This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] one-liner: deinline gcc/tree-vrp.c vrp_int_const_binop()
- From: Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>
- To: Denis Vlasenko <vda dot linux at googlemail dot com>
- Cc: Daniel Berlin <dannyb at google dot com>, H?kan Hjort <hakan at jasper-da dot com>, Richard Guenther <richard dot guenther at gmail dot com>, gcc-patches at gcc dot gnu dot org
- Date: Sat, 12 Aug 2006 16:02:47 +0200
- Subject: Re: [PATCH] one-liner: deinline gcc/tree-vrp.c vrp_int_const_binop()
- References: <200608121533.58521.vda.linux@googlemail.com>
Helo,
> * tree-vrp.c (vrp_int_const_binop): deinline it
please follow the conventions regarding the format of changelog
messages.
> There are two dozens of calls to that function.
> Before/after code size comparison:
>
> # size tree-vrp_inline.o tree-vrp_noinline.o
> text data bss dec hex filename
> 47147 52 24 47223 b877 tree-vrp_inline.o
> 44273 52 24 44349 ad3d tree-vrp_noinline.o
>
> This patch was compile-tested.
even trivial patches like this need full bootstrap and regtesting. The
reason is that even though this change is obviously harmless, it might
uncover some latent bug (the things will be optimized differently
depending on whether vrp_int_const_binop is inlined or not).
Zdenek
> --
> vda
> diff -urpN gcc-4.2-20060805.org/gcc/tree-vrp.c gcc-4.2-20060805/gcc/tree-vrp.c
> --- gcc-4.2-20060805.org/gcc/tree-vrp.c 2006-08-01 02:47:49.000000000 +0200
> +++ gcc-4.2-20060805/gcc/tree-vrp.c 2006-08-12 15:16:18.000000000 +0200
> @@ -1226,7 +1226,7 @@ extract_range_from_ssa_name (value_range
> are not using wrapping arithmetic, then adjust the result to be
> -INF or +INF depending on CODE, VAL1 and VAL2. */
>
> -static inline tree
> +static tree
> vrp_int_const_binop (enum tree_code code, tree val1, tree val2)
> {
> tree res;