src/mc/remote/RemoteSimulation.cpp

   1 /* Copyright (c) 2014-2020. 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/RemoteSimulation.hpp"
   9
  10 #include "src/mc/mc_smx.hpp"
  11 #include "src/mc/sosp/Snapshot.hpp"
  12 #include "xbt/file.hpp"
  13 #include "xbt/log.h"
  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     "libbrotlicommon",
  51     "libbrotlidec",
  52     "libbz2",
  53     "libc",
  54     "libc++",
  55     "libcdt",
  56     "libcgraph",
  57     "libcom_err",
  58     "libcrypt",
  59     "libcrypto",
  60     "libcurl",
  61     "libcxxrt",
  62     "libdebuginfod",
  63     "libdl",
  64     "libdw",
  65     "libelf",
  66     "libevent",
  67     "libexecinfo",
  68     "libflang",
  69     "libflangrti",
  70     "libgcc_s",
  71     "libgfortran",
  72     "libgssapi_krb5",
  73     "libidn2",
  74     "libimf",
  75     "libintlc",
  76     "libirng",
  77     "libk5crypto",
  78     "libkeyutils",
  79     "libkrb5",
  80     "libkrb5support", /*odd behaviour on fedora rawhide ... remove these when fixed*/
  81     "liblber",
  82     "libldap",
  83     "liblua5.1",
  84     "liblua5.3",
  85     "liblzma",
  86     "libm",
  87     "libnghttp2",
  88     "libomp",
  89     "libpapi",
  90     "libpcre2",
  91     "libpfm",
  92     "libpgmath",
  93     "libpsl",
  94     "libpthread",
  95     "libquadmath",
  96     "libresolv",
  97     "librt",
  98     "libsasl2",
  99     "libselinux",
 100     "libssh",
 101     "libssh2",
 102     "libssl",
 103     "libstdc++",
 104     "libsvml",
 105     "libtsan",  /* gcc sanitizers */
 106     "libubsan", /* gcc sanitizers */
 107     "libunistring",
 108     "libunwind",
 109     "libunwind-ptrace",
 110     "libunwind-x86",
 111     "libunwind-x86_64",
 112     "libz",
 113     "libzstd"};
 114
 115 static bool is_simgrid_lib(const std::string& libname)
 116 {
 117   return libname == "libsimgrid";
 118 }
 119
 120 static bool is_filtered_lib(const std::string& libname)
 121 {
 122   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
 123 }
 124
 125 static std::string get_lib_name(const std::string& pathname)
 126 {
 127   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
 128   std::string libname;
 129
 130   size_t pos = map_basename.rfind(".so");
 131   if (pos != std::string::npos) {
 132     // strip the extension (matching regex "\.so.*$")
 133     libname.assign(map_basename, 0, pos);
 134
 135     // strip the version suffix (matching regex "-[.0-9-]*$")
 136     while (true) {
 137       pos = libname.rfind('-');
 138       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
 139         break;
 140       libname.erase(pos);
 141     }
 142   }
 143
 144   return libname;
 145 }
 146
 147 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 148 {
 149   char* buffer       = (char*)buf;
 150   ssize_t real_count = count;
 151   while (count) {
 152     ssize_t res = pread(fd, buffer, count, offset);
 153     if (res > 0) {
 154       count -= res;
 155       buffer += res;
 156       offset += res;
 157     } else if (res == 0)
 158       return -1;
 159     else if (errno != EINTR) {
 160       perror("pread_whole");
 161       return -1;
 162     }
 163   }
 164   return real_count;
 165 }
 166
 167 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 168 {
 169   const char* buffer = (const char*)buf;
 170   ssize_t real_count = count;
 171   while (count) {
 172     ssize_t res = pwrite(fd, buffer, count, offset);
 173     if (res > 0) {
 174       count -= res;
 175       buffer += res;
 176       offset += res;
 177     } else if (res == 0)
 178       return -1;
 179     else if (errno != EINTR)
 180       return -1;
 181   }
 182   return real_count;
 183 }
 184
 185 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 186 static const void* zero_buffer;
 187 static const size_t zero_buffer_size = 10 * 4096;
 188
 189 static void zero_buffer_init()
 190 {
 191   int fd = open("/dev/zero", O_RDONLY);
 192   if (fd < 0)
 193     xbt_die("Could not open /dev/zero");
 194   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 195   if (zero_buffer == MAP_FAILED)
 196     xbt_die("Could not map the zero buffer");
 197   close(fd);
 198 }
 199
 200 int open_vm(pid_t pid, int flags)
 201 {
 202   const size_t buffer_size = 30;
 203   char buffer[buffer_size];
 204   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 205   if (res < 0 || (size_t)res >= buffer_size) {
 206     errno = ENAMETOOLONG;
 207     return -1;
 208   }
 209   return open(buffer, flags);
 210 }
 211
 212 // ***** RemoteSimulation
 213
 214 RemoteSimulation::RemoteSimulation(pid_t pid) : AddressSpace(this), pid_(pid), running_(true) {}
 215
 216 void RemoteSimulation::init()
 217 {
 218   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 219   this->init_memory_map_info();
 220
 221   int fd = open_vm(this->pid_, O_RDWR);
 222   xbt_assert(fd >= 0, "Could not open file for process virtual address space");
 223   this->memory_file = fd;
 224
 225   // Read std_heap (is a struct mdesc*):
 226   const simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 227   xbt_assert(std_heap_var, "No heap information in the target process");
 228   xbt_assert(std_heap_var->address, "No constant address for this variable");
 229   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address));
 230
 231   this->smx_actors_infos.clear();
 232   this->smx_dead_actors_infos.clear();
 233   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 234   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 235   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 236 }
 237
 238 RemoteSimulation::~RemoteSimulation()
 239 {
 240   if (this->memory_file >= 0)
 241     close(this->memory_file);
 242
 243   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 244     if (this->unw_underlying_addr_space)
 245       unw_destroy_addr_space(this->unw_underlying_addr_space);
 246     if (this->unw_underlying_context)
 247       _UPT_destroy(this->unw_underlying_context);
 248   }
 249
 250   unw_destroy_addr_space(this->unw_addr_space);
 251 }
 252
 253 /** Refresh the information about the process
 254  *
 255  *  Do not use directly, this is used by the getters when appropriate
 256  *  in order to have fresh data.
 257  */
 258 void RemoteSimulation::refresh_heap()
 259 {
 260   // Read/dereference/refresh the std_heap pointer:
 261   if (not this->heap)
 262     this->heap.reset(new s_xbt_mheap_t());
 263   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address));
 264   this->cache_flags_ |= RemoteSimulation::cache_heap;
 265 }
 266
 267 /** Refresh the information about the process
 268  *
 269  *  Do not use directly, this is used by the getters when appropriate
 270  *  in order to have fresh data.
 271  * */
 272 void RemoteSimulation::refresh_malloc_info()
 273 {
 274   // Refresh process->heapinfo:
 275   if (this->cache_flags_ & RemoteSimulation::cache_malloc)
 276     return;
 277   size_t count = this->heap->heaplimit + 1;
 278   if (this->heap_info.size() < count)
 279     this->heap_info.resize(count);
 280   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo));
 281   this->cache_flags_ |= RemoteSimulation::cache_malloc;
 282 }
 283
 284 /** @brief Finds the range of the different memory segments and binary paths */
 285 void RemoteSimulation::init_memory_map_info()
 286 {
 287   XBT_DEBUG("Get debug information ...");
 288   this->maestro_stack_start_ = nullptr;
 289   this->maestro_stack_end_   = nullptr;
 290   this->object_infos.resize(0);
 291   this->binary_info     = nullptr;
 292   this->libsimgrid_info = nullptr;
 293
 294   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 295
 296   const char* current_name = nullptr;
 297
 298   this->object_infos.clear();
 299
 300   for (size_t i = 0; i < maps.size(); i++) {
 301     simgrid::xbt::VmMap const& reg = maps[i];
 302     const char* pathname           = maps[i].pathname.c_str();
 303
 304     // Nothing to do
 305     if (maps[i].pathname.empty()) {
 306       current_name = nullptr;
 307       continue;
 308     }
 309
 310     // [stack], [vvar], [vsyscall], [vdso] ...
 311     if (pathname[0] == '[') {
 312       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 313         this->maestro_stack_start_ = remote(reg.start_addr);
 314         this->maestro_stack_end_   = remote(reg.end_addr);
 315       }
 316       current_name = nullptr;
 317       continue;
 318     }
 319
 320     if (current_name && strcmp(current_name, pathname) == 0)
 321       continue;
 322
 323     current_name = pathname;
 324     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 325       continue;
 326
 327     const bool is_executable = not i;
 328     std::string libname;
 329     if (not is_executable) {
 330       libname = get_lib_name(pathname);
 331       if (is_filtered_lib(libname)) {
 332         continue;
 333       }
 334     }
 335
 336     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 337         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 338     this->object_infos.push_back(info);
 339     if (is_executable)
 340       this->binary_info = info;
 341     else if (is_simgrid_lib(libname))
 342       this->libsimgrid_info = info;
 343   }
 344
 345   // Resolve time (including across different objects):
 346   for (auto const& object_info : this->object_infos)
 347     postProcessObjectInformation(this, object_info.get());
 348
 349   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 350   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 351
 352   XBT_DEBUG("Get debug information done !");
 353 }
 354
 355 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteSimulation::find_object_info(RemotePtr<void> addr) const
 356 {
 357   for (auto const& object_info : this->object_infos)
 358     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 359       return object_info;
 360   return nullptr;
 361 }
 362
 363 std::shared_ptr<ObjectInformation> RemoteSimulation::find_object_info_exec(RemotePtr<void> addr) const
 364 {
 365   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 366     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
 367       return info;
 368   return nullptr;
 369 }
 370
 371 std::shared_ptr<ObjectInformation> RemoteSimulation::find_object_info_rw(RemotePtr<void> addr) const
 372 {
 373   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 374     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
 375       return info;
 376   return nullptr;
 377 }
 378
 379 simgrid::mc::Frame* RemoteSimulation::find_function(RemotePtr<void> ip) const
 380 {
 381   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 382   return info ? info->find_function((void*)ip.address()) : nullptr;
 383 }
 384
 385 /** Find (one occurrence of) the named variable definition
 386  */
 387 const simgrid::mc::Variable* RemoteSimulation::find_variable(const char* name) const
 388 {
 389   // First lookup the variable in the executable shared object.
 390   // A global variable used directly by the executable code from a library
 391   // is reinstantiated in the executable memory .data/.bss.
 392   // We need to look up the variable in the executable first.
 393   if (this->binary_info) {
 394     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 395     const simgrid::mc::Variable* var                            = info->find_variable(name);
 396     if (var)
 397       return var;
 398   }
 399
 400   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 401     const simgrid::mc::Variable* var = info->find_variable(name);
 402     if (var)
 403       return var;
 404   }
 405
 406   return nullptr;
 407 }
 408
 409 void RemoteSimulation::read_variable(const char* name, void* target, size_t size) const
 410 {
 411   const simgrid::mc::Variable* var = this->find_variable(name);
 412   xbt_assert(var, "Variable %s not found", name);
 413   xbt_assert(var->address, "No simple location for this variable");
 414   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 415   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 416              size, (size_t)var->type->full_type->byte_size);
 417   this->read_bytes(target, size, remote(var->address));
 418 }
 419
 420 std::string RemoteSimulation::read_string(RemotePtr<char> address) const
 421 {
 422   if (not address)
 423     return {};
 424
 425   std::vector<char> res(128);
 426   off_t off = 0;
 427
 428   while (true) {
 429     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 430     if (c == -1 && errno == EINTR)
 431       continue;
 432     xbt_assert(c > 0, "Could not read string from remote process");
 433
 434     const void* p = memchr(res.data() + off, '\0', c);
 435     if (p)
 436       return std::string(res.data());
 437
 438     off += c;
 439     if (off == (off_t)res.size())
 440       res.resize(res.size() * 2);
 441   }
 442 }
 443
 444 void* RemoteSimulation::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address,
 445                                    ReadOptions /*options*/) const
 446 {
 447   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 448     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 449   return buffer;
 450 }
 451
 452 /** Write data to a process memory
 453  *
 454  *  @param buffer   local memory address (source)
 455  *  @param len      data size
 456  *  @param address  target process memory address (target)
 457  */
 458 void RemoteSimulation::write_bytes(const void* buffer, size_t len, RemotePtr<void> address) const
 459 {
 460   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 461     xbt_die("Write to process %lli failed", (long long)this->pid_);
 462 }
 463
 464 void RemoteSimulation::clear_bytes(RemotePtr<void> address, size_t len) const
 465 {
 466   pthread_once(&zero_buffer_flag, zero_buffer_init);
 467   while (len) {
 468     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 469     this->write_bytes(zero_buffer, s, address);
 470     address = remote((char*)address.address() + s);
 471     len -= s;
 472   }
 473 }
 474
 475 void RemoteSimulation::ignore_region(std::uint64_t addr, std::size_t size)
 476 {
 477   IgnoredRegion region;
 478   region.addr = addr;
 479   region.size = size;
 480
 481   if (ignored_regions_.empty()) {
 482     ignored_regions_.push_back(region);
 483     return;
 484   }
 485
 486   unsigned int cursor                 = 0;
 487   const IgnoredRegion* current_region = nullptr;
 488
 489   int start = 0;
 490   int end   = ignored_regions_.size() - 1;
 491   while (start <= end) {
 492     cursor         = (start + end) / 2;
 493     current_region = &ignored_regions_[cursor];
 494     if (current_region->addr == addr) {
 495       if (current_region->size == size)
 496         return;
 497       else if (current_region->size < size)
 498         start = cursor + 1;
 499       else
 500         end = cursor - 1;
 501     } else if (current_region->addr < addr)
 502       start = cursor + 1;
 503     else
 504       end = cursor - 1;
 505   }
 506
 507   std::size_t position;
 508   if (current_region->addr == addr) {
 509     if (current_region->size < size)
 510       position = cursor + 1;
 511     else
 512       position = cursor;
 513   } else if (current_region->addr < addr)
 514     position = cursor + 1;
 515   else
 516     position = cursor;
 517   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 518 }
 519
 520 void RemoteSimulation::ignore_heap(IgnoredHeapRegion const& region)
 521 {
 522   if (ignored_heap_.empty()) {
 523     ignored_heap_.push_back(std::move(region));
 524     return;
 525   }
 526
 527   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 528
 529   size_type start = 0;
 530   size_type end   = ignored_heap_.size() - 1;
 531
 532   // Binary search the position of insertion:
 533   size_type cursor;
 534   while (start <= end) {
 535     cursor                     = start + (end - start) / 2;
 536     auto const& current_region = ignored_heap_[cursor];
 537     if (current_region.address == region.address)
 538       return;
 539     else if (current_region.address < region.address)
 540       start = cursor + 1;
 541     else if (cursor != 0)
 542       end = cursor - 1;
 543     // Avoid underflow:
 544     else
 545       break;
 546   }
 547
 548   // Insert it mc_heap_ignore_region_t:
 549   if (ignored_heap_[cursor].address < region.address)
 550     ++cursor;
 551   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 552 }
 553
 554 void RemoteSimulation::unignore_heap(void* address, size_t size)
 555 {
 556   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 557
 558   size_type start = 0;
 559   size_type end   = ignored_heap_.size() - 1;
 560
 561   // Binary search:
 562   size_type cursor;
 563   while (start <= end) {
 564     cursor             = (start + end) / 2;
 565     auto const& region = ignored_heap_[cursor];
 566     if (region.address < address)
 567       start = cursor + 1;
 568     else if ((char*)region.address <= ((char*)address + size)) {
 569       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 570       return;
 571     } else if (cursor != 0)
 572       end = cursor - 1;
 573     // Avoid underflow:
 574     else
 575       break;
 576   }
 577 }
 578
 579 void RemoteSimulation::ignore_local_variable(const char* var_name, const char* frame_name) const
 580 {
 581   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 582     frame_name = nullptr;
 583   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 584     info->remove_local_variable(var_name, frame_name);
 585 }
 586
 587 std::vector<simgrid::mc::ActorInformation>& RemoteSimulation::actors()
 588 {
 589   this->refresh_simix();
 590   return smx_actors_infos;
 591 }
 592
 593 std::vector<simgrid::mc::ActorInformation>& RemoteSimulation::dead_actors()
 594 {
 595   this->refresh_simix();
 596   return smx_dead_actors_infos;
 597 }
 598
 599 void RemoteSimulation::dump_stack() const
 600 {
 601   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 602   if (as == nullptr) {
 603     XBT_ERROR("Could not initialize ptrace address space");
 604     return;
 605   }
 606
 607   void* context = _UPT_create(this->pid_);
 608   if (context == nullptr) {
 609     unw_destroy_addr_space(as);
 610     XBT_ERROR("Could not initialize ptrace context");
 611     return;
 612   }
 613
 614   unw_cursor_t cursor;
 615   if (unw_init_remote(&cursor, as, context) != 0) {
 616     _UPT_destroy(context);
 617     unw_destroy_addr_space(as);
 618     XBT_ERROR("Could not initialiez ptrace cursor");
 619     return;
 620   }
 621
 622   simgrid::mc::dumpStack(stderr, &cursor);
 623
 624   _UPT_destroy(context);
 625   unw_destroy_addr_space(as);
 626 }
 627 } // namespace mc
 628 } // namespace simgrid