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_snapshot.hpp"
  39 #include "src/mc/mc_unw.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_unit_test_framework",
  78     "libc",
  79     "libc++",
  80     "libcdt",
  81     "libcgraph",
  82     "libcrypto",
  83     "libcxxrt",
  84     "libdl",
  85     "libdw",
  86     "libelf",
  87     "libevent",
  88     "libgcc_s",
  89     "libimf",
  90     "libintlc",
  91     "libirng",
  92     "liblua5.1",
  93     "liblua5.3",
  94     "liblzma",
  95     "libm",
  96     "libpthread",
  97     "librt",
  98     "libstdc++",
  99     "libsvml",
 100     "libunwind",
 101     "libunwind-x86_64",
 102     "libunwind-x86",
 103     "libunwind-ptrace",
 104     "libz"};
 105
 106 static bool is_simgrid_lib(const std::string& libname)
 107 {
 108   return libname == "libsimgrid";
 109 }
 110
 111 static bool is_filtered_lib(const std::string& libname)
 112 {
 113   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
 114 }
 115
 116 static std::string get_lib_name(const std::string& pathname)
 117 {
 118   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
 119   std::string libname;
 120
 121   size_t pos = map_basename.rfind(".so");
 122   if (pos != std::string::npos) {
 123     // strip the extension (matching regex "\.so.*$")
 124     libname.assign(map_basename, 0, pos);
 125
 126     // strip the version suffix (matching regex "-[.0-9-]*$")
 127     while (true) {
 128       pos = libname.rfind('-');
 129       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
 130         break;
 131       libname.erase(pos);
 132     }
 133   }
 134
 135   return libname;
 136 }
 137
 138 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 139 {
 140   char* buffer       = (char*)buf;
 141   ssize_t real_count = count;
 142   while (count) {
 143     ssize_t res = pread(fd, buffer, count, offset);
 144     if (res > 0) {
 145       count -= res;
 146       buffer += res;
 147       offset += res;
 148     } else if (res == 0)
 149       return -1;
 150     else if (errno != EINTR) {
 151       perror("pread_whole");
 152       return -1;
 153     }
 154   }
 155   return real_count;
 156 }
 157
 158 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 159 {
 160   const char* buffer = (const char*)buf;
 161   ssize_t real_count = count;
 162   while (count) {
 163     ssize_t res = pwrite(fd, buffer, count, offset);
 164     if (res > 0) {
 165       count -= res;
 166       buffer += res;
 167       offset += res;
 168     } else if (res == 0)
 169       return -1;
 170     else if (errno != EINTR)
 171       return -1;
 172   }
 173   return real_count;
 174 }
 175
 176 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 177 static const void* zero_buffer;
 178 static const size_t zero_buffer_size = 10 * 4096;
 179
 180 static void zero_buffer_init()
 181 {
 182   int fd = open("/dev/zero", O_RDONLY);
 183   if (fd < 0)
 184     xbt_die("Could not open /dev/zero");
 185   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 186   if (zero_buffer == MAP_FAILED)
 187     xbt_die("Could not map the zero buffer");
 188   close(fd);
 189 }
 190
 191 int open_vm(pid_t pid, int flags)
 192 {
 193   const size_t buffer_size = 30;
 194   char buffer[buffer_size];
 195   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 196   if (res < 0 || (size_t)res >= buffer_size) {
 197     errno = ENAMETOOLONG;
 198     return -1;
 199   }
 200   return open(buffer, flags);
 201 }
 202
 203 // ***** Process
 204
 205 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
 206 {
 207 }
 208
 209 void RemoteClient::init()
 210 {
 211   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 212   this->init_memory_map_info();
 213
 214   int fd = open_vm(this->pid_, O_RDWR);
 215   if (fd < 0)
 216     xbt_die("Could not open file for process virtual address space");
 217   this->memory_file = fd;
 218
 219   // Read std_heap (is a struct mdesc*):
 220   simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 221   if (not std_heap_var)
 222     xbt_die("No heap information in the target process");
 223   if (not std_heap_var->address)
 224     xbt_die("No constant address for this variable");
 225   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address),
 226                    simgrid::mc::ProcessIndexDisabled);
 227
 228   this->smx_actors_infos.clear();
 229   this->smx_dead_actors_infos.clear();
 230   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 231   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 232   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 233 }
 234
 235 RemoteClient::~RemoteClient()
 236 {
 237   if (this->memory_file >= 0)
 238     close(this->memory_file);
 239
 240   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 241     if (this->unw_underlying_addr_space)
 242       unw_destroy_addr_space(this->unw_underlying_addr_space);
 243     if (this->unw_underlying_context)
 244       _UPT_destroy(this->unw_underlying_context);
 245   }
 246
 247   unw_destroy_addr_space(this->unw_addr_space);
 248 }
 249
 250 /** Refresh the information about the process
 251  *
 252  *  Do not use directly, this is used by the getters when appropriate
 253  *  in order to have fresh data.
 254  */
 255 void RemoteClient::refresh_heap()
 256 {
 257   // Read/dereference/refresh the std_heap pointer:
 258   if (not this->heap)
 259     this->heap = std::unique_ptr<s_xbt_mheap_t>(new s_xbt_mheap_t());
 260   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address), simgrid::mc::ProcessIndexDisabled);
 261   this->cache_flags_ |= RemoteClient::cache_heap;
 262 }
 263
 264 /** Refresh the information about the process
 265  *
 266  *  Do not use direclty, this is used by the getters when appropriate
 267  *  in order to have fresh data.
 268  * */
 269 void RemoteClient::refresh_malloc_info()
 270 {
 271   // Refresh process->heapinfo:
 272   if (this->cache_flags_ & RemoteClient::cache_malloc)
 273     return;
 274   size_t count = this->heap->heaplimit + 1;
 275   if (this->heap_info.size() < count)
 276     this->heap_info.resize(count);
 277   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo),
 278                    simgrid::mc::ProcessIndexDisabled);
 279   this->cache_flags_ |= RemoteClient::cache_malloc;
 280 }
 281
 282 /** @brief Finds the range of the different memory segments and binary paths */
 283 void RemoteClient::init_memory_map_info()
 284 {
 285   XBT_DEBUG("Get debug information ...");
 286   this->maestro_stack_start_ = nullptr;
 287   this->maestro_stack_end_   = nullptr;
 288   this->object_infos.resize(0);
 289   this->binary_info     = nullptr;
 290   this->libsimgrid_info = nullptr;
 291
 292   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 293
 294   const char* current_name = nullptr;
 295
 296   this->object_infos.clear();
 297
 298   for (size_t i = 0; i < maps.size(); i++) {
 299     simgrid::xbt::VmMap const& reg = maps[i];
 300     const char* pathname           = maps[i].pathname.c_str();
 301
 302     // Nothing to do
 303     if (maps[i].pathname.empty()) {
 304       current_name = nullptr;
 305       continue;
 306     }
 307
 308     // [stack], [vvar], [vsyscall], [vdso] ...
 309     if (pathname[0] == '[') {
 310       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 311         this->maestro_stack_start_ = remote(reg.start_addr);
 312         this->maestro_stack_end_   = remote(reg.end_addr);
 313       }
 314       current_name = nullptr;
 315       continue;
 316     }
 317
 318     if (current_name && strcmp(current_name, pathname) == 0)
 319       continue;
 320
 321     current_name = pathname;
 322     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 323       continue;
 324
 325     const bool is_executable = not i;
 326     std::string libname;
 327     if (not is_executable) {
 328       libname = get_lib_name(pathname);
 329       if (is_filtered_lib(libname)) {
 330         continue;
 331       }
 332     }
 333
 334     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 335         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 336     this->object_infos.push_back(info);
 337     if (is_executable)
 338       this->binary_info = info;
 339     else if (is_simgrid_lib(libname))
 340       this->libsimgrid_info = info;
 341   }
 342
 343   // Resolve time (including across different objects):
 344   for (auto const& object_info : this->object_infos)
 345     postProcessObjectInformation(this, object_info.get());
 346
 347   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 348   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 349
 350   XBT_DEBUG("Get debug information done !");
 351 }
 352
 353 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
 354 {
 355   for (auto const& object_info : this->object_infos)
 356     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 357       return object_info;
 358   return nullptr;
 359 }
 360
 361 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(RemotePtr<void> addr) const
 362 {
 363   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 364     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
 365       return info;
 366   return nullptr;
 367 }
 368
 369 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(RemotePtr<void> addr) const
 370 {
 371   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 372     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
 373       return info;
 374   return nullptr;
 375 }
 376
 377 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
 378 {
 379   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 380   return info ? info->find_function((void*)ip.address()) : nullptr;
 381 }
 382
 383 /** Find (one occurrence of) the named variable definition
 384  */
 385 simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
 386 {
 387   // First lookup the variable in the executable shared object.
 388   // A global variable used directly by the executable code from a library
 389   // is reinstanciated in the executable memory .data/.bss.
 390   // We need to look up the variable in the executable first.
 391   if (this->binary_info) {
 392     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 393     simgrid::mc::Variable* var                                  = info->find_variable(name);
 394     if (var)
 395       return var;
 396   }
 397
 398   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 399     simgrid::mc::Variable* var = info->find_variable(name);
 400     if (var)
 401       return var;
 402   }
 403
 404   return nullptr;
 405 }
 406
 407 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
 408 {
 409   simgrid::mc::Variable* var = this->find_variable(name);
 410   xbt_assert(var->address, "No simple location for this variable");
 411   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 412   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 413              size, (size_t)var->type->full_type->byte_size);
 414   this->read_bytes(target, size, remote(var->address));
 415 }
 416
 417 std::string RemoteClient::read_string(RemotePtr<char> address) const
 418 {
 419   if (not address)
 420     return {};
 421
 422   std::vector<char> res(128);
 423   off_t off = 0;
 424
 425   while (1) {
 426     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 427     if (c == -1) {
 428       if (errno == EINTR)
 429         continue;
 430       else
 431         xbt_die("Could not read from from remote process");
 432     }
 433     if (c == 0)
 434       xbt_die("Could not read string from remote process");
 435
 436     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 const void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, int process_index,
 447                                      ReadOptions options) const
 448 {
 449   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 450     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->find_object_info_rw(address);
 451 // Segment overlap is not handled.
 452 #if HAVE_SMPI
 453     if (info.get() && this->privatized(*info)) {
 454       if (process_index < 0)
 455         xbt_die("Missing process index");
 456       if (process_index >= (int)MC_smpi_process_count())
 457         xbt_die("Invalid process index");
 458
 459       // Read smpi_privatization_regions from MCed:
 460       smpi_privatization_region_t remote_smpi_privatization_regions =
 461           mc_model_checker->process().read_variable<smpi_privatization_region_t>("smpi_privatization_regions");
 462
 463       s_smpi_privatization_region_t privatization_region =
 464           mc_model_checker->process().read<s_smpi_privatization_region_t>(
 465               remote(remote_smpi_privatization_regions + process_index));
 466
 467       // Address translation in the privatization segment:
 468       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 469       address       = remote((char*)privatization_region.address + offset);
 470     }
 471 #endif
 472   }
 473   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 474     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 475   return buffer;
 476 }
 477
 478 /** Write data to a process memory
 479  *
 480  *  @param buffer   local memory address (source)
 481  *  @param len      data size
 482  *  @param address  target process memory address (target)
 483  */
 484 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 485 {
 486   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 487     xbt_die("Write to process %lli failed", (long long)this->pid_);
 488 }
 489
 490 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 491 {
 492   pthread_once(&zero_buffer_flag, zero_buffer_init);
 493   while (len) {
 494     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 495     this->write_bytes(zero_buffer, s, address);
 496     address = remote((char*)address.address() + s);
 497     len -= s;
 498   }
 499 }
 500
 501 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 502 {
 503   IgnoredRegion region;
 504   region.addr = addr;
 505   region.size = size;
 506
 507   if (ignored_regions_.empty()) {
 508     ignored_regions_.push_back(region);
 509     return;
 510   }
 511
 512   unsigned int cursor           = 0;
 513   IgnoredRegion* current_region = nullptr;
 514
 515   int start = 0;
 516   int end   = ignored_regions_.size() - 1;
 517   while (start <= end) {
 518     cursor         = (start + end) / 2;
 519     current_region = &ignored_regions_[cursor];
 520     if (current_region->addr == addr) {
 521       if (current_region->size == size)
 522         return;
 523       else if (current_region->size < size)
 524         start = cursor + 1;
 525       else
 526         end = cursor - 1;
 527     } else if (current_region->addr < addr)
 528       start = cursor + 1;
 529     else
 530       end = cursor - 1;
 531   }
 532
 533   std::size_t position;
 534   if (current_region->addr == addr) {
 535     if (current_region->size < size)
 536       position = cursor + 1;
 537     else
 538       position = cursor;
 539   } else if (current_region->addr < addr)
 540     position = cursor + 1;
 541   else
 542     position = cursor;
 543   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 544 }
 545
 546 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 547 {
 548   if (ignored_heap_.empty()) {
 549     ignored_heap_.push_back(std::move(region));
 550     return;
 551   }
 552
 553   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 554
 555   size_type start = 0;
 556   size_type end   = ignored_heap_.size() - 1;
 557
 558   // Binary search the position of insertion:
 559   size_type cursor;
 560   while (start <= end) {
 561     cursor               = start + (end - start) / 2;
 562     auto& current_region = ignored_heap_[cursor];
 563     if (current_region.address == region.address)
 564       return;
 565     else if (current_region.address < region.address)
 566       start = cursor + 1;
 567     else if (cursor != 0)
 568       end = cursor - 1;
 569     // Avoid underflow:
 570     else
 571       break;
 572   }
 573
 574   // Insert it mc_heap_ignore_region_t:
 575   if (ignored_heap_[cursor].address < region.address)
 576     ++cursor;
 577   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 578 }
 579
 580 void RemoteClient::unignore_heap(void* address, size_t size)
 581 {
 582   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 583
 584   size_type start = 0;
 585   size_type end   = ignored_heap_.size() - 1;
 586
 587   // Binary search:
 588   size_type cursor;
 589   while (start <= end) {
 590     cursor       = (start + end) / 2;
 591     auto& region = ignored_heap_[cursor];
 592     if (region.address < address)
 593       start = cursor + 1;
 594     else if ((char*)region.address <= ((char*)address + size)) {
 595       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 596       return;
 597     } else if (cursor != 0)
 598       end = cursor - 1;
 599     // Avoid underflow:
 600     else
 601       break;
 602   }
 603 }
 604
 605 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 606 {
 607   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 608     frame_name = nullptr;
 609   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 610     info->remove_local_variable(var_name, frame_name);
 611 }
 612
 613 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 614 {
 615   this->refresh_simix();
 616   return smx_actors_infos;
 617 }
 618
 619 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 620 {
 621   this->refresh_simix();
 622   return smx_dead_actors_infos;
 623 }
 624
 625 void RemoteClient::dumpStack()
 626 {
 627   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 628   if (as == nullptr) {
 629     XBT_ERROR("Could not initialize ptrace address space");
 630     return;
 631   }
 632
 633   void* context = _UPT_create(this->pid_);
 634   if (context == nullptr) {
 635     unw_destroy_addr_space(as);
 636     XBT_ERROR("Could not initialize ptrace context");
 637     return;
 638   }
 639
 640   unw_cursor_t cursor;
 641   if (unw_init_remote(&cursor, as, context) != 0) {
 642     _UPT_destroy(context);
 643     unw_destroy_addr_space(as);
 644     XBT_ERROR("Could not initialiez ptrace cursor");
 645     return;
 646   }
 647
 648   simgrid::mc::dumpStack(stderr, cursor);
 649
 650   _UPT_destroy(context);
 651   unw_destroy_addr_space(as);
 652 }
 653
 654 bool RemoteClient::actor_is_enabled(aid_t pid)
 655 {
 656   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 657   process()->getChannel().send(msg);
 658   char buff[MC_MESSAGE_LENGTH];
 659   ssize_t received = process()->getChannel().receive(buff, MC_MESSAGE_LENGTH, true);
 660   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
 661   return ((s_mc_message_int_t*)buff)->value;
 662 }
 663 }
 664 }