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_addr2line",
  46     "libboost_stacktrace_backtrace",
  47     "libboost_system",
  48     "libboost_thread",
  49     "libboost_timer",
  50     "libbz2",
  51     "libc",
  52     "libc++",
  53     "libcdt",
  54     "libcgraph",
  55     "libcrypto",
  56     "libcxxrt",
  57     "libdl",
  58     "libdw",
  59     "libelf",
  60     "libevent",
  61     "libexecinfo",
  62     "libflang",
  63     "libflangrti",
  64     "libgcc_s",
  65     "libgfortran",
  66     "libimf",
  67     "libintlc",
  68     "libirng",
  69     "liblua5.1",
  70     "liblua5.3",
  71     "liblzma",
  72     "libm",
  73     "libomp",
  74     "libpapi",
  75     "libpfm",
  76     "libpgmath",
  77     "libpthread",
  78     "libquadmath",
  79     "librt",
  80     "libstdc++",
  81     "libsvml",
  82     "libtsan",  /* gcc sanitizers */
  83     "libubsan", /* gcc sanitizers */
  84     "libunwind",
  85     "libunwind-ptrace",
  86     "libunwind-x86",
  87     "libunwind-x86_64",
  88     "libz"};
  89
  90 static bool is_simgrid_lib(const std::string& libname)
  91 {
  92   return libname == "libsimgrid";
  93 }
  94
  95 static bool is_filtered_lib(const std::string& libname)
  96 {
  97   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
  98 }
  99
 100 static std::string get_lib_name(const std::string& pathname)
 101 {
 102   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
 103   std::string libname;
 104
 105   size_t pos = map_basename.rfind(".so");
 106   if (pos != std::string::npos) {
 107     // strip the extension (matching regex "\.so.*$")
 108     libname.assign(map_basename, 0, pos);
 109
 110     // strip the version suffix (matching regex "-[.0-9-]*$")
 111     while (true) {
 112       pos = libname.rfind('-');
 113       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
 114         break;
 115       libname.erase(pos);
 116     }
 117   }
 118
 119   return libname;
 120 }
 121
 122 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 123 {
 124   char* buffer       = (char*)buf;
 125   ssize_t real_count = count;
 126   while (count) {
 127     ssize_t res = pread(fd, buffer, count, offset);
 128     if (res > 0) {
 129       count -= res;
 130       buffer += res;
 131       offset += res;
 132     } else if (res == 0)
 133       return -1;
 134     else if (errno != EINTR) {
 135       perror("pread_whole");
 136       return -1;
 137     }
 138   }
 139   return real_count;
 140 }
 141
 142 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 143 {
 144   const char* buffer = (const char*)buf;
 145   ssize_t real_count = count;
 146   while (count) {
 147     ssize_t res = pwrite(fd, buffer, count, offset);
 148     if (res > 0) {
 149       count -= res;
 150       buffer += res;
 151       offset += res;
 152     } else if (res == 0)
 153       return -1;
 154     else if (errno != EINTR)
 155       return -1;
 156   }
 157   return real_count;
 158 }
 159
 160 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 161 static const void* zero_buffer;
 162 static const size_t zero_buffer_size = 10 * 4096;
 163
 164 static void zero_buffer_init()
 165 {
 166   int fd = open("/dev/zero", O_RDONLY);
 167   if (fd < 0)
 168     xbt_die("Could not open /dev/zero");
 169   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 170   if (zero_buffer == MAP_FAILED)
 171     xbt_die("Could not map the zero buffer");
 172   close(fd);
 173 }
 174
 175 int open_vm(pid_t pid, int flags)
 176 {
 177   const size_t buffer_size = 30;
 178   char buffer[buffer_size];
 179   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 180   if (res < 0 || (size_t)res >= buffer_size) {
 181     errno = ENAMETOOLONG;
 182     return -1;
 183   }
 184   return open(buffer, flags);
 185 }
 186
 187 // ***** Process
 188
 189 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
 190 {
 191 }
 192
 193 void RemoteClient::init()
 194 {
 195   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 196   this->init_memory_map_info();
 197
 198   int fd = open_vm(this->pid_, O_RDWR);
 199   if (fd < 0)
 200     xbt_die("Could not open file for process virtual address space");
 201   this->memory_file = fd;
 202
 203   // Read std_heap (is a struct mdesc*):
 204   const simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 205   if (not std_heap_var)
 206     xbt_die("No heap information in the target process");
 207   if (not std_heap_var->address)
 208     xbt_die("No constant address for this variable");
 209   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address),
 210                    simgrid::mc::ProcessIndexDisabled);
 211
 212   this->smx_actors_infos.clear();
 213   this->smx_dead_actors_infos.clear();
 214   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 215   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 216   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 217 }
 218
 219 RemoteClient::~RemoteClient()
 220 {
 221   if (this->memory_file >= 0)
 222     close(this->memory_file);
 223
 224   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 225     if (this->unw_underlying_addr_space)
 226       unw_destroy_addr_space(this->unw_underlying_addr_space);
 227     if (this->unw_underlying_context)
 228       _UPT_destroy(this->unw_underlying_context);
 229   }
 230
 231   unw_destroy_addr_space(this->unw_addr_space);
 232 }
 233
 234 /** Refresh the information about the process
 235  *
 236  *  Do not use directly, this is used by the getters when appropriate
 237  *  in order to have fresh data.
 238  */
 239 void RemoteClient::refresh_heap()
 240 {
 241   // Read/dereference/refresh the std_heap pointer:
 242   if (not this->heap)
 243     this->heap.reset(new s_xbt_mheap_t());
 244   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address), simgrid::mc::ProcessIndexDisabled);
 245   this->cache_flags_ |= RemoteClient::cache_heap;
 246 }
 247
 248 /** Refresh the information about the process
 249  *
 250  *  Do not use direclty, this is used by the getters when appropriate
 251  *  in order to have fresh data.
 252  * */
 253 void RemoteClient::refresh_malloc_info()
 254 {
 255   // Refresh process->heapinfo:
 256   if (this->cache_flags_ & RemoteClient::cache_malloc)
 257     return;
 258   size_t count = this->heap->heaplimit + 1;
 259   if (this->heap_info.size() < count)
 260     this->heap_info.resize(count);
 261   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo),
 262                    simgrid::mc::ProcessIndexDisabled);
 263   this->cache_flags_ |= RemoteClient::cache_malloc;
 264 }
 265
 266 /** @brief Finds the range of the different memory segments and binary paths */
 267 void RemoteClient::init_memory_map_info()
 268 {
 269   XBT_DEBUG("Get debug information ...");
 270   this->maestro_stack_start_ = nullptr;
 271   this->maestro_stack_end_   = nullptr;
 272   this->object_infos.resize(0);
 273   this->binary_info     = nullptr;
 274   this->libsimgrid_info = nullptr;
 275
 276   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 277
 278   const char* current_name = nullptr;
 279
 280   this->object_infos.clear();
 281
 282   for (size_t i = 0; i < maps.size(); i++) {
 283     simgrid::xbt::VmMap const& reg = maps[i];
 284     const char* pathname           = maps[i].pathname.c_str();
 285
 286     // Nothing to do
 287     if (maps[i].pathname.empty()) {
 288       current_name = nullptr;
 289       continue;
 290     }
 291
 292     // [stack], [vvar], [vsyscall], [vdso] ...
 293     if (pathname[0] == '[') {
 294       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 295         this->maestro_stack_start_ = remote(reg.start_addr);
 296         this->maestro_stack_end_   = remote(reg.end_addr);
 297       }
 298       current_name = nullptr;
 299       continue;
 300     }
 301
 302     if (current_name && strcmp(current_name, pathname) == 0)
 303       continue;
 304
 305     current_name = pathname;
 306     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 307       continue;
 308
 309     const bool is_executable = not i;
 310     std::string libname;
 311     if (not is_executable) {
 312       libname = get_lib_name(pathname);
 313       if (is_filtered_lib(libname)) {
 314         continue;
 315       }
 316     }
 317
 318     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 319         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 320     this->object_infos.push_back(info);
 321     if (is_executable)
 322       this->binary_info = info;
 323     else if (is_simgrid_lib(libname))
 324       this->libsimgrid_info = info;
 325   }
 326
 327   // Resolve time (including across different objects):
 328   for (auto const& object_info : this->object_infos)
 329     postProcessObjectInformation(this, object_info.get());
 330
 331   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 332   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 333
 334   XBT_DEBUG("Get debug information done !");
 335 }
 336
 337 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
 338 {
 339   for (auto const& object_info : this->object_infos)
 340     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 341       return object_info;
 342   return nullptr;
 343 }
 344
 345 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(RemotePtr<void> addr) const
 346 {
 347   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 348     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
 349       return info;
 350   return nullptr;
 351 }
 352
 353 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(RemotePtr<void> addr) const
 354 {
 355   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 356     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
 357       return info;
 358   return nullptr;
 359 }
 360
 361 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
 362 {
 363   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 364   return info ? info->find_function((void*)ip.address()) : nullptr;
 365 }
 366
 367 /** Find (one occurrence of) the named variable definition
 368  */
 369 const simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
 370 {
 371   // First lookup the variable in the executable shared object.
 372   // A global variable used directly by the executable code from a library
 373   // is reinstanciated in the executable memory .data/.bss.
 374   // We need to look up the variable in the executable first.
 375   if (this->binary_info) {
 376     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 377     const simgrid::mc::Variable* var                            = info->find_variable(name);
 378     if (var)
 379       return var;
 380   }
 381
 382   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 383     const simgrid::mc::Variable* var = info->find_variable(name);
 384     if (var)
 385       return var;
 386   }
 387
 388   return nullptr;
 389 }
 390
 391 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
 392 {
 393   const simgrid::mc::Variable* var = this->find_variable(name);
 394   xbt_assert(var->address, "No simple location for this variable");
 395   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 396   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 397              size, (size_t)var->type->full_type->byte_size);
 398   this->read_bytes(target, size, remote(var->address));
 399 }
 400
 401 std::string RemoteClient::read_string(RemotePtr<char> address) const
 402 {
 403   if (not address)
 404     return {};
 405
 406   std::vector<char> res(128);
 407   off_t off = 0;
 408
 409   while (1) {
 410     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 411     if (c == -1) {
 412       if (errno == EINTR)
 413         continue;
 414       else
 415         xbt_die("Could not read from from remote process");
 416     }
 417     if (c == 0)
 418       xbt_die("Could not read string from remote process");
 419
 420     void* p = memchr(res.data() + off, '\0', c);
 421     if (p)
 422       return std::string(res.data());
 423
 424     off += c;
 425     if (off == (off_t)res.size())
 426       res.resize(res.size() * 2);
 427   }
 428 }
 429
 430 const void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, int process_index,
 431                                      ReadOptions /*options*/) const
 432 {
 433   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 434     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 435   return buffer;
 436 }
 437
 438 /** Write data to a process memory
 439  *
 440  *  @param buffer   local memory address (source)
 441  *  @param len      data size
 442  *  @param address  target process memory address (target)
 443  */
 444 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 445 {
 446   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 447     xbt_die("Write to process %lli failed", (long long)this->pid_);
 448 }
 449
 450 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 451 {
 452   pthread_once(&zero_buffer_flag, zero_buffer_init);
 453   while (len) {
 454     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 455     this->write_bytes(zero_buffer, s, address);
 456     address = remote((char*)address.address() + s);
 457     len -= s;
 458   }
 459 }
 460
 461 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 462 {
 463   IgnoredRegion region;
 464   region.addr = addr;
 465   region.size = size;
 466
 467   if (ignored_regions_.empty()) {
 468     ignored_regions_.push_back(region);
 469     return;
 470   }
 471
 472   unsigned int cursor           = 0;
 473   IgnoredRegion* current_region = nullptr;
 474
 475   int start = 0;
 476   int end   = ignored_regions_.size() - 1;
 477   while (start <= end) {
 478     cursor         = (start + end) / 2;
 479     current_region = &ignored_regions_[cursor];
 480     if (current_region->addr == addr) {
 481       if (current_region->size == size)
 482         return;
 483       else if (current_region->size < size)
 484         start = cursor + 1;
 485       else
 486         end = cursor - 1;
 487     } else if (current_region->addr < addr)
 488       start = cursor + 1;
 489     else
 490       end = cursor - 1;
 491   }
 492
 493   std::size_t position;
 494   if (current_region->addr == addr) {
 495     if (current_region->size < size)
 496       position = cursor + 1;
 497     else
 498       position = cursor;
 499   } else if (current_region->addr < addr)
 500     position = cursor + 1;
 501   else
 502     position = cursor;
 503   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 504 }
 505
 506 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 507 {
 508   if (ignored_heap_.empty()) {
 509     ignored_heap_.push_back(std::move(region));
 510     return;
 511   }
 512
 513   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 514
 515   size_type start = 0;
 516   size_type end   = ignored_heap_.size() - 1;
 517
 518   // Binary search the position of insertion:
 519   size_type cursor;
 520   while (start <= end) {
 521     cursor               = start + (end - start) / 2;
 522     auto& current_region = ignored_heap_[cursor];
 523     if (current_region.address == region.address)
 524       return;
 525     else if (current_region.address < region.address)
 526       start = cursor + 1;
 527     else if (cursor != 0)
 528       end = cursor - 1;
 529     // Avoid underflow:
 530     else
 531       break;
 532   }
 533
 534   // Insert it mc_heap_ignore_region_t:
 535   if (ignored_heap_[cursor].address < region.address)
 536     ++cursor;
 537   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 538 }
 539
 540 void RemoteClient::unignore_heap(void* address, size_t size)
 541 {
 542   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 543
 544   size_type start = 0;
 545   size_type end   = ignored_heap_.size() - 1;
 546
 547   // Binary search:
 548   size_type cursor;
 549   while (start <= end) {
 550     cursor       = (start + end) / 2;
 551     auto& region = ignored_heap_[cursor];
 552     if (region.address < address)
 553       start = cursor + 1;
 554     else if ((char*)region.address <= ((char*)address + size)) {
 555       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 556       return;
 557     } else if (cursor != 0)
 558       end = cursor - 1;
 559     // Avoid underflow:
 560     else
 561       break;
 562   }
 563 }
 564
 565 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 566 {
 567   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 568     frame_name = nullptr;
 569   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 570     info->remove_local_variable(var_name, frame_name);
 571 }
 572
 573 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 574 {
 575   this->refresh_simix();
 576   return smx_actors_infos;
 577 }
 578
 579 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 580 {
 581   this->refresh_simix();
 582   return smx_dead_actors_infos;
 583 }
 584
 585 void RemoteClient::dumpStack()
 586 {
 587   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 588   if (as == nullptr) {
 589     XBT_ERROR("Could not initialize ptrace address space");
 590     return;
 591   }
 592
 593   void* context = _UPT_create(this->pid_);
 594   if (context == nullptr) {
 595     unw_destroy_addr_space(as);
 596     XBT_ERROR("Could not initialize ptrace context");
 597     return;
 598   }
 599
 600   unw_cursor_t cursor;
 601   if (unw_init_remote(&cursor, as, context) != 0) {
 602     _UPT_destroy(context);
 603     unw_destroy_addr_space(as);
 604     XBT_ERROR("Could not initialiez ptrace cursor");
 605     return;
 606   }
 607
 608   simgrid::mc::dumpStack(stderr, std::move(cursor));
 609
 610   _UPT_destroy(context);
 611   unw_destroy_addr_space(as);
 612 }
 613
 614 bool RemoteClient::actor_is_enabled(aid_t pid)
 615 {
 616   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 617   process()->getChannel().send(msg);
 618   char buff[MC_MESSAGE_LENGTH];
 619   ssize_t received = process()->getChannel().receive(buff, MC_MESSAGE_LENGTH, true);
 620   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
 621   return ((s_mc_message_int_t*)buff)->value;
 622 }
 623 }
 624 }