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