[PATCH] Simplify builtins when result ignored.

[Zack Weinberg]

Roger Sayle <roger@www.eyesopen.com> writes:

> This patch avoids expanding built-in functions for strlen, memcmp,
> strcmp, strncmp and similar functions when its known the result is
> ignored and the pointer arguments are not volatile.  This reduces
> the amount of initial RTL we generate and the usual good things.
> These calls are eliminated by life analysis in the RTL optimizers
> (if the "pure" attribute machinery is working :), but less useless
> RTL means a smaller memory foot-print, less fragmentation, less
> work for the early RTL optimizers, etc...

This is a good idea, but can't you make it a general optimization done
for all calls to functions with attribute pure or const, rather than
applying it individually in all these expanders?

Also it might be a good idea to do it at "fold" time as well, which
would avoid even more work.

> Whilst there, I also added the expansion-time optimization of strlen
> for constant strings.  This is typically handled by "fold", but
> performing the check in expand_builtin_strlen provides both a good
> pedagogical example and reduces the dependencies between GCC passes.
> For example, if tree-SSA or a language front-end didn't call "fold"
> on an expression before RTL expansion.

This piece is approved for mainline -- but only as a separate patch
from the other changes.

zw