I need to run this simple inline assembly code:
#include <stdio.h>
int count;
int main() {
count = 0;
for (int i = 0; i < 10; i++) {
asm volatile ("incl count"); // count++
}
printf("count=%d\n", count);
return 0;
}
It works fine (printing count=10) until I turn on optimization (gcc -O1), in which case it prints count=0. I read that the "volatile" qualifier will prevent the optimizer to put the code out of loop. But it seems to have no effect here.
Problem found. I had to define the global variable "count" as volatile. The problem was not putting asm block out of the loop. The compiler simply replaced count with 0 in printf.
volatile int count;
int main() {
count = 0;
for (int i = 0; i < 10; i++) {
asm volatile ("incl count"); // count++
}
printf("count=%d\n", count);
return 0;
}
volatile will kill the C parts completely. It's completely confusing me, why you are pushing these tools somewhere where they shouldn't work, and are in docs recommended to not push them like that, why don't you write rather pure assembly functions if you want to learn assembly, which is much easier than learning assembly and on top of that also learning the gcc inline assembly rules, which are far from trivial.
++count;, it will end better (faster, shorter, etc..).
asm ("incl %0" : "+rm"(count));should work whethercountis a local or global variable . The manual states Accessing data from C programs without using input/output operands (such as by using global symbols directly from the assembler template) may not work as expected.countis being affected, so the optimizer is free to assume it never changes from the initial value of 0. From the docs: GCC does not parse the assembler instructions themselves and does not know what they mean or even whether they are valid assembler input.