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