diff options
author | Jing Yu <jingyu@google.com> | 2011-01-30 22:18:29 -0800 |
---|---|---|
committer | Jing Yu <jingyu@google.com> | 2011-01-30 22:18:29 -0800 |
commit | 4a66e756636cb8364582ea503abd10d76f5b4aa3 (patch) | |
tree | 9660204ec085888a0601a6460c967b204a63d5f3 /gcc-4.4.3/gcc/passes.c | |
parent | b6be42e837844cce5283f42fcfac31e6d66a277d (diff) | |
download | toolchain_gcc-4a66e756636cb8364582ea503abd10d76f5b4aa3.tar.gz toolchain_gcc-4a66e756636cb8364582ea503abd10d76f5b4aa3.tar.bz2 toolchain_gcc-4a66e756636cb8364582ea503abd10d76f5b4aa3.zip |
Upgrade gcc-4.4.3 for Android toolchain.
- Backport upstream patches to support arm hardfp.
- Backport gcc-4.5 patches to support -march=atom. Now it is
able to build atom toolchain with glibc from this branch
- Develop a bunch of optimizations
- Fix a few arm dejagnu failures
To-do list:
- Support Android/atom
- Fix ia32 bootstrap failure
Change-Id: I5e10dcd21620d4d8ca984d1d1707a76067e61691
Diffstat (limited to 'gcc-4.4.3/gcc/passes.c')
-rw-r--r-- | gcc-4.4.3/gcc/passes.c | 18 |
1 files changed, 17 insertions, 1 deletions
diff --git a/gcc-4.4.3/gcc/passes.c b/gcc-4.4.3/gcc/passes.c index e4765bf66..efc5370ab 100644 --- a/gcc-4.4.3/gcc/passes.c +++ b/gcc-4.4.3/gcc/passes.c @@ -581,6 +581,18 @@ init_optimization_passes (void) } NEXT_PASS (pass_ipa_increase_alignment); NEXT_PASS (pass_ipa_matrix_reorg); + /* The ipa multiversion_dispatch pass must run before ipa_cp. Otherwise, + it must restore ipa_node_params data structures for the clones + set by ipa_cp. */ + NEXT_PASS (pass_ipa_multiversion_dispatch); + { + struct opt_pass **p = &pass_ipa_multiversion_dispatch.pass.sub; + NEXT_PASS (pass_tree_convert_builtin_dispatch); + /* Rebuilding cgraph edges is necessary as the above passes change + the call graph. Otherwise, future optimizations use the old + call graph and make wrong decisions sometimes.*/ + NEXT_PASS (pass_rebuild_cgraph_edges); + } NEXT_PASS (pass_ipa_cp); NEXT_PASS (pass_ipa_inline); NEXT_PASS (pass_ipa_reference); @@ -594,6 +606,10 @@ init_optimization_passes (void) output to the assembler file. */ p = &all_passes; NEXT_PASS (pass_direct_call_profile); + /* Synthesize variables of union type whose fields are local variable + that can share stack slots. Do this before transformations that + perform code motion. */ + NEXT_PASS (pass_stack_overlay); NEXT_PASS (pass_all_optimizations); { struct opt_pass **p = &pass_all_optimizations.pass.sub; @@ -743,7 +759,6 @@ init_optimization_passes (void) NEXT_PASS (pass_cse); NEXT_PASS (pass_rtl_fwprop); NEXT_PASS (pass_gcse); - NEXT_PASS (pass_simplify_got); NEXT_PASS (pass_rtl_ifcvt); /* Perform loop optimizations. It might be better to do them a bit sooner, but we want the profile feedback to work more @@ -759,6 +774,7 @@ init_optimization_passes (void) NEXT_PASS (pass_rtl_loop_done); *p = NULL; } + NEXT_PASS (pass_simplify_got); NEXT_PASS (pass_web); NEXT_PASS (pass_jump_bypass); NEXT_PASS (pass_cse2); |