On Wed, Feb 23, 2022 at 11:36:26AM +0100, Richard Biener wrote:
> > The following patch avoids infinite recursion during generic folding.
> > The (cmp (bswap @0) INTEGER_CST@1) simplification relies on
> > (bswap @1) actually being simplified, if it is not simplified, we just
> > move the bswap from one operand to the other and if @0 is also INTEGER_CST,
> > we apply the same rule next.
> >
> > The reason why bswap @1 isn't folded to INTEGER_CST is that the INTEGER_CST
> > has TREE_OVERFLOW set on it and fold-const-call.cc predicate punts in
> > such cases:
> > static inline bool
> > integer_cst_p (tree t)
> > {
> > return TREE_CODE (t) == INTEGER_CST && !TREE_OVERFLOW (t);
> > }
> > The patch uses ! modifier to ensure the bswap is simplified, but because
> > ! is only supported in gimple-match, guards it also with #if GIMPLE.
> >
> > Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
>
> Does it recurse on GIMPLE as well? If not, the ! is not necessary
> since you guard with GIMPLE already (but it deserves a comment I guess).
It could if TREE_OVERFLOW would somehow appear in there, but given that the
gimplifier and other spots drop_tree_overflow, it is only theoretical.
Note, I've posted a different version of the patch that allows optimizing
this even if GENERIC in most cases. I think I prefer that version.
Jakub
