LG

• Installation en parallèle

Ubuntu ne fournit pas de méthode 100% automatique. Il faut passer manuellement par update-alternatives.

sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-11 110 --slave /usr/bin/g++ g++ /usr/bin/g++-11

Utiliser pour switcher entre les versions :

sudo update-alternatives --config gcc

class VirtualBase
{
  public :
    virtual int tick(int n) = 0;
};
 
class VirtualDerived : public VirtualBase
{
  public :
    int m_counter;
  public :
    VirtualDerived() : m_counter(0) {}
    int tick(int n) { m_counter += n; return m_counter; }
};

• Jusqu'à gcc 7 : g++ -fdump-class-hierarchy -c main.cc
• A partir de gcc 8 : g++ -fdump-lang-class -c main.cc

Vtable for VirtualBase
VirtualBase::_ZTV11VirtualBase: 3 entries
0     (int (*)(...))0
8     (int (*)(...))(& _ZTI11VirtualBase)
16    (int (*)(...))__cxa_pure_virtual

Class VirtualBase
   size=8 align=8
   base size=8 base align=8
VirtualBase (0x0x6ffffbe0480) 0 nearly-empty
    vptr=((& VirtualBase::_ZTV11VirtualBase) + 16)

Vtable for VirtualDerived
VirtualDerived::_ZTV14VirtualDerived: 3 entries
0     (int (*)(...))0
8     (int (*)(...))(& _ZTI14VirtualDerived)
16    (int (*)(...))VirtualDerived::tick

Class VirtualDerived
   size=16 align=8
   base size=12 base align=8
VirtualDerived (0x0x6ffffc301a0) 0
    vptr=((& VirtualDerived::_ZTV14VirtualDerived) + 16)
  VirtualBase (0x0x6ffffbe04e0) 0 nearly-empty
      primary-for VirtualDerived (0x0x6ffffc301a0)

• clang : clang -cc1 -emit-llvm -fdump-record-layouts main.cc

Ci-dessous, on affiche que le contenu des Dumping AST Record Layout.

*** Dumping AST Record Layout
         0 | class VirtualBase
         0 |   (VirtualBase vtable pointer)
           | [sizeof=8, dsize=8, align=8,
           |  nvsize=8, nvalign=8]

*** Dumping AST Record Layout
         0 | class VirtualDerived
         0 |   class VirtualBase (primary base)
         0 |     (VirtualBase vtable pointer)
         8 |   int m_counter
           | [sizeof=16, dsize=12, align=8,
           |  nvsize=12, nvalign=8]

• Table virtuelle uniquement : clang -cc1 -emit-llvm -fdump-vtable-layouts main.cc

Vtable for 'VirtualDerived' (3 entries).
   0 | offset_to_top (0)
   1 | VirtualDerived RTTI
       -- (VirtualBase, 0) vtable address --
       -- (VirtualDerived, 0) vtable address --
   2 | int VirtualDerived::tick(int)

VTable indices for 'VirtualDerived' (1 entries).
   0 | int VirtualDerived::tick(int)

Vtable for 'VirtualBase' (3 entries).
   0 | offset_to_top (0)
   1 | VirtualBase RTTI
       -- (VirtualBase, 0) vtable address --
   2 | int VirtualBase::tick(int) [pure]

VTable indices for 'VirtualBase' (1 entries).
   0 | int VirtualBase::tick(int)

Génère un graphe de l'utilisation des fonctions d'un programme (nécessite l'option de compilation -pg).

GPROF Tutorial – How to use Linux GNU GCC Profiling Tool Archive du 20/08/2012 le 24/02/2020

Options à ajouter à la compilation : -O0 --coverage

Options à ajouter au lieur : --coverage

Puis exécuter le ou les programmes. Attention, pas d'exécution en parallèle. Les fichiers .gcno ne le supportent pas.

lcov --capture --directory build --gcov-tool /usr/bin/gcov-10 --rc lcov_branch_coverage=1 --output-file build/coverage.info
lcov --remove build/coverage.info "/usr/include/*" --rc lcov_branch_coverage=1 -o build/coverage2.info
genhtml --output-directory build/coverage --demangle-cpp --num-spaces 2 --sort --function-coverage --branch-coverage --legend build/coverage2.info
find build/coverage -name "*.html" -exec sed -i -r "s#headerValue""\>.*-.*-.* .*:.*:.*td#headerValue""></td#g" {} \;

Attention, la version de gcov doit correspondre à celle de gcc.

En option, on supprime la date pour qu'un diff simple puisse se faire.

Il existe -fsanitize=address,thread,undefined mais address et thread ne sont pas compatible simultanément.

Il détecte des erreurs de type global-buffer-overflow, heap-use-after-free. CppCon 2015: Kostya Serebryany “Beyond Sanitizers, Fuzzing and Hardening your C++ apps for Security and Reliability”

• global-buffer-overflow

main.c

int global_array[100] = {-1};
 
int main(int argc, char **argv) {
  return global_array[argc+100];
}

gcc main.c -g -fsanitize=address -o main && ./main

=================================================================
==2500==ERROR: AddressSanitizer: global-buffer-overflow on address 0x000000740cf4 at pc 0x00000050d9bf bp 0x7ffeabb89b70 sp 0x7ffeabb89b68
READ of size 4 at 0x000000740cf4 thread T0
    #0 0x50d9be in main /tmp/main.c:4:10
    #1 0x7f1208faa461 in __libc_start_main .../sys-libs/glibc-2.25-r9/work/glibc-2.25/csu/../csu/libc-start.c:295
    #2 0x419709 in _start (/tmp/main+0x419709)

0x000000740cf4 is located 4 bytes to the right of global variable 'global_array' defined in 'main.c:1:5' (0x740b60) of size 400
SUMMARY: AddressSanitizer: global-buffer-overflow /tmp/main.c:4:10 in main
Shadow bytes around the buggy address:
  0x0000800e0140: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0000800e0150: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0000800e0160: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0000800e0170: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0000800e0180: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
=>0x0000800e0190: 00 00 00 00 00 00 00 00 00 00 00 00 00 00[f9]f9
  0x0000800e01a0: f9 f9 f9 f9 f9 f9 f9 f9 f9 f9 f9 f9 00 00 00 00
  0x0000800e01b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0000800e01c0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0000800e01d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0000800e01e0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Shadow byte legend (one shadow byte represents 8 application bytes):
  Addressable:           00
  Partially addressable: 01 02 03 04 05 06 07 
  Heap left redzone:       fa
  Freed heap region:       fd
  Stack left redzone:      f1
  Stack mid redzone:       f2
  Stack right redzone:     f3
  Stack after return:      f5
  Stack use after scope:   f8
  Global redzone:          f9
  Global init order:       f6
  Poisoned by user:        f7
  Container overflow:      fc
  Array cookie:            ac
  Intra object redzone:    bb
  ASan internal:           fe
  Left alloca redzone:     ca
  Right alloca redzone:    cb
==2500==ABORTING

valgrind ne détecte pas l'erreur.

• heap-use-after-free

main2.cc

int main(int argc, char **argv) {
  int *array = new int[100];
  delete [] array;
  return array[argc];
}

g++ main2.c -g -fsanitize=address -o main2 && ./main2

=================================================================
==2765==ERROR: AddressSanitizer: heap-use-after-free on address 0x614000000044 at pc 0x000000512445 bp 0x7ffd51684f60 sp 0x7ffd51684f58
READ of size 4 at 0x614000000044 thread T0
    #0 0x512444 in main /tmp/main2.c:4:10
    #1 0x7fd8d0086461 in __libc_start_main .../sys-libs/glibc-2.25-r9/work/glibc-2.25/csu/../csu/libc-start.c:295
    #2 0x419d29 in _start (/tmp/main2+0x419d29)

0x614000000044 is located 4 bytes inside of 400-byte region [0x614000000040,0x6140000001d0)
freed by thread T0 here:
    #0 0x50ef30 in operator delete[](void*) .../sys-libs/compiler-rt-sanitizers-5.0.0/work/compiler-rt-5.0.0.src/lib/asan/asan_new_delete.cc:141
    #1 0x5123f6 in main /tmp/main2.c:3:3
    #2 0x7fd8d0086461 in __libc_start_main .../sys-libs/glibc-2.25-r9/work/glibc-2.25/csu/../csu/libc-start.c:295
    #3 0x419d29 in _start (/tmp/main2+0x419d29)
  
previously allocated by thread T0 here:
    #0 0x50e1c8 in operator new[](unsigned long) .../sys-libs/compiler-rt-  sanitizers-5.0.0/work/compiler-rt-5.0.0.src/lib/asan/asan_new_delete.cc:95
    #1 0x5123d4 in main /tmp/main2.c:2:16
    #2 0x7fd8d0086461 in __libc_start_main .../sys-libs/glibc-2.25-r9/work/glibc-2.25/csu/../csu/libc-start.c:295
    #3 0x419d29 in _start (/tmp/main2+0x419d29)

SUMMARY: AddressSanitizer: heap-use-after-free /tmp/main2.c:4:10 in main
Shadow bytes around the buggy address:
  0x0c287fff7fb0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0c287fff7fc0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0c287fff7fd0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0c287fff7fe0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0c287fff7ff0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
=>0x0c287fff8000: fa fa fa fa fa fa fa fa[fd]fd fd fd fd fd fd fd
  0x0c287fff8010: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd
  0x0c287fff8020: fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd fd
  0x0c287fff8030: fd fd fd fd fd fd fd fd fd fd fa fa fa fa fa fa
  0x0c287fff8040: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
  0x0c287fff8050: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa

valgrind détecte l'erreur.

==2787== Command: ./main2
==2787== 
==2787== Invalid read of size 4
==2787==    at 0x4005AF: main (main2.c:4)
==2787==  Address 0x5b25c84 is 4 bytes inside a block of size 400 free'd
==2787==    at 0x4C2CAA5: operator delete[](void*) (vg_replace_malloc.c:621)
==2787==    by 0x4005A6: main (main2.c:3)
==2787==  Block was alloc'd at
==2787==    at 0x4C2BAC8: operator new[](unsigned long) (vg_replace_malloc.c:423)
==2787==    by 0x400584: main (main2.c:2)

• stack-use-after-return

main3.c

int *g;
 
void LeakLocal() {
  int local;
  g = &local;
}
 
int main(){
  LeakLocal();
  return *g;
}

gcc main3.c -g -o main3 -fsanitize=address
ASAN_OPTIONS=detect_stack_use_after_return=1 ./main3

L'utilisation de ASAN_OPTIONS=detect_stack_use_after_return=1 est nécessaire car cette option peut créer des faux positifs.

=================================================================
==2907==ERROR: AddressSanitizer: stack-use-after-return on address 0x7f57b2a00020 at pc 0x00000050db25 bp 0x7ffe8dfc7330 sp 0x7ffe8dfc7328
READ of size 4 at 0x7f57b2a00020 thread T0
    #0 0x50db24 in main /tmp/main3.c:10:10
    #1 0x7f57b604d461 in __libc_start_main .../sys-libs/glibc-2.25-r9/work/glibc-2.25/csu/../csu/libc-start.c:295
    #2 0x419709 in _start (/tmp/main3+0x419709)

Address 0x7f57b2a00020 is located in stack of thread T0 at offset 32 in frame
    #0 0x50d95f in LeakLocal /tmp/main3.c:3

  This frame has 1 object(s):
    [32, 36) 'local' (line 4) <== Memory access at offset 32 is inside this variable
HINT: this may be a false positive if your program uses some custom stack unwind mechanism or swapcontext
      (longjmp and C++ exceptions *are* supported)
SUMMARY: AddressSanitizer: stack-use-after-return /tmp/main3.c:10:10 in main
Shadow bytes around the buggy address:
  0x0feb76537fb0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0feb76537fc0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0feb76537fd0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0feb76537fe0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0feb76537ff0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
=>0x0feb76538000: f5 f5 f5 f5[f5]f5 f5 f5 00 00 00 00 00 00 00 00
  0x0feb76538010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0feb76538020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0feb76538030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0feb76538040: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x0feb76538050: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

valgrind détecte l'erreur.

valgrind --track-origins=yes ./main3

==2961== Syscall param exit_group(status) contains uninitialised byte(s)
==2961==    at 0x4F002B8: _Exit (_exit.c:31)
==2961==    by 0x4E70423: __run_exit_handlers (exit.c:98)
==2961==    by 0x4E704DC: exit (exit.c:105)
==2961==    by 0x4E58468: (below main) (libc-start.c:329)
==2961==  Uninitialised value was created by a stack allocation
==2961==    at 0x4004A4: main (main3.c:10)

main4.cc

#include <thread>
 
int main() {
  int x;
  std::thread t([&]{x=42;});
  x = 43;
  t.join();
 
  return 0;
}

g++ -std=c++11 main4.c -g -o main4 -fsanitize=thread

==================
WARNING: ThreadSanitizer: data race (pid=3226)
  Write of size 4 at 0x7ffd04d5b144 by thread T1:
    #0 operator() /tmp/main4.c:5 (main4+0x000000400d0b)
    #1 __invoke_impl<void, main()::<lambda()> > /usr/lib/gcc/x86_64-pc-linux-gnu/7.2.0/include/g++-v7/bits/invoke.h:60 (main4+0x000000401171)
    #2 __invoke<main()::<lambda()> > /usr/lib/gcc/x86_64-pc-linux-gnu/7.2.0/include/g++-v7/bits/invoke.h:95 (main4+0x000000400e30)
    #3 _M_invoke<0> /usr/lib/gcc/x86_64-pc-linux-gnu/7.2.0/include/g++-v7/thread:234 (main4+0x0000004014b5)
    #4 operator() /usr/lib/gcc/x86_64-pc-linux-gnu/7.2.0/include/g++-v7/thread:243 (main4+0x000000401448)
    #5 _M_run /usr/lib/gcc/x86_64-pc-linux-gnu/7.2.0/include/g++-v7/thread:186 (main4+0x0000004013ee)
    #6 <null> <null> (libstdc++.so.6+0x0000000e251e)

  Previous write of size 4 at 0x7ffd04d5b144 by main thread:
    #0 main /tmp/main4.c:6 (main4+0x000000400d75)

  Location is stack of main thread.

  Thread T1 (tid=3228, running) created by main thread at:
    #0 pthread_create <null> (libtsan.so.0+0x00000002917a)
    #1 std::thread::_M_start_thread(std::unique_ptr<std::thread::_State, std::default_delete<std::thread::_State> >, void (*)()) <null> (libstdc++.so.6+0x0000000e288c)
    #2 main /tmp/main4.c:5 (main4+0x000000400d69)

SUMMARY: ThreadSanitizer: data race /tmp/main4.c:5 in operator()
==================
ThreadSanitizer: reported 1 warnings

valgrind ne détecte pas l'erreur.

main5.c

int main(int argc, char **argv) {
  int t = argc << 16;
  return t*t;
}

gcc -fsanitize=undefined main5.c -g -o main5

main6.c:3:11: runtime error: signed integer overflow: 65536 * 65536 cannot be represented in type 'int'

valgrind ne détecte pas l'erreur.