https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111151
--- Comment #14 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #12)
> The following testcase at least reproduces the unsigned multiplication
> issue, but doesn't reproduce the signed multiplication nor division by -1.
> int
> main ()
> {
> unsigned a = (1U + __INT_MAX__) / 2U;
> unsigned b = 1U;
> unsigned c = (a * 2U > b * 2U ? a * 2U : b * 2U) * 2U;
> if (c != 0U)
> __builtin_abort ();
> int d = (-__INT_MAX__ - 1) / 2;
> int e = 10;
> int f = (d * 2 > e * 2 ? d * 2 : e * 2) * 6;
> if (f != 120)
> __builtin_abort ();
> int g = (-__INT_MAX__ - 1) / 2;
> int h = 0;
> int i = (g * 2 > h * 2 ? g * 2 : h * 2) / -1;
> if (i != 0)
> __builtin_abort ();
> }
Ah, the reason it doesn't fail for the f and i cases is that for the signed
type cases, we actually don't create a MIN_EXPR or MAX_EXPR but COND_EXPR which
just compares the vars and performs multiplication only in the COND_EXPR
second/third arguments.
So it is kind of hard trying to make it trigger for the problematic cases where
the recursive calls would extract something.
Will see in full bootstrap/regtest with logging how often does the patch
trigger.
