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