This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] Fix ACATS failures in GCC 3.4
- From: Eric Botcazou <ebotcazou at act-europe dot fr>
- To: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Thu, 26 Feb 2004 14:59:04 +0100
- Subject: Re: [PATCH] Fix ACATS failures in GCC 3.4
- References: <10402261335.AA27442@vlsi1.ultra.nyu.edu>
> Define "strip". If you mean in the tree, no, it's not: the types of
> the operands of an arithmetic expression must *exactly* match that of
> the expression itself.
OK. Pretty strong requirement.
> If you mean in setting ARG0 and ARG1, then yes, it's safe to strip them
> because those values aren't supposed to be put back into the tree without
> being converted to the type of the tree.
Then I'm under the impression someone should rewrite fold :-)
Is this valid (fold-const.c:5482) ?
t = fold (build (code == BIT_AND_EXPR ? TRUTH_AND_EXPR
: code == BIT_IOR_EXPR ? TRUTH_OR_EXPR
type, arg0, arg1));
if (code == EQ_EXPR)
t = invert_truthvalue (t);