src/mc/remote/RemoteClient.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/RemoteClient.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
 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 // ***** Process
 213
 214 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
 215 {
 216 }
 217
 218 void RemoteClient::init()
 219 {
 220   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 221   this->init_memory_map_info();
 222
 223   int fd = open_vm(this->pid_, O_RDWR);
 224   xbt_assert(fd >= 0, "Could not open file for process virtual address space");
 225   this->memory_file = fd;
 226
 227   // Read std_heap (is a struct mdesc*):
 228   const simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 229   xbt_assert(std_heap_var, "No heap information in the target process");
 230   xbt_assert(std_heap_var->address, "No constant address for this variable");
 231   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address));
 232
 233   this->smx_actors_infos.clear();
 234   this->smx_dead_actors_infos.clear();
 235   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 236   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 237   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 238 }
 239
 240 RemoteClient::~RemoteClient()
 241 {
 242   if (this->memory_file >= 0)
 243     close(this->memory_file);
 244
 245   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 246     if (this->unw_underlying_addr_space)
 247       unw_destroy_addr_space(this->unw_underlying_addr_space);
 248     if (this->unw_underlying_context)
 249       _UPT_destroy(this->unw_underlying_context);
 250   }
 251
 252   unw_destroy_addr_space(this->unw_addr_space);
 253 }
 254
 255 /** Refresh the information about the process
 256  *
 257  *  Do not use directly, this is used by the getters when appropriate
 258  *  in order to have fresh data.
 259  */
 260 void RemoteClient::refresh_heap()
 261 {
 262   // Read/dereference/refresh the std_heap pointer:
 263   if (not this->heap)
 264     this->heap.reset(new s_xbt_mheap_t());
 265   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address));
 266   this->cache_flags_ |= RemoteClient::cache_heap;
 267 }
 268
 269 /** Refresh the information about the process
 270  *
 271  *  Do not use directly, this is used by the getters when appropriate
 272  *  in order to have fresh data.
 273  * */
 274 void RemoteClient::refresh_malloc_info()
 275 {
 276   // Refresh process->heapinfo:
 277   if (this->cache_flags_ & RemoteClient::cache_malloc)
 278     return;
 279   size_t count = this->heap->heaplimit + 1;
 280   if (this->heap_info.size() < count)
 281     this->heap_info.resize(count);
 282   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo));
 283   this->cache_flags_ |= RemoteClient::cache_malloc;
 284 }
 285
 286 /** @brief Finds the range of the different memory segments and binary paths */
 287 void RemoteClient::init_memory_map_info()
 288 {
 289   XBT_DEBUG("Get debug information ...");
 290   this->maestro_stack_start_ = nullptr;
 291   this->maestro_stack_end_   = nullptr;
 292   this->object_infos.resize(0);
 293   this->binary_info     = nullptr;
 294   this->libsimgrid_info = nullptr;
 295
 296   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 297
 298   const char* current_name = nullptr;
 299
 300   this->object_infos.clear();
 301
 302   for (size_t i = 0; i < maps.size(); i++) {
 303     simgrid::xbt::VmMap const& reg = maps[i];
 304     const char* pathname           = maps[i].pathname.c_str();
 305
 306     // Nothing to do
 307     if (maps[i].pathname.empty()) {
 308       current_name = nullptr;
 309       continue;
 310     }
 311
 312     // [stack], [vvar], [vsyscall], [vdso] ...
 313     if (pathname[0] == '[') {
 314       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 315         this->maestro_stack_start_ = remote(reg.start_addr);
 316         this->maestro_stack_end_   = remote(reg.end_addr);
 317       }
 318       current_name = nullptr;
 319       continue;
 320     }
 321
 322     if (current_name && strcmp(current_name, pathname) == 0)
 323       continue;
 324
 325     current_name = pathname;
 326     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 327       continue;
 328
 329     const bool is_executable = not i;
 330     std::string libname;
 331     if (not is_executable) {
 332       libname = get_lib_name(pathname);
 333       if (is_filtered_lib(libname)) {
 334         continue;
 335       }
 336     }
 337
 338     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 339         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 340     this->object_infos.push_back(info);
 341     if (is_executable)
 342       this->binary_info = info;
 343     else if (is_simgrid_lib(libname))
 344       this->libsimgrid_info = info;
 345   }
 346
 347   // Resolve time (including across different objects):
 348   for (auto const& object_info : this->object_infos)
 349     postProcessObjectInformation(this, object_info.get());
 350
 351   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 352   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 353
 354   XBT_DEBUG("Get debug information done !");
 355 }
 356
 357 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
 358 {
 359   for (auto const& object_info : this->object_infos)
 360     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 361       return object_info;
 362   return nullptr;
 363 }
 364
 365 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(RemotePtr<void> addr) const
 366 {
 367   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 368     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
 369       return info;
 370   return nullptr;
 371 }
 372
 373 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(RemotePtr<void> addr) const
 374 {
 375   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 376     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
 377       return info;
 378   return nullptr;
 379 }
 380
 381 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
 382 {
 383   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 384   return info ? info->find_function((void*)ip.address()) : nullptr;
 385 }
 386
 387 /** Find (one occurrence of) the named variable definition
 388  */
 389 const simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
 390 {
 391   // First lookup the variable in the executable shared object.
 392   // A global variable used directly by the executable code from a library
 393   // is reinstantiated in the executable memory .data/.bss.
 394   // We need to look up the variable in the executable first.
 395   if (this->binary_info) {
 396     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 397     const simgrid::mc::Variable* var                            = info->find_variable(name);
 398     if (var)
 399       return var;
 400   }
 401
 402   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 403     const simgrid::mc::Variable* var = info->find_variable(name);
 404     if (var)
 405       return var;
 406   }
 407
 408   return nullptr;
 409 }
 410
 411 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
 412 {
 413   const simgrid::mc::Variable* var = this->find_variable(name);
 414   xbt_assert(var, "Variable %s not found", name);
 415   xbt_assert(var->address, "No simple location for this variable");
 416   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 417   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 418              size, (size_t)var->type->full_type->byte_size);
 419   this->read_bytes(target, size, remote(var->address));
 420 }
 421
 422 std::string RemoteClient::read_string(RemotePtr<char> address) const
 423 {
 424   if (not address)
 425     return {};
 426
 427   std::vector<char> res(128);
 428   off_t off = 0;
 429
 430   while (1) {
 431     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 432     if (c == -1 && errno == EINTR)
 433       continue;
 434     xbt_assert(c > 0, "Could not read string from remote process");
 435
 436     const void* p = memchr(res.data() + off, '\0', c);
 437     if (p)
 438       return std::string(res.data());
 439
 440     off += c;
 441     if (off == (off_t)res.size())
 442       res.resize(res.size() * 2);
 443   }
 444 }
 445
 446 void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, ReadOptions /*options*/) const
 447 {
 448   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 449     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 450   return buffer;
 451 }
 452
 453 /** Write data to a process memory
 454  *
 455  *  @param buffer   local memory address (source)
 456  *  @param len      data size
 457  *  @param address  target process memory address (target)
 458  */
 459 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 460 {
 461   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 462     xbt_die("Write to process %lli failed", (long long)this->pid_);
 463 }
 464
 465 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 466 {
 467   pthread_once(&zero_buffer_flag, zero_buffer_init);
 468   while (len) {
 469     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 470     this->write_bytes(zero_buffer, s, address);
 471     address = remote((char*)address.address() + s);
 472     len -= s;
 473   }
 474 }
 475
 476 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 477 {
 478   IgnoredRegion region;
 479   region.addr = addr;
 480   region.size = size;
 481
 482   if (ignored_regions_.empty()) {
 483     ignored_regions_.push_back(region);
 484     return;
 485   }
 486
 487   unsigned int cursor           = 0;
 488   const IgnoredRegion* current_region = nullptr;
 489
 490   int start = 0;
 491   int end   = ignored_regions_.size() - 1;
 492   while (start <= end) {
 493     cursor         = (start + end) / 2;
 494     current_region = &ignored_regions_[cursor];
 495     if (current_region->addr == addr) {
 496       if (current_region->size == size)
 497         return;
 498       else if (current_region->size < size)
 499         start = cursor + 1;
 500       else
 501         end = cursor - 1;
 502     } else if (current_region->addr < addr)
 503       start = cursor + 1;
 504     else
 505       end = cursor - 1;
 506   }
 507
 508   std::size_t position;
 509   if (current_region->addr == addr) {
 510     if (current_region->size < size)
 511       position = cursor + 1;
 512     else
 513       position = cursor;
 514   } else if (current_region->addr < addr)
 515     position = cursor + 1;
 516   else
 517     position = cursor;
 518   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 519 }
 520
 521 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 522 {
 523   if (ignored_heap_.empty()) {
 524     ignored_heap_.push_back(std::move(region));
 525     return;
 526   }
 527
 528   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 529
 530   size_type start = 0;
 531   size_type end   = ignored_heap_.size() - 1;
 532
 533   // Binary search the position of insertion:
 534   size_type cursor;
 535   while (start <= end) {
 536     cursor               = start + (end - start) / 2;
 537     auto const& current_region = ignored_heap_[cursor];
 538     if (current_region.address == region.address)
 539       return;
 540     else if (current_region.address < region.address)
 541       start = cursor + 1;
 542     else if (cursor != 0)
 543       end = cursor - 1;
 544     // Avoid underflow:
 545     else
 546       break;
 547   }
 548
 549   // Insert it mc_heap_ignore_region_t:
 550   if (ignored_heap_[cursor].address < region.address)
 551     ++cursor;
 552   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 553 }
 554
 555 void RemoteClient::unignore_heap(void* address, size_t size)
 556 {
 557   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 558
 559   size_type start = 0;
 560   size_type end   = ignored_heap_.size() - 1;
 561
 562   // Binary search:
 563   size_type cursor;
 564   while (start <= end) {
 565     cursor       = (start + end) / 2;
 566     auto const& region = ignored_heap_[cursor];
 567     if (region.address < address)
 568       start = cursor + 1;
 569     else if ((char*)region.address <= ((char*)address + size)) {
 570       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 571       return;
 572     } else if (cursor != 0)
 573       end = cursor - 1;
 574     // Avoid underflow:
 575     else
 576       break;
 577   }
 578 }
 579
 580 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 581 {
 582   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 583     frame_name = nullptr;
 584   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 585     info->remove_local_variable(var_name, frame_name);
 586 }
 587
 588 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 589 {
 590   this->refresh_simix();
 591   return smx_actors_infos;
 592 }
 593
 594 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 595 {
 596   this->refresh_simix();
 597   return smx_dead_actors_infos;
 598 }
 599
 600 void RemoteClient::dump_stack()
 601 {
 602   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 603   if (as == nullptr) {
 604     XBT_ERROR("Could not initialize ptrace address space");
 605     return;
 606   }
 607
 608   void* context = _UPT_create(this->pid_);
 609   if (context == nullptr) {
 610     unw_destroy_addr_space(as);
 611     XBT_ERROR("Could not initialize ptrace context");
 612     return;
 613   }
 614
 615   unw_cursor_t cursor;
 616   if (unw_init_remote(&cursor, as, context) != 0) {
 617     _UPT_destroy(context);
 618     unw_destroy_addr_space(as);
 619     XBT_ERROR("Could not initialiez ptrace cursor");
 620     return;
 621   }
 622
 623   simgrid::mc::dumpStack(stderr, &cursor);
 624
 625   _UPT_destroy(context);
 626   unw_destroy_addr_space(as);
 627 }
 628
 629 bool RemoteClient::actor_is_enabled(aid_t pid)
 630 {
 631   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 632   process()->get_channel().send(msg);
 633   char buff[MC_MESSAGE_LENGTH];
 634   ssize_t received = process()->get_channel().receive(buff, MC_MESSAGE_LENGTH, true);
 635   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
 636   return ((s_mc_message_int_t*)buff)->value;
 637 }
 638 }
 639 }