Logo AND Algorithmique Numérique Distribuée

Public GIT Repository
e8000dad036fbee4eacbf9ae40adfcede1013bc1
[simgrid.git] / src / mc / remote / RemoteClient.cpp
1 /* Copyright (c) 2014-2019. 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     "libpgmath",
104     "libpthread",
105     "libquadmath",
106     "librt",
107     "libstdc++",
108     "libsvml",
109     "libtsan",  /* gcc sanitizers */
110     "libubsan", /* gcc sanitizers */
111     "libunwind",
112     "libunwind-ptrace",
113     "libunwind-x86",
114     "libunwind-x86_64",
115     "libz"};
116
117 static bool is_simgrid_lib(const std::string& libname)
118 {
119   return libname == "libsimgrid";
120 }
121
122 static bool is_filtered_lib(const std::string& libname)
123 {
124   return std::find(begin(filtered_libraries), end(filtered_libraries), libname) != end(filtered_libraries);
125 }
126
127 static std::string get_lib_name(const std::string& pathname)
128 {
129   std::string map_basename = simgrid::xbt::Path(pathname).get_base_name();
130   std::string libname;
131
132   size_t pos = map_basename.rfind(".so");
133   if (pos != std::string::npos) {
134     // strip the extension (matching regex "\.so.*$")
135     libname.assign(map_basename, 0, pos);
136
137     // strip the version suffix (matching regex "-[.0-9-]*$")
138     while (true) {
139       pos = libname.rfind('-');
140       if (pos == std::string::npos || libname.find_first_not_of(".0123456789", pos + 1) != std::string::npos)
141         break;
142       libname.erase(pos);
143     }
144   }
145
146   return libname;
147 }
148
149 static ssize_t pread_whole(int fd, void* buf, size_t count, off_t offset)
150 {
151   char* buffer       = (char*)buf;
152   ssize_t real_count = count;
153   while (count) {
154     ssize_t res = pread(fd, buffer, count, offset);
155     if (res > 0) {
156       count -= res;
157       buffer += res;
158       offset += res;
159     } else if (res == 0)
160       return -1;
161     else if (errno != EINTR) {
162       perror("pread_whole");
163       return -1;
164     }
165   }
166   return real_count;
167 }
168
169 static ssize_t pwrite_whole(int fd, const void* buf, size_t count, off_t offset)
170 {
171   const char* buffer = (const char*)buf;
172   ssize_t real_count = count;
173   while (count) {
174     ssize_t res = pwrite(fd, buffer, count, offset);
175     if (res > 0) {
176       count -= res;
177       buffer += res;
178       offset += res;
179     } else if (res == 0)
180       return -1;
181     else if (errno != EINTR)
182       return -1;
183   }
184   return real_count;
185 }
186
187 static pthread_once_t zero_buffer_flag = PTHREAD_ONCE_INIT;
188 static const void* zero_buffer;
189 static const size_t zero_buffer_size = 10 * 4096;
190
191 static void zero_buffer_init()
192 {
193   int fd = open("/dev/zero", O_RDONLY);
194   if (fd < 0)
195     xbt_die("Could not open /dev/zero");
196   zero_buffer = mmap(nullptr, zero_buffer_size, PROT_READ, MAP_SHARED, fd, 0);
197   if (zero_buffer == MAP_FAILED)
198     xbt_die("Could not map the zero buffer");
199   close(fd);
200 }
201
202 int open_vm(pid_t pid, int flags)
203 {
204   const size_t buffer_size = 30;
205   char buffer[buffer_size];
206   int res = snprintf(buffer, buffer_size, "/proc/%lli/mem", (long long)pid);
207   if (res < 0 || (size_t)res >= buffer_size) {
208     errno = ENAMETOOLONG;
209     return -1;
210   }
211   return open(buffer, flags);
212 }
213
214 // ***** Process
215
216 RemoteClient::RemoteClient(pid_t pid, int sockfd) : AddressSpace(this), pid_(pid), channel_(sockfd), running_(true)
217 {
218 }
219
220 void RemoteClient::init()
221 {
222   this->memory_map_ = simgrid::xbt::get_memory_map(this->pid_);
223   this->init_memory_map_info();
224
225   int fd = open_vm(this->pid_, O_RDWR);
226   if (fd < 0)
227     xbt_die("Could not open file for process virtual address space");
228   this->memory_file = fd;
229
230   // Read std_heap (is a struct mdesc*):
231   simgrid::mc::Variable* std_heap_var = this->find_variable("__mmalloc_default_mdp");
232   if (not std_heap_var)
233     xbt_die("No heap information in the target process");
234   if (not std_heap_var->address)
235     xbt_die("No constant address for this variable");
236   this->read_bytes(&this->heap_address, sizeof(mdesc*), remote(std_heap_var->address),
237                    simgrid::mc::ProcessIndexDisabled);
238
239   this->smx_actors_infos.clear();
240   this->smx_dead_actors_infos.clear();
241   this->unw_addr_space            = simgrid::mc::UnwindContext::createUnwindAddressSpace();
242   this->unw_underlying_addr_space = simgrid::unw::create_addr_space();
243   this->unw_underlying_context    = simgrid::unw::create_context(this->unw_underlying_addr_space, this->pid_);
244 }
245
246 RemoteClient::~RemoteClient()
247 {
248   if (this->memory_file >= 0)
249     close(this->memory_file);
250
251   if (this->unw_underlying_addr_space != unw_local_addr_space) {
252     if (this->unw_underlying_addr_space)
253       unw_destroy_addr_space(this->unw_underlying_addr_space);
254     if (this->unw_underlying_context)
255       _UPT_destroy(this->unw_underlying_context);
256   }
257
258   unw_destroy_addr_space(this->unw_addr_space);
259 }
260
261 /** Refresh the information about the process
262  *
263  *  Do not use directly, this is used by the getters when appropriate
264  *  in order to have fresh data.
265  */
266 void RemoteClient::refresh_heap()
267 {
268   // Read/dereference/refresh the std_heap pointer:
269   if (not this->heap)
270     this->heap = std::unique_ptr<s_xbt_mheap_t>(new s_xbt_mheap_t());
271   this->read_bytes(this->heap.get(), sizeof(mdesc), remote(this->heap_address), simgrid::mc::ProcessIndexDisabled);
272   this->cache_flags_ |= RemoteClient::cache_heap;
273 }
274
275 /** Refresh the information about the process
276  *
277  *  Do not use direclty, this is used by the getters when appropriate
278  *  in order to have fresh data.
279  * */
280 void RemoteClient::refresh_malloc_info()
281 {
282   // Refresh process->heapinfo:
283   if (this->cache_flags_ & RemoteClient::cache_malloc)
284     return;
285   size_t count = this->heap->heaplimit + 1;
286   if (this->heap_info.size() < count)
287     this->heap_info.resize(count);
288   this->read_bytes(this->heap_info.data(), count * sizeof(malloc_info), remote(this->heap->heapinfo),
289                    simgrid::mc::ProcessIndexDisabled);
290   this->cache_flags_ |= RemoteClient::cache_malloc;
291 }
292
293 /** @brief Finds the range of the different memory segments and binary paths */
294 void RemoteClient::init_memory_map_info()
295 {
296   XBT_DEBUG("Get debug information ...");
297   this->maestro_stack_start_ = nullptr;
298   this->maestro_stack_end_   = nullptr;
299   this->object_infos.resize(0);
300   this->binary_info     = nullptr;
301   this->libsimgrid_info = nullptr;
302
303   std::vector<simgrid::xbt::VmMap> const& maps = this->memory_map_;
304
305   const char* current_name = nullptr;
306
307   this->object_infos.clear();
308
309   for (size_t i = 0; i < maps.size(); i++) {
310     simgrid::xbt::VmMap const& reg = maps[i];
311     const char* pathname           = maps[i].pathname.c_str();
312
313     // Nothing to do
314     if (maps[i].pathname.empty()) {
315       current_name = nullptr;
316       continue;
317     }
318
319     // [stack], [vvar], [vsyscall], [vdso] ...
320     if (pathname[0] == '[') {
321       if ((reg.prot & PROT_WRITE) && not memcmp(pathname, "[stack]", 7)) {
322         this->maestro_stack_start_ = remote(reg.start_addr);
323         this->maestro_stack_end_   = remote(reg.end_addr);
324       }
325       current_name = nullptr;
326       continue;
327     }
328
329     if (current_name && strcmp(current_name, pathname) == 0)
330       continue;
331
332     current_name = pathname;
333     if (not(reg.prot & PROT_READ) && (reg.prot & PROT_EXEC))
334       continue;
335
336     const bool is_executable = not i;
337     std::string libname;
338     if (not is_executable) {
339       libname = get_lib_name(pathname);
340       if (is_filtered_lib(libname)) {
341         continue;
342       }
343     }
344
345     std::shared_ptr<simgrid::mc::ObjectInformation> info =
346         simgrid::mc::createObjectInformation(this->memory_map_, pathname);
347     this->object_infos.push_back(info);
348     if (is_executable)
349       this->binary_info = info;
350     else if (is_simgrid_lib(libname))
351       this->libsimgrid_info = info;
352   }
353
354   // Resolve time (including across different objects):
355   for (auto const& object_info : this->object_infos)
356     postProcessObjectInformation(this, object_info.get());
357
358   xbt_assert(this->maestro_stack_start_, "Did not find maestro_stack_start");
359   xbt_assert(this->maestro_stack_end_, "Did not find maestro_stack_end");
360
361   XBT_DEBUG("Get debug information done !");
362 }
363
364 std::shared_ptr<simgrid::mc::ObjectInformation> RemoteClient::find_object_info(RemotePtr<void> addr) const
365 {
366   for (auto const& object_info : this->object_infos)
367     if (addr.address() >= (std::uint64_t)object_info->start && addr.address() <= (std::uint64_t)object_info->end)
368       return object_info;
369   return nullptr;
370 }
371
372 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_exec(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_exec && addr.address() <= (std::uint64_t)info->end_exec)
376       return info;
377   return nullptr;
378 }
379
380 std::shared_ptr<ObjectInformation> RemoteClient::find_object_info_rw(RemotePtr<void> addr) const
381 {
382   for (std::shared_ptr<ObjectInformation> const& info : this->object_infos)
383     if (addr.address() >= (std::uint64_t)info->start_rw && addr.address() <= (std::uint64_t)info->end_rw)
384       return info;
385   return nullptr;
386 }
387
388 simgrid::mc::Frame* RemoteClient::find_function(RemotePtr<void> ip) const
389 {
390   std::shared_ptr<simgrid::mc::ObjectInformation> info = this->find_object_info_exec(ip);
391   return info ? info->find_function((void*)ip.address()) : nullptr;
392 }
393
394 /** Find (one occurrence of) the named variable definition
395  */
396 simgrid::mc::Variable* RemoteClient::find_variable(const char* name) const
397 {
398   // First lookup the variable in the executable shared object.
399   // A global variable used directly by the executable code from a library
400   // is reinstanciated in the executable memory .data/.bss.
401   // We need to look up the variable in the executable first.
402   if (this->binary_info) {
403     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->binary_info;
404     simgrid::mc::Variable* var                                  = info->find_variable(name);
405     if (var)
406       return var;
407   }
408
409   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos) {
410     simgrid::mc::Variable* var = info->find_variable(name);
411     if (var)
412       return var;
413   }
414
415   return nullptr;
416 }
417
418 void RemoteClient::read_variable(const char* name, void* target, size_t size) const
419 {
420   simgrid::mc::Variable* var = this->find_variable(name);
421   xbt_assert(var->address, "No simple location for this variable");
422   xbt_assert(var->type->full_type, "Partial type for %s, cannot check size", name);
423   xbt_assert((size_t)var->type->full_type->byte_size == size, "Unexpected size for %s (expected %zu, was %zu)", name,
424              size, (size_t)var->type->full_type->byte_size);
425   this->read_bytes(target, size, remote(var->address));
426 }
427
428 std::string RemoteClient::read_string(RemotePtr<char> address) const
429 {
430   if (not address)
431     return {};
432
433   std::vector<char> res(128);
434   off_t off = 0;
435
436   while (1) {
437     ssize_t c = pread(this->memory_file, res.data() + off, res.size() - off, (off_t)address.address() + off);
438     if (c == -1) {
439       if (errno == EINTR)
440         continue;
441       else
442         xbt_die("Could not read from from remote process");
443     }
444     if (c == 0)
445       xbt_die("Could not read string from remote process");
446
447     void* p = memchr(res.data() + off, '\0', c);
448     if (p)
449       return std::string(res.data());
450
451     off += c;
452     if (off == (off_t)res.size())
453       res.resize(res.size() * 2);
454   }
455 }
456
457 const void* RemoteClient::read_bytes(void* buffer, std::size_t size, RemotePtr<void> address, int process_index,
458                                      ReadOptions options) const
459 {
460 #if HAVE_SMPI
461   if (process_index != simgrid::mc::ProcessIndexDisabled) {
462     std::shared_ptr<simgrid::mc::ObjectInformation> const& info = this->find_object_info_rw(address);
463     // Segment overlap is not handled.
464     if (info.get() && this->privatized(*info)) {
465       if (process_index < 0)
466         xbt_die("Missing process index");
467       if (process_index >= (int)MC_smpi_process_count())
468         xbt_die("Invalid process index");
469
470       // Read smpi_privatization_regions from MCed:
471       smpi_privatization_region_t remote_smpi_privatization_regions =
472           mc_model_checker->process().read_variable<smpi_privatization_region_t>("smpi_privatization_regions");
473
474       s_smpi_privatization_region_t privatization_region =
475           mc_model_checker->process().read<s_smpi_privatization_region_t>(
476               remote(remote_smpi_privatization_regions + process_index));
477
478       // Address translation in the privatization segment:
479       size_t offset = address.address() - (std::uint64_t)info->start_rw;
480       address       = remote((char*)privatization_region.address + offset);
481     }
482   }
483 #endif
484   if (pread_whole(this->memory_file, buffer, size, (size_t)address.address()) < 0)
485     xbt_die("Read at %p from process %lli failed", (void*)address.address(), (long long)this->pid_);
486   return buffer;
487 }
488
489 /** Write data to a process memory
490  *
491  *  @param buffer   local memory address (source)
492  *  @param len      data size
493  *  @param address  target process memory address (target)
494  */
495 void RemoteClient::write_bytes(const void* buffer, size_t len, RemotePtr<void> address)
496 {
497   if (pwrite_whole(this->memory_file, buffer, len, (size_t)address.address()) < 0)
498     xbt_die("Write to process %lli failed", (long long)this->pid_);
499 }
500
501 void RemoteClient::clear_bytes(RemotePtr<void> address, size_t len)
502 {
503   pthread_once(&zero_buffer_flag, zero_buffer_init);
504   while (len) {
505     size_t s = len > zero_buffer_size ? zero_buffer_size : len;
506     this->write_bytes(zero_buffer, s, address);
507     address = remote((char*)address.address() + s);
508     len -= s;
509   }
510 }
511
512 void RemoteClient::ignore_region(std::uint64_t addr, std::size_t size)
513 {
514   IgnoredRegion region;
515   region.addr = addr;
516   region.size = size;
517
518   if (ignored_regions_.empty()) {
519     ignored_regions_.push_back(region);
520     return;
521   }
522
523   unsigned int cursor           = 0;
524   IgnoredRegion* current_region = nullptr;
525
526   int start = 0;
527   int end   = ignored_regions_.size() - 1;
528   while (start <= end) {
529     cursor         = (start + end) / 2;
530     current_region = &ignored_regions_[cursor];
531     if (current_region->addr == addr) {
532       if (current_region->size == size)
533         return;
534       else if (current_region->size < size)
535         start = cursor + 1;
536       else
537         end = cursor - 1;
538     } else if (current_region->addr < addr)
539       start = cursor + 1;
540     else
541       end = cursor - 1;
542   }
543
544   std::size_t position;
545   if (current_region->addr == addr) {
546     if (current_region->size < size)
547       position = cursor + 1;
548     else
549       position = cursor;
550   } else if (current_region->addr < addr)
551     position = cursor + 1;
552   else
553     position = cursor;
554   ignored_regions_.insert(ignored_regions_.begin() + position, region);
555 }
556
557 void RemoteClient::ignore_heap(IgnoredHeapRegion const& region)
558 {
559   if (ignored_heap_.empty()) {
560     ignored_heap_.push_back(std::move(region));
561     return;
562   }
563
564   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
565
566   size_type start = 0;
567   size_type end   = ignored_heap_.size() - 1;
568
569   // Binary search the position of insertion:
570   size_type cursor;
571   while (start <= end) {
572     cursor               = start + (end - start) / 2;
573     auto& current_region = ignored_heap_[cursor];
574     if (current_region.address == region.address)
575       return;
576     else if (current_region.address < region.address)
577       start = cursor + 1;
578     else if (cursor != 0)
579       end = cursor - 1;
580     // Avoid underflow:
581     else
582       break;
583   }
584
585   // Insert it mc_heap_ignore_region_t:
586   if (ignored_heap_[cursor].address < region.address)
587     ++cursor;
588   ignored_heap_.insert(ignored_heap_.begin() + cursor, region);
589 }
590
591 void RemoteClient::unignore_heap(void* address, size_t size)
592 {
593   typedef std::vector<IgnoredHeapRegion>::size_type size_type;
594
595   size_type start = 0;
596   size_type end   = ignored_heap_.size() - 1;
597
598   // Binary search:
599   size_type cursor;
600   while (start <= end) {
601     cursor       = (start + end) / 2;
602     auto& region = ignored_heap_[cursor];
603     if (region.address < address)
604       start = cursor + 1;
605     else if ((char*)region.address <= ((char*)address + size)) {
606       ignored_heap_.erase(ignored_heap_.begin() + cursor);
607       return;
608     } else if (cursor != 0)
609       end = cursor - 1;
610     // Avoid underflow:
611     else
612       break;
613   }
614 }
615
616 void RemoteClient::ignore_local_variable(const char* var_name, const char* frame_name)
617 {
618   if (frame_name != nullptr && strcmp(frame_name, "*") == 0)
619     frame_name = nullptr;
620   for (std::shared_ptr<simgrid::mc::ObjectInformation> const& info : this->object_infos)
621     info->remove_local_variable(var_name, frame_name);
622 }
623
624 std::vector<simgrid::mc::ActorInformation>& RemoteClient::actors()
625 {
626   this->refresh_simix();
627   return smx_actors_infos;
628 }
629
630 std::vector<simgrid::mc::ActorInformation>& RemoteClient::dead_actors()
631 {
632   this->refresh_simix();
633   return smx_dead_actors_infos;
634 }
635
636 void RemoteClient::dumpStack()
637 {
638   unw_addr_space_t as = unw_create_addr_space(&_UPT_accessors, BYTE_ORDER);
639   if (as == nullptr) {
640     XBT_ERROR("Could not initialize ptrace address space");
641     return;
642   }
643
644   void* context = _UPT_create(this->pid_);
645   if (context == nullptr) {
646     unw_destroy_addr_space(as);
647     XBT_ERROR("Could not initialize ptrace context");
648     return;
649   }
650
651   unw_cursor_t cursor;
652   if (unw_init_remote(&cursor, as, context) != 0) {
653     _UPT_destroy(context);
654     unw_destroy_addr_space(as);
655     XBT_ERROR("Could not initialiez ptrace cursor");
656     return;
657   }
658
659   simgrid::mc::dumpStack(stderr, cursor);
660
661   _UPT_destroy(context);
662   unw_destroy_addr_space(as);
663 }
664
665 bool RemoteClient::actor_is_enabled(aid_t pid)
666 {
667   s_mc_message_actor_enabled_t msg{MC_MESSAGE_ACTOR_ENABLED, pid};
668   process()->getChannel().send(msg);
669   char buff[MC_MESSAGE_LENGTH];
670   ssize_t received = process()->getChannel().receive(buff, MC_MESSAGE_LENGTH, true);
671   xbt_assert(received == sizeof(s_mc_message_int_t), "Unexpected size in answer to ACTOR_ENABLED");
672   return ((s_mc_message_int_t*)buff)->value;
673 }
674 }
675 }