complibs/ppl: build only C and C++ interfaces for PPL
By default, PPL wants to build interfaces for any of a variety of
langauges it finds on the local host (python, java, possibly perl, also
more esoteric languages such as ocaml and prolog).
These extra interfaces can double the compile time for the library. For
single-process builds, I found a savings of more than 40%:
default / j1: 716s total, 143.2s avg, 0.52s stdev
just_c / j1: 406s total, 81.2s avg, 0.33s stdev
just_c_cpp / j1: 413s total, 82.6s avg, 0.22s stdev
And for multi-process builds, it approached 50%:
default / j4: 625s total, 125.0s avg, 0.57s stdev
just_c / j4: 338s total, 67.6s avg, 1.25s stdev
just_c_cpp / j4: 327s total, 65.4s avg, 0.36s stdev
Since the PPL we build within ct-ng is only used by GCC, we only need to
build the C and C++ interfaces.
Signed-Off-By: Anthony Foiani <anthony.foiani@gmail.com>
1 Original patch from gentoo: gentoo/src/patchsets/gcc/4.3.1/gentoo/18_all_904-flatten-switch-stmt-00.patch
2 http://gcc.gnu.org/ml/gcc-patches/2007-04/msg00927.html
6 The attached patch makes sure that we create smaller object code for
7 simple switch statements. We just make sure to flatten the switch
8 statement into an if-else chain, basically.
10 This fixes a size-regression as compared to gcc-3.4, as can be seen
13 2007-04-15 Bernhard Fischer <..>
15 * stmt.c (expand_case): Do not create a complex binary tree when
16 optimizing for size but rather use the simple ordered list.
17 (emit_case_nodes): do not emit jumps to the default_label when
23 Attached is the test switch.c mentioned below.
25 $ for i in 2.95 3.3 3.4 4.0 4.1 4.2.orig-HEAD 4.3.orig-HEAD 4.3-HEAD;do
26 gcc-$i -DCHAIN -Os -o switch-CHAIN-$i.o -c switch.c ;done
27 $ for i in 2.95 3.3 3.4 4.0 4.1 4.2.orig-HEAD 4.3.orig-HEAD 4.3-HEAD;do
28 gcc-$i -UCHAIN -Os -o switch-$i.o -c switch.c ;done
31 text data bss dec hex filename
32 169 0 0 169 a9 switch-2.95.o
33 115 0 0 115 73 switch-3.3.o
34 103 0 0 103 67 switch-3.4.o
35 124 0 0 124 7c switch-4.0.o
36 124 0 0 124 7c switch-4.1.o
37 124 0 0 124 7c switch-4.2.orig-HEAD.o
38 95 0 0 95 5f switch-4.3-HEAD.o
39 124 0 0 124 7c switch-4.3.orig-HEAD.o
40 166 0 0 166 a6 switch-CHAIN-2.95.o
41 111 0 0 111 6f switch-CHAIN-3.3.o
42 95 0 0 95 5f switch-CHAIN-3.4.o
43 95 0 0 95 5f switch-CHAIN-4.0.o
44 95 0 0 95 5f switch-CHAIN-4.1.o
45 95 0 0 95 5f switch-CHAIN-4.2.orig-HEAD.o
46 95 0 0 95 5f switch-CHAIN-4.3-HEAD.o
47 95 0 0 95 5f switch-CHAIN-4.3.orig-HEAD.o
50 Content-Type: text/x-diff; charset=us-ascii
51 Content-Disposition: attachment; filename="gcc-4.3.gcc-flatten-switch-stmt.00.diff"
53 diff -durN gcc-4.3.1.orig/gcc/stmt.c gcc-4.3.1/gcc/stmt.c
54 --- gcc-4.3.1.orig/gcc/stmt.c 2008-05-09 20:12:13.000000000 +0200
55 +++ gcc-4.3.1/gcc/stmt.c 2008-06-10 14:57:58.000000000 +0200
56 @@ -2509,7 +2509,11 @@
58 = (TREE_CODE (orig_type) != ENUMERAL_TYPE
59 && estimate_case_costs (case_list));
60 - balance_case_nodes (&case_list, NULL);
61 + /* When optimizing for size, we want a straight list to avoid
62 + jumps as much as possible. This basically creates an if-else
65 + balance_case_nodes (&case_list, NULL);
66 emit_case_nodes (index, case_list, default_label, index_type);
67 emit_jump (default_label);
71 if (!node_has_low_bound (node, index_type))
73 + if (!optimize_size) /* don't jl to the .default_label. */
74 emit_cmp_and_jump_insns (index,