src/mc/remote/RemoteClient.cpp

   1 /* Copyright (c) 2014-2020. 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 "src/mc/remote/RemoteClient.hpp"
   9
  10 #include "src/mc/mc_smx.hpp"
  11 #include "src/mc/sosp/Snapshot.hpp"
  12 #include "xbt/file.hpp"
  13 #include "xbt/log.h"
  14
  15 #include <fcntl.h>
  16 #include <libunwind-ptrace.h>
  17 #include <sys/mman.h> // PROT_*
  18
  19 using simgrid::mc::remote;
  20
  21 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(mc_process, mc, "MC process information");
  22
  23 namespace simgrid {
  24 namespace mc {
  25
  26 // ***** Helper stuff
  27
  28 // List of library which memory segments are not considered:
  29 static const std::vector<std::string> filtered_libraries = {
  30 #ifdef __linux__
  31     "ld",
  32 #elif defined __FreeBSD__
  33     "ld-elf",
  34     "ld-elf32",
  35     "libkvm",      /* kernel data access library */
  36     "libprocstat", /* process and file information retrieval */
  37     "libthr",      /* thread library */
  38     "libutil",
  39 #endif
  40     "libargp", /* workarounds for glibc-less systems */
  41     "libasan", /* gcc sanitizers */
  42     "libboost_chrono",
  43     "libboost_context",
  44     "libboost_context-mt",
  45     "libboost_stacktrace_addr2line",
  46     "libboost_stacktrace_backtrace",
  47     "libboost_system",
  48     "libboost_thread",
  49     "libboost_timer",
  50     "libbz2",
  51     "libc",
  52     "libc++",
  53     "libcdt",
  54     "libcgraph",
  55     "libcrypto",
  56     "libcxxrt",
  57     "libdl",
  58     "libdw",
  59     "libelf",
  60     "libevent",
  61     "libexecinfo",
  62     "libflang",
  63     "libflangrti",
  64     "libgcc_s",
  65     "libgfortran",
  66     "libimf",
  67     "libintlc",
  68     "libirng",
  69     "liblua5.1",
  70     "liblua5.3",
  71     "liblzma",
  72     "libm",
  73     "libomp",
  74     "libpapi",
  75     "libpcre2",
  76     "libpfm",
  77     "libpgmath",
  78     "libpthread",
  79     "libquadmath",
  80     "librt",
  81     "libstdc++",
  82     "libsvml",
  83     "libtsan",  /* gcc sanitizers */
  84     "libubsan", /* gcc sanitizers */
  85     "libunwind",
  86     "libunwind-ptrace",
  87     "libunwind-x86",
  88     "libunwind-x86_64",
  89     "libz",
  90     "libkrb5support", /*odd behaviour on fedora rawhide ... remove these when fixed*/
  91     "libkeyutils",
  92     "libunistring",
  93     "libbrotlidec",
  94     "liblber",
  95     "libldap",
  96     "libcom_err",
  97     "libk5crypto",
  98     "libkrb5",
  99     "libgssapi_krb5",
 100     "libssl",
 101     "libpsl",
 102     "libssh",
 103     "libidn2",
 104     "libnghttp2",
 105     "libcurl",
 106     "libdebuginfod",
 107     "libbrotlicommon",
 108     "libsasl2",
 109     "libresolv",
 110     "libcrypt",
 111     "libselinux"
 112 };
 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   xbt_assert(fd >= 0, "Could not open file for process virtual address space");
 224   this->memory_file = fd;
 225
 226   // Read std_heap (is a struct mdesc*):
 227   const simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
 228   xbt_assert(std_heap_var, "No heap information in the target process");
 229   xbt_assert(std_heap_var->address, "No constant address for this variable");
 230   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address));
 231
 232   this->smx_actors_infos.clear();
 233   this->smx_dead_actors_infos.clear();
 234   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
 235   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
 236   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
 237 }
 238
 239 RemoteClient::~RemoteClient()
 240 {
 241   if (this->memory_file >= 0)
 242     close(this->memory_file);
 243
 244   if (this->unw_underlying_addr_space != unw_local_addr_space) {
 245     if (this->unw_underlying_addr_space)
 246       unw_destroy_addr_space(this->unw_underlying_addr_space);
 247     if (this->unw_underlying_context)
 248       _UPT_destroy(this->unw_underlying_context);
 249   }
 250
 251   unw_destroy_addr_space(this->unw_addr_space);
 252 }
 253
 254 /** Refresh the information about the process
 255  *
 256  *  Do not use directly, this is used by the getters when appropriate
 257  *  in order to have fresh data.
 258  */
 259 void RemoteClient::refresh_heap()
 260 {
 261   // Read/dereference/refresh the std_heap pointer:
 262   if (not this->heap)
 263     this->heap.reset(new s_xbt_mheap_t());
 264   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address));
 265   this->cache_flags_ |= RemoteClient::cache_heap;
 266 }
 267
 268 /** Refresh the information about the process
 269  *
 270  *  Do not use directly, this is used by the getters when appropriate
 271  *  in order to have fresh data.
 272  * */
 273 void RemoteClient::refresh_malloc_info()
 274 {
 275   // Refresh process->heapinfo:
 276   if (this->cache_flags_ & RemoteClient::cache_malloc)
 277     return;
 278   size_t count = this->heap->heaplimit + 1;
 279   if (this->heap_info.size() < count)
 280     this->heap_info.resize(count);
 281   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo));
 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 const 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 reinstantiated 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     const 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     const 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   const simgrid::mc::Variable* var = this->find_variable(name);
 413   xbt_assert(var, "Variable %s not found", name);
 414   xbt_assert(var->address, "No simple location for this variable");
 415   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
 416   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
 417              size, (size_t)var->type->full_type->byte_size);
 418   this->read_bytes(target, size, remote(var->address));
 419 }
 420
 421 std::string RemoteClient::read_string(RemotePtr<char> address) const
 422 {
 423   if (not address)
 424     return {};
 425
 426   std::vector<char> res(128);
 427   off_t off = 0;
 428
 429   while (1) {
 430     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
 431     if (c == -1 && errno == EINTR)
 432       continue;
 433     xbt_assert(c > 0, "Could not read string from remote process");
 434
 435     const void* p = memchr(res.data() + off, '\0', c);
 436     if (p)
 437       return std::string(res.data());
 438
 439     off += c;
 440     if (off == (off_t)res.size())
 441       res.resize(res.size() * 2);
 442   }
 443 }
 444
 445 void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, ReadOptions /*options*/) const
 446 {
 447   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
 448     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
 449   return buffer;
 450 }
 451
 452 /** Write data to a process memory
 453  *
 454  *  @param buffer   local memory address (source)
 455  *  @param len      data size
 456  *  @param address  target process memory address (target)
 457  */
 458 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
 459 {
 460   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
 461     xbt_die("Write to process %lli failed", (long long)this->pid_);
 462 }
 463
 464 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
 465 {
 466   pthread_once(&zero_buffer_flag, zero_buffer_init);
 467   while (len) {
 468     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
 469     this->write_bytes(zero_buffer, s, address);
 470     address = remote((char*)address.address() + s);
 471     len -= s;
 472   }
 473 }
 474
 475 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
 476 {
 477   IgnoredRegion region;
 478   region.addr = addr;
 479   region.size = size;
 480
 481   if (ignored_regions_.empty()) {
 482     ignored_regions_.push_back(region);
 483     return;
 484   }
 485
 486   unsigned int cursor           = 0;
 487   const IgnoredRegion* current_region = nullptr;
 488
 489   int start = 0;
 490   int end   = ignored_regions_.size() - 1;
 491   while (start <= end) {
 492     cursor         = (start + end) / 2;
 493     current_region = &ignored_regions_[cursor];
 494     if (current_region->addr == addr) {
 495       if (current_region->size == size)
 496         return;
 497       else if (current_region->size < size)
 498         start = cursor + 1;
 499       else
 500         end = cursor - 1;
 501     } else if (current_region->addr < addr)
 502       start = cursor + 1;
 503     else
 504       end = cursor - 1;
 505   }
 506
 507   std::size_t position;
 508   if (current_region->addr == addr) {
 509     if (current_region->size < size)
 510       position = cursor + 1;
 511     else
 512       position = cursor;
 513   } else if (current_region->addr < addr)
 514     position = cursor + 1;
 515   else
 516     position = cursor;
 517   ignored_regions_.insert(ignored_regions_.begin() + position, region);
 518 }
 519
 520 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
 521 {
 522   if (ignored_heap_.empty()) {
 523     ignored_heap_.push_back(std::move(region));
 524     return;
 525   }
 526
 527   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
 528
 529   size_type start = 0;
 530   size_type end   = ignored_heap_.size() - 1;
 531
 532   // Binary search the position of insertion:
 533   size_type cursor;
 534   while (start <= end) {
 535     cursor               = start + (end - start) / 2;
 536     auto const& current_region = ignored_heap_[cursor];
 537     if (current_region.address == region.address)
 538       return;
 539     else if (current_region.address < region.address)
 540       start = cursor + 1;
 541     else if (cursor != 0)
 542       end = cursor - 1;
 543     // Avoid underflow:
 544     else
 545       break;
 546   }
 547
 548   // Insert it mc_heap_ignore_region_t:
 549   if (ignored_heap_[cursor].address < region.address)
 550     ++cursor;
 551   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
 552 }
 553
 554 void RemoteClient::unignore_heap(void* address, size_t size)
 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:
 562   size_type cursor;
 563   while (start <= end) {
 564     cursor       = (start + end) / 2;
 565     auto const& region = ignored_heap_[cursor];
 566     if (region.address < address)
 567       start = cursor + 1;
 568     else if ((char*)region.address <= ((char*)address + size)) {
 569       ignored_heap_.erase(ignored_heap_.begin() + cursor);
 570       return;
 571     } else if (cursor != 0)
 572       end = cursor - 1;
 573     // Avoid underflow:
 574     else
 575       break;
 576   }
 577 }
 578
 579 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
 580 {
 581   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
 582     frame_name = nullptr;
 583   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
 584     info->remove_local_variable(var_name, frame_name);
 585 }
 586
 587 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
 588 {
 589   this->refresh_simix();
 590   return smx_actors_infos;
 591 }
 592
 593 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
 594 {
 595   this->refresh_simix();
 596   return smx_dead_actors_infos;
 597 }
 598
 599 void RemoteClient::dump_stack()
 600 {
 601   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
 602   if (as == nullptr) {
 603     XBT_ERROR("Could not initialize ptrace address space");
 604     return;
 605   }
 606
 607   void* context = _UPT_create(this->pid_);
 608   if (context == nullptr) {
 609     unw_destroy_addr_space(as);
 610     XBT_ERROR("Could not initialize ptrace context");
 611     return;
 612   }
 613
 614   unw_cursor_t cursor;
 615   if (unw_init_remote(&cursor, as, context) != 0) {
 616     _UPT_destroy(context);
 617     unw_destroy_addr_space(as);
 618     XBT_ERROR("Could not initialiez ptrace cursor");
 619     return;
 620   }
 621
 622   simgrid::mc::dumpStack(stderr, &cursor);
 623
 624   _UPT_destroy(context);
 625   unw_destroy_addr_space(as);
 626 }
 627
 628 bool RemoteClient::actor_is_enabled(aid_t pid)
 629 {
 630   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
 631   process()->get_channel().send(msg);
 632   char buff[MC_MESSAGE_LENGTH];
 633   ssize_t received = process()->get_channel().receive(buff, MC_MESSAGE_LENGTH, true);
 634   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
 635   return ((s_mc_message_int_t*)buff)->value;
 636 }
 637 }
 638 }