr_space);
 274   process->unw_addr_space = NULL;
 275
 276   process->cache_flags = MC_PROCESS_CACHE_FLAG_NONE;
 277
 278   free(process->heap);
 279   process->heap = NULL;
 280
 281   free(process->heap_info);
 282   process->heap_info = NULL;
 283 }
 284
 285 /** Refresh the information about the process
 286  *
 287  *  Do not use direclty, this is used by the getters when appropriate
 288  *  in order to have fresh data.
 289  */
 290 void Process::refresh_heap()
 291 {
 292   xbt_assert(mc_mode == MC_MODE_SERVER);
 293   xbt_assert(!this->is_self());
 294   // Read/dereference/refresh the std_heap pointer:
 295   if (!this->heap) {
 296     this->heap = (struct mdesc*) malloc(sizeof(struct mdesc));
 297   }
 298   this->read_bytes(this->heap, sizeof(struct mdesc), remote(this->heap_address),
 299     simgrid::mc::ProcessIndexDisabled);
 300   this->cache_flags |= MC_PROCESS_CACHE_FLAG_HEAP;
 301 }
 302
 303 /** Refresh the information about the process
 304  *
 305  *  Do not use direclty, this is used by the getters when appropriate
 306  *  in order to have fresh data.
 307  * */
 308 void Process::refresh_malloc_info()
 309 {
 310   xbt_assert(mc_mode == MC_MODE_SERVER);
 311   xbt_assert(!this->is_self());
 312   if (!(this->cache_flags & MC_PROCESS_CACHE_FLAG_HEAP))
 313     this->refresh_heap();
 314   // Refresh process->heapinfo:
 315   size_t malloc_info_bytesize =
 316     (this->heap->heaplimit + 1) * sizeof(malloc_info);
 317   this->heap_info = (malloc_info*) realloc(this->heap_info, malloc_info_bytesize);
 318   this->read_bytes(this->heap_info, malloc_info_bytesize,
 319     remote(this->heap->heapinfo), simgrid::mc::ProcessIndexDisabled);
 320   this->cache_flags |= MC_PROCESS_CACHE_FLAG_MALLOC_INFO;
 321 }
 322
 323 /** @brief Finds the range of the different memory segments and binary paths */
 324 void Process::init_memory_map_info()
 325 {
 326   XBT_DEBUG("Get debug information ...");
 327   this->maestro_stack_start_ = nullptr;
 328   this->maestro_stack_end_ = nullptr;
 329   this->object_infos = NULL;
 330   this->object_infos_size = 0;
 331   this->binary_info = NULL;
 332   this->libsimgrid_info = NULL;
 333
 334   struct s_mc_memory_map_re res;
 335
 336   if(regcomp(&res.so_re, SO_RE, 0) || regcomp(&res.version_re, VERSION_RE, 0))
 337     xbt_die(".so regexp did not compile");
 338
 339   std::vector<simgrid::mc::VmMap> const& maps = this->memory_map_;
 340
 341   const char* current_name = NULL;
 342
 343   for (size_t i=0; i < maps.size(); i++) {
 344     simgrid::mc::VmMap const& reg = maps[i];
 345     const char* pathname = maps[i].pathname.c_str();
 346
 347     // Nothing to do
 348     if (maps[i].pathname.empty()) {
 349       current_name = NULL;
 350       continue;
 351     }
 352
 353     // [stack], [vvar], [vsyscall], [vdso] ...
 354     if (pathname[0] == '[') {
 355       if ((reg.prot & PROT_WRITE) && !memcmp(pathname, "[stack]", 7)) {
 356         this->maestro_stack_start_ = remote(reg.start_addr);
 357         this->maestro_stack_end_ = remote(reg.end_addr);
 358       }
 359       current_name = NULL;
 360       continue;
 361     }
 362
 363     if (current_name && strcmp(current_name, pathname)==0)
 364       continue;
 365
 366     current_name = pathname;
 367     if (!(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 368       continue;
 369
 370     const bool is_executable = !i;
 371     char* libname = NULL;
 372     if (!is_executable) {
 373       libname = MC_get_lib_name(pathname, &res);
 374       if(!libname)
 375         continue;
 376       if (MC_is_filtered_lib(libname)) {
 377         free(libname);
 378         continue;
 379       }
 380     }
 381
 382     mc_object_info_t info =
 383       MC_find_object_info(this->memory_map_, pathname, is_executable);
 384     this->object_infos = (mc_object_info_t*) realloc(this->object_infos,
 385       (this->object_infos_size+1) * sizeof(mc_object_info_t));
 386     this->object_infos[this->object_infos_size] = info;
 387     this->object_infos_size++;
 388     if (is_executable)
 389       this->binary_info = info;
 390     else if (libname && MC_is_simgrid_lib(libname))
 391       this->libsimgrid_info = info;
 392     free(libname);
 393   }
 394
 395   regfree(&res.so_re);
 396   regfree(&res.version_re);
 397
 398   // Resolve time (including accross differents objects):
 399   for (size_t i=0; i!=this->object_infos_size; ++i)
 400     MC_post_process_object_info(this, this->object_infos[i]);
 401
 402   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 403   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 404
 405   XBT_DEBUG("Get debug information done !");
 406 }
 407
 408 mc_object_info_t Process::find_object_info(remote_ptr<void> addr) const
 409 {
 410   size_t i;
 411   for (i = 0; i != this->object_infos_size; ++i) {
 412     if (addr.address() >= (std::uint64_t)this->object_infos[i]->start
 413         && addr.address() <= (std::uint64_t)this->object_infos[i]->end) {
 414       return this->object_infos[i];
 415     }
 416   }
 417   return NULL;
 418 }
 419
 420 mc_object_info_t Process::find_object_info_exec(remote_ptr<void> addr) const
 421 {
 422   size_t i;
 423   for (i = 0; i != this->object_infos_size; ++i) {
 424     if (addr.address() >= (std::uint64_t)this->object_infos[i]->start_exec
 425         && addr.address() <= (std::uint64_t)this->object_infos[i]->end_exec) {
 426       return this->object_infos[i];
 427     }
 428   }
 429   return NULL;
 430 }
 431
 432 mc_object_info_t Process::find_object_info_rw(remote_ptr<void> addr) const
 433 {
 434   size_t i;
 435   for (i = 0; i != this->object_infos_size; ++i) {
 436     if (addr.address() >= (std::uint64_t)this->object_infos[i]->start_rw
 437         && addr.address() <= (std::uint64_t)this->object_infos[i]->end_rw) {
 438       return this->object_infos[i];
 439     }
 440   }
 441   return NULL;
 442 }
 443
 444 dw_frame_t Process::find_function(remote_ptr<void> ip) const
 445 {
 446   mc_object_info_t info = this->find_object_info_exec(ip);
 447   if (info == NULL)
 448     return NULL;
 449   else
 450     return MC_file_object_info_find_function(info, (void*) ip.address());
 451 }
 452
 453 /** Find (one occurence of) the named variable definition
 454  */
 455 dw_variable_t Process::find_variable(const char* name) const
 456 {
 457   const size_t n = this->object_infos_size;
 458   size_t i;
 459
 460   // First lookup the variable in the executable shared object.
 461   // A global variable used directly by the executable code from a library
 462   // is reinstanciated in the executable memory .data/.bss.
 463   // We need to look up the variable in the execvutable first.
 464   if (this->binary_info) {
 465     mc_object_info_t info = this->binary_info;
 466     dw_variable_t var = MC_file_object_info_find_variable_by_name(info, name);
 467     if (var)
 468       return var;
 469   }
 470
 471   for (i=0; i!=n; ++i) {
 472     mc_object_info_t info = this->object_infos[i];
 473     dw_variable_t var = MC_file_object_info_find_variable_by_name(info, name);
 474     if (var)
 475       return var;
 476   }
 477
 478   return NULL;
 479 }
 480
 481 void Process::read_variable(const char* name, void* target, size_t size) const
 482 {
 483   dw_variable_t var = this->find_variable(name);
 484   if (!var->address)
 485     xbt_die("No simple location for this variable");
 486   if (!var->type->full_type)
 487     xbt_die("Partial type for %s, cannot check size", name);
 488   if ((size_t) var->type->full_type->byte_size != size)
 489     xbt_die("Unexpected size for %s (expected %zi, was %zi)",
 490       name, size, (size_t) var->type->full_type->byte_size);
 491   this->read_bytes(target, size, remote(var->address));
 492 }
 493
 494 char* Process::read_string(remote_ptr<void> address) const
 495 {
 496   if (!address)
 497     return NULL;
 498   if (this->is_self())
 499     return strdup((char*) address.address());
 500
 501   off_t len = 128;
 502   char* res = (char*) malloc(len);
 503   off_t off = 0;
 504
 505   while (1) {
 506     ssize_t c = pread(this->memory_file, res + off, len - off, (off_t) address.address() + off);
 507     if (c == -1) {
 508       if (errno == EINTR)
 509         continue;
 510       else
 511         xbt_die("Could not read from from remote process");
 512     }
 513     if (c==0)
 514       xbt_die("Could not read string from remote process");
 515
 516     void* p = memchr(res + off, '\0', c);
 517     if (p)
 518       return res;
 519
 520     off += c;
 521     if (off == len) {
 522       len *= 2;
 523       res = (char*) realloc(res, len);
 524     }
 525   }
 526 }
 527
 528 const void *Process::read_bytes(void* buffer, std::size_t size,
 529   remote_ptr<void> address, int process_index,
 530   AddressSpace::ReadMode mode) const
 531 {
 532   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 533     mc_object_info_t info = this->find_object_info_rw((void*)address.address());
 534     // Segment overlap is not handled.
 535     if (MC_object_info_is_privatized(info)) {
 536       if (process_index < 0)
 537         xbt_die("Missing process index");
 538       if (process_index >= (int) MC_smpi_process_count())
 539         xbt_die("Invalid process index");
 540
 541       // Read smpi_privatisation_regions from MCed:
 542       smpi_privatisation_region_t remote_smpi_privatisation_regions =
 543         mc_model_checker->process().read_variable<smpi_privatisation_region_t>(
 544           "smpi_privatisation_regions");
 545
 546       s_smpi_privatisation_region_t privatisation_region =
 547         mc_model_checker->process().read<s_smpi_privatisation_region_t>(
 548           remote(remote_smpi_privatisation_regions + process_index));
 549
 550       // Address translation in the privaization segment:
 551       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 552       address = remote((char*)privatisation_region.address + offset);
 553     }
 554   }
 555
 556   if (this->is_self()) {
 557     if (mode == simgrid::mc::AddressSpace::Lazy)
 558       return (void*)address.address();
 559     else {
 560       memcpy(buffer, (void*)address.address(), size);
 561       return buffer;
 562     }
 563   } else {
 564     if (pread_whole(this->memory_file, buffer, size, (off_t) address.address()) < 0)
 565       xbt_die("Read from process %lli failed", (long long) this->pid_);
 566     return buffer;
 567   }
 568 }
 569
 570 /** Write data to a process memory
 571  *
 572  *  @param process the process
 573  *  @param local   local memory address (source)
 574  *  @param remote  target process memory address (target)
 575  *  @param len     data size
 576  */
 577 void Process::write_bytes(const void* buffer, size_t len, remote_ptr<void> address)
 578 {
 579   if (this->is_self()) {
 580     memcpy((void*)address.address(), buffer, len);
 581   } else {
 582     if (pwrite_whole(this->memory_file, buffer, len, address.address()) < 0)
 583       xbt_die("Write to process %lli failed", (long long) this->pid_);
 584   }
 585 }
 586
 587 void Process::clear_bytes(remote_ptr<void> address, size_t len)
 588 {
 589   if (this->is_self()) {
 590     memset((void*)address.address(), 0, len);
 591   } else {
 592     pthread_once(&zero_buffer_flag, MC_zero_buffer_init);
 593     while (len) {
 594       size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 595       this->write_bytes(zero_buffer, s, address);
 596       address = remote((char*) address.address() + s);
 597       len -= s;
 598     }
 599   }
 600 }
 601
 602 void Process::ignore_region(std::uint64_t addr, std::size_t size)
 603 {
 604   IgnoredRegion region;
 605   region.addr = addr;
 606   region.size = size;
 607
 608   if (ignored_regions_.empty()) {
 609     ignored_regions_.push_back(region);
 610     return;
 611   }
 612
 613   unsigned int cursor = 0;
 614   IgnoredRegion* current_region = nullptr;
 615
 616   int start = 0;
 617   int end = ignored_regions_.size() - 1;
 618   while (start <= end) {
 619     cursor = (start + end) / 2;
 620     current_region = &ignored_regions_[cursor];
 621     if (current_region->addr == addr) {
 622       if (current_region->size == size)
 623         return;
 624       else if (current_region->size < size)
 625         start = cursor + 1;
 626       else
 627         end = cursor - 1;
 628     } else if (current_region->addr < addr)
 629       start = cursor + 1;
 630     else
 631       end = cursor - 1;
 632   }
 633
 634   std::size_t position;
 635   if (current_region->addr == addr) {
 636     if (current_region->size < size) {
 637       position = cursor + 1;
 638     } else {
 639       position = cursor;
 640     }
 641   } else if (current_region->addr < addr) {
 642     position = cursor + 1;
 643   } else {
 644     position = cursor;
 645   }
 646   ignored_regions_.insert(
 647     ignored_regions_.begin() + position, region);
 648 }
 649
 650 }
 651 }