src/mc/remote/RemoteClient.cpp

   1 /* Copyright (c) 2014-2017. 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 <regex>
  14 #include <string>
  15 #include <vector>
  16
  17 #include <sys/ptrace.h>
  18
  19 #include <cstdio>
  20
  21 #include <fcntl.h>
  22 #include <sys/mman.h> // PROT_*
  23 #include <sys/types.h>
  24 #include <unistd.h>
  25
  26 #include <pthread.h>
  27
  28 #include <libgen.h>
  29
  30 #include <libunwind-ptrace.h>
  31 #include <libunwind.h>
  32
  33 #include "xbt/base.h"
  34 #include "xbt/file.hpp"
  35 #include "xbt/log.h"
  36 #include <xbt/mmalloc.h>
  37
  38 #include "src/mc/mc_smx.hpp"
  39 #include "src/mc/mc_snapshot.hpp"
  40 #include "src/mc/mc_unw.hpp"
  41
  42 #include "src/mc/AddressSpace.hpp"
  43 #include "src/mc/ObjectInformation.hpp"
  44 #include "src/mc/Variable.hpp"
  45 #include "src/mc/remote/RemoteClient.hpp"
  46
  47 using simgrid::mc::remote;
  48
  49 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(mc_process, mc, "MC process information");
  50
  51 namespace simgrid {
  52 namespace mc {
  53
  54 // ***** Helper stuff
  55
  56 // List of library which memory segments are not considered:
  57 static const std::vector<std::string> filtered_libraries = {
  58 #ifdef __linux__
  59     "ld",
  60 #elif defined __FreeBSD__
  61     "ld-elf",
  62     "ld-elf32",
  63     "libkvm",      /* kernel data access library */
  64     "libprocstat", /* process and file information retrieval */
  65     "libthr",      /* thread library */
  66     "libutil",
  67 #endif
  68     "libasan", /* gcc sanitizers */
  69     "libargp", /* workarounds for glibc-less systems */
  70     "libtsan",
  71     "libubsan",
  72     "libbz2",
  73     "libboost_chrono",
  74     "libboost_context",
  75     "libboost_context-mt",
  76     "libboost_system",
  77     "libboost_thread",
  78     "libc",
  79     "libc++",
  80     "libcdt",
  81     "libcgraph",
  82     "libcrypto",
  83     "libcxxrt",
  84     "libdl",
  85     "libdw",
  86     "libelf",
  87     "libevent",
  88     "libgcc_s",
  89     "liblua5.1",
  90     "liblua5.3",
  91     "liblzma",
  92     "libm",
  93     "libpthread",
  94     "librt",
  95     "libstdc++",
  96     "libunwind",
  97     "libunwind-x86_64",
  98     "libunwind-x86",
  99     "libunwind-ptrace",
 100     "libz"};
 101
 102 static bool is_simgrid_lib(const std::string& libname)
 103 {
 104   return libname == "libsimgrid";
 105 }
 106
 107 static bool is_filtered_lib(const std::string& libname)
 108 {
 109   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
 110 }
 111
 112 static std::string get_lib_name(const std::string& pathname)
 113 {
 114   static const std::regex so_re("\\.so[\\.0-9]*$", std::regex_constants::basic);
 115   static const std::regex version_re("-[\\.0-9-]*$", std::regex_constants::basic);
 116   std::string map_basename = simgrid::xbt::Path(pathname).getBasename();
 117   std::string libname;
 118
 119   std::smatch match;
 120   if (std::regex_search(map_basename, match, so_re)) {
 121     libname = match.prefix();
 122
 123     // Strip the version suffix:
 124     if (std::regex_search(libname, match, version_re))
 125       libname = match.prefix();
 126   }
 127
 128   return libname;
 129 }
 130
 131 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 132 {
 133   char* buffer       = (char*)buf;
 134   ssize_t real_count = count;
 135   while (count) {
 136     ssize_t res = pread(fd, buffer, count, offset);
 137     if (res > 0) {
 138       count -= res;
 139       buffer += res;
 140       offset += res;
 141     } else if (res == 0)
 142       return -1;
 143     else if (errno != EINTR) {
 144       perror("pread_whole");
 145       return -1;
 146     }
 147   }
 148   return real_count;
 149 }
 150
 151 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 152 {
 153   const char* buffer = (const char*)buf;
 154   ssize_t real_count = count;
 155   while (count) {
 156     ssize_t res = pwrite(fd, buffer, count, offset);
 157     if (res > 0) {
 158       count -= res;
 159       buffer += res;
 160       offset += res;
 161     } else if (res == 0)
 162       return -1;
 163     else if (errno != EINTR)
 164       return -1;
 165   }
 166   return real_count;
 167 }
 168
 169 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 170 static const void* zero_buffer;
 171 static const size_t zero_buffer_size = 10 * 4096;
 172
 173 static void zero_buffer_init()
 174 {
 175   int fd = open("/dev/zero", O_RDONLY);
 176   if (fd < 0)
 177     xbt_die("Could not open /dev/zero");
 178   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 179   if (zero_buffer == MAP_FAILED)
 180     xbt_die("Could not map the zero buffer");
 181   close(fd);
 182 }
 183
 184 int open_vm(pid_t pid, int flags)
 185 {
 186   const size_t buffer_size = 30;
 187   char buffer[buffer_size];
 188   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 189   if (res < 0 || (size_t)res >= buffer_size) {
 190     errno = ENAMETOOLONG;
 191     return -1;
 192   }
 193   return open(buffer, flags);
 194 }
 195
 196 // ***** Process
 197
 198 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
 199 {
 200 }
 201
 202 void RemoteClient::init()
 203 {
 204   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 205   this->init_memory_map_info();
 206
 207   int fd = open_vm(this->pid_, O_RDWR);
 208   if (fd < 0)
 209     xbt_die("Could not open file for process virtual address space");
 210   this->memory_file = fd;
 211
 212   // Read std_heap (is a struct mdesc*):
 213   simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 214   if (not std_heap_var)
 215     xbt_die("No heap information in the target process");
 216   if (not std_heap_var->address)
 217     xbt_die("No constant address for this variable");
 218   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address),
 219                    simgrid::mc::ProcessIndexDisabled);
 220
 221   this->smx_actors_infos.clear();
 222   this->smx_dead_actors_infos.clear();
 223   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 224   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 225   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 226 }
 227
 228 RemoteClient::~RemoteClient()
 229 {
 230   if (this->memory_file >= 0)
 231     close(this->memory_file);
 232
 233   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 234     if (this->unw_underlying_addr_space)
 235       unw_destroy_addr_space(this->unw_underlying_addr_space);
 236     if (this->unw_underlying_context)
 237       _UPT_destroy(this->unw_underlying_context);
 238   }
 239
 240   unw_destroy_addr_space(this->unw_addr_space);
 241 }
 242
 243 /** Refresh the information about the process
 244  *
 245  *  Do not use directly, this is used by the getters when appropriate
 246  *  in order to have fresh data.
 247  */
 248 void RemoteClient::refresh_heap()
 249 {
 250   // Read/dereference/refresh the std_heap pointer:
 251   if (not this->heap)
 252     this->heap = std::unique_ptr<s_xbt_mheap_t>(new s_xbt_mheap_t());
 253   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address), simgrid::mc::ProcessIndexDisabled);
 254   this->cache_flags_ |= RemoteClient::cache_heap;
 255 }
 256
 257 /** Refresh the information about the process
 258  *
 259  *  Do not use direclty, this is used by the getters when appropriate
 260  *  in order to have fresh data.
 261  * */
 262 void RemoteClient::refresh_malloc_info()
 263 {
 264   // Refresh process->heapinfo:
 265   if (this->cache_flags_ & RemoteClient::cache_malloc)
 266     return;
 267   size_t count = this->heap->heaplimit + 1;
 268   if (this->heap_info.size() < count)
 269     this->heap_info.resize(count);
 270   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo),
 271                    simgrid::mc::ProcessIndexDisabled);
 272   this->cache_flags_ |= RemoteClient::cache_malloc;
 273 }
 274
 275 /** @brief Finds the range of the different memory segments and binary paths */
 276 void RemoteClient::init_memory_map_info()
 277 {
 278   XBT_DEBUG("Get debug information ...");
 279   this->maestro_stack_start_ = nullptr;
 280   this->maestro_stack_end_   = nullptr;
 281   this->object_infos.resize(0);
 282   this->binary_info     = nullptr;
 283   this->libsimgrid_info = nullptr;
 284
 285   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 286
 287   const char* current_name = nullptr;
 288
 289   this->object_infos.clear();
 290
 291   for (size_t i = 0; i < maps.size(); i++) {
 292     simgrid::xbt::VmMap const& reg = maps[i];
 293     const char* pathname           = maps[i].pathname.c_str();
 294
 295     // Nothing to do
 296     if (maps[i].pathname.empty()) {
 297       current_name = nullptr;
 298       continue;
 299     }
 300
 301     // [stack], [vvar], [vsyscall], [vdso] ...
 302     if (pathname[0] == '[') {
 303       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 304         this->maestro_stack_start_ = remote(reg.start_addr);
 305         this->maestro_stack_end_   = remote(reg.end_addr);
 306       }
 307       current_name = nullptr;
 308       continue;
 309     }
 310
 311     if (current_name && strcmp(current_name, pathname) == 0)
 312       continue;
 313
 314     current_name = pathname;
 315     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 316       continue;
 317
 318     const bool is_executable = not i;
 319     std::string libname;
 320     if (not is_executable) {
 321       libname = get_lib_name(pathname);
 322       if (is_filtered_lib(libname)) {
 323         continue;
 324       }
 325     }
 326
 327     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 328         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 329     this->object_infos.push_back(info);
 330     if (is_executable)
 331       this->binary_info = info;
 332     else if (is_simgrid_lib(libname))
 333       this->libsimgrid_info = info;
 334   }
 335
 336   // Resolve time (including across different objects):
 337   for (auto const& object_info : this->object_infos)
 338     postProcessObjectInformation(this, object_info.get());
 339
 340   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 341   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 342
 343   XBT_DEBUG("Get debug information done !");
 344 }
 345
 346 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
 347 {
 348   for (auto const& object_info : this->object_infos)
 349     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 350       return object_info;
 351   return nullptr;
 352 }
 353
 354 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(RemotePtr<void> addr) const
 355 {
 356   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 357     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
 358       return info;
 359   return nullptr;
 360 }
 361
 362 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(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_rw && addr.address() <= (std::uint64_t)info->end_rw)
 366       return info;
 367   return nullptr;
 368 }
 369
 370 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
 371 {
 372   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 373   return info ? info->find_function((void*)ip.address()) : nullptr;
 374 }
 375
 376 /** Find (one occurrence of) the named variable definition
 377  */
 378 simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
 379 {
 380   // First lookup the variable in the executable shared object.
 381   // A global variable used directly by the executable code from a library
 382   // is reinstanciated in the executable memory .data/.bss.
 383   // We need to look up the variable in the executable first.
 384   if (this->binary_info) {
 385     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 386     simgrid::mc::Variable* var                                  = info->find_variable(name);
 387     if (var)
 388       return var;
 389   }
 390
 391   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 392     simgrid::mc::Variable* var = info->find_variable(name);
 393     if (var)
 394       return var;
 395   }
 396
 397   return nullptr;
 398 }
 399
 400 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
 401 {
 402   simgrid::mc::Variable* var = this->find_variable(name);
 403   xbt_assert(var->address, "No simple location for this variable");
 404   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 405   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 406              size, (size_t)var->type->full_type->byte_size);
 407   this->read_bytes(target, size, remote(var->address));
 408 }
 409
 410 std::string RemoteClient::read_string(RemotePtr<char> address) const
 411 {
 412   if (not address)
 413     return {};
 414
 415   // TODO, use std::vector with .data() in C++17 to avoid useless copies
 416   std::vector<char> res(128);
 417   off_t off = 0;
 418
 419   while (1) {
 420     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 421     if (c == -1) {
 422       if (errno == EINTR)
 423         continue;
 424       else
 425         xbt_die("Could not read from from remote process");
 426     }
 427     if (c == 0)
 428       xbt_die("Could not read string from remote process");
 429
 430     void* p = memchr(res.data() + off, '\0', c);
 431     if (p)
 432       return std::string(res.data());
 433
 434     off += c;
 435     if (off == (off_t)res.size())
 436       res.resize(res.size() * 2);
 437   }
 438 }
 439
 440 const void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, int process_index,
 441                                      ReadOptions options) const
 442 {
 443   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 444     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->find_object_info_rw((void*)address.address());
 445 // Segment overlap is not handled.
 446 #if HAVE_SMPI
 447     if (info.get() && this->privatized(*info)) {
 448       if (process_index < 0)
 449         xbt_die("Missing process index");
 450       if (process_index >= (int)MC_smpi_process_count())
 451         xbt_die("Invalid process index");
 452
 453       // Read smpi_privatization_regions from MCed:
 454       smpi_privatization_region_t remote_smpi_privatization_regions =
 455           mc_model_checker->process().read_variable<smpi_privatization_region_t>("smpi_privatization_regions");
 456
 457       s_smpi_privatization_region_t privatization_region =
 458           mc_model_checker->process().read<s_smpi_privatization_region_t>(
 459               remote(remote_smpi_privatization_regions + process_index));
 460
 461       // Address translation in the privatization segment:
 462       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 463       address       = remote((char*)privatization_region.address + offset);
 464     }
 465 #endif
 466   }
 467   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 468     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 469   return buffer;
 470 }
 471
 472 /** Write data to a process memory
 473  *
 474  *  @param buffer   local memory address (source)
 475  *  @param len      data size
 476  *  @param address  target process memory address (target)
 477  */
 478 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 479 {
 480   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 481     xbt_die("Write to process %lli failed", (long long)this->pid_);
 482 }
 483
 484 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 485 {
 486   pthread_once(&zero_buffer_flag, zero_buffer_init);
 487   while (len) {
 488     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 489     this->write_bytes(zero_buffer, s, address);
 490     address = remote((char*)address.address() + s);
 491     len -= s;
 492   }
 493 }
 494
 495 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 496 {
 497   IgnoredRegion region;
 498   region.addr = addr;
 499   region.size = size;
 500
 501   if (ignored_regions_.empty()) {
 502     ignored_regions_.push_back(region);
 503     return;
 504   }
 505
 506   unsigned int cursor           = 0;
 507   IgnoredRegion* current_region = nullptr;
 508
 509   int start = 0;
 510   int end   = ignored_regions_.size() - 1;
 511   while (start <= end) {
 512     cursor         = (start + end) / 2;
 513     current_region = &ignored_regions_[cursor];
 514     if (current_region->addr == addr) {
 515       if (current_region->size == size)
 516         return;
 517       else if (current_region->size < size)
 518         start = cursor + 1;
 519       else
 520         end = cursor - 1;
 521     } else if (current_region->addr < addr)
 522       start = cursor + 1;
 523     else
 524       end = cursor - 1;
 525   }
 526
 527   std::size_t position;
 528   if (current_region->addr == addr) {
 529     if (current_region->size < size)
 530       position = cursor + 1;
 531     else
 532       position = cursor;
 533   } else if (current_region->addr < addr)
 534     position = cursor + 1;
 535   else
 536     position = cursor;
 537   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 538 }
 539
 540 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 541 {
 542   if (ignored_heap_.empty()) {
 543     ignored_heap_.push_back(std::move(region));
 544     return;
 545   }
 546
 547   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 548
 549   size_type start = 0;
 550   size_type end   = ignored_heap_.size() - 1;
 551
 552   // Binary search the position of insertion:
 553   size_type cursor;
 554   while (start <= end) {
 555     cursor               = start + (end - start) / 2;
 556     auto& current_region = ignored_heap_[cursor];
 557     if (current_region.address == region.address)
 558       return;
 559     else if (current_region.address < region.address)
 560       start = cursor + 1;
 561     else if (cursor != 0)
 562       end = cursor - 1;
 563     // Avoid underflow:
 564     else
 565       break;
 566   }
 567
 568   // Insert it mc_heap_ignore_region_t:
 569   if (ignored_heap_[cursor].address < region.address)
 570     ++cursor;
 571   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 572 }
 573
 574 void RemoteClient::unignore_heap(void* address, size_t size)
 575 {
 576   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 577
 578   size_type start = 0;
 579   size_type end   = ignored_heap_.size() - 1;
 580
 581   // Binary search:
 582   size_type cursor;
 583   while (start <= end) {
 584     cursor       = (start + end) / 2;
 585     auto& region = ignored_heap_[cursor];
 586     if (region.address == address) {
 587       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 588       return;
 589     } else if (region.address < address)
 590       start = cursor + 1;
 591     else if ((char*)region.address <= ((char*)address + size)) {
 592       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 593       return;
 594     } else if (cursor != 0)
 595       end = cursor - 1;
 596     // Avoid underflow:
 597     else
 598       break;
 599   }
 600 }
 601
 602 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 603 {
 604   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 605     frame_name = nullptr;
 606   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 607     info->remove_local_variable(var_name, frame_name);
 608 }
 609
 610 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 611 {
 612   this->refresh_simix();
 613   return smx_actors_infos;
 614 }
 615
 616 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 617 {
 618   this->refresh_simix();
 619   return smx_dead_actors_infos;
 620 }
 621
 622 void RemoteClient::dumpStack()
 623 {
 624   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 625   if (as == nullptr) {
 626     XBT_ERROR("Could not initialize ptrace address space");
 627     return;
 628   }
 629
 630   void* context = _UPT_create(this->pid_);
 631   if (context == nullptr) {
 632     unw_destroy_addr_space(as);
 633     XBT_ERROR("Could not initialize ptrace context");
 634     return;
 635   }
 636
 637   unw_cursor_t cursor;
 638   if (unw_init_remote(&cursor, as, context) != 0) {
 639     _UPT_destroy(context);
 640     unw_destroy_addr_space(as);
 641     XBT_ERROR("Could not initialiez ptrace cursor");
 642     return;
 643   }
 644
 645   simgrid::mc::dumpStack(stderr, cursor);
 646
 647   _UPT_destroy(context);
 648   unw_destroy_addr_space(as);
 649   return;
 650 }
 651
 652 bool RemoteClient::actor_is_enabled(aid_t pid)
 653 {
 654   s_mc_message_actor_enabled msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 655   process()->getChannel().send(msg);
 656   char buff[MC_MESSAGE_LENGTH];
 657   ssize_t received = process()->getChannel().receive(buff, MC_MESSAGE_LENGTH, true);
 658   xbt_assert(received == sizeof(s_mc_message_int), "Unexpected size in answer to ACTOR_ENABLED");
 659   return ((mc_message_int_t*)buff)->value;
 660 }
 661 }
 662 }