Strict aliasing-related bug even with optimizations disabled

[Andrew Pinski]

On Sat, Mar 17, 2018 at 11:40 AM, Jonathon Reinhart
<jonathon.reinhart@gmail.com> wrote:
> This was originally posted on Stack Overflow:
> https://stackoverflow.com/a/49339771/119527
>
> The following program:
>
>     #include <stdio.h>
>
>     static void pshort(short val)
>     {
>        printf("0x%hx ", val);
>     }
>
>     int main(void)
>     {
>        short A[] = {1, 2, 3, 4, 5, 6};
>
>     #define EXP ((short*)((char*)A + 7))
>
>        short *p = EXP;
>        short q = *EXP;
>
>        pshort(*p);
>        pshort(q);
>        pshort(*EXP);
>        printf("\n");
>
>        return 0;
>     }
>
> ...when compiled on x86-64 with
>
>     gcc -O0 -fno-strict-aliasing -g -Wall -Werror endian.c
>
> ...produces the following unexpected output
>
>     0x500 0x500 0x4
>
> ...on all available versions of GCC, from 4.9 thru 7.3.1.
>
> It appears that GCC is actually generating different code when the
> expression is used directly as an argument, versus when used with
> intermediate variables.
>
> The language lawyers declared that UB is UB. But -fno-strict-aliasing
> seems to be ineffective here, and this feels like a bug.


As mentioned this is the right mailing list for this kind of question.
But this is undefined not due to aliasing but alignment reasons.
short has an alignment requirement of 2 while you made it unaligned.

Thanks,
Andrew

>
> Jonathon Reinhart