[Bug middle-end/18908] Missed folding opportunities with bools

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18908

--- Comment #15 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-07-19 08:33:23 UTC ---
On x86_64 we have

f3:
.LFB2:
        .cfi_startproc
        xorb    $1, (%rdi)
        andb    $1, (%rdi)
        ret

already, thus only a truncation to bitfield precision is left.

As we expand ~*p as *p ^ 1 folding on the tree level isn't strictly necessary
so this is a missed opt on RTL or expand.