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 #include <memory>
  20
  21 using simgrid::mc::remote;
  22
  23 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(mc_process, mc, "MC process information");
  24
  25 namespace simgrid {
  26 namespace mc {
  27
  28 // ***** Helper stuff
  29
  30 // List of library which memory segments are not considered:
  31 static const std::vector<std::string> filtered_libraries = {
  32 #ifdef __linux__
  33     "ld",
  34 #elif defined __FreeBSD__
  35     "ld-elf",
  36     "ld-elf32",
  37     "libkvm",      /* kernel data access library */
  38     "libprocstat", /* process and file information retrieval */
  39     "libthr",      /* thread library */
  40     "libutil",
  41 #endif
  42     "libargp", /* workarounds for glibc-less systems */
  43     "libasan", /* gcc sanitizers */
  44     "libboost_chrono",
  45     "libboost_context",
  46     "libboost_context-mt",
  47     "libboost_stacktrace_addr2line",
  48     "libboost_stacktrace_backtrace",
  49     "libboost_system",
  50     "libboost_thread",
  51     "libboost_timer",
  52     "libbrotlicommon",
  53     "libbrotlidec",
  54     "libbz2",
  55     "libc",
  56     "libc++",
  57     "libcdt",
  58     "libcgraph",
  59     "libcom_err",
  60     "libcrypt",
  61     "libcrypto",
  62     "libcurl",
  63     "libcxxrt",
  64     "libdebuginfod",
  65     "libdl",
  66     "libdw",
  67     "libelf",
  68     "libevent",
  69     "libexecinfo",
  70     "libffi",
  71     "libflang",
  72     "libflangrti",
  73     "libgcc_s",
  74     "libgfortran",
  75     "libgssapi_krb5",
  76     "libidn2",
  77     "libimf",
  78     "libintlc",
  79     "libirng",
  80     "libk5crypto",
  81     "libkeyutils",
  82     "libkrb5",
  83     "libkrb5support", /*odd behaviour on fedora rawhide ... remove these when fixed*/
  84     "liblber",
  85     "libldap",
  86     "liblua5.1",
  87     "liblua5.3",
  88     "liblzma",
  89     "libm",
  90     "libmd",
  91     "libnghttp2",
  92     "libomp",
  93     "libpapi",
  94     "libpcre2",
  95     "libpfm",
  96     "libpgmath",
  97     "libpsl",
  98     "libpthread",
  99     "libquadmath",
 100     "libresolv",
 101     "librt",
 102     "libsasl2",
 103     "libselinux",
 104     "libssh",
 105     "libssh2",
 106     "libssl",
 107     "libstdc++",
 108     "libsvml",
 109     "libtsan",  /* gcc sanitizers */
 110     "libubsan", /* gcc sanitizers */
 111     "libunistring",
 112     "libunwind",
 113     "libunwind-ptrace",
 114     "libunwind-x86",
 115     "libunwind-x86_64",
 116     "libz",
 117     "libzstd"};
 118
 119 static bool is_simgrid_lib(const std::string& libname)
 120 {
 121   return libname == "libsimgrid";
 122 }
 123
 124 static bool is_filtered_lib(const std::string& libname)
 125 {
 126   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
 127 }
 128
 129 static std::string get_lib_name(const std::string& pathname)
 130 {
 131   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
 132   std::string libname;
 133
 134   size_t pos = map_basename.rfind(".so");
 135   if (pos != std::string::npos) {
 136     // strip the extension (matching regex "\.so.*$")
 137     libname.assign(map_basename, 0, pos);
 138
 139     // strip the version suffix (matching regex "-[.0-9-]*$")
 140     while (true) {
 141       pos = libname.rfind('-');
 142       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
 143         break;
 144       libname.erase(pos);
 145     }
 146   }
 147
 148   return libname;
 149 }
 150
 151 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 152 {
 153   auto* buffer       = static_cast<char*>(buf);
 154   ssize_t real_count = count;
 155   while (count) {
 156     ssize_t res = pread(fd, buffer, count, offset);
 157     if (res > 0) {
 158       count -= res;
 159       buffer += res;
 160       offset += res;
 161     } else if (res == 0)
 162       return -1;
 163     else if (errno != EINTR) {
 164       perror("pread_whole");
 165       return -1;
 166     }
 167   }
 168   return real_count;
 169 }
 170
 171 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 172 {
 173   const auto* buffer = static_cast<const char*>(buf);
 174   ssize_t real_count = count;
 175   while (count) {
 176     ssize_t res = pwrite(fd, buffer, count, offset);
 177     if (res > 0) {
 178       count -= res;
 179       buffer += res;
 180       offset += res;
 181     } else if (res == 0)
 182       return -1;
 183     else if (errno != EINTR)
 184       return -1;
 185   }
 186   return real_count;
 187 }
 188
 189 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 190 static const void* zero_buffer;
 191 static const size_t zero_buffer_size = 10 * 4096;
 192
 193 static void zero_buffer_init()
 194 {
 195   int fd = open("/dev/zero", O_RDONLY);
 196   if (fd < 0)
 197     xbt_die("Could not open /dev/zero");
 198   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 199   if (zero_buffer == MAP_FAILED)
 200     xbt_die("Could not map the zero buffer");
 201   close(fd);
 202 }
 203
 204 int open_vm(pid_t pid, int flags)
 205 {
 206   const size_t buffer_size = 30;
 207   char buffer[buffer_size];
 208   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 209   if (res < 0 || (size_t)res >= buffer_size) {
 210     errno = ENAMETOOLONG;
 211     return -1;
 212   }
 213   return open(buffer, flags);
 214 }
 215
 216 // ***** RemoteSimulation
 217
 218 RemoteSimulation::RemoteSimulation(pid_t pid) : AddressSpace(this), pid_(pid), running_(true) {}
 219
 220 void RemoteSimulation::init()
 221 {
 222   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 223   this->init_memory_map_info();
 224
 225   int fd = open_vm(this->pid_, O_RDWR);
 226   xbt_assert(fd >= 0, "Could not open file for process virtual address space");
 227   this->memory_file = fd;
 228
 229   // Read std_heap (is a struct mdesc*):
 230   const simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 231   xbt_assert(std_heap_var, "No heap information in the target process");
 232   xbt_assert(std_heap_var->address, "No constant address for this variable");
 233   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address));
 234
 235   this->smx_actors_infos.clear();
 236   this->smx_dead_actors_infos.clear();
 237   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 238   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 239   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 240 }
 241
 242 RemoteSimulation::~RemoteSimulation()
 243 {
 244   if (this->memory_file >= 0)
 245     close(this->memory_file);
 246
 247   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 248     if (this->unw_underlying_addr_space)
 249       unw_destroy_addr_space(this->unw_underlying_addr_space);
 250     if (this->unw_underlying_context)
 251       _UPT_destroy(this->unw_underlying_context);
 252   }
 253
 254   unw_destroy_addr_space(this->unw_addr_space);
 255 }
 256
 257 /** Refresh the information about the process
 258  *
 259  *  Do not use directly, this is used by the getters when appropriate
 260  *  in order to have fresh data.
 261  */
 262 void RemoteSimulation::refresh_heap()
 263 {
 264   // Read/dereference/refresh the std_heap pointer:
 265   if (not this->heap)
 266     this->heap = std::make_unique<s_xbt_mheap_t>();
 267   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address));
 268   this->cache_flags_ |= RemoteSimulation::cache_heap;
 269 }
 270
 271 /** Refresh the information about the process
 272  *
 273  *  Do not use directly, this is used by the getters when appropriate
 274  *  in order to have fresh data.
 275  * */
 276 void RemoteSimulation::refresh_malloc_info()
 277 {
 278   // Refresh process->heapinfo:
 279   if (this->cache_flags_ & RemoteSimulation::cache_malloc)
 280     return;
 281   size_t count = this->heap->heaplimit + 1;
 282   if (this->heap_info.size() < count)
 283     this->heap_info.resize(count);
 284   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo));
 285   this->cache_flags_ |= RemoteSimulation::cache_malloc;
 286 }
 287
 288 /** @brief Finds the range of the different memory segments and binary paths */
 289 void RemoteSimulation::init_memory_map_info()
 290 {
 291   XBT_DEBUG("Get debug information ...");
 292   this->maestro_stack_start_ = nullptr;
 293   this->maestro_stack_end_   = nullptr;
 294   this->object_infos.resize(0);
 295   this->binary_info     = nullptr;
 296   this->libsimgrid_info = nullptr;
 297
 298   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 299
 300   const char* current_name = nullptr;
 301
 302   this->object_infos.clear();
 303
 304   for (size_t i = 0; i < maps.size(); i++) {
 305     simgrid::xbt::VmMap const& reg = maps[i];
 306     const char* pathname           = maps[i].pathname.c_str();
 307
 308     // Nothing to do
 309     if (maps[i].pathname.empty()) {
 310       current_name = nullptr;
 311       continue;
 312     }
 313
 314     // [stack], [vvar], [vsyscall], [vdso] ...
 315     if (pathname[0] == '[') {
 316       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 317         this->maestro_stack_start_ = remote(reg.start_addr);
 318         this->maestro_stack_end_   = remote(reg.end_addr);
 319       }
 320       current_name = nullptr;
 321       continue;
 322     }
 323
 324     if (current_name && strcmp(current_name, pathname) == 0)
 325       continue;
 326
 327     current_name = pathname;
 328     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 329       continue;
 330
 331     const bool is_executable = not i;
 332     std::string libname;
 333     if (not is_executable) {
 334       libname = get_lib_name(pathname);
 335       if (is_filtered_lib(libname)) {
 336         continue;
 337       }
 338     }
 339
 340     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 341         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 342     this->object_infos.push_back(info);
 343     if (is_executable)
 344       this->binary_info = info;
 345     else if (is_simgrid_lib(libname))
 346       this->libsimgrid_info = info;
 347   }
 348
 349   // Resolve time (including across different objects):
 350   for (auto const& object_info : this->object_infos)
 351     postProcessObjectInformation(this, object_info.get());
 352
 353   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 354   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 355
 356   XBT_DEBUG("Get debug information done !");
 357 }
 358
 359 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteSimulation::find_object_info(RemotePtr<void> addr) const
 360 {
 361   for (auto const& object_info : this->object_infos)
 362     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 363       return object_info;
 364   return nullptr;
 365 }
 366
 367 std::shared_ptr<ObjectInformation> RemoteSimulation::find_object_info_exec(RemotePtr<void> addr) const
 368 {
 369   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 370     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
 371       return info;
 372   return nullptr;
 373 }
 374
 375 std::shared_ptr<ObjectInformation> RemoteSimulation::find_object_info_rw(RemotePtr<void> addr) const
 376 {
 377   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 378     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
 379       return info;
 380   return nullptr;
 381 }
 382
 383 simgrid::mc::Frame* RemoteSimulation::find_function(RemotePtr<void> ip) const
 384 {
 385   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 386   return info ? info->find_function((void*)ip.address()) : nullptr;
 387 }
 388
 389 /** Find (one occurrence of) the named variable definition
 390  */
 391 const simgrid::mc::Variable* RemoteSimulation::find_variable(const char* name) const
 392 {
 393   // First lookup the variable in the executable shared object.
 394   // A global variable used directly by the executable code from a library
 395   // is reinstantiated in the executable memory .data/.bss.
 396   // We need to look up the variable in the executable first.
 397   if (this->binary_info) {
 398     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 399     const simgrid::mc::Variable* var                            = info->find_variable(name);
 400     if (var)
 401       return var;
 402   }
 403
 404   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 405     const simgrid::mc::Variable* var = info->find_variable(name);
 406     if (var)
 407       return var;
 408   }
 409
 410   return nullptr;
 411 }
 412
 413 void RemoteSimulation::read_variable(const char* name, void* target, size_t size) const
 414 {
 415   const simgrid::mc::Variable* var = this->find_variable(name);
 416   xbt_assert(var, "Variable %s not found", name);
 417   xbt_assert(var->address, "No simple location for this variable");
 418   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 419   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 420              size, (size_t)var->type->full_type->byte_size);
 421   this->read_bytes(target, size, remote(var->address));
 422 }
 423
 424 std::string RemoteSimulation::read_string(RemotePtr<char> address) const
 425 {
 426   if (not address)
 427     return {};
 428
 429   std::vector<char> res(128);
 430   off_t off = 0;
 431
 432   while (true) {
 433     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 434     if (c == -1 && errno == EINTR)
 435       continue;
 436     xbt_assert(c > 0, "Could not read string from remote process");
 437
 438     const void* p = memchr(res.data() + off, '\0', c);
 439     if (p)
 440       return std::string(res.data());
 441
 442     off += c;
 443     if (off == (off_t)res.size())
 444       res.resize(res.size() * 2);
 445   }
 446 }
 447
 448 void* RemoteSimulation::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address,
 449                                    ReadOptions /*options*/) const
 450 {
 451   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 452     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 453   return buffer;
 454 }
 455
 456 /** Write data to a process memory
 457  *
 458  *  @param buffer   local memory address (source)
 459  *  @param len      data size
 460  *  @param address  target process memory address (target)
 461  */
 462 void RemoteSimulation::write_bytes(const void* buffer, size_t len, RemotePtr<void> address) const
 463 {
 464   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 465     xbt_die("Write to process %lli failed", (long long)this->pid_);
 466 }
 467
 468 void RemoteSimulation::clear_bytes(RemotePtr<void> address, size_t len) const
 469 {
 470   pthread_once(&zero_buffer_flag, zero_buffer_init);
 471   while (len) {
 472     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 473     this->write_bytes(zero_buffer, s, address);
 474     address = remote((char*)address.address() + s);
 475     len -= s;
 476   }
 477 }
 478
 479 void RemoteSimulation::ignore_region(std::uint64_t addr, std::size_t size)
 480 {
 481   IgnoredRegion region;
 482   region.addr = addr;
 483   region.size = size;
 484
 485   if (ignored_regions_.empty()) {
 486     ignored_regions_.push_back(region);
 487     return;
 488   }
 489
 490   unsigned int cursor                 = 0;
 491   const IgnoredRegion* current_region = nullptr;
 492
 493   int start = 0;
 494   int end   = ignored_regions_.size() - 1;
 495   while (start <= end) {
 496     cursor         = (start + end) / 2;
 497     current_region = &ignored_regions_[cursor];
 498     if (current_region->addr == addr) {
 499       if (current_region->size == size)
 500         return;
 501       else if (current_region->size < size)
 502         start = cursor + 1;
 503       else
 504         end = cursor - 1;
 505     } else if (current_region->addr < addr)
 506       start = cursor + 1;
 507     else
 508       end = cursor - 1;
 509   }
 510
 511   std::size_t position;
 512   if (current_region->addr == addr) {
 513     if (current_region->size < size)
 514       position = cursor + 1;
 515     else
 516       position = cursor;
 517   } else if (current_region->addr < addr)
 518     position = cursor + 1;
 519   else
 520     position = cursor;
 521   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 522 }
 523
 524 void RemoteSimulation::ignore_heap(IgnoredHeapRegion const& region)
 525 {
 526   if (ignored_heap_.empty()) {
 527     ignored_heap_.push_back(std::move(region));
 528     return;
 529   }
 530
 531   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 532
 533   size_type start = 0;
 534   size_type end   = ignored_heap_.size() - 1;
 535
 536   // Binary search the position of insertion:
 537   size_type cursor;
 538   while (start <= end) {
 539     cursor                     = start + (end - start) / 2;
 540     auto const& current_region = ignored_heap_[cursor];
 541     if (current_region.address == region.address)
 542       return;
 543     else if (current_region.address < region.address)
 544       start = cursor + 1;
 545     else if (cursor != 0)
 546       end = cursor - 1;
 547     // Avoid underflow:
 548     else
 549       break;
 550   }
 551
 552   // Insert it mc_heap_ignore_region_t:
 553   if (ignored_heap_[cursor].address < region.address)
 554     ++cursor;
 555   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 556 }
 557
 558 void RemoteSimulation::unignore_heap(void* address, size_t size)
 559 {
 560   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 561
 562   size_type start = 0;
 563   size_type end   = ignored_heap_.size() - 1;
 564
 565   // Binary search:
 566   size_type cursor;
 567   while (start <= end) {
 568     cursor             = (start + end) / 2;
 569     auto const& region = ignored_heap_[cursor];
 570     if (region.address < address)
 571       start = cursor + 1;
 572     else if ((char*)region.address <= ((char*)address + size)) {
 573       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 574       return;
 575     } else if (cursor != 0)
 576       end = cursor - 1;
 577     // Avoid underflow:
 578     else
 579       break;
 580   }
 581 }
 582
 583 void RemoteSimulation::ignore_local_variable(const char* var_name, const char* frame_name) const
 584 {
 585   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 586     frame_name = nullptr;
 587   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 588     info->remove_local_variable(var_name, frame_name);
 589 }
 590
 591 std::vector<simgrid::mc::ActorInformation>& RemoteSimulation::actors()
 592 {
 593   this->refresh_simix();
 594   return smx_actors_infos;
 595 }
 596
 597 std::vector<simgrid::mc::ActorInformation>& RemoteSimulation::dead_actors()
 598 {
 599   this->refresh_simix();
 600   return smx_dead_actors_infos;
 601 }
 602
 603 void RemoteSimulation::dump_stack() const
 604 {
 605   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 606   if (as == nullptr) {
 607     XBT_ERROR("Could not initialize ptrace address space");
 608     return;
 609   }
 610
 611   void* context = _UPT_create(this->pid_);
 612   if (context == nullptr) {
 613     unw_destroy_addr_space(as);
 614     XBT_ERROR("Could not initialize ptrace context");
 615     return;
 616   }
 617
 618   unw_cursor_t cursor;
 619   if (unw_init_remote(&cursor, as, context) != 0) {
 620     _UPT_destroy(context);
 621     unw_destroy_addr_space(as);
 622     XBT_ERROR("Could not initialiez ptrace cursor");
 623     return;
 624   }
 625
 626   simgrid::mc::dumpStack(stderr, &cursor);
 627
 628   _UPT_destroy(context);
 629   unw_destroy_addr_space(as);
 630 }
 631 } // namespace mc
 632 } // namespace simgrid