src/mc/remote/RemoteClient.cpp

   1 /* Copyright (c) 2014-2019. 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     "libpcre2-8",
  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
  91 static bool is_simgrid_lib(const std::string& libname)
  92 {
  93   return libname == "libsimgrid";
  94 }
  95
  96 static bool is_filtered_lib(const std::string& libname)
  97 {
  98   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
  99 }
 100
 101 static std::string get_lib_name(const std::string& pathname)
 102 {
 103   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
 104   std::string libname;
 105
 106   size_t pos = map_basename.rfind(".so");
 107   if (pos != std::string::npos) {
 108     // strip the extension (matching regex "\.so.*$")
 109     libname.assign(map_basename, 0, pos);
 110
 111     // strip the version suffix (matching regex "-[.0-9-]*$")
 112     while (true) {
 113       pos = libname.rfind('-');
 114       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
 115         break;
 116       libname.erase(pos);
 117     }
 118   }
 119
 120   return libname;
 121 }
 122
 123 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 124 {
 125   char* buffer       = (char*)buf;
 126   ssize_t real_count = count;
 127   while (count) {
 128     ssize_t res = pread(fd, buffer, count, offset);
 129     if (res > 0) {
 130       count -= res;
 131       buffer += res;
 132       offset += res;
 133     } else if (res == 0)
 134       return -1;
 135     else if (errno != EINTR) {
 136       perror("pread_whole");
 137       return -1;
 138     }
 139   }
 140   return real_count;
 141 }
 142
 143 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 144 {
 145   const char* buffer = (const char*)buf;
 146   ssize_t real_count = count;
 147   while (count) {
 148     ssize_t res = pwrite(fd, buffer, count, offset);
 149     if (res > 0) {
 150       count -= res;
 151       buffer += res;
 152       offset += res;
 153     } else if (res == 0)
 154       return -1;
 155     else if (errno != EINTR)
 156       return -1;
 157   }
 158   return real_count;
 159 }
 160
 161 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 162 static const void* zero_buffer;
 163 static const size_t zero_buffer_size = 10 * 4096;
 164
 165 static void zero_buffer_init()
 166 {
 167   int fd = open("/dev/zero", O_RDONLY);
 168   if (fd < 0)
 169     xbt_die("Could not open /dev/zero");
 170   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 171   if (zero_buffer == MAP_FAILED)
 172     xbt_die("Could not map the zero buffer");
 173   close(fd);
 174 }
 175
 176 int open_vm(pid_t pid, int flags)
 177 {
 178   const size_t buffer_size = 30;
 179   char buffer[buffer_size];
 180   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 181   if (res < 0 || (size_t)res >= buffer_size) {
 182     errno = ENAMETOOLONG;
 183     return -1;
 184   }
 185   return open(buffer, flags);
 186 }
 187
 188 // ***** Process
 189
 190 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
 191 {
 192 }
 193
 194 void RemoteClient::init()
 195 {
 196   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 197   this->init_memory_map_info();
 198
 199   int fd = open_vm(this->pid_, O_RDWR);
 200   if (fd < 0)
 201     xbt_die("Could not open file for process virtual address space");
 202   this->memory_file = fd;
 203
 204   // Read std_heap (is a struct mdesc*):
 205   const simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 206   if (not std_heap_var)
 207     xbt_die("No heap information in the target process");
 208   if (not std_heap_var->address)
 209     xbt_die("No constant address for this variable");
 210   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address));
 211
 212   this->smx_actors_infos.clear();
 213   this->smx_dead_actors_infos.clear();
 214   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 215   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 216   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 217 }
 218
 219 RemoteClient::~RemoteClient()
 220 {
 221   if (this->memory_file >= 0)
 222     close(this->memory_file);
 223
 224   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 225     if (this->unw_underlying_addr_space)
 226       unw_destroy_addr_space(this->unw_underlying_addr_space);
 227     if (this->unw_underlying_context)
 228       _UPT_destroy(this->unw_underlying_context);
 229   }
 230
 231   unw_destroy_addr_space(this->unw_addr_space);
 232 }
 233
 234 /** Refresh the information about the process
 235  *
 236  *  Do not use directly, this is used by the getters when appropriate
 237  *  in order to have fresh data.
 238  */
 239 void RemoteClient::refresh_heap()
 240 {
 241   // Read/dereference/refresh the std_heap pointer:
 242   if (not this->heap)
 243     this->heap.reset(new s_xbt_mheap_t());
 244   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address));
 245   this->cache_flags_ |= RemoteClient::cache_heap;
 246 }
 247
 248 /** Refresh the information about the process
 249  *
 250  *  Do not use directly, this is used by the getters when appropriate
 251  *  in order to have fresh data.
 252  * */
 253 void RemoteClient::refresh_malloc_info()
 254 {
 255   // Refresh process->heapinfo:
 256   if (this->cache_flags_ & RemoteClient::cache_malloc)
 257     return;
 258   size_t count = this->heap->heaplimit + 1;
 259   if (this->heap_info.size() < count)
 260     this->heap_info.resize(count);
 261   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo));
 262   this->cache_flags_ |= RemoteClient::cache_malloc;
 263 }
 264
 265 /** @brief Finds the range of the different memory segments and binary paths */
 266 void RemoteClient::init_memory_map_info()
 267 {
 268   XBT_DEBUG("Get debug information ...");
 269   this->maestro_stack_start_ = nullptr;
 270   this->maestro_stack_end_   = nullptr;
 271   this->object_infos.resize(0);
 272   this->binary_info     = nullptr;
 273   this->libsimgrid_info = nullptr;
 274
 275   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 276
 277   const char* current_name = nullptr;
 278
 279   this->object_infos.clear();
 280
 281   for (size_t i = 0; i < maps.size(); i++) {
 282     simgrid::xbt::VmMap const& reg = maps[i];
 283     const char* pathname           = maps[i].pathname.c_str();
 284
 285     // Nothing to do
 286     if (maps[i].pathname.empty()) {
 287       current_name = nullptr;
 288       continue;
 289     }
 290
 291     // [stack], [vvar], [vsyscall], [vdso] ...
 292     if (pathname[0] == '[') {
 293       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 294         this->maestro_stack_start_ = remote(reg.start_addr);
 295         this->maestro_stack_end_   = remote(reg.end_addr);
 296       }
 297       current_name = nullptr;
 298       continue;
 299     }
 300
 301     if (current_name && strcmp(current_name, pathname) == 0)
 302       continue;
 303
 304     current_name = pathname;
 305     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 306       continue;
 307
 308     const bool is_executable = not i;
 309     std::string libname;
 310     if (not is_executable) {
 311       libname = get_lib_name(pathname);
 312       if (is_filtered_lib(libname)) {
 313         continue;
 314       }
 315     }
 316
 317     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 318         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 319     this->object_infos.push_back(info);
 320     if (is_executable)
 321       this->binary_info = info;
 322     else if (is_simgrid_lib(libname))
 323       this->libsimgrid_info = info;
 324   }
 325
 326   // Resolve time (including across different objects):
 327   for (auto const& object_info : this->object_infos)
 328     postProcessObjectInformation(this, object_info.get());
 329
 330   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 331   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 332
 333   XBT_DEBUG("Get debug information done !");
 334 }
 335
 336 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
 337 {
 338   for (auto const& object_info : this->object_infos)
 339     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 340       return object_info;
 341   return nullptr;
 342 }
 343
 344 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(RemotePtr<void> addr) const
 345 {
 346   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 347     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
 348       return info;
 349   return nullptr;
 350 }
 351
 352 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(RemotePtr<void> addr) const
 353 {
 354   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 355     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
 356       return info;
 357   return nullptr;
 358 }
 359
 360 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
 361 {
 362   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 363   return info ? info->find_function((void*)ip.address()) : nullptr;
 364 }
 365
 366 /** Find (one occurrence of) the named variable definition
 367  */
 368 const simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
 369 {
 370   // First lookup the variable in the executable shared object.
 371   // A global variable used directly by the executable code from a library
 372   // is reinstantiated in the executable memory .data/.bss.
 373   // We need to look up the variable in the executable first.
 374   if (this->binary_info) {
 375     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 376     const simgrid::mc::Variable* var                            = info->find_variable(name);
 377     if (var)
 378       return var;
 379   }
 380
 381   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 382     const simgrid::mc::Variable* var = info->find_variable(name);
 383     if (var)
 384       return var;
 385   }
 386
 387   return nullptr;
 388 }
 389
 390 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
 391 {
 392   const simgrid::mc::Variable* var = this->find_variable(name);
 393   xbt_assert(var, "Variable %s not found", name);
 394   xbt_assert(var->address, "No simple location for this variable");
 395   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 396   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 397              size, (size_t)var->type->full_type->byte_size);
 398   this->read_bytes(target, size, remote(var->address));
 399 }
 400
 401 std::string RemoteClient::read_string(RemotePtr<char> address) const
 402 {
 403   if (not address)
 404     return {};
 405
 406   std::vector<char> res(128);
 407   off_t off = 0;
 408
 409   while (1) {
 410     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 411     if (c == -1 && errno == EINTR)
 412       continue;
 413     xbt_assert(c > 0, "Could not read string from remote process");
 414
 415     void* p = memchr(res.data() + off, '\0', c);
 416     if (p)
 417       return std::string(res.data());
 418
 419     off += c;
 420     if (off == (off_t)res.size())
 421       res.resize(res.size() * 2);
 422   }
 423 }
 424
 425 void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, ReadOptions /*options*/) const
 426 {
 427   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 428     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 429   return buffer;
 430 }
 431
 432 /** Write data to a process memory
 433  *
 434  *  @param buffer   local memory address (source)
 435  *  @param len      data size
 436  *  @param address  target process memory address (target)
 437  */
 438 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 439 {
 440   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 441     xbt_die("Write to process %lli failed", (long long)this->pid_);
 442 }
 443
 444 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 445 {
 446   pthread_once(&zero_buffer_flag, zero_buffer_init);
 447   while (len) {
 448     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 449     this->write_bytes(zero_buffer, s, address);
 450     address = remote((char*)address.address() + s);
 451     len -= s;
 452   }
 453 }
 454
 455 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 456 {
 457   IgnoredRegion region;
 458   region.addr = addr;
 459   region.size = size;
 460
 461   if (ignored_regions_.empty()) {
 462     ignored_regions_.push_back(region);
 463     return;
 464   }
 465
 466   unsigned int cursor           = 0;
 467   IgnoredRegion* current_region = nullptr;
 468
 469   int start = 0;
 470   int end   = ignored_regions_.size() - 1;
 471   while (start <= end) {
 472     cursor         = (start + end) / 2;
 473     current_region = &ignored_regions_[cursor];
 474     if (current_region->addr == addr) {
 475       if (current_region->size == size)
 476         return;
 477       else if (current_region->size < size)
 478         start = cursor + 1;
 479       else
 480         end = cursor - 1;
 481     } else if (current_region->addr < addr)
 482       start = cursor + 1;
 483     else
 484       end = cursor - 1;
 485   }
 486
 487   std::size_t position;
 488   if (current_region->addr == addr) {
 489     if (current_region->size < size)
 490       position = cursor + 1;
 491     else
 492       position = cursor;
 493   } else if (current_region->addr < addr)
 494     position = cursor + 1;
 495   else
 496     position = cursor;
 497   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 498 }
 499
 500 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 501 {
 502   if (ignored_heap_.empty()) {
 503     ignored_heap_.push_back(std::move(region));
 504     return;
 505   }
 506
 507   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 508
 509   size_type start = 0;
 510   size_type end   = ignored_heap_.size() - 1;
 511
 512   // Binary search the position of insertion:
 513   size_type cursor;
 514   while (start <= end) {
 515     cursor               = start + (end - start) / 2;
 516     auto& current_region = ignored_heap_[cursor];
 517     if (current_region.address == region.address)
 518       return;
 519     else if (current_region.address < region.address)
 520       start = cursor + 1;
 521     else if (cursor != 0)
 522       end = cursor - 1;
 523     // Avoid underflow:
 524     else
 525       break;
 526   }
 527
 528   // Insert it mc_heap_ignore_region_t:
 529   if (ignored_heap_[cursor].address < region.address)
 530     ++cursor;
 531   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 532 }
 533
 534 void RemoteClient::unignore_heap(void* address, size_t size)
 535 {
 536   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 537
 538   size_type start = 0;
 539   size_type end   = ignored_heap_.size() - 1;
 540
 541   // Binary search:
 542   size_type cursor;
 543   while (start <= end) {
 544     cursor       = (start + end) / 2;
 545     auto& region = ignored_heap_[cursor];
 546     if (region.address < address)
 547       start = cursor + 1;
 548     else if ((char*)region.address <= ((char*)address + size)) {
 549       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 550       return;
 551     } else if (cursor != 0)
 552       end = cursor - 1;
 553     // Avoid underflow:
 554     else
 555       break;
 556   }
 557 }
 558
 559 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 560 {
 561   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 562     frame_name = nullptr;
 563   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 564     info->remove_local_variable(var_name, frame_name);
 565 }
 566
 567 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 568 {
 569   this->refresh_simix();
 570   return smx_actors_infos;
 571 }
 572
 573 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 574 {
 575   this->refresh_simix();
 576   return smx_dead_actors_infos;
 577 }
 578
 579 void RemoteClient::dump_stack()
 580 {
 581   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 582   if (as == nullptr) {
 583     XBT_ERROR("Could not initialize ptrace address space");
 584     return;
 585   }
 586
 587   void* context = _UPT_create(this->pid_);
 588   if (context == nullptr) {
 589     unw_destroy_addr_space(as);
 590     XBT_ERROR("Could not initialize ptrace context");
 591     return;
 592   }
 593
 594   unw_cursor_t cursor;
 595   if (unw_init_remote(&cursor, as, context) != 0) {
 596     _UPT_destroy(context);
 597     unw_destroy_addr_space(as);
 598     XBT_ERROR("Could not initialiez ptrace cursor");
 599     return;
 600   }
 601
 602   simgrid::mc::dumpStack(stderr, std::move(cursor));
 603
 604   _UPT_destroy(context);
 605   unw_destroy_addr_space(as);
 606 }
 607
 608 bool RemoteClient::actor_is_enabled(aid_t pid)
 609 {
 610   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 611   process()->get_channel().send(msg);
 612   char buff[MC_MESSAGE_LENGTH];
 613   ssize_t received = process()->get_channel().receive(buff, MC_MESSAGE_LENGTH, true);
 614   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
 615   return ((s_mc_message_int_t*)buff)->value;
 616 }
 617 }
 618 }