I've applied the attached patch to fix PR target/62111 which is a sh64 specific 4.9/5 regression. It seems that *zero_extendhisi2_media makes reload unhappy with permitting (truncate (pseudo_reg)) when pseudo_reg can be substituted by mem. See
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62111 for details. Tested on sh64-elf and sh4-unknown-linux-gnu with no new failures. I'll backport this to 4.9 in a week or so. Regards, kaz -- 2014-08-25 Kaz Kojima <kkoj...@gcc.gnu.org> PR target/62111 * config/sh/predicates.md (general_extend_operand): Disable TRUNCATE before reload completes. --- gcc/config/sh/predicates.md.orig 2014-08-02 11:55:29.228875715 +0900 +++ gcc/config/sh/predicates.md 2014-08-17 08:30:20.439326569 +0900 @@ -398,7 +398,7 @@ (define_predicate "general_extend_operand" (match_code "subreg,reg,mem,truncate") { - if (GET_CODE (op) == TRUNCATE) + if (reload_completed && GET_CODE (op) == TRUNCATE) return arith_operand (op, mode); if (MEM_P (op) || (GET_CODE (op) == SUBREG && MEM_P (SUBREG_REG (op))))
