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