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