The binary is a flag printer - but it takes a lot of time process the input which is under layers of xor.
The binary `realloc`s a buffer for which size is multiplied by 2 many times. With time this operation gets costly or probably oob and null ptr derefernce
A simple way is to patch the multiplication to addition of 32 - the length of the flag.
Or just F5(decompile) in IDA copy, edit and rebuild since the decompilation was neat.

```c
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

void shuffle(char *s, int seed, int mul) {
srand(seed);
int v6 = 0LL, v7 = 0, v4 = 0;
for (int i = 0; i <= 31; ++i) {
v7 = rand() + v6;
v6 *= mul;
v4 = s[v7 % 32];
s[v7 % 32] = s[v6 % 32];
s[v6 % 32] = v4;
}
}
char res[0x200];
unsigned long long v3[4];
int main() {
char *s = "actf{l_plus_ratio_plus_no_flags}";
char *orig = strdup(s);
int ops[] = {7, 2, 3, 2, 2, 6, 2, 9, 2, 6, 2, 2, 1, 2};
int residx = 0;
for (int o = 0; o < sizeof(ops) / sizeof(*ops); o++) {
if (ops[o] == 2) {
memcpy(&res[32 * residx++], orig, 0x20);
} else if (ops[o] == 3) {
shuffle(orig, 2, 2);
} else if (ops[o] == 6) {
shuffle(orig, 0, 0);
} else if (ops[o] == 1) {
shuffle(orig, 1, 1);
} else if (ops[o] == 9) {
for (int i = 0; i < sizeof(res) / sizeof(*res); i++) {
res[i] *= 9;
}
}
}
for (int i = 0; i < 0x20; i++) {
orig[i] = 0;
}
for (int i = 0; i < 0x200; i++) {
orig[i % 32] ^= res[i];
}
v3[0] = 0x5646564D6104635FULL;
v3[1] = 0xD2DA506ACC801521ULL;
v3[2] = 0x660A7262DAB5E734ULL;
v3[3] = 0x8C0A3EFC87E57E8CULL;
char *x = (char *)v3;
for (int i = 0; i <= 31; ++i)
orig[i] ^= x[i];
printf("%32s\n", orig);
}
```