diff options
| author | Jerry DeLisle <jvdelisle@gcc.gnu.org> | 2025-09-02 15:58:26 -0700 |
|---|---|---|
| committer | Jerry DeLisle <jvdelisle@gcc.gnu.org> | 2025-09-02 15:58:26 -0700 |
| commit | 071b4126c613881f4cb25b4e5c39032964827f88 (patch) | |
| tree | 7ed805786566918630d1d617b1ed8f7310f5fd8e /gcc/config/avr/avr-passes.def | |
| parent | 845d23f3ea08ba873197c275a8857eee7edad996 (diff) | |
| parent | caa1c2f42691d68af4d894a5c3e700ecd2dba080 (diff) | |
| download | gcc-devel/gfortran-test.zip gcc-devel/gfortran-test.tar.gz gcc-devel/gfortran-test.tar.bz2 | |
Merge branch 'master' into gfortran-testdevel/gfortran-test
Diffstat (limited to 'gcc/config/avr/avr-passes.def')
| -rw-r--r-- | gcc/config/avr/avr-passes.def | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/gcc/config/avr/avr-passes.def b/gcc/config/avr/avr-passes.def index eb60a93..d668c7f 100644 --- a/gcc/config/avr/avr-passes.def +++ b/gcc/config/avr/avr-passes.def @@ -74,6 +74,14 @@ INSERT_PASS_BEFORE (pass_free_cfg, 1, avr_pass_recompute_notes); INSERT_PASS_AFTER (pass_expand, 1, avr_pass_casesi); +/* Insn combine may come up with superfluous reg-reg moves, where the combine + people say that these are no problem since reg-alloc is supposed to optimize + them. The issue is that the lower-subreg pass sitting between combine and + reg-alloc may split such moves, coming up with a zoo of subregs which are + only handled poorly by the register allocator. */ + +INSERT_PASS_AFTER (pass_combine, 1, avr_pass_2moves); + /* Some combine insns have nonzero_bits() in their condition, though insns should not use such stuff in their condition. Therefore, we split such insn into something without nonzero_bits() in their condition right after |