1 /* Copyright (c) 2014-2021. The SimGrid Team. All rights reserved. */
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. */
6 #define _FILE_OFFSET_BITS 64 /* needed for pread_whole to work as expected on 32bits */
8 #include "src/mc/remote/RemoteProcess.hpp"
10 #include "src/mc/sosp/Snapshot.hpp"
11 #include "xbt/file.hpp"
15 #include <libunwind-ptrace.h>
16 #include <sys/mman.h> // PROT_*
24 using simgrid::mc::remote;
26 XBT_LOG_NEW_DEFAULT_SUBCATEGORY(mc_process, mc, "MC process information");
33 // List of library which memory segments are not considered:
34 static const std::vector<std::string> filtered_libraries = {
37 #elif defined __FreeBSD__
40 "libkvm", /* kernel data access library */
41 "libprocstat", /* process and file information retrieval */
42 "libthr", /* thread library */
45 "libargp", /* workarounds for glibc-less systems */
46 "libasan", /* gcc sanitizers */
50 "libboost_context-mt",
51 "libboost_stacktrace_addr2line",
52 "libboost_stacktrace_backtrace",
98 "libkrb5support", /*odd behaviour on fedora rawhide ... remove these when fixed*/
131 "libtsan", /* gcc sanitizers */
132 "libubsan", /* gcc sanitizers */
142 static bool is_filtered_lib(const std::string& libname)
144 return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
147 static std::string get_lib_name(const std::string& pathname)
149 std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
152 size_t pos = map_basename.rfind(".so");
153 if (pos != std::string::npos) {
154 // strip the extension (matching regex "\.so.*$")
155 libname.assign(map_basename, 0, pos);
157 // strip the version suffix (matching regex "-[.0-9-]*$")
159 pos = libname.rfind('-');
160 if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
169 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
171 auto* buffer = static_cast<char*>(buf);
172 ssize_t real_count = count;
174 ssize_t res = pread(fd, buffer, count, offset);
181 else if (errno != EINTR) {
182 XBT_ERROR("pread_whole: %s", strerror(errno));
189 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
191 const auto* buffer = static_cast<const char*>(buf);
192 ssize_t real_count = count;
194 ssize_t res = pwrite(fd, buffer, count, offset);
201 else if (errno != EINTR) {
202 XBT_ERROR("pwrite_whole: %s", strerror(errno));
209 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
210 static const void* zero_buffer;
211 static const size_t zero_buffer_size = 10 * 4096;
213 static void zero_buffer_init()
215 int fd = open("/dev/zero", O_RDONLY);
217 xbt_die("Could not open /dev/zero");
218 zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
219 if (zero_buffer == MAP_FAILED)
220 xbt_die("Could not map the zero buffer");
224 int open_vm(pid_t pid, int flags)
226 std::string buffer = "/proc/" + std::to_string(pid) + "/mem";
227 return open(buffer.c_str(), flags);
230 // ***** RemoteProcess
232 RemoteProcess::RemoteProcess(pid_t pid) : AddressSpace(this), pid_(pid), running_(true) {}
234 void RemoteProcess::init()
236 this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
237 this->init_memory_map_info();
239 int fd = open_vm(this->pid_, O_RDWR);
240 xbt_assert(fd >= 0, "Could not open file for process virtual address space");
241 this->memory_file = fd;
243 // Read std_heap (is a struct mdesc*):
244 const simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
245 xbt_assert(std_heap_var, "No heap information in the target process");
246 xbt_assert(std_heap_var->address, "No constant address for this variable");
247 this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address));
249 this->smx_actors_infos.clear();
250 this->smx_dead_actors_infos.clear();
251 this->unw_addr_space = simgrid::mc::UnwindContext::createUnwindAddressSpace();
252 this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
253 this->unw_underlying_context = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
256 RemoteProcess::~RemoteProcess()
258 if (this->memory_file >= 0)
259 close(this->memory_file);
261 if (this->unw_underlying_addr_space != unw_local_addr_space) {
262 if (this->unw_underlying_addr_space)
263 unw_destroy_addr_space(this->unw_underlying_addr_space);
264 if (this->unw_underlying_context)
265 _UPT_destroy(this->unw_underlying_context);
268 unw_destroy_addr_space(this->unw_addr_space);
271 /** Refresh the information about the process
273 * Do not use directly, this is used by the getters when appropriate
274 * in order to have fresh data.
276 void RemoteProcess::refresh_heap()
278 // Read/dereference/refresh the std_heap pointer:
280 this->heap = std::make_unique<s_xbt_mheap_t>();
281 this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address));
282 this->cache_flags_ |= RemoteProcess::cache_heap;
285 /** Refresh the information about the process
287 * Do not use directly, this is used by the getters when appropriate
288 * in order to have fresh data.
290 void RemoteProcess::refresh_malloc_info()
292 // Refresh process->heapinfo:
293 if (this->cache_flags_ & RemoteProcess::cache_malloc)
295 size_t count = this->heap->heaplimit + 1;
296 if (this->heap_info.size() < count)
297 this->heap_info.resize(count);
298 this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo));
299 this->cache_flags_ |= RemoteProcess::cache_malloc;
302 /** @brief Finds the range of the different memory segments and binary paths */
303 void RemoteProcess::init_memory_map_info()
305 XBT_DEBUG("Get debug information ...");
306 this->maestro_stack_start_ = nullptr;
307 this->maestro_stack_end_ = nullptr;
308 this->object_infos.resize(0);
309 this->binary_info = nullptr;
311 std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
313 const char* current_name = nullptr;
315 this->object_infos.clear();
317 for (size_t i = 0; i < maps.size(); i++) {
318 simgrid::xbt::VmMap const& reg = maps[i];
319 const char* pathname = maps[i].pathname.c_str();
322 if (maps[i].pathname.empty()) {
323 current_name = nullptr;
327 // [stack], [vvar], [vsyscall], [vdso] ...
328 if (pathname[0] == '[') {
329 if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
330 this->maestro_stack_start_ = remote(reg.start_addr);
331 this->maestro_stack_end_ = remote(reg.end_addr);
333 current_name = nullptr;
337 if (current_name && strcmp(current_name, pathname) == 0)
340 current_name = pathname;
341 if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
344 const bool is_executable = not i;
346 if (not is_executable) {
347 libname = get_lib_name(pathname);
348 if (is_filtered_lib(libname)) {
353 std::shared_ptr<simgrid::mc::ObjectInformation> info =
354 simgrid::mc::createObjectInformation(this->memory_map_, pathname);
355 this->object_infos.push_back(info);
357 this->binary_info = info;
360 xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
361 xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
363 XBT_DEBUG("Get debug information done !");
366 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteProcess::find_object_info(RemotePtr<void> addr) const
368 for (auto const& object_info : this->object_infos)
369 if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
374 std::shared_ptr<ObjectInformation> RemoteProcess::find_object_info_exec(RemotePtr<void> addr) const
376 for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
377 if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
382 std::shared_ptr<ObjectInformation> RemoteProcess::find_object_info_rw(RemotePtr<void> addr) const
384 for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
385 if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
390 simgrid::mc::Frame* RemoteProcess::find_function(RemotePtr<void> ip) const
392 std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
393 return info ? info->find_function((void*)ip.address()) : nullptr;
396 /** Find (one occurrence of) the named variable definition
398 const simgrid::mc::Variable* RemoteProcess::find_variable(const char* name) const
400 // First lookup the variable in the executable shared object.
401 // A global variable used directly by the executable code from a library
402 // is reinstantiated in the executable memory .data/.bss.
403 // We need to look up the variable in the executable first.
404 if (this->binary_info) {
405 std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
406 const simgrid::mc::Variable* var = info->find_variable(name);
411 for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
412 const simgrid::mc::Variable* var = info->find_variable(name);
420 void RemoteProcess::read_variable(const char* name, void* target, size_t size) const
422 const simgrid::mc::Variable* var = this->find_variable(name);
423 xbt_assert(var, "Variable %s not found", name);
424 xbt_assert(var->address, "No simple location for this variable");
426 if (not var->type->full_type) // Try to resolve this type. The needed ObjectInfo was maybe (lazily) loaded recently
427 for (auto const& object_info : this->object_infos)
428 postProcessObjectInformation(this, object_info.get());
429 xbt_assert(var->type->full_type, "Partial type for %s (even after re-resolving types), cannot retrieve its size.",
431 xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, received %zu).",
432 name, size, (size_t)var->type->full_type->byte_size);
433 this->read_bytes(target, size, remote(var->address));
436 std::string RemoteProcess::read_string(RemotePtr<char> address) const
441 std::vector<char> res(128);
445 ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
446 if (c == -1 && errno == EINTR)
448 xbt_assert(c > 0, "Could not read string from remote process");
450 const void* p = memchr(res.data() + off, '\0', c);
452 return std::string(res.data());
455 if (off == (off_t)res.size())
456 res.resize(res.size() * 2);
460 void* RemoteProcess::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, ReadOptions /*options*/) const
462 if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
463 xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
467 /** Write data to a process memory
469 * @param buffer local memory address (source)
470 * @param len data size
471 * @param address target process memory address (target)
473 void RemoteProcess::write_bytes(const void* buffer, size_t len, RemotePtr<void> address) const
475 if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
476 xbt_die("Write to process %lli failed", (long long)this->pid_);
479 void RemoteProcess::clear_bytes(RemotePtr<void> address, size_t len) const
481 pthread_once(&zero_buffer_flag, zero_buffer_init);
483 size_t s = len > zero_buffer_size ? zero_buffer_size : len;
484 this->write_bytes(zero_buffer, s, address);
485 address = remote((char*)address.address() + s);
490 void RemoteProcess::ignore_region(std::uint64_t addr, std::size_t size)
492 IgnoredRegion region;
496 auto pos = std::lower_bound(ignored_regions_.begin(), ignored_regions_.end(), region,
497 [](auto const& reg1, auto const& reg2) {
498 return reg1.addr < reg2.addr || (reg1.addr == reg2.addr && reg1.size < reg2.size);
500 if (pos == ignored_regions_.end() || pos->addr != addr || pos->size != size)
501 ignored_regions_.insert(pos, region);
504 void RemoteProcess::ignore_heap(IgnoredHeapRegion const& region)
506 // Binary search the position of insertion:
507 auto pos = std::lower_bound(ignored_heap_.begin(), ignored_heap_.end(), region.address,
508 [](auto const& reg, auto const* addr) { return reg.address < addr; });
509 if (pos == ignored_heap_.end() || pos->address != region.address) {
511 ignored_heap_.insert(pos, region);
515 void RemoteProcess::unignore_heap(void* address, size_t size)
518 auto pos = std::lower_bound(ignored_heap_.begin(), ignored_heap_.end(), address,
519 [](auto const& reg, auto const* addr) { return reg.address < addr; });
520 if (pos != ignored_heap_.end() && static_cast<char*>(pos->address) <= static_cast<char*>(address) + size)
521 ignored_heap_.erase(pos);
524 void RemoteProcess::ignore_local_variable(const char* var_name, const char* frame_name) const
526 if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
527 frame_name = nullptr;
528 for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
529 info->remove_local_variable(var_name, frame_name);
532 std::vector<simgrid::mc::ActorInformation>& RemoteProcess::actors()
534 this->refresh_simix();
535 return smx_actors_infos;
538 std::vector<simgrid::mc::ActorInformation>& RemoteProcess::dead_actors()
540 this->refresh_simix();
541 return smx_dead_actors_infos;
544 void RemoteProcess::dump_stack() const
546 unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
548 XBT_ERROR("Could not initialize ptrace address space");
552 void* context = _UPT_create(this->pid_);
553 if (context == nullptr) {
554 unw_destroy_addr_space(as);
555 XBT_ERROR("Could not initialize ptrace context");
560 if (unw_init_remote(&cursor, as, context) != 0) {
561 _UPT_destroy(context);
562 unw_destroy_addr_space(as);
563 XBT_ERROR("Could not initialiez ptrace cursor");
567 simgrid::mc::dumpStack(stderr, &cursor);
569 _UPT_destroy(context);
570 unw_destroy_addr_space(as);
573 unsigned long RemoteProcess::get_maxpid() const
575 static const char* name = nullptr;
577 name = "simgrid::kernel::actor::maxpid";
578 if (find_variable(name) == nullptr)
579 name = "maxpid"; // We seem to miss the namespaces when compiling with GCC
581 unsigned long maxpid;
582 read_variable(name, &maxpid, sizeof(maxpid));
586 void RemoteProcess::get_actor_vectors(RemotePtr<s_xbt_dynar_t>& actors, RemotePtr<s_xbt_dynar_t>& dead_actors)
588 static_assert(std::is_same<std::unique_ptr<simgrid::simix::Global>, decltype(simix_global)>::value,
589 "Unexpected type for simix_global");
590 static_assert(sizeof(simix_global) == sizeof(simgrid::simix::Global*), "Bad size for simix_global");
592 // TODO, avoid to reload `&simix_global`, `simix_global`, `*simix_global`
593 RemotePtr<simgrid::simix::Global> simix_global_p{this->read_variable<simgrid::simix::Global*>("simix_global")};
594 Remote<simgrid::simix::Global> simix_global = this->read<simgrid::simix::Global>(simix_global_p);
596 actors = remote(simix_global.get_buffer()->actors_vector);
597 dead_actors = remote(simix_global.get_buffer()->dead_actors_vector);
600 } // namespace simgrid