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     "libexecinfo",
  89     "libflang",
  90     "libevent",
  91     "libgcc_s",
  92     "libgfortran",
  93     "libimf",
  94     "libintlc",
  95     "libirng",
  96     "liblua5.1",
  97     "liblua5.3",
  98     "liblzma",
  99     "libm",
 100     "libpapi",
 101     "libpfm",
 102     "libpthread",
 103     "libquadmath",
 104     "librt",
 105     "libstdc++",
 106     "libsvml",
 107     "libunwind",
 108     "libunwind-x86_64",
 109     "libunwind-x86",
 110     "libunwind-ptrace",
 111     "libz"};
 112
 113 static bool is_simgrid_lib(const std::string& libname)
 114 {
 115   return libname == "libsimgrid";
 116 }
 117
 118 static bool is_filtered_lib(const std::string& libname)
 119 {
 120   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
 121 }
 122
 123 static std::string get_lib_name(const std::string& pathname)
 124 {
 125   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
 126   std::string libname;
 127
 128   size_t pos = map_basename.rfind(".so");
 129   if (pos != std::string::npos) {
 130     // strip the extension (matching regex "\.so.*$")
 131     libname.assign(map_basename, 0, pos);
 132
 133     // strip the version suffix (matching regex "-[.0-9-]*$")
 134     while (true) {
 135       pos = libname.rfind('-');
 136       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
 137         break;
 138       libname.erase(pos);
 139     }
 140   }
 141
 142   return libname;
 143 }
 144
 145 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 146 {
 147   char* buffer       = (char*)buf;
 148   ssize_t real_count = count;
 149   while (count) {
 150     ssize_t res = pread(fd, buffer, count, offset);
 151     if (res > 0) {
 152       count -= res;
 153       buffer += res;
 154       offset += res;
 155     } else if (res == 0)
 156       return -1;
 157     else if (errno != EINTR) {
 158       perror("pread_whole");
 159       return -1;
 160     }
 161   }
 162   return real_count;
 163 }
 164
 165 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 166 {
 167   const char* buffer = (const char*)buf;
 168   ssize_t real_count = count;
 169   while (count) {
 170     ssize_t res = pwrite(fd, buffer, count, offset);
 171     if (res > 0) {
 172       count -= res;
 173       buffer += res;
 174       offset += res;
 175     } else if (res == 0)
 176       return -1;
 177     else if (errno != EINTR)
 178       return -1;
 179   }
 180   return real_count;
 181 }
 182
 183 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 184 static const void* zero_buffer;
 185 static const size_t zero_buffer_size = 10 * 4096;
 186
 187 static void zero_buffer_init()
 188 {
 189   int fd = open("/dev/zero", O_RDONLY);
 190   if (fd < 0)
 191     xbt_die("Could not open /dev/zero");
 192   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 193   if (zero_buffer == MAP_FAILED)
 194     xbt_die("Could not map the zero buffer");
 195   close(fd);
 196 }
 197
 198 int open_vm(pid_t pid, int flags)
 199 {
 200   const size_t buffer_size = 30;
 201   char buffer[buffer_size];
 202   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 203   if (res < 0 || (size_t)res >= buffer_size) {
 204     errno = ENAMETOOLONG;
 205     return -1;
 206   }
 207   return open(buffer, flags);
 208 }
 209
 210 // ***** Process
 211
 212 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
 213 {
 214 }
 215
 216 void RemoteClient::init()
 217 {
 218   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 219   this->init_memory_map_info();
 220
 221   int fd = open_vm(this->pid_, O_RDWR);
 222   if (fd < 0)
 223     xbt_die("Could not open file for process virtual address space");
 224   this->memory_file = fd;
 225
 226   // Read std_heap (is a struct mdesc*):
 227   simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 228   if (not std_heap_var)
 229     xbt_die("No heap information in the target process");
 230   if (not std_heap_var->address)
 231     xbt_die("No constant address for this variable");
 232   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address),
 233                    simgrid::mc::ProcessIndexDisabled);
 234
 235   this->smx_actors_infos.clear();
 236   this->smx_dead_actors_infos.clear();
 237   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 238   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 239   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 240 }
 241
 242 RemoteClient::~RemoteClient()
 243 {
 244   if (this->memory_file >= 0)
 245     close(this->memory_file);
 246
 247   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 248     if (this->unw_underlying_addr_space)
 249       unw_destroy_addr_space(this->unw_underlying_addr_space);
 250     if (this->unw_underlying_context)
 251       _UPT_destroy(this->unw_underlying_context);
 252   }
 253
 254   unw_destroy_addr_space(this->unw_addr_space);
 255 }
 256
 257 /** Refresh the information about the process
 258  *
 259  *  Do not use directly, this is used by the getters when appropriate
 260  *  in order to have fresh data.
 261  */
 262 void RemoteClient::refresh_heap()
 263 {
 264   // Read/dereference/refresh the std_heap pointer:
 265   if (not this->heap)
 266     this->heap = std::unique_ptr<s_xbt_mheap_t>(new s_xbt_mheap_t());
 267   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address), simgrid::mc::ProcessIndexDisabled);
 268   this->cache_flags_ |= RemoteClient::cache_heap;
 269 }
 270
 271 /** Refresh the information about the process
 272  *
 273  *  Do not use direclty, this is used by the getters when appropriate
 274  *  in order to have fresh data.
 275  * */
 276 void RemoteClient::refresh_malloc_info()
 277 {
 278   // Refresh process->heapinfo:
 279   if (this->cache_flags_ & RemoteClient::cache_malloc)
 280     return;
 281   size_t count = this->heap->heaplimit + 1;
 282   if (this->heap_info.size() < count)
 283     this->heap_info.resize(count);
 284   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo),
 285                    simgrid::mc::ProcessIndexDisabled);
 286   this->cache_flags_ |= RemoteClient::cache_malloc;
 287 }
 288
 289 /** @brief Finds the range of the different memory segments and binary paths */
 290 void RemoteClient::init_memory_map_info()
 291 {
 292   XBT_DEBUG("Get debug information ...");
 293   this->maestro_stack_start_ = nullptr;
 294   this->maestro_stack_end_   = nullptr;
 295   this->object_infos.resize(0);
 296   this->binary_info     = nullptr;
 297   this->libsimgrid_info = nullptr;
 298
 299   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 300
 301   const char* current_name = nullptr;
 302
 303   this->object_infos.clear();
 304
 305   for (size_t i = 0; i < maps.size(); i++) {
 306     simgrid::xbt::VmMap const& reg = maps[i];
 307     const char* pathname           = maps[i].pathname.c_str();
 308
 309     // Nothing to do
 310     if (maps[i].pathname.empty()) {
 311       current_name = nullptr;
 312       continue;
 313     }
 314
 315     // [stack], [vvar], [vsyscall], [vdso] ...
 316     if (pathname[0] == '[') {
 317       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 318         this->maestro_stack_start_ = remote(reg.start_addr);
 319         this->maestro_stack_end_   = remote(reg.end_addr);
 320       }
 321       current_name = nullptr;
 322       continue;
 323     }
 324
 325     if (current_name && strcmp(current_name, pathname) == 0)
 326       continue;
 327
 328     current_name = pathname;
 329     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 330       continue;
 331
 332     const bool is_executable = not i;
 333     std::string libname;
 334     if (not is_executable) {
 335       libname = get_lib_name(pathname);
 336       if (is_filtered_lib(libname)) {
 337         continue;
 338       }
 339     }
 340
 341     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 342         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 343     this->object_infos.push_back(info);
 344     if (is_executable)
 345       this->binary_info = info;
 346     else if (is_simgrid_lib(libname))
 347       this->libsimgrid_info = info;
 348   }
 349
 350   // Resolve time (including across different objects):
 351   for (auto const& object_info : this->object_infos)
 352     postProcessObjectInformation(this, object_info.get());
 353
 354   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 355   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 356
 357   XBT_DEBUG("Get debug information done !");
 358 }
 359
 360 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
 361 {
 362   for (auto const& object_info : this->object_infos)
 363     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 364       return object_info;
 365   return nullptr;
 366 }
 367
 368 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(RemotePtr<void> addr) const
 369 {
 370   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 371     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
 372       return info;
 373   return nullptr;
 374 }
 375
 376 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(RemotePtr<void> addr) const
 377 {
 378   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 379     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
 380       return info;
 381   return nullptr;
 382 }
 383
 384 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
 385 {
 386   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 387   return info ? info->find_function((void*)ip.address()) : nullptr;
 388 }
 389
 390 /** Find (one occurrence of) the named variable definition
 391  */
 392 simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
 393 {
 394   // First lookup the variable in the executable shared object.
 395   // A global variable used directly by the executable code from a library
 396   // is reinstanciated in the executable memory .data/.bss.
 397   // We need to look up the variable in the executable first.
 398   if (this->binary_info) {
 399     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 400     simgrid::mc::Variable* var                                  = info->find_variable(name);
 401     if (var)
 402       return var;
 403   }
 404
 405   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 406     simgrid::mc::Variable* var = info->find_variable(name);
 407     if (var)
 408       return var;
 409   }
 410
 411   return nullptr;
 412 }
 413
 414 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
 415 {
 416   simgrid::mc::Variable* var = this->find_variable(name);
 417   xbt_assert(var->address, "No simple location for this variable");
 418   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 419   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 420              size, (size_t)var->type->full_type->byte_size);
 421   this->read_bytes(target, size, remote(var->address));
 422 }
 423
 424 std::string RemoteClient::read_string(RemotePtr<char> address) const
 425 {
 426   if (not address)
 427     return {};
 428
 429   std::vector<char> res(128);
 430   off_t off = 0;
 431
 432   while (1) {
 433     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 434     if (c == -1) {
 435       if (errno == EINTR)
 436         continue;
 437       else
 438         xbt_die("Could not read from from remote process");
 439     }
 440     if (c == 0)
 441       xbt_die("Could not read string from remote process");
 442
 443     void* p = memchr(res.data() + off, '\0', c);
 444     if (p)
 445       return std::string(res.data());
 446
 447     off += c;
 448     if (off == (off_t)res.size())
 449       res.resize(res.size() * 2);
 450   }
 451 }
 452
 453 const void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, int process_index,
 454                                      ReadOptions options) const
 455 {
 456   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 457     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->find_object_info_rw(address);
 458 // Segment overlap is not handled.
 459 #if HAVE_SMPI
 460     if (info.get() && this->privatized(*info)) {
 461       if (process_index < 0)
 462         xbt_die("Missing process index");
 463       if (process_index >= (int)MC_smpi_process_count())
 464         xbt_die("Invalid process index");
 465
 466       // Read smpi_privatization_regions from MCed:
 467       smpi_privatization_region_t remote_smpi_privatization_regions =
 468           mc_model_checker->process().read_variable<smpi_privatization_region_t>("smpi_privatization_regions");
 469
 470       s_smpi_privatization_region_t privatization_region =
 471           mc_model_checker->process().read<s_smpi_privatization_region_t>(
 472               remote(remote_smpi_privatization_regions + process_index));
 473
 474       // Address translation in the privatization segment:
 475       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 476       address       = remote((char*)privatization_region.address + offset);
 477     }
 478 #endif
 479   }
 480   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 481     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 482   return buffer;
 483 }
 484
 485 /** Write data to a process memory
 486  *
 487  *  @param buffer   local memory address (source)
 488  *  @param len      data size
 489  *  @param address  target process memory address (target)
 490  */
 491 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 492 {
 493   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 494     xbt_die("Write to process %lli failed", (long long)this->pid_);
 495 }
 496
 497 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 498 {
 499   pthread_once(&zero_buffer_flag, zero_buffer_init);
 500   while (len) {
 501     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 502     this->write_bytes(zero_buffer, s, address);
 503     address = remote((char*)address.address() + s);
 504     len -= s;
 505   }
 506 }
 507
 508 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 509 {
 510   IgnoredRegion region;
 511   region.addr = addr;
 512   region.size = size;
 513
 514   if (ignored_regions_.empty()) {
 515     ignored_regions_.push_back(region);
 516     return;
 517   }
 518
 519   unsigned int cursor           = 0;
 520   IgnoredRegion* current_region = nullptr;
 521
 522   int start = 0;
 523   int end   = ignored_regions_.size() - 1;
 524   while (start <= end) {
 525     cursor         = (start + end) / 2;
 526     current_region = &ignored_regions_[cursor];
 527     if (current_region->addr == addr) {
 528       if (current_region->size == size)
 529         return;
 530       else if (current_region->size < size)
 531         start = cursor + 1;
 532       else
 533         end = cursor - 1;
 534     } else if (current_region->addr < addr)
 535       start = cursor + 1;
 536     else
 537       end = cursor - 1;
 538   }
 539
 540   std::size_t position;
 541   if (current_region->addr == addr) {
 542     if (current_region->size < size)
 543       position = cursor + 1;
 544     else
 545       position = cursor;
 546   } else if (current_region->addr < addr)
 547     position = cursor + 1;
 548   else
 549     position = cursor;
 550   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 551 }
 552
 553 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 554 {
 555   if (ignored_heap_.empty()) {
 556     ignored_heap_.push_back(std::move(region));
 557     return;
 558   }
 559
 560   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 561
 562   size_type start = 0;
 563   size_type end   = ignored_heap_.size() - 1;
 564
 565   // Binary search the position of insertion:
 566   size_type cursor;
 567   while (start <= end) {
 568     cursor               = start + (end - start) / 2;
 569     auto& current_region = ignored_heap_[cursor];
 570     if (current_region.address == region.address)
 571       return;
 572     else if (current_region.address < region.address)
 573       start = cursor + 1;
 574     else if (cursor != 0)
 575       end = cursor - 1;
 576     // Avoid underflow:
 577     else
 578       break;
 579   }
 580
 581   // Insert it mc_heap_ignore_region_t:
 582   if (ignored_heap_[cursor].address < region.address)
 583     ++cursor;
 584   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 585 }
 586
 587 void RemoteClient::unignore_heap(void* address, size_t size)
 588 {
 589   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 590
 591   size_type start = 0;
 592   size_type end   = ignored_heap_.size() - 1;
 593
 594   // Binary search:
 595   size_type cursor;
 596   while (start <= end) {
 597     cursor       = (start + end) / 2;
 598     auto& region = ignored_heap_[cursor];
 599     if (region.address < address)
 600       start = cursor + 1;
 601     else if ((char*)region.address <= ((char*)address + size)) {
 602       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 603       return;
 604     } else if (cursor != 0)
 605       end = cursor - 1;
 606     // Avoid underflow:
 607     else
 608       break;
 609   }
 610 }
 611
 612 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 613 {
 614   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 615     frame_name = nullptr;
 616   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 617     info->remove_local_variable(var_name, frame_name);
 618 }
 619
 620 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 621 {
 622   this->refresh_simix();
 623   return smx_actors_infos;
 624 }
 625
 626 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 627 {
 628   this->refresh_simix();
 629   return smx_dead_actors_infos;
 630 }
 631
 632 void RemoteClient::dumpStack()
 633 {
 634   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 635   if (as == nullptr) {
 636     XBT_ERROR("Could not initialize ptrace address space");
 637     return;
 638   }
 639
 640   void* context = _UPT_create(this->pid_);
 641   if (context == nullptr) {
 642     unw_destroy_addr_space(as);
 643     XBT_ERROR("Could not initialize ptrace context");
 644     return;
 645   }
 646
 647   unw_cursor_t cursor;
 648   if (unw_init_remote(&cursor, as, context) != 0) {
 649     _UPT_destroy(context);
 650     unw_destroy_addr_space(as);
 651     XBT_ERROR("Could not initialiez ptrace cursor");
 652     return;
 653   }
 654
 655   simgrid::mc::dumpStack(stderr, cursor);
 656
 657   _UPT_destroy(context);
 658   unw_destroy_addr_space(as);
 659 }
 660
 661 bool RemoteClient::actor_is_enabled(aid_t pid)
 662 {
 663   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 664   process()->getChannel().send(msg);
 665   char buff[MC_MESSAGE_LENGTH];
 666   ssize_t received = process()->getChannel().receive(buff, MC_MESSAGE_LENGTH, true);
 667   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
 668   return ((s_mc_message_int_t*)buff)->value;
 669 }
 670 }
 671 }