Logo AND Algorithmique Numérique Distribuée

Public GIT Repository
c876ccea2a7b151a969d12569fb1ec341148ca82
[simgrid.git] / 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_stacktrace_backtrace",
73     "libboost_system",
74     "libboost_thread",
75     "libboost_timer",
76     "libboost_unit_test_framework",
77     "libbz2",
78     "libc",
79     "libc++",
80     "libcdt",
81     "libcgraph",
82     "libcrypto",
83     "libcxxrt",
84     "libdl",
85     "libdw",
86     "libelf",
87     "libevent",
88     "libexecinfo",
89     "libflang",
90     "libflangrti",
91     "libgcc_s",
92     "libgfortran",
93     "libimf",
94     "libintlc",
95     "libirng",
96     "liblua5.1",
97     "liblua5.3",
98     "liblzma",
99     "libm",
100     "libomp",
101     "libpapi",
102     "libpfm",
103     "libpthread",
104     "libquadmath",
105     "librt",
106     "libstdc++",
107     "libsvml",
108     "libtsan",  /* gcc sanitizers */
109     "libubsan", /* gcc sanitizers */
110     "libunwind",
111     "libunwind-ptrace",
112     "libunwind-x86",
113     "libunwind-x86_64",
114     "libz"};
115
116 static bool is_simgrid_lib(const std::string& libname)
117 {
118   return libname == "libsimgrid";
119 }
120
121 static bool is_filtered_lib(const std::string& libname)
122 {
123   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
124 }
125
126 static std::string get_lib_name(const std::string& pathname)
127 {
128   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
129   std::string libname;
130
131   size_t pos = map_basename.rfind(".so");
132   if (pos != std::string::npos) {
133     // strip the extension (matching regex "\.so.*$")
134     libname.assign(map_basename, 0, pos);
135
136     // strip the version suffix (matching regex "-[.0-9-]*$")
137     while (true) {
138       pos = libname.rfind('-');
139       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
140         break;
141       libname.erase(pos);
142     }
143   }
144
145   return libname;
146 }
147
148 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
149 {
150   char* buffer       = (char*)buf;
151   ssize_t real_count = count;
152   while (count) {
153     ssize_t res = pread(fd, buffer, count, offset);
154     if (res > 0) {
155       count -= res;
156       buffer += res;
157       offset += res;
158     } else if (res == 0)
159       return -1;
160     else if (errno != EINTR) {
161       perror("pread_whole");
162       return -1;
163     }
164   }
165   return real_count;
166 }
167
168 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
169 {
170   const char* buffer = (const char*)buf;
171   ssize_t real_count = count;
172   while (count) {
173     ssize_t res = pwrite(fd, buffer, count, offset);
174     if (res > 0) {
175       count -= res;
176       buffer += res;
177       offset += res;
178     } else if (res == 0)
179       return -1;
180     else if (errno != EINTR)
181       return -1;
182   }
183   return real_count;
184 }
185
186 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
187 static const void* zero_buffer;
188 static const size_t zero_buffer_size = 10 * 4096;
189
190 static void zero_buffer_init()
191 {
192   int fd = open("/dev/zero", O_RDONLY);
193   if (fd < 0)
194     xbt_die("Could not open /dev/zero");
195   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
196   if (zero_buffer == MAP_FAILED)
197     xbt_die("Could not map the zero buffer");
198   close(fd);
199 }
200
201 int open_vm(pid_t pid, int flags)
202 {
203   const size_t buffer_size = 30;
204   char buffer[buffer_size];
205   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
206   if (res < 0 || (size_t)res >= buffer_size) {
207     errno = ENAMETOOLONG;
208     return -1;
209   }
210   return open(buffer, flags);
211 }
212
213 // ***** Process
214
215 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
216 {
217 }
218
219 void RemoteClient::init()
220 {
221   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
222   this->init_memory_map_info();
223
224   int fd = open_vm(this->pid_, O_RDWR);
225   if (fd < 0)
226     xbt_die("Could not open file for process virtual address space");
227   this->memory_file = fd;
228
229   // Read std_heap (is a struct mdesc*):
230   simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
231   if (not std_heap_var)
232     xbt_die("No heap information in the target process");
233   if (not std_heap_var->address)
234     xbt_die("No constant address for this variable");
235   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address),
236                    simgrid::mc::ProcessIndexDisabled);
237
238   this->smx_actors_infos.clear();
239   this->smx_dead_actors_infos.clear();
240   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
241   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
242   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
243 }
244
245 RemoteClient::~RemoteClient()
246 {
247   if (this->memory_file >= 0)
248     close(this->memory_file);
249
250   if (this->unw_underlying_addr_space != unw_local_addr_space) {
251     if (this->unw_underlying_addr_space)
252       unw_destroy_addr_space(this->unw_underlying_addr_space);
253     if (this->unw_underlying_context)
254       _UPT_destroy(this->unw_underlying_context);
255   }
256
257   unw_destroy_addr_space(this->unw_addr_space);
258 }
259
260 /** Refresh the information about the process
261  *
262  *  Do not use directly, this is used by the getters when appropriate
263  *  in order to have fresh data.
264  */
265 void RemoteClient::refresh_heap()
266 {
267   // Read/dereference/refresh the std_heap pointer:
268   if (not this->heap)
269     this->heap = std::unique_ptr<s_xbt_mheap_t>(new s_xbt_mheap_t());
270   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address), simgrid::mc::ProcessIndexDisabled);
271   this->cache_flags_ |= RemoteClient::cache_heap;
272 }
273
274 /** Refresh the information about the process
275  *
276  *  Do not use direclty, this is used by the getters when appropriate
277  *  in order to have fresh data.
278  * */
279 void RemoteClient::refresh_malloc_info()
280 {
281   // Refresh process->heapinfo:
282   if (this->cache_flags_ & RemoteClient::cache_malloc)
283     return;
284   size_t count = this->heap->heaplimit + 1;
285   if (this->heap_info.size() < count)
286     this->heap_info.resize(count);
287   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo),
288                    simgrid::mc::ProcessIndexDisabled);
289   this->cache_flags_ |= RemoteClient::cache_malloc;
290 }
291
292 /** @brief Finds the range of the different memory segments and binary paths */
293 void RemoteClient::init_memory_map_info()
294 {
295   XBT_DEBUG("Get debug information ...");
296   this->maestro_stack_start_ = nullptr;
297   this->maestro_stack_end_   = nullptr;
298   this->object_infos.resize(0);
299   this->binary_info     = nullptr;
300   this->libsimgrid_info = nullptr;
301
302   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
303
304   const char* current_name = nullptr;
305
306   this->object_infos.clear();
307
308   for (size_t i = 0; i < maps.size(); i++) {
309     simgrid::xbt::VmMap const& reg = maps[i];
310     const char* pathname           = maps[i].pathname.c_str();
311
312     // Nothing to do
313     if (maps[i].pathname.empty()) {
314       current_name = nullptr;
315       continue;
316     }
317
318     // [stack], [vvar], [vsyscall], [vdso] ...
319     if (pathname[0] == '[') {
320       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
321         this->maestro_stack_start_ = remote(reg.start_addr);
322         this->maestro_stack_end_   = remote(reg.end_addr);
323       }
324       current_name = nullptr;
325       continue;
326     }
327
328     if (current_name && strcmp(current_name, pathname) == 0)
329       continue;
330
331     current_name = pathname;
332     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
333       continue;
334
335     const bool is_executable = not i;
336     std::string libname;
337     if (not is_executable) {
338       libname = get_lib_name(pathname);
339       if (is_filtered_lib(libname)) {
340         continue;
341       }
342     }
343
344     std::shared_ptr<simgrid::mc::ObjectInformation> info =
345         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
346     this->object_infos.push_back(info);
347     if (is_executable)
348       this->binary_info = info;
349     else if (is_simgrid_lib(libname))
350       this->libsimgrid_info = info;
351   }
352
353   // Resolve time (including across different objects):
354   for (auto const& object_info : this->object_infos)
355     postProcessObjectInformation(this, object_info.get());
356
357   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
358   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
359
360   XBT_DEBUG("Get debug information done !");
361 }
362
363 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
364 {
365   for (auto const& object_info : this->object_infos)
366     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
367       return object_info;
368   return nullptr;
369 }
370
371 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(RemotePtr<void> addr) const
372 {
373   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
374     if (addr.address() >= (std::uint64_t)info->start_exec && addr.address() <= (std::uint64_t)info->end_exec)
375       return info;
376   return nullptr;
377 }
378
379 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(RemotePtr<void> addr) const
380 {
381   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
382     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
383       return info;
384   return nullptr;
385 }
386
387 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
388 {
389   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
390   return info ? info->find_function((void*)ip.address()) : nullptr;
391 }
392
393 /** Find (one occurrence of) the named variable definition
394  */
395 simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
396 {
397   // First lookup the variable in the executable shared object.
398   // A global variable used directly by the executable code from a library
399   // is reinstanciated in the executable memory .data/.bss.
400   // We need to look up the variable in the executable first.
401   if (this->binary_info) {
402     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
403     simgrid::mc::Variable* var                                  = info->find_variable(name);
404     if (var)
405       return var;
406   }
407
408   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
409     simgrid::mc::Variable* var = info->find_variable(name);
410     if (var)
411       return var;
412   }
413
414   return nullptr;
415 }
416
417 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
418 {
419   simgrid::mc::Variable* var = this->find_variable(name);
420   xbt_assert(var->address, "No simple location for this variable");
421   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
422   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
423              size, (size_t)var->type->full_type->byte_size);
424   this->read_bytes(target, size, remote(var->address));
425 }
426
427 std::string RemoteClient::read_string(RemotePtr<char> address) const
428 {
429   if (not address)
430     return {};
431
432   std::vector<char> res(128);
433   off_t off = 0;
434
435   while (1) {
436     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
437     if (c == -1) {
438       if (errno == EINTR)
439         continue;
440       else
441         xbt_die("Could not read from from remote process");
442     }
443     if (c == 0)
444       xbt_die("Could not read string from remote process");
445
446     void* p = memchr(res.data() + off, '\0', c);
447     if (p)
448       return std::string(res.data());
449
450     off += c;
451     if (off == (off_t)res.size())
452       res.resize(res.size() * 2);
453   }
454 }
455
456 const void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, int process_index,
457                                      ReadOptions options) const
458 {
459 #if HAVE_SMPI
460   if (process_index != simgrid::mc::ProcessIndexDisabled) {
461     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->find_object_info_rw(address);
462     // Segment overlap is not handled.
463     if (info.get() && this->privatized(*info)) {
464       if (process_index < 0)
465         xbt_die("Missing process index");
466       if (process_index >= (int)MC_smpi_process_count())
467         xbt_die("Invalid process index");
468
469       // Read smpi_privatization_regions from MCed:
470       smpi_privatization_region_t remote_smpi_privatization_regions =
471           mc_model_checker->process().read_variable<smpi_privatization_region_t>("smpi_privatization_regions");
472
473       s_smpi_privatization_region_t privatization_region =
474           mc_model_checker->process().read<s_smpi_privatization_region_t>(
475               remote(remote_smpi_privatization_regions + process_index));
476
477       // Address translation in the privatization segment:
478       size_t offset = address.address() - (std::uint64_t)info->start_rw;
479       address       = remote((char*)privatization_region.address + offset);
480     }
481   }
482 #endif
483   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
484     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
485   return buffer;
486 }
487
488 /** Write data to a process memory
489  *
490  *  @param buffer   local memory address (source)
491  *  @param len      data size
492  *  @param address  target process memory address (target)
493  */
494 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
495 {
496   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
497     xbt_die("Write to process %lli failed", (long long)this->pid_);
498 }
499
500 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
501 {
502   pthread_once(&zero_buffer_flag, zero_buffer_init);
503   while (len) {
504     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
505     this->write_bytes(zero_buffer, s, address);
506     address = remote((char*)address.address() + s);
507     len -= s;
508   }
509 }
510
511 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
512 {
513   IgnoredRegion region;
514   region.addr = addr;
515   region.size = size;
516
517   if (ignored_regions_.empty()) {
518     ignored_regions_.push_back(region);
519     return;
520   }
521
522   unsigned int cursor           = 0;
523   IgnoredRegion* current_region = nullptr;
524
525   int start = 0;
526   int end   = ignored_regions_.size() - 1;
527   while (start <= end) {
528     cursor         = (start + end) / 2;
529     current_region = &ignored_regions_[cursor];
530     if (current_region->addr == addr) {
531       if (current_region->size == size)
532         return;
533       else if (current_region->size < size)
534         start = cursor + 1;
535       else
536         end = cursor - 1;
537     } else if (current_region->addr < addr)
538       start = cursor + 1;
539     else
540       end = cursor - 1;
541   }
542
543   std::size_t position;
544   if (current_region->addr == addr) {
545     if (current_region->size < size)
546       position = cursor + 1;
547     else
548       position = cursor;
549   } else if (current_region->addr < addr)
550     position = cursor + 1;
551   else
552     position = cursor;
553   ignored_regions_.insert(ignored_regions_.begin() + position, region);
554 }
555
556 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
557 {
558   if (ignored_heap_.empty()) {
559     ignored_heap_.push_back(std::move(region));
560     return;
561   }
562
563   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
564
565   size_type start = 0;
566   size_type end   = ignored_heap_.size() - 1;
567
568   // Binary search the position of insertion:
569   size_type cursor;
570   while (start <= end) {
571     cursor               = start + (end - start) / 2;
572     auto& current_region = ignored_heap_[cursor];
573     if (current_region.address == region.address)
574       return;
575     else if (current_region.address < region.address)
576       start = cursor + 1;
577     else if (cursor != 0)
578       end = cursor - 1;
579     // Avoid underflow:
580     else
581       break;
582   }
583
584   // Insert it mc_heap_ignore_region_t:
585   if (ignored_heap_[cursor].address < region.address)
586     ++cursor;
587   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
588 }
589
590 void RemoteClient::unignore_heap(void* address, size_t size)
591 {
592   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
593
594   size_type start = 0;
595   size_type end   = ignored_heap_.size() - 1;
596
597   // Binary search:
598   size_type cursor;
599   while (start <= end) {
600     cursor       = (start + end) / 2;
601     auto& region = ignored_heap_[cursor];
602     if (region.address < address)
603       start = cursor + 1;
604     else if ((char*)region.address <= ((char*)address + size)) {
605       ignored_heap_.erase(ignored_heap_.begin() + cursor);
606       return;
607     } else if (cursor != 0)
608       end = cursor - 1;
609     // Avoid underflow:
610     else
611       break;
612   }
613 }
614
615 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
616 {
617   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
618     frame_name = nullptr;
619   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
620     info->remove_local_variable(var_name, frame_name);
621 }
622
623 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
624 {
625   this->refresh_simix();
626   return smx_actors_infos;
627 }
628
629 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
630 {
631   this->refresh_simix();
632   return smx_dead_actors_infos;
633 }
634
635 void RemoteClient::dumpStack()
636 {
637   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
638   if (as == nullptr) {
639     XBT_ERROR("Could not initialize ptrace address space");
640     return;
641   }
642
643   void* context = _UPT_create(this->pid_);
644   if (context == nullptr) {
645     unw_destroy_addr_space(as);
646     XBT_ERROR("Could not initialize ptrace context");
647     return;
648   }
649
650   unw_cursor_t cursor;
651   if (unw_init_remote(&cursor, as, context) != 0) {
652     _UPT_destroy(context);
653     unw_destroy_addr_space(as);
654     XBT_ERROR("Could not initialiez ptrace cursor");
655     return;
656   }
657
658   simgrid::mc::dumpStack(stderr, cursor);
659
660   _UPT_destroy(context);
661   unw_destroy_addr_space(as);
662 }
663
664 bool RemoteClient::actor_is_enabled(aid_t pid)
665 {
666   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
667   process()->getChannel().send(msg);
668   char buff[MC_MESSAGE_LENGTH];
669   ssize_t received = process()->getChannel().receive(buff, MC_MESSAGE_LENGTH, true);
670   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
671   return ((s_mc_message_int_t*)buff)->value;
672 }
673 }
674 }