https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120578
Bug ID: 120578
Summary: Loop termination check eliminated under -O2/-O3/Os
when int overflow occurs
Product: gcc
Version: 15.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: 2023091106 at cauc dot edu.cn
Target Milestone: ---
in gcc11/12/13/14/15,When using optimization flags such as -O2, -O3, or -Os,
the program may fall into an infinite loop. Upon inspecting the generated
assembly code, the cmp and jl (compare and jump) instructions are missing,
indicating that the condition i < 10 has been optimized away by the compiler.
However, adding a branch like if (i == 2), or changing value += i * k to value
+= (long long)i * k, causes the program to execute correctly. This suggests
that under these optimization levels, GCC performs early evaluation of
arithmetic expressions and assumes int as the default type.
This behavior results in the for loop’s termination condition being affected by
internal computations, violating the Semantic Preservation Principle. Until a
more appropriate constraint relationship is established, such optimizations
should be avoided, especially when internal calculations can interfere with
loop bounds.
========the gcc code========
#include <stdio.h>
#include <stdlib.h>
int main() {
int value=0;
int k = 1024*1024*1024;
for (int i = 0; i < 10; i++) {
if (i == 3){ // when set if (i ==2), the code will work normal
break;
}
value += i * k;
printf("i: %d\n", i);
}
return value;
}
========the gcc output========
when use -O2/O3/Os, the code won't stop
$ gcc -Wall -Wextra -O0 -o test test.c
$ ./test
i: 0
i: 1
i: 2
$ gcc-14 -Wall -Wextra -O2 -o test test.c
$ ./test
i: 0
i: 1
i: 2
i: 3
...
========the gcc versions=========
$ gcc-14 -v
Using built-in specs.
COLLECT_GCC=gcc-14
COLLECT_LTO_WRAPPER=/opt/gcc-14/libexec/gcc/x86_64-pc-linux-gnu/14.1.0/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: ../configure --prefix=/opt/gcc-14 --enable-languages=c,c++
--disable-multilib
Thread model: posix
Supported LTO compression algorithms: zlib
gcc version 14.1.0 (GCC)
========the assembly code=========
when use -O2/O3/Os, the cmp and jl (compare and jump) instructions are missing
$in -O0
.LC0:
.string "i: %d\n"
main:
push rbx
mov ebx, 0
.L2:
mov esi, ebx
mov edi, OFFSET FLAT:.LC0
mov eax, 0
call printf
add ebx, 1
cmp ebx, 3
je .L4
cmp ebx, 9
jle .L2
.L4:
mov eax, 0
pop rbx
ret
$in -O2
.LC0:
.string "i: %d\n"
main:
push rbx
xor ebx, ebx
.L2:
mov esi, ebx
mov edi, OFFSET FLAT:.LC0
xor eax, eax
add ebx, 1
call printf
jmp .L2
