[Bug sanitizer/108085] New: gcc trunk's ASAN at -O3 missed a stack-use-after-scope

[Bug sanitizer/108085] New: gcc trunk's ASAN at -O3 missed a stack-use-after-scope

[shaohua.li at inf dot ethz.ch]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108085

            Bug ID: 108085
           Summary: gcc trunk's ASAN at -O3 missed a stack-use-after-scope
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: sanitizer
          Assignee: unassigned at gcc dot gnu.org
          Reporter: shaohua.li at inf dot ethz.ch
                CC: dodji at gcc dot gnu.org, dvyukov at gcc dot gnu.org,
                    jakub at gcc dot gnu.org, kcc at gcc dot gnu.org, marxin at gcc dot gnu.org
  Target Milestone: ---

For the following code, ASAN at -O3 failed to report the stack-use-after-scope,
other opt levels detected it successfully.

GCC-8 works fine. 
Clang at all optimization levels reported it.

Compiler explorer: https://godbolt.org/z/jbsThYWa6

% cat a.c
int a;
void b() {
  int c[1];
  c;
}
int main() {
  int d[1]={1};
  int *e = d;
  a = 0;
  for (; a <= 5; ++a) {
    int f[1]={};
    e = f;
    a || (b(), 1);
  }
  return *e;
}
%
% gcc-tk -O2 -fsanitize=address -g a.c && ./a.out
=================================================================
==780198==ERROR: AddressSanitizer: stack-use-after-scope on address
0x7f2307600030 at pc 0x0000004011c5 bp 0x7ffdd19ecd80 sp 0x7ffdd19ecd78
READ of size 4 at 0x7f2307600030 thread T0
    #0 0x4011c4 in main /a.c:15
    #1 0x7f2309f41082 in __libc_start_main
(/lib/x86_64-linux-gnu/libc.so.6+0x24082) (BuildId:
1878e6b475720c7c51969e69ab2d276fae6d1dee)
    #2 0x40126d in _start (/a.out+0x40126d)

Address 0x7f2307600030 is located in stack of thread T0 at offset 48 in frame
    #0 0x4010bf in main /a.c:6

  This frame has 3 object(s):
    [32, 36) 'd' (line 7)
    [48, 52) 'f' (line 11) <== Memory access at offset 48 is inside this
variable
    [64, 68) 'c' (line 3)
...
%
% gcc-tk -O3 -fsanitize=address -g a.c && ./a.out
%

[Bug sanitizer/108085] gcc trunk's ASAN at -O3 missed a stack-use-after-scope

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108085

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |marxin at gcc dot gnu.org
   Last reconfirmed|                            |2022-12-14
             Status|UNCONFIRMED                 |ASSIGNED
     Ever confirmed|0                           |1

--- Comment #1 from Martin Liška <marxin at gcc dot gnu.org> ---
I can take a look.

Note I can't reproduce it with Clang with any of -O1+ options:

$ clang pr108085.c -fsanitize=address -w && ./a.out
=================================================================
==20150==ERROR: AddressSanitizer: stack-use-after-scope on address
0x7ffff5e00030 at pc 0x55555566a2aa bp 0x7fffffffd510 sp 0x7fffffffd508
READ of size 4 at 0x7ffff5e00030 thread T0
    #0 0x55555566a2a9 in main pr108085.c
    #1 0x7ffff7ca35af in __libc_start_call_main (/lib64/libc.so.6+0x275af)
(BuildId: 3f7f40d4302fa8a9568a057065c023bf137fceb7)
    #2 0x7ffff7ca3678 in __libc_start_main@GLIBC_2.2.5
(/lib64/libc.so.6+0x27678) (BuildId: 3f7f40d4302fa8a9568a057065c023bf137fceb7)
    #3 0x5555555a9d64 in _start
/home/abuild/rpmbuild/BUILD/glibc-2.36/csu/../sysdeps/x86_64/start.S:115

Address 0x7ffff5e00030 is located in stack of thread T0 at offset 48 in frame
    #0 0x55555566a11f in main pr108085.c

  This frame has 2 object(s):
    [32, 36) 'd'
    [48, 52) 'f' <== Memory access at offset 48 is inside this variable
HINT: this may be a false positive if your program uses some custom stack
unwind mechanism, swapcontext or vfork
      (longjmp and C++ exceptions *are* supported)
SUMMARY: AddressSanitizer: stack-use-after-scope pr108085.c in main
Shadow bytes around the buggy address:
  0x10007ebb7fb0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x10007ebb7fc0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x10007ebb7fd0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x10007ebb7fe0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x10007ebb7ff0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
=>0x10007ebb8000: f1 f1 f1 f1 04 f2[f8]f3 00 00 00 00 00 00 00 00
  0x10007ebb8010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x10007ebb8020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x10007ebb8030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x10007ebb8040: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  0x10007ebb8050: 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
==20150==ABORTING

$ clang pr108085.c -fsanitize=address -w -O1 && ./a.out
(no output)

[Bug sanitizer/108085] gcc trunk's ASAN at -O3 missed a stack-use-after-scope

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108085

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
I bet we unroll the loop and then elide f[] at -O3.

[Bug sanitizer/108085] gcc trunk's ASAN at -O3 missed a stack-use-after-scope

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108085

--- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
Created attachment 54153
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=54153&action=edit
pr108085.c.216t.uncprop1.dot.svg

So no, it's a real issue where we optimize out .ASAN_CHECK (6, &f, 4, 8); in
the exit block. As seen in the dump file, we have the very ASAN_CHECK in bb_3:
.ASAN_CHECK (7, &f, 4, 8), however, there are 2 ASAN_MARK (POISON, &f, 4) calls
that are on the path from bb_3 to the exit block.

@Jakub: Can you please take a look at the optimization algorithm why the check
is not preserved?

[Bug sanitizer/108085] gcc trunk's ASAN at -O3 missed a stack-use-after-scope

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108085

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|marxin at gcc dot gnu.org          |unassigned at gcc dot gnu.org
             Status|ASSIGNED                    |NEW

[Bug sanitizer/108085] gcc trunk's ASAN at -O3 missed a stack-use-after-scope

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108085

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Hmm, using the C++ front-end, the use-after-scope still happens at -O3 but not
with the C front-end.

[Bug sanitizer/108085] gcc trunk's ASAN at -O3 missed a stack-use-after-scope

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108085

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
The difference between the two front-ends is at the original.
The C++ front-end adds a BLOCK around the loop while the C front-end does not.

This difference changes where the ASAN_MARK is placed with respect to the
return statement.