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   "ld",
  51   "libbz2",
  52   "libboost_context",
  53   "libc",
  54   "libc++",
  55   "libcdt",
  56   "libcgraph",
  57   "libdl",
  58   "libdw",
  59   "libelf",
  60   "libgcc_s",
  61   "liblua5.1",
  62   "liblzma",
  63   "libm",
  64   "libpthread",
  65   "librt",
  66   "libstdc++",
  67   "libunwind",
  68   "libunwind-x86_64",
  69   "libunwind-x86",
  70   "libunwind-ptrace",
  71   "libz"
  72 };
  73
  74 static bool MC_is_simgrid_lib(const char* libname)
  75 {
  76   return !strcmp(libname, "libsimgrid");
  77 }
  78
  79 static bool MC_is_filtered_lib(const char* libname)
  80 {
  81   for (const char* filtered_lib : FILTERED_LIBS)
  82     if (strcmp(libname, filtered_lib)==0)
  83       return true;
  84   return false;
  85 }
  86
  87 struct s_mc_memory_map_re {
  88   regex_t so_re;
  89   regex_t version_re;
  90 };
  91
  92 static char* MC_get_lib_name(const char* pathname, struct s_mc_memory_map_re* res)
  93 {
  94   const char* map_basename = basename((char*) pathname);
  95
  96   regmatch_t match;
  97   if(regexec(&res->so_re, map_basename, 1, &match, 0))
  98     return NULL;
  99
 100   char* libname = strndup(map_basename, match.rm_so);
 101
 102   // Strip the version suffix:
 103   if(libname && !regexec(&res->version_re, libname, 1, &match, 0)) {
 104     char* temp = libname;
 105     libname = strndup(temp, match.rm_so);
 106     free(temp);
 107   }
 108
 109   return libname;
 110 }
 111
 112 static ssize_t pread_whole(int fd, void *buf, size_t count, off_t offset)
 113 {
 114   char* buffer = (char*) buf;
 115   ssize_t real_count = count;
 116   while (count) {
 117     ssize_t res = pread(fd, buffer, count, offset);
 118     if (res > 0) {
 119       count  -= res;
 120       buffer += res;
 121       offset += res;
 122     } else if (res==0) {
 123       return -1;
 124     } else if (errno != EINTR) {
 125       return -1;
 126     }
 127   }
 128   return real_count;
 129 }
 130
 131 static ssize_t pwrite_whole(int fd, const void *buf, size_t count, off_t offset)
 132 {
 133   const char* buffer = (const char*) buf;
 134   ssize_t real_count = count;
 135   while (count) {
 136     ssize_t res = pwrite(fd, buffer, count, offset);
 137     if (res > 0) {
 138       count  -= res;
 139       buffer += res;
 140       offset += res;
 141     } else if (res==0) {
 142       return -1;
 143     } else if (errno != EINTR) {
 144       return -1;
 145     }
 146   }
 147   return real_count;
 148 }
 149
 150 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 151 static const void* zero_buffer;
 152 static const size_t zero_buffer_size = 10 * 4096;
 153
 154 static void MC_zero_buffer_init(void)
 155 {
 156   int fd = open("/dev/zero", O_RDONLY);
 157   if (fd<0)
 158     xbt_die("Could not open /dev/zero");
 159   zero_buffer = mmap(NULL, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 160   if (zero_buffer == MAP_FAILED)
 161     xbt_die("Could not map the zero buffer");
 162   close(fd);
 163 }
 164
 165 }
 166
 167 namespace simgrid {
 168 namespace mc {
 169
 170 int open_vm(pid_t pid, int flags)
 171 {
 172   const size_t buffer_size = 30;
 173   char buffer[buffer_size];
 174   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long) pid);
 175   if (res < 0 || (size_t) res >= buffer_size) {
 176     errno = ENAMETOOLONG;
 177     return -1;
 178   }
 179   return open(buffer, flags);
 180 }
 181
 182
 183 }
 184 }
 185
 186 // ***** Process
 187
 188 namespace simgrid {
 189 namespace mc {
 190
 191 Process::Process(pid_t pid, int sockfd)
 192 {
 193   Process* process = this;
 194
 195   process->process_flags = MC_PROCESS_NO_FLAG;
 196   process->socket_ = sockfd;
 197   process->pid_ = pid;
 198   if (pid==getpid())
 199     process->process_flags |= MC_PROCESS_SELF_FLAG;
 200   process->running_ = true;
 201   process->status_ = 0;
 202   process->memory_map_ = get_memory_map(pid);
 203   process->cache_flags = MC_PROCESS_CACHE_FLAG_NONE;
 204   process->heap = NULL;
 205   process->heap_info = NULL;
 206   process->init_memory_map_info();
 207
 208   // Open the memory file
 209   if (process->is_self())
 210     process->memory_file = -1;
 211   else {
 212     int fd = open_vm(process->pid_, O_RDWR);
 213     if (fd<0)
 214       xbt_die("Could not open file for process virtual address space");
 215     process->memory_file = fd;
 216   }
 217
 218   // Read std_heap (is a struct mdesc*):
 219   dw_variable_t std_heap_var = process->find_variable("__mmalloc_default_mdp");
 220   if (!std_heap_var)
 221     xbt_die("No heap information in the target process");
 222   if(!std_heap_var->address)
 223     xbt_die("No constant address for this variable");
 224   process->read_bytes(&process->heap_address, sizeof(struct mdesc*),
 225     remote(std_heap_var->address),
 226     simgrid::mc::ProcessIndexDisabled);
 227
 228   process->smx_process_infos = MC_smx_process_info_list_new();
 229   process->smx_old_process_infos = MC_smx_process_info_list_new();
 230
 231   process->unw_addr_space = unw_create_addr_space(&mc_unw_accessors  , __BYTE_ORDER);
 232   if (process->process_flags & MC_PROCESS_SELF_FLAG) {
 233     process->unw_underlying_addr_space = unw_local_addr_space;
 234     process->unw_underlying_context = NULL;
 235   } else {
 236     process->unw_underlying_addr_space = unw_create_addr_space(&mc_unw_vmread_accessors, __BYTE_ORDER);
 237     process->unw_underlying_context = _UPT_create(pid);
 238   }
 239 }
 240
 241 Process::~Process()
 242 {
 243   Process* process = this;
 244
 245   if (this->socket_ >= 0 && close(this->socket_) < 0)
 246     xbt_die("Could not close communication socket");
 247
 248   process->process_flags = MC_PROCESS_NO_FLAG;
 249   process->pid_ = 0;
 250
 251   process->maestro_stack_start_ = nullptr;
 252   process->maestro_stack_end_ = nullptr;
 253
 254   xbt_dynar_free(&process->smx_process_infos);
 255   xbt_dynar_free(&process->smx_old_process_infos);
 256
 257   if (process->memory_file >= 0) {
 258     close(process->memory_file);
 259   }
 260
 261   if (process->unw_underlying_addr_space != unw_local_addr_space) {
 262     unw_destroy_addr_space(process->unw_underlying_addr_space);
 263     _UPT_destroy(process->unw_underlying_context);
 264   }
 265   process->unw_underlying_context = NULL;
 266   process->unw_underlying_addr_space = NULL;
 267
 268   unw_destroy_addr_space(process->unw_addr_space);
 269   process->unw_addr_space = NULL;
 270
 271   process->cache_flags = MC_PROCESS_CACHE_FLAG_NONE;
 272
 273   free(process->heap);
 274   process->heap = NULL;
 275
 276   free(process->heap_info);
 277   process->heap_info = NULL;
 278 }
 279
 280 /** Refresh the information about the process
 281  *
 282  *  Do not use direclty, this is used by the getters when appropriate
 283  *  in order to have fresh data.
 284  */
 285 void Process::refresh_heap()
 286 {
 287   xbt_assert(mc_mode == MC_MODE_SERVER);
 288   xbt_assert(!this->is_self());
 289   // Read/dereference/refresh the std_heap pointer:
 290   if (!this->heap) {
 291     this->heap = (struct mdesc*) malloc(sizeof(struct mdesc));
 292   }
 293   this->read_bytes(this->heap, sizeof(struct mdesc), remote(this->heap_address),
 294     simgrid::mc::ProcessIndexDisabled);
 295   this->cache_flags |= MC_PROCESS_CACHE_FLAG_HEAP;
 296 }
 297
 298 /** Refresh the information about the process
 299  *
 300  *  Do not use direclty, this is used by the getters when appropriate
 301  *  in order to have fresh data.
 302  * */
 303 void Process::refresh_malloc_info()
 304 {
 305   xbt_assert(mc_mode == MC_MODE_SERVER);
 306   xbt_assert(!this->is_self());
 307   if (!(this->cache_flags & MC_PROCESS_CACHE_FLAG_HEAP))
 308     this->refresh_heap();
 309   // Refresh process->heapinfo:
 310   size_t malloc_info_bytesize =
 311     (this->heap->heaplimit + 1) * sizeof(malloc_info);
 312   this->heap_info = (malloc_info*) realloc(this->heap_info, malloc_info_bytesize);
 313   this->read_bytes(this->heap_info, malloc_info_bytesize,
 314     remote(this->heap->heapinfo), simgrid::mc::ProcessIndexDisabled);
 315   this->cache_flags |= MC_PROCESS_CACHE_FLAG_MALLOC_INFO;
 316 }
 317
 318 /** @brief Finds the range of the different memory segments and binary paths */
 319 void Process::init_memory_map_info()
 320 {
 321   XBT_DEBUG("Get debug information ...");
 322   this->maestro_stack_start_ = nullptr;
 323   this->maestro_stack_end_ = nullptr;
 324   this->object_infos.resize(0);
 325   this->binary_info = NULL;
 326   this->libsimgrid_info = NULL;
 327
 328   struct s_mc_memory_map_re res;
 329
 330   if(regcomp(&res.so_re, SO_RE, 0) || regcomp(&res.version_re, VERSION_RE, 0))
 331     xbt_die(".so regexp did not compile");
 332
 333   std::vector<simgrid::mc::VmMap> const& maps = this->memory_map_;
 334
 335   const char* current_name = NULL;
 336
 337   this->object_infos.resize(0);
 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     std::shared_ptr<s_mc_object_info_t> info =
 379       MC_find_object_info(this->memory_map_, pathname, is_executable);
 380     this->object_infos.push_back(info);
 381     if (is_executable)
 382       this->binary_info = info;
 383     else if (libname && MC_is_simgrid_lib(libname))
 384       this->libsimgrid_info = info;
 385     free(libname);
 386   }
 387
 388   regfree(&res.so_re);
 389   regfree(&res.version_re);
 390
 391   // Resolve time (including accross differents objects):
 392   for (auto const& object_info : this->object_infos)
 393     MC_post_process_object_info(this, object_info.get());
 394
 395   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 396   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 397
 398   XBT_DEBUG("Get debug information done !");
 399 }
 400
 401 std::shared_ptr<s_mc_object_info_t> Process::find_object_info(remote_ptr<void> addr) const
 402 {
 403   for (auto const& object_info : this->object_infos) {
 404     if (addr.address() >= (std::uint64_t)object_info->start
 405         && addr.address() <= (std::uint64_t)object_info->end) {
 406       return object_info;
 407     }
 408   }
 409   return NULL;
 410 }
 411
 412 std::shared_ptr<s_mc_object_info_t> Process::find_object_info_exec(remote_ptr<void> addr) const
 413 {
 414   for (std::shared_ptr<s_mc_object_info> const& info : this->object_infos) {
 415     if (addr.address() >= (std::uint64_t) info->start_exec
 416         && addr.address() <= (std::uint64_t) info->end_exec) {
 417       return info;
 418     }
 419   }
 420   return nullptr;
 421 }
 422
 423 std::shared_ptr<s_mc_object_info_t> Process::find_object_info_rw(remote_ptr<void> addr) const
 424 {
 425   for (std::shared_ptr<s_mc_object_info> const& info : this->object_infos) {
 426     if (addr.address() >= (std::uint64_t)info->start_rw
 427         && addr.address() <= (std::uint64_t)info->end_rw) {
 428       return info;
 429     }
 430   }
 431   return nullptr;
 432 }
 433
 434 dw_frame_t Process::find_function(remote_ptr<void> ip) const
 435 {
 436   std::shared_ptr<s_mc_object_info_t> info = this->find_object_info_exec(ip);
 437   if (!info)
 438     return nullptr;
 439   else
 440     return MC_file_object_info_find_function(info.get(), (void*) ip.address());
 441 }
 442
 443 /** Find (one occurence of) the named variable definition
 444  */
 445 dw_variable_t Process::find_variable(const char* name) const
 446 {
 447   // First lookup the variable in the executable shared object.
 448   // A global variable used directly by the executable code from a library
 449   // is reinstanciated in the executable memory .data/.bss.
 450   // We need to look up the variable in the execvutable first.
 451   if (this->binary_info) {
 452     std::shared_ptr<s_mc_object_info_t> const& info = this->binary_info;
 453     dw_variable_t var = MC_file_object_info_find_variable_by_name(info.get(), name);
 454     if (var)
 455       return var;
 456   }
 457
 458   for (std::shared_ptr<s_mc_object_info_t> const& info : this->object_infos) {
 459     dw_variable_t var = MC_file_object_info_find_variable_by_name(info.get(), name);
 460     if (var)
 461       return var;
 462   }
 463
 464   return NULL;
 465 }
 466
 467 void Process::read_variable(const char* name, void* target, size_t size) const
 468 {
 469   dw_variable_t var = this->find_variable(name);
 470   if (!var->address)
 471     xbt_die("No simple location for this variable");
 472   if (!var->type->full_type)
 473     xbt_die("Partial type for %s, cannot check size", name);
 474   if ((size_t) var->type->full_type->byte_size != size)
 475     xbt_die("Unexpected size for %s (expected %zi, was %zi)",
 476       name, size, (size_t) var->type->full_type->byte_size);
 477   this->read_bytes(target, size, remote(var->address));
 478 }
 479
 480 char* Process::read_string(remote_ptr<void> address) const
 481 {
 482   if (!address)
 483     return NULL;
 484   if (this->is_self())
 485     return xbt_strdup((char*) address.address());
 486
 487   off_t len = 128;
 488   char* res = (char*) malloc(len);
 489   off_t off = 0;
 490
 491   while (1) {
 492     ssize_t c = pread(this->memory_file, res + off, len - off, (off_t) address.address() + off);
 493     if (c == -1) {
 494       if (errno == EINTR)
 495         continue;
 496       else
 497         xbt_die("Could not read from from remote process");
 498     }
 499     if (c==0)
 500       xbt_die("Could not read string from remote process");
 501
 502     void* p = memchr(res + off, '\0', c);
 503     if (p)
 504       return res;
 505
 506     off += c;
 507     if (off == len) {
 508       len *= 2;
 509       res = (char*) realloc(res, len);
 510     }
 511   }
 512 }
 513
 514 const void *Process::read_bytes(void* buffer, std::size_t size,
 515   remote_ptr<void> address, int process_index,
 516   AddressSpace::ReadMode mode) const
 517 {
 518   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 519     std::shared_ptr<s_mc_object_info_t> const& info =
 520       this->find_object_info_rw((void*)address.address());
 521     // Segment overlap is not handled.
 522     if (MC_object_info_is_privatized(info.get())) {
 523       if (process_index < 0)
 524         xbt_die("Missing process index");
 525       if (process_index >= (int) MC_smpi_process_count())
 526         xbt_die("Invalid process index");
 527
 528       // Read smpi_privatisation_regions from MCed:
 529       smpi_privatisation_region_t remote_smpi_privatisation_regions =
 530         mc_model_checker->process().read_variable<smpi_privatisation_region_t>(
 531           "smpi_privatisation_regions");
 532
 533       s_smpi_privatisation_region_t privatisation_region =
 534         mc_model_checker->process().read<s_smpi_privatisation_region_t>(
 535           remote(remote_smpi_privatisation_regions + process_index));
 536
 537       // Address translation in the privaization segment:
 538       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 539       address = remote((char*)privatisation_region.address + offset);
 540     }
 541   }
 542
 543   if (this->is_self()) {
 544     if (mode == simgrid::mc::AddressSpace::Lazy)
 545       return (void*)address.address();
 546     else {
 547       memcpy(buffer, (void*)address.address(), size);
 548       return buffer;
 549     }
 550   } else {
 551     if (pread_whole(this->memory_file, buffer, size, (off_t) address.address()) < 0)
 552       xbt_die("Read from process %lli failed", (long long) this->pid_);
 553     return buffer;
 554   }
 555 }
 556
 557 /** Write data to a process memory
 558  *
 559  *  @param process the process
 560  *  @param local   local memory address (source)
 561  *  @param remote  target process memory address (target)
 562  *  @param len     data size
 563  */
 564 void Process::write_bytes(const void* buffer, size_t len, remote_ptr<void> address)
 565 {
 566   if (this->is_self()) {
 567     memcpy((void*)address.address(), buffer, len);
 568   } else {
 569     if (pwrite_whole(this->memory_file, buffer, len, address.address()) < 0)
 570       xbt_die("Write to process %lli failed", (long long) this->pid_);
 571   }
 572 }
 573
 574 void Process::clear_bytes(remote_ptr<void> address, size_t len)
 575 {
 576   if (this->is_self()) {
 577     memset((void*)address.address(), 0, len);
 578   } else {
 579     pthread_once(&zero_buffer_flag, MC_zero_buffer_init);
 580     while (len) {
 581       size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 582       this->write_bytes(zero_buffer, s, address);
 583       address = remote((char*) address.address() + s);
 584       len -= s;
 585     }
 586   }
 587 }
 588
 589 void Process::ignore_region(std::uint64_t addr, std::size_t size)
 590 {
 591   IgnoredRegion region;
 592   region.addr = addr;
 593   region.size = size;
 594
 595   if (ignored_regions_.empty()) {
 596     ignored_regions_.push_back(region);
 597     return;
 598   }
 599
 600   unsigned int cursor = 0;
 601   IgnoredRegion* current_region = nullptr;
 602
 603   int start = 0;
 604   int end = ignored_regions_.size() - 1;
 605   while (start <= end) {
 606     cursor = (start + end) / 2;
 607     current_region = &ignored_regions_[cursor];
 608     if (current_region->addr == addr) {
 609       if (current_region->size == size)
 610         return;
 611       else if (current_region->size < size)
 612         start = cursor + 1;
 613       else
 614         end = cursor - 1;
 615     } else if (current_region->addr < addr)
 616       start = cursor + 1;
 617     else
 618       end = cursor - 1;
 619   }
 620
 621   std::size_t position;
 622   if (current_region->addr == addr) {
 623     if (current_region->size < size) {
 624       position = cursor + 1;
 625     } else {
 626       position = cursor;
 627     }
 628   } else if (current_region->addr < addr) {
 629     position = cursor + 1;
 630   } else {
 631     position = cursor;
 632   }
 633   ignored_regions_.insert(
 634     ignored_regions_.begin() + position, region);
 635 }
 636
 637 }
 638 }