src/mc/remote/RemoteClient.cpp

   1 /* Copyright (c) 2014-2019. The SimGrid Team. All rights reserved.          */
   2
   3 /* This program is free software; you can redistribute it and/or modify it
   4  * under the terms of the license (GNU LGPL) which comes with this package. */
   5
   6 #define _FILE_OFFSET_BITS 64 /* needed for pread_whole to work as expected on 32bits */
   7
   8 #include "src/mc/remote/RemoteClient.hpp"
   9
  10 #include "xbt/file.hpp"
  11 #include "xbt/log.h"
  12 #include "src/mc/mc_smx.hpp"
  13 #include "src/mc/sosp/mc_snapshot.hpp"
  14
  15 #include <fcntl.h>
  16 #include <libunwind-ptrace.h>
  17 #include <sys/mman.h> // PROT_*
  18
  19 using simgrid::mc::remote;
  20
  21 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(mc_process, mc, "MC process information");
  22
  23 namespace simgrid {
  24 namespace mc {
  25
  26 // ***** Helper stuff
  27
  28 // List of library which memory segments are not considered:
  29 static const std::vector<std::string> filtered_libraries = {
  30 #ifdef __linux__
  31     "ld",
  32 #elif defined __FreeBSD__
  33     "ld-elf",
  34     "ld-elf32",
  35     "libkvm",      /* kernel data access library */
  36     "libprocstat", /* process and file information retrieval */
  37     "libthr",      /* thread library */
  38     "libutil",
  39 #endif
  40     "libargp", /* workarounds for glibc-less systems */
  41     "libasan", /* gcc sanitizers */
  42     "libboost_chrono",
  43     "libboost_context",
  44     "libboost_context-mt",
  45     "libboost_stacktrace_backtrace",
  46     "libboost_system",
  47     "libboost_thread",
  48     "libboost_timer",
  49     "libbz2",
  50     "libc",
  51     "libc++",
  52     "libcdt",
  53     "libcgraph",
  54     "libcrypto",
  55     "libcxxrt",
  56     "libdl",
  57     "libdw",
  58     "libelf",
  59     "libevent",
  60     "libexecinfo",
  61     "libflang",
  62     "libflangrti",
  63     "libgcc_s",
  64     "libgfortran",
  65     "libimf",
  66     "libintlc",
  67     "libirng",
  68     "liblua5.1",
  69     "liblua5.3",
  70     "liblzma",
  71     "libm",
  72     "libomp",
  73     "libpapi",
  74     "libpfm",
  75     "libpgmath",
  76     "libpthread",
  77     "libquadmath",
  78     "librt",
  79     "libstdc++",
  80     "libsvml",
  81     "libtsan",  /* gcc sanitizers */
  82     "libubsan", /* gcc sanitizers */
  83     "libunwind",
  84     "libunwind-ptrace",
  85     "libunwind-x86",
  86     "libunwind-x86_64",
  87     "libz"};
  88
  89 static bool is_simgrid_lib(const std::string& libname)
  90 {
  91   return libname == "libsimgrid";
  92 }
  93
  94 static bool is_filtered_lib(const std::string& libname)
  95 {
  96   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
  97 }
  98
  99 static std::string get_lib_name(const std::string& pathname)
 100 {
 101   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
 102   std::string libname;
 103
 104   size_t pos = map_basename.rfind(".so");
 105   if (pos != std::string::npos) {
 106     // strip the extension (matching regex "\.so.*$")
 107     libname.assign(map_basename, 0, pos);
 108
 109     // strip the version suffix (matching regex "-[.0-9-]*$")
 110     while (true) {
 111       pos = libname.rfind('-');
 112       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
 113         break;
 114       libname.erase(pos);
 115     }
 116   }
 117
 118   return libname;
 119 }
 120
 121 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 122 {
 123   char* buffer       = (char*)buf;
 124   ssize_t real_count = count;
 125   while (count) {
 126     ssize_t res = pread(fd, buffer, count, offset);
 127     if (res > 0) {
 128       count -= res;
 129       buffer += res;
 130       offset += res;
 131     } else if (res == 0)
 132       return -1;
 133     else if (errno != EINTR) {
 134       perror("pread_whole");
 135       return -1;
 136     }
 137   }
 138   return real_count;
 139 }
 140
 141 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 142 {
 143   const char* buffer = (const char*)buf;
 144   ssize_t real_count = count;
 145   while (count) {
 146     ssize_t res = pwrite(fd, buffer, count, offset);
 147     if (res > 0) {
 148       count -= res;
 149       buffer += res;
 150       offset += res;
 151     } else if (res == 0)
 152       return -1;
 153     else if (errno != EINTR)
 154       return -1;
 155   }
 156   return real_count;
 157 }
 158
 159 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 160 static const void* zero_buffer;
 161 static const size_t zero_buffer_size = 10 * 4096;
 162
 163 static void zero_buffer_init()
 164 {
 165   int fd = open("/dev/zero", O_RDONLY);
 166   if (fd < 0)
 167     xbt_die("Could not open /dev/zero");
 168   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 169   if (zero_buffer == MAP_FAILED)
 170     xbt_die("Could not map the zero buffer");
 171   close(fd);
 172 }
 173
 174 int open_vm(pid_t pid, int flags)
 175 {
 176   const size_t buffer_size = 30;
 177   char buffer[buffer_size];
 178   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 179   if (res < 0 || (size_t)res >= buffer_size) {
 180     errno = ENAMETOOLONG;
 181     return -1;
 182   }
 183   return open(buffer, flags);
 184 }
 185
 186 // ***** Process
 187
 188 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
 189 {
 190 }
 191
 192 void RemoteClient::init()
 193 {
 194   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 195   this->init_memory_map_info();
 196
 197   int fd = open_vm(this->pid_, O_RDWR);
 198   if (fd < 0)
 199     xbt_die("Could not open file for process virtual address space");
 200   this->memory_file = fd;
 201
 202   // Read std_heap (is a struct mdesc*):
 203   simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 204   if (not std_heap_var)
 205     xbt_die("No heap information in the target process");
 206   if (not std_heap_var->address)
 207     xbt_die("No constant address for this variable");
 208   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address),
 209                    simgrid::mc::ProcessIndexDisabled);
 210
 211   this->smx_actors_infos.clear();
 212   this->smx_dead_actors_infos.clear();
 213   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 214   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 215   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 216 }
 217
 218 RemoteClient::~RemoteClient()
 219 {
 220   if (this->memory_file >= 0)
 221     close(this->memory_file);
 222
 223   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 224     if (this->unw_underlying_addr_space)
 225       unw_destroy_addr_space(this->unw_underlying_addr_space);
 226     if (this->unw_underlying_context)
 227       _UPT_destroy(this->unw_underlying_context);
 228   }
 229
 230   unw_destroy_addr_space(this->unw_addr_space);
 231 }
 232
 233 /** Refresh the information about the process
 234  *
 235  *  Do not use directly, this is used by the getters when appropriate
 236  *  in order to have fresh data.
 237  */
 238 void RemoteClient::refresh_heap()
 239 {
 240   // Read/dereference/refresh the std_heap pointer:
 241   if (not this->heap)
 242     this->heap.reset(new s_xbt_mheap_t());
 243   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address), simgrid::mc::ProcessIndexDisabled);
 244   this->cache_flags_ |= RemoteClient::cache_heap;
 245 }
 246
 247 /** Refresh the information about the process
 248  *
 249  *  Do not use direclty, this is used by the getters when appropriate
 250  *  in order to have fresh data.
 251  * */
 252 void RemoteClient::refresh_malloc_info()
 253 {
 254   // Refresh process->heapinfo:
 255   if (this->cache_flags_ & RemoteClient::cache_malloc)
 256     return;
 257   size_t count = this->heap->heaplimit + 1;
 258   if (this->heap_info.size() < count)
 259     this->heap_info.resize(count);
 260   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo),
 261                    simgrid::mc::ProcessIndexDisabled);
 262   this->cache_flags_ |= RemoteClient::cache_malloc;
 263 }
 264
 265 /** @brief Finds the range of the different memory segments and binary paths */
 266 void RemoteClient::init_memory_map_info()
 267 {
 268   XBT_DEBUG("Get debug information ...");
 269   this->maestro_stack_start_ = nullptr;
 270   this->maestro_stack_end_   = nullptr;
 271   this->object_infos.resize(0);
 272   this->binary_info     = nullptr;
 273   this->libsimgrid_info = nullptr;
 274
 275   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 276
 277   const char* current_name = nullptr;
 278
 279   this->object_infos.clear();
 280
 281   for (size_t i = 0; i < maps.size(); i++) {
 282     simgrid::xbt::VmMap const& reg = maps[i];
 283     const char* pathname           = maps[i].pathname.c_str();
 284
 285     // Nothing to do
 286     if (maps[i].pathname.empty()) {
 287       current_name = nullptr;
 288       continue;
 289     }
 290
 291     // [stack], [vvar], [vsyscall], [vdso] ...
 292     if (pathname[0] == '[') {
 293       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 294         this->maestro_stack_start_ = remote(reg.start_addr);
 295         this->maestro_stack_end_   = remote(reg.end_addr);
 296       }
 297       current_name = nullptr;
 298       continue;
 299     }
 300
 301     if (current_name && strcmp(current_name, pathname) == 0)
 302       continue;
 303
 304     current_name = pathname;
 305     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 306       continue;
 307
 308     const bool is_executable = not i;
 309     std::string libname;
 310     if (not is_executable) {
 311       libname = get_lib_name(pathname);
 312       if (is_filtered_lib(libname)) {
 313         continue;
 314       }
 315     }
 316
 317     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 318         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 319     this->object_infos.push_back(info);
 320     if (is_executable)
 321       this->binary_info = info;
 322     else if (is_simgrid_lib(libname))
 323       this->libsimgrid_info = info;
 324   }
 325
 326   // Resolve time (including across different objects):
 327   for (auto const& object_info : this->object_infos)
 328     postProcessObjectInformation(this, object_info.get());
 329
 330   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 331   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 332
 333   XBT_DEBUG("Get debug information done !");
 334 }
 335
 336 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
 337 {
 338   for (auto const& object_info : this->object_infos)
 339     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 340       return object_info;
 341   return nullptr;
 342 }
 343
 344 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(RemotePtr<void> addr) const
 345 {
 346   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 347     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
 348       return info;
 349   return nullptr;
 350 }
 351
 352 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(RemotePtr<void> addr) const
 353 {
 354   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 355     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
 356       return info;
 357   return nullptr;
 358 }
 359
 360 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
 361 {
 362   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 363   return info ? info->find_function((void*)ip.address()) : nullptr;
 364 }
 365
 366 /** Find (one occurrence of) the named variable definition
 367  */
 368 simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
 369 {
 370   // First lookup the variable in the executable shared object.
 371   // A global variable used directly by the executable code from a library
 372   // is reinstanciated in the executable memory .data/.bss.
 373   // We need to look up the variable in the executable first.
 374   if (this->binary_info) {
 375     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 376     simgrid::mc::Variable* var                                  = info->find_variable(name);
 377     if (var)
 378       return var;
 379   }
 380
 381   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 382     simgrid::mc::Variable* var = info->find_variable(name);
 383     if (var)
 384       return var;
 385   }
 386
 387   return nullptr;
 388 }
 389
 390 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
 391 {
 392   simgrid::mc::Variable* var = this->find_variable(name);
 393   xbt_assert(var->address, "No simple location for this variable");
 394   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 395   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 396              size, (size_t)var->type->full_type->byte_size);
 397   this->read_bytes(target, size, remote(var->address));
 398 }
 399
 400 std::string RemoteClient::read_string(RemotePtr<char> address) const
 401 {
 402   if (not address)
 403     return {};
 404
 405   std::vector<char> res(128);
 406   off_t off = 0;
 407
 408   while (1) {
 409     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 410     if (c == -1) {
 411       if (errno == EINTR)
 412         continue;
 413       else
 414         xbt_die("Could not read from from remote process");
 415     }
 416     if (c == 0)
 417       xbt_die("Could not read string from remote process");
 418
 419     void* p = memchr(res.data() + off, '\0', c);
 420     if (p)
 421       return std::string(res.data());
 422
 423     off += c;
 424     if (off == (off_t)res.size())
 425       res.resize(res.size() * 2);
 426   }
 427 }
 428
 429 const void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, int process_index,
 430                                      ReadOptions /*options*/) const
 431 {
 432 #if HAVE_SMPI
 433   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 434     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->find_object_info_rw(address);
 435     // Segment overlap is not handled.
 436     if (info.get() && this->privatized(*info)) {
 437       if (process_index < 0)
 438         xbt_die("Missing process index");
 439       if (process_index >= (int)MC_smpi_process_count())
 440         xbt_die("Invalid process index");
 441
 442       // Read smpi_privatization_regions from MCed:
 443       smpi_privatization_region_t remote_smpi_privatization_regions =
 444           mc_model_checker->process().read_variable<smpi_privatization_region_t>("smpi_privatization_regions");
 445
 446       s_smpi_privatization_region_t privatization_region =
 447           mc_model_checker->process().read<s_smpi_privatization_region_t>(
 448               remote(remote_smpi_privatization_regions + process_index));
 449
 450       // Address translation in the privatization segment:
 451       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 452       address       = remote((char*)privatization_region.address + offset);
 453     }
 454   }
 455 #endif
 456   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 457     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 458   return buffer;
 459 }
 460
 461 /** Write data to a process memory
 462  *
 463  *  @param buffer   local memory address (source)
 464  *  @param len      data size
 465  *  @param address  target process memory address (target)
 466  */
 467 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 468 {
 469   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 470     xbt_die("Write to process %lli failed", (long long)this->pid_);
 471 }
 472
 473 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 474 {
 475   pthread_once(&zero_buffer_flag, zero_buffer_init);
 476   while (len) {
 477     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 478     this->write_bytes(zero_buffer, s, address);
 479     address = remote((char*)address.address() + s);
 480     len -= s;
 481   }
 482 }
 483
 484 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 485 {
 486   IgnoredRegion region;
 487   region.addr = addr;
 488   region.size = size;
 489
 490   if (ignored_regions_.empty()) {
 491     ignored_regions_.push_back(region);
 492     return;
 493   }
 494
 495   unsigned int cursor           = 0;
 496   IgnoredRegion* current_region = nullptr;
 497
 498   int start = 0;
 499   int end   = ignored_regions_.size() - 1;
 500   while (start <= end) {
 501     cursor         = (start + end) / 2;
 502     current_region = &ignored_regions_[cursor];
 503     if (current_region->addr == addr) {
 504       if (current_region->size == size)
 505         return;
 506       else if (current_region->size < size)
 507         start = cursor + 1;
 508       else
 509         end = cursor - 1;
 510     } else if (current_region->addr < addr)
 511       start = cursor + 1;
 512     else
 513       end = cursor - 1;
 514   }
 515
 516   std::size_t position;
 517   if (current_region->addr == addr) {
 518     if (current_region->size < size)
 519       position = cursor + 1;
 520     else
 521       position = cursor;
 522   } else if (current_region->addr < addr)
 523     position = cursor + 1;
 524   else
 525     position = cursor;
 526   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 527 }
 528
 529 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 530 {
 531   if (ignored_heap_.empty()) {
 532     ignored_heap_.push_back(std::move(region));
 533     return;
 534   }
 535
 536   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 537
 538   size_type start = 0;
 539   size_type end   = ignored_heap_.size() - 1;
 540
 541   // Binary search the position of insertion:
 542   size_type cursor;
 543   while (start <= end) {
 544     cursor               = start + (end - start) / 2;
 545     auto& current_region = ignored_heap_[cursor];
 546     if (current_region.address == region.address)
 547       return;
 548     else if (current_region.address < region.address)
 549       start = cursor + 1;
 550     else if (cursor != 0)
 551       end = cursor - 1;
 552     // Avoid underflow:
 553     else
 554       break;
 555   }
 556
 557   // Insert it mc_heap_ignore_region_t:
 558   if (ignored_heap_[cursor].address < region.address)
 559     ++cursor;
 560   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 561 }
 562
 563 void RemoteClient::unignore_heap(void* address, size_t size)
 564 {
 565   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 566
 567   size_type start = 0;
 568   size_type end   = ignored_heap_.size() - 1;
 569
 570   // Binary search:
 571   size_type cursor;
 572   while (start <= end) {
 573     cursor       = (start + end) / 2;
 574     auto& region = ignored_heap_[cursor];
 575     if (region.address < address)
 576       start = cursor + 1;
 577     else if ((char*)region.address <= ((char*)address + size)) {
 578       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 579       return;
 580     } else if (cursor != 0)
 581       end = cursor - 1;
 582     // Avoid underflow:
 583     else
 584       break;
 585   }
 586 }
 587
 588 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 589 {
 590   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 591     frame_name = nullptr;
 592   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 593     info->remove_local_variable(var_name, frame_name);
 594 }
 595
 596 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 597 {
 598   this->refresh_simix();
 599   return smx_actors_infos;
 600 }
 601
 602 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 603 {
 604   this->refresh_simix();
 605   return smx_dead_actors_infos;
 606 }
 607
 608 void RemoteClient::dumpStack()
 609 {
 610   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 611   if (as == nullptr) {
 612     XBT_ERROR("Could not initialize ptrace address space");
 613     return;
 614   }
 615
 616   void* context = _UPT_create(this->pid_);
 617   if (context == nullptr) {
 618     unw_destroy_addr_space(as);
 619     XBT_ERROR("Could not initialize ptrace context");
 620     return;
 621   }
 622
 623   unw_cursor_t cursor;
 624   if (unw_init_remote(&cursor, as, context) != 0) {
 625     _UPT_destroy(context);
 626     unw_destroy_addr_space(as);
 627     XBT_ERROR("Could not initialiez ptrace cursor");
 628     return;
 629   }
 630
 631   simgrid::mc::dumpStack(stderr, cursor);
 632
 633   _UPT_destroy(context);
 634   unw_destroy_addr_space(as);
 635 }
 636
 637 bool RemoteClient::actor_is_enabled(aid_t pid)
 638 {
 639   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 640   process()->getChannel().send(msg);
 641   char buff[MC_MESSAGE_LENGTH];
 642   ssize_t received = process()->getChannel().receive(buff, MC_MESSAGE_LENGTH, true);
 643   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
 644   return ((s_mc_message_int_t*)buff)->value;
 645 }
 646 }
 647 }