src/mc/remote/RemoteClient.cpp

   1 /* Copyright (c) 2014-2018. 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 <algorithm>
   9 #include <cassert>
  10 #include <cerrno>
  11 #include <cstddef>
  12 #include <cstdint>
  13 #include <string>
  14 #include <vector>
  15
  16 #include <sys/ptrace.h>
  17
  18 #include <cstdio>
  19
  20 #include <fcntl.h>
  21 #include <sys/mman.h> // PROT_*
  22 #include <sys/types.h>
  23 #include <unistd.h>
  24
  25 #include <pthread.h>
  26
  27 #include <libgen.h>
  28
  29 #include <libunwind-ptrace.h>
  30 #include <libunwind.h>
  31
  32 #include "xbt/base.h"
  33 #include "xbt/file.hpp"
  34 #include "xbt/log.h"
  35 #include <xbt/mmalloc.h>
  36
  37 #include "src/mc/mc_smx.hpp"
  38 #include "src/mc/mc_unw.hpp"
  39 #include "src/mc/sosp/mc_snapshot.hpp"
  40
  41 #include "src/mc/AddressSpace.hpp"
  42 #include "src/mc/ObjectInformation.hpp"
  43 #include "src/mc/Variable.hpp"
  44 #include "src/mc/remote/RemoteClient.hpp"
  45
  46 using simgrid::mc::remote;
  47
  48 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(mc_process, mc, "MC process information");
  49
  50 namespace simgrid {
  51 namespace mc {
  52
  53 // ***** Helper stuff
  54
  55 // List of library which memory segments are not considered:
  56 static const std::vector<std::string> filtered_libraries = {
  57 #ifdef __linux__
  58     "ld",
  59 #elif defined __FreeBSD__
  60     "ld-elf",
  61     "ld-elf32",
  62     "libkvm",      /* kernel data access library */
  63     "libprocstat", /* process and file information retrieval */
  64     "libthr",      /* thread library */
  65     "libutil",
  66 #endif
  67     "libasan", /* gcc sanitizers */
  68     "libargp", /* workarounds for glibc-less systems */
  69     "libtsan",
  70     "libubsan",
  71     "libbz2",
  72     "libboost_chrono",
  73     "libboost_context",
  74     "libboost_context-mt",
  75     "libboost_system",
  76     "libboost_thread",
  77     "libboost_timer",
  78     "libboost_unit_test_framework",
  79     "libc",
  80     "libc++",
  81     "libcdt",
  82     "libcgraph",
  83     "libcrypto",
  84     "libcxxrt",
  85     "libdl",
  86     "libdw",
  87     "libelf",
  88     "libevent",
  89     "libgcc_s",
  90     "libimf",
  91     "libintlc",
  92     "libirng",
  93     "liblua5.1",
  94     "liblua5.3",
  95     "liblzma",
  96     "libm",
  97     "libpthread",
  98     "librt",
  99     "libstdc++",
 100     "libsvml",
 101     "libunwind",
 102     "libunwind-x86_64",
 103     "libunwind-x86",
 104     "libunwind-ptrace",
 105     "libz"};
 106
 107 static bool is_simgrid_lib(const std::string& libname)
 108 {
 109   return libname == "libsimgrid";
 110 }
 111
 112 static bool is_filtered_lib(const std::string& libname)
 113 {
 114   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
 115 }
 116
 117 static std::string get_lib_name(const std::string& pathname)
 118 {
 119   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
 120   std::string libname;
 121
 122   size_t pos = map_basename.rfind(".so");
 123   if (pos != std::string::npos) {
 124     // strip the extension (matching regex "\.so.*$")
 125     libname.assign(map_basename, 0, pos);
 126
 127     // strip the version suffix (matching regex "-[.0-9-]*$")
 128     while (true) {
 129       pos = libname.rfind('-');
 130       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
 131         break;
 132       libname.erase(pos);
 133     }
 134   }
 135
 136   return libname;
 137 }
 138
 139 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 140 {
 141   char* buffer       = (char*)buf;
 142   ssize_t real_count = count;
 143   while (count) {
 144     ssize_t res = pread(fd, buffer, count, offset);
 145     if (res > 0) {
 146       count -= res;
 147       buffer += res;
 148       offset += res;
 149     } else if (res == 0)
 150       return -1;
 151     else if (errno != EINTR) {
 152       perror("pread_whole");
 153       return -1;
 154     }
 155   }
 156   return real_count;
 157 }
 158
 159 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 160 {
 161   const char* buffer = (const char*)buf;
 162   ssize_t real_count = count;
 163   while (count) {
 164     ssize_t res = pwrite(fd, buffer, count, offset);
 165     if (res > 0) {
 166       count -= res;
 167       buffer += res;
 168       offset += res;
 169     } else if (res == 0)
 170       return -1;
 171     else if (errno != EINTR)
 172       return -1;
 173   }
 174   return real_count;
 175 }
 176
 177 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 178 static const void* zero_buffer;
 179 static const size_t zero_buffer_size = 10 * 4096;
 180
 181 static void zero_buffer_init()
 182 {
 183   int fd = open("/dev/zero", O_RDONLY);
 184   if (fd < 0)
 185     xbt_die("Could not open /dev/zero");
 186   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 187   if (zero_buffer == MAP_FAILED)
 188     xbt_die("Could not map the zero buffer");
 189   close(fd);
 190 }
 191
 192 int open_vm(pid_t pid, int flags)
 193 {
 194   const size_t buffer_size = 30;
 195   char buffer[buffer_size];
 196   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 197   if (res < 0 || (size_t)res >= buffer_size) {
 198     errno = ENAMETOOLONG;
 199     return -1;
 200   }
 201   return open(buffer, flags);
 202 }
 203
 204 // ***** Process
 205
 206 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
 207 {
 208 }
 209
 210 void RemoteClient::init()
 211 {
 212   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 213   this->init_memory_map_info();
 214
 215   int fd = open_vm(this->pid_, O_RDWR);
 216   if (fd < 0)
 217     xbt_die("Could not open file for process virtual address space");
 218   this->memory_file = fd;
 219
 220   // Read std_heap (is a struct mdesc*):
 221   simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 222   if (not std_heap_var)
 223     xbt_die("No heap information in the target process");
 224   if (not std_heap_var->address)
 225     xbt_die("No constant address for this variable");
 226   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address),
 227                    simgrid::mc::ProcessIndexDisabled);
 228
 229   this->smx_actors_infos.clear();
 230   this->smx_dead_actors_infos.clear();
 231   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 232   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 233   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 234 }
 235
 236 RemoteClient::~RemoteClient()
 237 {
 238   if (this->memory_file >= 0)
 239     close(this->memory_file);
 240
 241   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 242     if (this->unw_underlying_addr_space)
 243       unw_destroy_addr_space(this->unw_underlying_addr_space);
 244     if (this->unw_underlying_context)
 245       _UPT_destroy(this->unw_underlying_context);
 246   }
 247
 248   unw_destroy_addr_space(this->unw_addr_space);
 249 }
 250
 251 /** Refresh the information about the process
 252  *
 253  *  Do not use directly, this is used by the getters when appropriate
 254  *  in order to have fresh data.
 255  */
 256 void RemoteClient::refresh_heap()
 257 {
 258   // Read/dereference/refresh the std_heap pointer:
 259   if (not this->heap)
 260     this->heap = std::unique_ptr<s_xbt_mheap_t>(new s_xbt_mheap_t());
 261   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address), simgrid::mc::ProcessIndexDisabled);
 262   this->cache_flags_ |= RemoteClient::cache_heap;
 263 }
 264
 265 /** Refresh the information about the process
 266  *
 267  *  Do not use direclty, this is used by the getters when appropriate
 268  *  in order to have fresh data.
 269  * */
 270 void RemoteClient::refresh_malloc_info()
 271 {
 272   // Refresh process->heapinfo:
 273   if (this->cache_flags_ & RemoteClient::cache_malloc)
 274     return;
 275   size_t count = this->heap->heaplimit + 1;
 276   if (this->heap_info.size() < count)
 277     this->heap_info.resize(count);
 278   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo),
 279                    simgrid::mc::ProcessIndexDisabled);
 280   this->cache_flags_ |= RemoteClient::cache_malloc;
 281 }
 282
 283 /** @brief Finds the range of the different memory segments and binary paths */
 284 void RemoteClient::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> RemoteClient::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> RemoteClient::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> RemoteClient::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* RemoteClient::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 simgrid::mc::Variable* RemoteClient::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 reinstanciated 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     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     simgrid::mc::Variable* var = info->find_variable(name);
 401     if (var)
 402       return var;
 403   }
 404
 405   return nullptr;
 406 }
 407
 408 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
 409 {
 410   simgrid::mc::Variable* var = this->find_variable(name);
 411   xbt_assert(var->address, "No simple location for this variable");
 412   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 413   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 414              size, (size_t)var->type->full_type->byte_size);
 415   this->read_bytes(target, size, remote(var->address));
 416 }
 417
 418 std::string RemoteClient::read_string(RemotePtr<char> address) const
 419 {
 420   if (not address)
 421     return {};
 422
 423   std::vector<char> res(128);
 424   off_t off = 0;
 425
 426   while (1) {
 427     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 428     if (c == -1) {
 429       if (errno == EINTR)
 430         continue;
 431       else
 432         xbt_die("Could not read from from remote process");
 433     }
 434     if (c == 0)
 435       xbt_die("Could not read string from remote process");
 436
 437     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 const void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, int process_index,
 448                                      ReadOptions options) const
 449 {
 450   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 451     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->find_object_info_rw(address);
 452 // Segment overlap is not handled.
 453 #if HAVE_SMPI
 454     if (info.get() && this->privatized(*info)) {
 455       if (process_index < 0)
 456         xbt_die("Missing process index");
 457       if (process_index >= (int)MC_smpi_process_count())
 458         xbt_die("Invalid process index");
 459
 460       // Read smpi_privatization_regions from MCed:
 461       smpi_privatization_region_t remote_smpi_privatization_regions =
 462           mc_model_checker->process().read_variable<smpi_privatization_region_t>("smpi_privatization_regions");
 463
 464       s_smpi_privatization_region_t privatization_region =
 465           mc_model_checker->process().read<s_smpi_privatization_region_t>(
 466               remote(remote_smpi_privatization_regions + process_index));
 467
 468       // Address translation in the privatization segment:
 469       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 470       address       = remote((char*)privatization_region.address + offset);
 471     }
 472 #endif
 473   }
 474   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 475     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 476   return buffer;
 477 }
 478
 479 /** Write data to a process memory
 480  *
 481  *  @param buffer   local memory address (source)
 482  *  @param len      data size
 483  *  @param address  target process memory address (target)
 484  */
 485 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 486 {
 487   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 488     xbt_die("Write to process %lli failed", (long long)this->pid_);
 489 }
 490
 491 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 492 {
 493   pthread_once(&zero_buffer_flag, zero_buffer_init);
 494   while (len) {
 495     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 496     this->write_bytes(zero_buffer, s, address);
 497     address = remote((char*)address.address() + s);
 498     len -= s;
 499   }
 500 }
 501
 502 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 503 {
 504   IgnoredRegion region;
 505   region.addr = addr;
 506   region.size = size;
 507
 508   if (ignored_regions_.empty()) {
 509     ignored_regions_.push_back(region);
 510     return;
 511   }
 512
 513   unsigned int cursor           = 0;
 514   IgnoredRegion* current_region = nullptr;
 515
 516   int start = 0;
 517   int end   = ignored_regions_.size() - 1;
 518   while (start <= end) {
 519     cursor         = (start + end) / 2;
 520     current_region = &ignored_regions_[cursor];
 521     if (current_region->addr == addr) {
 522       if (current_region->size == size)
 523         return;
 524       else if (current_region->size < size)
 525         start = cursor + 1;
 526       else
 527         end = cursor - 1;
 528     } else if (current_region->addr < addr)
 529       start = cursor + 1;
 530     else
 531       end = cursor - 1;
 532   }
 533
 534   std::size_t position;
 535   if (current_region->addr == addr) {
 536     if (current_region->size < size)
 537       position = cursor + 1;
 538     else
 539       position = cursor;
 540   } else if (current_region->addr < addr)
 541     position = cursor + 1;
 542   else
 543     position = cursor;
 544   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 545 }
 546
 547 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 548 {
 549   if (ignored_heap_.empty()) {
 550     ignored_heap_.push_back(std::move(region));
 551     return;
 552   }
 553
 554   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 555
 556   size_type start = 0;
 557   size_type end   = ignored_heap_.size() - 1;
 558
 559   // Binary search the position of insertion:
 560   size_type cursor;
 561   while (start <= end) {
 562     cursor               = start + (end - start) / 2;
 563     auto& current_region = ignored_heap_[cursor];
 564     if (current_region.address == region.address)
 565       return;
 566     else if (current_region.address < region.address)
 567       start = cursor + 1;
 568     else if (cursor != 0)
 569       end = cursor - 1;
 570     // Avoid underflow:
 571     else
 572       break;
 573   }
 574
 575   // Insert it mc_heap_ignore_region_t:
 576   if (ignored_heap_[cursor].address < region.address)
 577     ++cursor;
 578   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 579 }
 580
 581 void RemoteClient::unignore_heap(void* address, size_t size)
 582 {
 583   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 584
 585   size_type start = 0;
 586   size_type end   = ignored_heap_.size() - 1;
 587
 588   // Binary search:
 589   size_type cursor;
 590   while (start <= end) {
 591     cursor       = (start + end) / 2;
 592     auto& region = ignored_heap_[cursor];
 593     if (region.address < address)
 594       start = cursor + 1;
 595     else if ((char*)region.address <= ((char*)address + size)) {
 596       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 597       return;
 598     } else if (cursor != 0)
 599       end = cursor - 1;
 600     // Avoid underflow:
 601     else
 602       break;
 603   }
 604 }
 605
 606 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 607 {
 608   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 609     frame_name = nullptr;
 610   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 611     info->remove_local_variable(var_name, frame_name);
 612 }
 613
 614 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 615 {
 616   this->refresh_simix();
 617   return smx_actors_infos;
 618 }
 619
 620 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 621 {
 622   this->refresh_simix();
 623   return smx_dead_actors_infos;
 624 }
 625
 626 void RemoteClient::dumpStack()
 627 {
 628   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 629   if (as == nullptr) {
 630     XBT_ERROR("Could not initialize ptrace address space");
 631     return;
 632   }
 633
 634   void* context = _UPT_create(this->pid_);
 635   if (context == nullptr) {
 636     unw_destroy_addr_space(as);
 637     XBT_ERROR("Could not initialize ptrace context");
 638     return;
 639   }
 640
 641   unw_cursor_t cursor;
 642   if (unw_init_remote(&cursor, as, context) != 0) {
 643     _UPT_destroy(context);
 644     unw_destroy_addr_space(as);
 645     XBT_ERROR("Could not initialiez ptrace cursor");
 646     return;
 647   }
 648
 649   simgrid::mc::dumpStack(stderr, cursor);
 650
 651   _UPT_destroy(context);
 652   unw_destroy_addr_space(as);
 653 }
 654
 655 bool RemoteClient::actor_is_enabled(aid_t pid)
 656 {
 657   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 658   process()->getChannel().send(msg);
 659   char buff[MC_MESSAGE_LENGTH];
 660   ssize_t received = process()->getChannel().receive(buff, MC_MESSAGE_LENGTH, true);
 661   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
 662   return ((s_mc_message_int_t*)buff)->value;
 663 }
 664 }
 665 }