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 #define _FILE_OFFSET_BITS 64
   8
   9 #include <assert.h>
  10 #include <stddef.h>
  11 #include <stdbool.h>
  12 #include <stdint.h>
  13 #include <errno.h>
  14
  15 #include <sys/types.h>
  16 #include <fcntl.h>
  17 #include <unistd.h>
  18 #include <regex.h>
  19 #include <sys/mman.h> // PROT_*
  20
  21 #include <pthread.h>
  22
  23 #include <libgen.h>
  24
  25 #include <libunwind.h>
  26 #include <libunwind-ptrace.h>
  27
  28 #include <xbt/mmalloc.h>
  29
  30 #include "mc_process.h"
  31 #include "mc_object_info.h"
  32 #include "AddressSpace.hpp"
  33 #include "mc_unw.h"
  34 #include "mc_snapshot.h"
  35 #include "mc_ignore.h"
  36 #include "mc_smx.h"
  37 #include "mc_server.h"
  38
  39 using simgrid::mc::remote;
  40
  41 extern "C" {
  42
  43 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(mc_process, mc,
  44                                 "MC process information");
  45
  46 // ***** Helper stuff
  47
  48 #define SO_RE "\\.so[\\.0-9]*$"
  49 #define VERSION_RE "-[\\.0-9-]*$"
  50
  51 static const char *const FILTERED_LIBS[] = {
  52   "ld",
  53   "libbz2",
  54   "libboost_context",
  55   "libc",
  56   "libc++",
  57   "libcdt",
  58   "libcgraph",
  59   "libdl",
  60   "libdw",
  61   "libelf",
  62   "libgcc_s",
  63   "liblua5.1",
  64   "liblzma",
  65   "libm",
  66   "libpthread",
  67   "librt",
  68   "libstdc++",
  69   "libunwind",
  70   "libunwind-x86_64",
  71   "libunwind-x86",
  72   "libunwind-ptrace",
  73   "libz"
  74 };
  75
  76 static bool MC_is_simgrid_lib(const char* libname)
  77 {
  78   return !strcmp(libname, "libsimgrid");
  79 }
  80
  81 static bool MC_is_filtered_lib(const char* libname)
  82 {
  83   for (const char* filtered_lib : FILTERED_LIBS)
  84     if (strcmp(libname, filtered_lib)==0)
  85       return true;
  86   return false;
  87 }
  88
  89 struct s_mc_memory_map_re {
  90   regex_t so_re;
  91   regex_t version_re;
  92 };
  93
  94 static char* MC_get_lib_name(const char* pathname, struct s_mc_memory_map_re* res)
  95 {
  96   const char* map_basename = basename((char*) pathname);
  97
  98   regmatch_t match;
  99   if(regexec(&res->so_re, map_basename, 1, &match, 0))
 100     return NULL;
 101
 102   char* libname = strndup(map_basename, match.rm_so);
 103
 104   // Strip the version suffix:
 105   if(libname && !regexec(&res->version_re, libname, 1, &match, 0)) {
 106     char* temp = libname;
 107     libname = strndup(temp, match.rm_so);
 108     free(temp);
 109   }
 110
 111   return libname;
 112 }
 113
 114 static ssize_t pread_whole(int fd, void *buf, size_t count, std::uint64_t offset)
 115 {
 116   char* buffer = (char*) buf;
 117   ssize_t real_count = count;
 118   while (count) {
 119     ssize_t res = pread(fd, buffer, count, (off_t)(std::int64_t) offset);
 120     if (res > 0) {
 121       count  -= res;
 122       buffer += res;
 123       offset += res;
 124     } else if (res==0) {
 125       return -1;
 126     } else if (errno != EINTR) {
 127       return -1;
 128     }
 129   }
 130   return real_count;
 131 }
 132
 133 static ssize_t pwrite_whole(int fd, const void *buf, size_t count, off_t offset)
 134 {
 135   const char* buffer = (const char*) buf;
 136   ssize_t real_count = count;
 137   while (count) {
 138     ssize_t res = pwrite(fd, buffer, count, offset);
 139     if (res > 0) {
 140       count  -= res;
 141       buffer += res;
 142       offset += res;
 143     } else if (res==0) {
 144       return -1;
 145     } else if (errno != EINTR) {
 146       return -1;
 147     }
 148   }
 149   return real_count;
 150 }
 151
 152 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 153 static const void* zero_buffer;
 154 static const size_t zero_buffer_size = 10 * 4096;
 155
 156 static void MC_zero_buffer_init(void)
 157 {
 158   int fd = open("/dev/zero", O_RDONLY);
 159   if (fd<0)
 160     xbt_die("Could not open /dev/zero");
 161   zero_buffer = mmap(NULL, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 162   if (zero_buffer == MAP_FAILED)
 163     xbt_die("Could not map the zero buffer");
 164   close(fd);
 165 }
 166
 167 }
 168
 169 namespace simgrid {
 170 namespace mc {
 171
 172 int open_vm(pid_t pid, int flags)
 173 {
 174   const size_t buffer_size = 30;
 175   char buffer[buffer_size];
 176   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long) pid);
 177   if (res < 0 || (size_t) res >= buffer_size) {
 178     errno = ENAMETOOLONG;
 179     return -1;
 180   }
 181   return open(buffer, flags);
 182 }
 183
 184
 185 }
 186 }
 187
 188 // ***** Process
 189
 190 namespace simgrid {
 191 namespace mc {
 192
 193 Process::Process(pid_t pid, int sockfd)
 194 {
 195   Process* process = this;
 196
 197   process->process_flags = MC_PROCESS_NO_FLAG;
 198   process->socket_ = sockfd;
 199   process->pid_ = pid;
 200   if (pid==getpid())
 201     process->process_flags |= MC_PROCESS_SELF_FLAG;
 202   process->running_ = true;
 203   process->status_ = 0;
 204   process->memory_map_ = get_memory_map(pid);
 205   process->cache_flags = MC_PROCESS_CACHE_FLAG_NONE;
 206   process->heap = NULL;
 207   process->heap_info = NULL;
 208   process->init_memory_map_info();
 209
 210   // Open the memory file
 211   if (process->is_self())
 212     process->memory_file = -1;
 213   else {
 214     int fd = open_vm(process->pid_, O_RDWR);
 215     if (fd<0)
 216       xbt_die("Could not open file for process virtual address space");
 217     process->memory_file = fd;
 218   }
 219
 220   // Read std_heap (is a struct mdesc*):
 221   dw_variable_t std_heap_var = process->find_variable("__mmalloc_default_mdp");
 222   if (!std_heap_var)
 223     xbt_die("No heap information in the target process");
 224   if(!std_heap_var->address)
 225     xbt_die("No constant address for this variable");
 226   process->read_bytes(&process->heap_address, sizeof(struct mdesc*),
 227     remote(std_heap_var->address),
 228     simgrid::mc::ProcessIndexDisabled);
 229
 230   process->smx_process_infos = MC_smx_process_info_list_new();
 231   process->smx_old_process_infos = MC_smx_process_info_list_new();
 232
 233   process->unw_addr_space = unw_create_addr_space(&mc_unw_accessors  , __BYTE_ORDER);
 234   if (process->process_flags & MC_PROCESS_SELF_FLAG) {
 235     process->unw_underlying_addr_space = unw_local_addr_space;
 236     process->unw_underlying_context = NULL;
 237   } else {
 238     process->unw_underlying_addr_space = unw_create_addr_space(&mc_unw_vmread_accessors, __BYTE_ORDER);
 239     process->unw_underlying_context = _UPT_create(pid);
 240   }
 241 }
 242
 243 Process::~Process()
 244 {
 245   Process* process = this;
 246
 247   if (this->socket_ >= 0 && close(this->socket_) < 0)
 248     xbt_die("Could not close communication socket");
 249
 250   process->process_flags = MC_PROCESS_NO_FLAG;
 251   process->pid_ = 0;
 252
 253   process->maestro_stack_start_ = nullptr;
 254   process->maestro_stack_end_ = nullptr;
 255
 256   xbt_dynar_free(&process->smx_process_infos);
 257   xbt_dynar_free(&process->smx_old_process_infos);
 258
 259   if (process->memory_file >= 0) {
 260     close(process->memory_file);
 261   }
 262
 263   if (process->unw_underlying_addr_space != unw_local_addr_space) {
 264     unw_destroy_addr_space(process->unw_underlying_addr_space);
 265     _UPT_destroy(process->unw_underlying_context);
 266   }
 267   process->unw_underlying_context = NULL;
 268   process->unw_underlying_addr_space = NULL;
 269
 270   unw_destroy_addr_space(process->unw_addr_space);
 271   process->unw_addr_space = NULL;
 272
 273   process->cache_flags = MC_PROCESS_CACHE_FLAG_NONE;
 274
 275   free(process->heap);
 276   process->heap = NULL;
 277
 278   free(process->heap_info);
 279   process->heap_info = NULL;
 280 }
 281
 282 /** Refresh the information about the process
 283  *
 284  *  Do not use direclty, this is used by the getters when appropriate
 285  *  in order to have fresh data.
 286  */
 287 void Process::refresh_heap()
 288 {
 289   xbt_assert(mc_mode == MC_MODE_SERVER);
 290   xbt_assert(!this->is_self());
 291   // Read/dereference/refresh the std_heap pointer:
 292   if (!this->heap) {
 293     this->heap = (struct mdesc*) malloc(sizeof(struct mdesc));
 294   }
 295   this->read_bytes(this->heap, sizeof(struct mdesc), remote(this->heap_address),
 296     simgrid::mc::ProcessIndexDisabled);
 297   this->cache_flags |= MC_PROCESS_CACHE_FLAG_HEAP;
 298 }
 299
 300 /** Refresh the information about the process
 301  *
 302  *  Do not use direclty, this is used by the getters when appropriate
 303  *  in order to have fresh data.
 304  * */
 305 void Process::refresh_malloc_info()
 306 {
 307   xbt_assert(mc_mode == MC_MODE_SERVER);
 308   xbt_assert(!this->is_self());
 309   if (!(this->cache_flags & MC_PROCESS_CACHE_FLAG_HEAP))
 310     this->refresh_heap();
 311   // Refresh process->heapinfo:
 312   size_t malloc_info_bytesize =
 313     (this->heap->heaplimit + 1) * sizeof(malloc_info);
 314   this->heap_info = (malloc_info*) realloc(this->heap_info, malloc_info_bytesize);
 315   this->read_bytes(this->heap_info, malloc_info_bytesize,
 316     remote(this->heap->heapinfo), simgrid::mc::ProcessIndexDisabled);
 317   this->cache_flags |= MC_PROCESS_CACHE_FLAG_MALLOC_INFO;
 318 }
 319
 320 /** @brief Finds the range of the different memory segments and binary paths */
 321 void Process::init_memory_map_info()
 322 {
 323   XBT_DEBUG("Get debug information ...");
 324   this->maestro_stack_start_ = nullptr;
 325   this->maestro_stack_end_ = nullptr;
 326   this->object_infos.resize(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   this->object_infos.resize(0);
 340
 341   for (size_t i=0; i < maps.size(); i++) {
 342     simgrid::mc::VmMap const& reg = maps[i];
 343     const char* pathname = maps[i].pathname.c_str();
 344
 345     // Nothing to do
 346     if (maps[i].pathname.empty()) {
 347       current_name = NULL;
 348       continue;
 349     }
 350
 351     // [stack], [vvar], [vsyscall], [vdso] ...
 352     if (pathname[0] == '[') {
 353       if ((reg.prot & PROT_WRITE) && !memcmp(pathname, "[stack]", 7)) {
 354         this->maestro_stack_start_ = remote(reg.start_addr);
 355         this->maestro_stack_end_ = remote(reg.end_addr);
 356       }
 357       current_name = NULL;
 358       continue;
 359     }
 360
 361     if (current_name && strcmp(current_name, pathname)==0)
 362       continue;
 363
 364     current_name = pathname;
 365     if (!(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 366       continue;
 367
 368     const bool is_executable = !i;
 369     char* libname = NULL;
 370     if (!is_executable) {
 371       libname = MC_get_lib_name(pathname, &res);
 372       if(!libname)
 373         continue;
 374       if (MC_is_filtered_lib(libname)) {
 375         free(libname);
 376         continue;
 377       }
 378     }
 379
 380     std::shared_ptr<s_mc_object_info_t> info =
 381       MC_find_object_info(this->memory_map_, pathname, is_executable);
 382     this->object_infos.push_back(info);
 383     if (is_executable)
 384       this->binary_info = info;
 385     else if (libname && MC_is_simgrid_lib(libname))
 386       this->libsimgrid_info = info;
 387     free(libname);
 388   }
 389
 390   regfree(&res.so_re);
 391   regfree(&res.version_re);
 392
 393   // Resolve time (including accross differents objects):
 394   for (auto const& object_info : this->object_infos)
 395     MC_post_process_object_info(this, object_info.get());
 396
 397   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 398   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 399
 400   XBT_DEBUG("Get debug information done !");
 401 }
 402
 403 std::shared_ptr<s_mc_object_info_t> Process::find_object_info(remote_ptr<void> addr) const
 404 {
 405   for (auto const& object_info : this->object_infos) {
 406     if (addr.address() >= (std::uint64_t)object_info->start
 407         && addr.address() <= (std::uint64_t)object_info->end) {
 408       return object_info;
 409     }
 410   }
 411   return NULL;
 412 }
 413
 414 std::shared_ptr<s_mc_object_info_t> Process::find_object_info_exec(remote_ptr<void> addr) const
 415 {
 416   for (std::shared_ptr<s_mc_object_info> const& info : this->object_infos) {
 417     if (addr.address() >= (std::uint64_t) info->start_exec
 418         && addr.address() <= (std::uint64_t) info->end_exec) {
 419       return info;
 420     }
 421   }
 422   return nullptr;
 423 }
 424
 425 std::shared_ptr<s_mc_object_info_t> Process::find_object_info_rw(remote_ptr<void> addr) const
 426 {
 427   for (std::shared_ptr<s_mc_object_info> const& info : this->object_infos) {
 428     if (addr.address() >= (std::uint64_t)info->start_rw
 429         && addr.address() <= (std::uint64_t)info->end_rw) {
 430       return info;
 431     }
 432   }
 433   return nullptr;
 434 }
 435
 436 dw_frame_t Process::find_function(remote_ptr<void> ip) const
 437 {
 438   std::shared_ptr<s_mc_object_info_t> info = this->find_object_info_exec(ip);
 439   if (!info)
 440     return nullptr;
 441   else
 442     return MC_file_object_info_find_function(info.get(), (void*) ip.address());
 443 }
 444
 445 /** Find (one occurence of) the named variable definition
 446  */
 447 dw_variable_t Process::find_variable(const char* name) const
 448 {
 449   // First lookup the variable in the executable shared object.
 450   // A global variable used directly by the executable code from a library
 451   // is reinstanciated in the executable memory .data/.bss.
 452   // We need to look up the variable in the execvutable first.
 453   if (this->binary_info) {
 454     std::shared_ptr<s_mc_object_info_t> const& info = this->binary_info;
 455     dw_variable_t var = MC_file_object_info_find_variable_by_name(info.get(), name);
 456     if (var)
 457       return var;
 458   }
 459
 460   for (std::shared_ptr<s_mc_object_info_t> const& info : this->object_infos) {
 461     dw_variable_t var = MC_file_object_info_find_variable_by_name(info.get(), name);
 462     if (var)
 463       return var;
 464   }
 465
 466   return NULL;
 467 }
 468
 469 void Process::read_variable(const char* name, void* target, size_t size) const
 470 {
 471   dw_variable_t var = this->find_variable(name);
 472   if (!var->address)
 473     xbt_die("No simple location for this variable");
 474   if (!var->type->full_type)
 475     xbt_die("Partial type for %s, cannot check size", name);
 476   if ((size_t) var->type->full_type->byte_size != size)
 477     xbt_die("Unexpected size for %s (expected %zi, was %zi)",
 478       name, size, (size_t) var->type->full_type->byte_size);
 479   this->read_bytes(target, size, remote(var->address));
 480 }
 481
 482 char* Process::read_string(remote_ptr<void> address) const
 483 {
 484   if (!address)
 485     return NULL;
 486   if (this->is_self())
 487     return xbt_strdup((char*) address.address());
 488
 489   off_t len = 128;
 490   char* res = (char*) malloc(len);
 491   off_t off = 0;
 492
 493   while (1) {
 494     ssize_t c = pread(this->memory_file, res + off, len - off, (off_t) address.address() + off);
 495     if (c == -1) {
 496       if (errno == EINTR)
 497         continue;
 498       else
 499         xbt_die("Could not read from from remote process");
 500     }
 501     if (c==0)
 502       xbt_die("Could not read string from remote process");
 503
 504     void* p = memchr(res + off, '\0', c);
 505     if (p)
 506       return res;
 507
 508     off += c;
 509     if (off == len) {
 510       len *= 2;
 511       res = (char*) realloc(res, len);
 512     }
 513   }
 514 }
 515
 516 const void *Process::read_bytes(void* buffer, std::size_t size,
 517   remote_ptr<void> address, int process_index,
 518   AddressSpace::ReadMode mode) const
 519 {
 520   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 521     std::shared_ptr<s_mc_object_info_t> const& info =
 522       this->find_object_info_rw((void*)address.address());
 523     // Segment overlap is not handled.
 524     if (MC_object_info_is_privatized(info.get())) {
 525       if (process_index < 0)
 526         xbt_die("Missing process index");
 527       if (process_index >= (int) MC_smpi_process_count())
 528         xbt_die("Invalid process index");
 529
 530       // Read smpi_privatisation_regions from MCed:
 531       smpi_privatisation_region_t remote_smpi_privatisation_regions =
 532         mc_model_checker->process().read_variable<smpi_privatisation_region_t>(
 533           "smpi_privatisation_regions");
 534
 535       s_smpi_privatisation_region_t privatisation_region =
 536         mc_model_checker->process().read<s_smpi_privatisation_region_t>(
 537           remote(remote_smpi_privatisation_regions + process_index));
 538
 539       // Address translation in the privaization segment:
 540       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 541       address = remote((char*)privatisation_region.address + offset);
 542     }
 543   }
 544
 545   if (this->is_self()) {
 546     if (mode == simgrid::mc::AddressSpace::Lazy)
 547       return (void*)address.address();
 548     else {
 549       memcpy(buffer, (void*)address.address(), size);
 550       return buffer;
 551     }
 552   } else {
 553     if (pread_whole(this->memory_file, buffer, size, address.address()) < 0)
 554       xbt_die("Read from process %lli failed", (long long) this->pid_);
 555     return buffer;
 556   }
 557 }
 558
 559 /** Write data to a process memory
 560  *
 561  *  @param process the process
 562  *  @param local   local memory address (source)
 563  *  @param remote  target process memory address (target)
 564  *  @param len     data size
 565  */
 566 void Process::write_bytes(const void* buffer, size_t len, remote_ptr<void> address)
 567 {
 568   if (this->is_self()) {
 569     memcpy((void*)address.address(), buffer, len);
 570   } else {
 571     if (pwrite_whole(this->memory_file, buffer, len, address.address()) < 0)
 572       xbt_die("Write to process %lli failed", (long long) this->pid_);
 573   }
 574 }
 575
 576 void Process::clear_bytes(remote_ptr<void> address, size_t len)
 577 {
 578   if (this->is_self()) {
 579     memset((void*)address.address(), 0, len);
 580   } else {
 581     pthread_once(&zero_buffer_flag, MC_zero_buffer_init);
 582     while (len) {
 583       size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 584       this->write_bytes(zero_buffer, s, address);
 585       address = remote((char*) address.address() + s);
 586       len -= s;
 587     }
 588   }
 589 }
 590
 591 void Process::ignore_region(std::uint64_t addr, std::size_t size)
 592 {
 593   IgnoredRegion region;
 594   region.addr = addr;
 595   region.size = size;
 596
 597   if (ignored_regions_.empty()) {
 598     ignored_regions_.push_back(region);
 599     return;
 600   }
 601
 602   unsigned int cursor = 0;
 603   IgnoredRegion* current_region = nullptr;
 604
 605   int start = 0;
 606   int end = ignored_regions_.size() - 1;
 607   while (start <= end) {
 608     cursor = (start + end) / 2;
 609     current_region = &ignored_regions_[cursor];
 610     if (current_region->addr == addr) {
 611       if (current_region->size == size)
 612         return;
 613       else if (current_region->size < size)
 614         start = cursor + 1;
 615       else
 616         end = cursor - 1;
 617     } else if (current_region->addr < addr)
 618       start = cursor + 1;
 619     else
 620       end = cursor - 1;
 621   }
 622
 623   std::size_t position;
 624   if (current_region->addr == addr) {
 625     if (current_region->size < size) {
 626       position = cursor + 1;
 627     } else {
 628       position = cursor;
 629     }
 630   } else if (current_region->addr < addr) {
 631     position = cursor + 1;
 632   } else {
 633     position = cursor;
 634   }
 635   ignored_regions_.insert(
 636     ignored_regions_.begin() + position, region);
 637 }
 638
 639 }
 640 }