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     "libargp", /* workarounds for glibc-less systems */
  68     "libasan", /* gcc sanitizers */
  69     "libboost_chrono",
  70     "libboost_context",
  71     "libboost_context-mt",
  72     "libboost_system",
  73     "libboost_thread",
  74     "libboost_timer",
  75     "libboost_unit_test_framework",
  76     "libbz2",
  77     "libc",
  78     "libc++",
  79     "libcdt",
  80     "libcgraph",
  81     "libcrypto",
  82     "libcxxrt",
  83     "libdl",
  84     "libdw",
  85     "libelf",
  86     "libevent",
  87     "libexecinfo",
  88     "libflang",
  89     "libflangrti",
  90     "libgcc_s",
  91     "libgfortran",
  92     "libimf",
  93     "libintlc",
  94     "libirng",
  95     "liblua5.1",
  96     "liblua5.3",
  97     "liblzma",
  98     "libm",
  99     "libpapi",
 100     "libpfm",
 101     "libpthread",
 102     "libquadmath",
 103     "librt",
 104     "libstdc++",
 105     "libsvml",
 106     "libtsan",  /* gcc sanitizers */
 107     "libubsan", /* gcc sanitizers */
 108     "libunwind",
 109     "libunwind-ptrace",
 110     "libunwind-x86",
 111     "libunwind-x86_64",
 112     "libz"};
 113
 114 static bool is_simgrid_lib(const std::string& libname)
 115 {
 116   return libname == "libsimgrid";
 117 }
 118
 119 static bool is_filtered_lib(const std::string& libname)
 120 {
 121   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
 122 }
 123
 124 static std::string get_lib_name(const std::string& pathname)
 125 {
 126   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
 127   std::string libname;
 128
 129   size_t pos = map_basename.rfind(".so");
 130   if (pos != std::string::npos) {
 131     // strip the extension (matching regex "\.so.*$")
 132     libname.assign(map_basename, 0, pos);
 133
 134     // strip the version suffix (matching regex "-[.0-9-]*$")
 135     while (true) {
 136       pos = libname.rfind('-');
 137       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
 138         break;
 139       libname.erase(pos);
 140     }
 141   }
 142
 143   return libname;
 144 }
 145
 146 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
 147 {
 148   char* buffer       = (char*)buf;
 149   ssize_t real_count = count;
 150   while (count) {
 151     ssize_t res = pread(fd, buffer, count, offset);
 152     if (res > 0) {
 153       count -= res;
 154       buffer += res;
 155       offset += res;
 156     } else if (res == 0)
 157       return -1;
 158     else if (errno != EINTR) {
 159       perror("pread_whole");
 160       return -1;
 161     }
 162   }
 163   return real_count;
 164 }
 165
 166 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
 167 {
 168   const char* buffer = (const char*)buf;
 169   ssize_t real_count = count;
 170   while (count) {
 171     ssize_t res = pwrite(fd, buffer, count, offset);
 172     if (res > 0) {
 173       count -= res;
 174       buffer += res;
 175       offset += res;
 176     } else if (res == 0)
 177       return -1;
 178     else if (errno != EINTR)
 179       return -1;
 180   }
 181   return real_count;
 182 }
 183
 184 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
 185 static const void* zero_buffer;
 186 static const size_t zero_buffer_size = 10 * 4096;
 187
 188 static void zero_buffer_init()
 189 {
 190   int fd = open("/dev/zero", O_RDONLY);
 191   if (fd < 0)
 192     xbt_die("Could not open /dev/zero");
 193   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
 194   if (zero_buffer == MAP_FAILED)
 195     xbt_die("Could not map the zero buffer");
 196   close(fd);
 197 }
 198
 199 int open_vm(pid_t pid, int flags)
 200 {
 201   const size_t buffer_size = 30;
 202   char buffer[buffer_size];
 203   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
 204   if (res < 0 || (size_t)res >= buffer_size) {
 205     errno = ENAMETOOLONG;
 206     return -1;
 207   }
 208   return open(buffer, flags);
 209 }
 210
 211 // ***** Process
 212
 213 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
 214 {
 215 }
 216
 217 void RemoteClient::init()
 218 {
 219   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
 220   this->init_memory_map_info();
 221
 222   int fd = open_vm(this->pid_, O_RDWR);
 223   if (fd < 0)
 224     xbt_die("Could not open file for process virtual address space");
 225   this->memory_file = fd;
 226
 227   // Read std_heap (is a struct mdesc*):
 228   simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 229   if (not std_heap_var)
 230     xbt_die("No heap information in the target process");
 231   if (not std_heap_var->address)
 232     xbt_die("No constant address for this variable");
 233   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address),
 234                    simgrid::mc::ProcessIndexDisabled);
 235
 236   this->smx_actors_infos.clear();
 237   this->smx_dead_actors_infos.clear();
 238   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 239   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 240   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 241 }
 242
 243 RemoteClient::~RemoteClient()
 244 {
 245   if (this->memory_file >= 0)
 246     close(this->memory_file);
 247
 248   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 249     if (this->unw_underlying_addr_space)
 250       unw_destroy_addr_space(this->unw_underlying_addr_space);
 251     if (this->unw_underlying_context)
 252       _UPT_destroy(this->unw_underlying_context);
 253   }
 254
 255   unw_destroy_addr_space(this->unw_addr_space);
 256 }
 257
 258 /** Refresh the information about the process
 259  *
 260  *  Do not use directly, this is used by the getters when appropriate
 261  *  in order to have fresh data.
 262  */
 263 void RemoteClient::refresh_heap()
 264 {
 265   // Read/dereference/refresh the std_heap pointer:
 266   if (not this->heap)
 267     this->heap = std::unique_ptr<s_xbt_mheap_t>(new s_xbt_mheap_t());
 268   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address), simgrid::mc::ProcessIndexDisabled);
 269   this->cache_flags_ |= RemoteClient::cache_heap;
 270 }
 271
 272 /** Refresh the information about the process
 273  *
 274  *  Do not use direclty, this is used by the getters when appropriate
 275  *  in order to have fresh data.
 276  * */
 277 void RemoteClient::refresh_malloc_info()
 278 {
 279   // Refresh process->heapinfo:
 280   if (this->cache_flags_ & RemoteClient::cache_malloc)
 281     return;
 282   size_t count = this->heap->heaplimit + 1;
 283   if (this->heap_info.size() < count)
 284     this->heap_info.resize(count);
 285   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo),
 286                    simgrid::mc::ProcessIndexDisabled);
 287   this->cache_flags_ |= RemoteClient::cache_malloc;
 288 }
 289
 290 /** @brief Finds the range of the different memory segments and binary paths */
 291 void RemoteClient::init_memory_map_info()
 292 {
 293   XBT_DEBUG("Get debug information ...");
 294   this->maestro_stack_start_ = nullptr;
 295   this->maestro_stack_end_   = nullptr;
 296   this->object_infos.resize(0);
 297   this->binary_info     = nullptr;
 298   this->libsimgrid_info = nullptr;
 299
 300   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
 301
 302   const char* current_name = nullptr;
 303
 304   this->object_infos.clear();
 305
 306   for (size_t i = 0; i < maps.size(); i++) {
 307     simgrid::xbt::VmMap const& reg = maps[i];
 308     const char* pathname           = maps[i].pathname.c_str();
 309
 310     // Nothing to do
 311     if (maps[i].pathname.empty()) {
 312       current_name = nullptr;
 313       continue;
 314     }
 315
 316     // [stack], [vvar], [vsyscall], [vdso] ...
 317     if (pathname[0] == '[') {
 318       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
 319         this->maestro_stack_start_ = remote(reg.start_addr);
 320         this->maestro_stack_end_   = remote(reg.end_addr);
 321       }
 322       current_name = nullptr;
 323       continue;
 324     }
 325
 326     if (current_name && strcmp(current_name, pathname) == 0)
 327       continue;
 328
 329     current_name = pathname;
 330     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
 331       continue;
 332
 333     const bool is_executable = not i;
 334     std::string libname;
 335     if (not is_executable) {
 336       libname = get_lib_name(pathname);
 337       if (is_filtered_lib(libname)) {
 338         continue;
 339       }
 340     }
 341
 342     std::shared_ptr<simgrid::mc::ObjectInformation> info =
 343         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
 344     this->object_infos.push_back(info);
 345     if (is_executable)
 346       this->binary_info = info;
 347     else if (is_simgrid_lib(libname))
 348       this->libsimgrid_info = info;
 349   }
 350
 351   // Resolve time (including across different objects):
 352   for (auto const& object_info : this->object_infos)
 353     postProcessObjectInformation(this, object_info.get());
 354
 355   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
 356   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
 357
 358   XBT_DEBUG("Get debug information done !");
 359 }
 360
 361 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
 362 {
 363   for (auto const& object_info : this->object_infos)
 364     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
 365       return object_info;
 366   return nullptr;
 367 }
 368
 369 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(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_exec && addr.address() <= (std::uint64_t)info->end_exec)
 373       return info;
 374   return nullptr;
 375 }
 376
 377 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(RemotePtr<void> addr) const
 378 {
 379   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
 380     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
 381       return info;
 382   return nullptr;
 383 }
 384
 385 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
 386 {
 387   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
 388   return info ? info->find_function((void*)ip.address()) : nullptr;
 389 }
 390
 391 /** Find (one occurrence of) the named variable definition
 392  */
 393 simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
 394 {
 395   // First lookup the variable in the executable shared object.
 396   // A global variable used directly by the executable code from a library
 397   // is reinstanciated in the executable memory .data/.bss.
 398   // We need to look up the variable in the executable first.
 399   if (this->binary_info) {
 400     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
 401     simgrid::mc::Variable* var                                  = info->find_variable(name);
 402     if (var)
 403       return var;
 404   }
 405
 406   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
 407     simgrid::mc::Variable* var = info->find_variable(name);
 408     if (var)
 409       return var;
 410   }
 411
 412   return nullptr;
 413 }
 414
 415 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
 416 {
 417   simgrid::mc::Variable* var = this->find_variable(name);
 418   xbt_assert(var->address, "No simple location for this variable");
 419   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 420   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 421              size, (size_t)var->type->full_type->byte_size);
 422   this->read_bytes(target, size, remote(var->address));
 423 }
 424
 425 std::string RemoteClient::read_string(RemotePtr<char> address) const
 426 {
 427   if (not address)
 428     return {};
 429
 430   std::vector<char> res(128);
 431   off_t off = 0;
 432
 433   while (1) {
 434     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 435     if (c == -1) {
 436       if (errno == EINTR)
 437         continue;
 438       else
 439         xbt_die("Could not read from from remote process");
 440     }
 441     if (c == 0)
 442       xbt_die("Could not read string from remote process");
 443
 444     void* p = memchr(res.data() + off, '\0', c);
 445     if (p)
 446       return std::string(res.data());
 447
 448     off += c;
 449     if (off == (off_t)res.size())
 450       res.resize(res.size() * 2);
 451   }
 452 }
 453
 454 const void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, int process_index,
 455                                      ReadOptions options) const
 456 {
 457   if (process_index != simgrid::mc::ProcessIndexDisabled) {
 458     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->find_object_info_rw(address);
 459 // Segment overlap is not handled.
 460 #if HAVE_SMPI
 461     if (info.get() && this->privatized(*info)) {
 462       if (process_index < 0)
 463         xbt_die("Missing process index");
 464       if (process_index >= (int)MC_smpi_process_count())
 465         xbt_die("Invalid process index");
 466
 467       // Read smpi_privatization_regions from MCed:
 468       smpi_privatization_region_t remote_smpi_privatization_regions =
 469           mc_model_checker->process().read_variable<smpi_privatization_region_t>("smpi_privatization_regions");
 470
 471       s_smpi_privatization_region_t privatization_region =
 472           mc_model_checker->process().read<s_smpi_privatization_region_t>(
 473               remote(remote_smpi_privatization_regions + process_index));
 474
 475       // Address translation in the privatization segment:
 476       size_t offset = address.address() - (std::uint64_t)info->start_rw;
 477       address       = remote((char*)privatization_region.address + offset);
 478     }
 479 #endif
 480   }
 481   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 482     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 483   return buffer;
 484 }
 485
 486 /** Write data to a process memory
 487  *
 488  *  @param buffer   local memory address (source)
 489  *  @param len      data size
 490  *  @param address  target process memory address (target)
 491  */
 492 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 493 {
 494   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 495     xbt_die("Write to process %lli failed", (long long)this->pid_);
 496 }
 497
 498 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 499 {
 500   pthread_once(&zero_buffer_flag, zero_buffer_init);
 501   while (len) {
 502     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 503     this->write_bytes(zero_buffer, s, address);
 504     address = remote((char*)address.address() + s);
 505     len -= s;
 506   }
 507 }
 508
 509 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 510 {
 511   IgnoredRegion region;
 512   region.addr = addr;
 513   region.size = size;
 514
 515   if (ignored_regions_.empty()) {
 516     ignored_regions_.push_back(region);
 517     return;
 518   }
 519
 520   unsigned int cursor           = 0;
 521   IgnoredRegion* current_region = nullptr;
 522
 523   int start = 0;
 524   int end   = ignored_regions_.size() - 1;
 525   while (start <= end) {
 526     cursor         = (start + end) / 2;
 527     current_region = &ignored_regions_[cursor];
 528     if (current_region->addr == addr) {
 529       if (current_region->size == size)
 530         return;
 531       else if (current_region->size < size)
 532         start = cursor + 1;
 533       else
 534         end = cursor - 1;
 535     } else if (current_region->addr < addr)
 536       start = cursor + 1;
 537     else
 538       end = cursor - 1;
 539   }
 540
 541   std::size_t position;
 542   if (current_region->addr == addr) {
 543     if (current_region->size < size)
 544       position = cursor + 1;
 545     else
 546       position = cursor;
 547   } else if (current_region->addr < addr)
 548     position = cursor + 1;
 549   else
 550     position = cursor;
 551   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 552 }
 553
 554 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 555 {
 556   if (ignored_heap_.empty()) {
 557     ignored_heap_.push_back(std::move(region));
 558     return;
 559   }
 560
 561   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 562
 563   size_type start = 0;
 564   size_type end   = ignored_heap_.size() - 1;
 565
 566   // Binary search the position of insertion:
 567   size_type cursor;
 568   while (start <= end) {
 569     cursor               = start + (end - start) / 2;
 570     auto& current_region = ignored_heap_[cursor];
 571     if (current_region.address == region.address)
 572       return;
 573     else if (current_region.address < region.address)
 574       start = cursor + 1;
 575     else if (cursor != 0)
 576       end = cursor - 1;
 577     // Avoid underflow:
 578     else
 579       break;
 580   }
 581
 582   // Insert it mc_heap_ignore_region_t:
 583   if (ignored_heap_[cursor].address < region.address)
 584     ++cursor;
 585   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 586 }
 587
 588 void RemoteClient::unignore_heap(void* address, size_t size)
 589 {
 590   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 591
 592   size_type start = 0;
 593   size_type end   = ignored_heap_.size() - 1;
 594
 595   // Binary search:
 596   size_type cursor;
 597   while (start <= end) {
 598     cursor       = (start + end) / 2;
 599     auto& region = ignored_heap_[cursor];
 600     if (region.address < address)
 601       start = cursor + 1;
 602     else if ((char*)region.address <= ((char*)address + size)) {
 603       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 604       return;
 605     } else if (cursor != 0)
 606       end = cursor - 1;
 607     // Avoid underflow:
 608     else
 609       break;
 610   }
 611 }
 612
 613 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 614 {
 615   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 616     frame_name = nullptr;
 617   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 618     info->remove_local_variable(var_name, frame_name);
 619 }
 620
 621 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 622 {
 623   this->refresh_simix();
 624   return smx_actors_infos;
 625 }
 626
 627 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 628 {
 629   this->refresh_simix();
 630   return smx_dead_actors_infos;
 631 }
 632
 633 void RemoteClient::dumpStack()
 634 {
 635   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 636   if (as == nullptr) {
 637     XBT_ERROR("Could not initialize ptrace address space");
 638     return;
 639   }
 640
 641   void* context = _UPT_create(this->pid_);
 642   if (context == nullptr) {
 643     unw_destroy_addr_space(as);
 644     XBT_ERROR("Could not initialize ptrace context");
 645     return;
 646   }
 647
 648   unw_cursor_t cursor;
 649   if (unw_init_remote(&cursor, as, context) != 0) {
 650     _UPT_destroy(context);
 651     unw_destroy_addr_space(as);
 652     XBT_ERROR("Could not initialiez ptrace cursor");
 653     return;
 654   }
 655
 656   simgrid::mc::dumpStack(stderr, cursor);
 657
 658   _UPT_destroy(context);
 659   unw_destroy_addr_space(as);
 660 }
 661
 662 bool RemoteClient::actor_is_enabled(aid_t pid)
 663 {
 664   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 665   process()->getChannel().send(msg);
 666   char buff[MC_MESSAGE_LENGTH];
 667   ssize_t received = process()->getChannel().receive(buff, MC_MESSAGE_LENGTH, true);
 668   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
 669   return ((s_mc_message_int_t*)buff)->value;
 670 }
 671 }
 672 }