Hello! 2016-04-29 Uros Bizjak <ubiz...@gmail.com>
* config/i386/i386.md (Load+RegOp to Mov+MemOp peephole2): Use SWI mode iterator. Use general_reg_operand predicate. (Load+RegOp to Mov+MemOp peephole2 with vector regs): Split peephole to MMX and SSE part. Use mmx_reg_operand and sse_reg_operand predicates. Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}. Committed to mainline SVN. Uros.
Index: i386.md =================================================================== --- i386.md (revision 235619) +++ i386.md (working copy) @@ -17905,20 +17905,20 @@ (match_op_dup 3 [(match_dup 2) (match_dup 0)])) (clobber (reg:CC FLAGS_REG))])]) -;; Prefer Load+RegOp to Mov+MemOp. Watch out for cases when the memory address -;; refers to the destination of the load! +;; Prefer Load+RegOp to Mov+MemOp. Watch out for cases when +;; the memory address refers to the destination of the load! (define_peephole2 - [(set (match_operand:SI 0 "register_operand") - (match_operand:SI 1 "register_operand")) + [(set (match_operand:SWI 0 "general_reg_operand") + (match_operand:SWI 1 "general_reg_operand")) (parallel [(set (match_dup 0) - (match_operator:SI 3 "commutative_operator" + (match_operator:SWI 3 "commutative_operator" [(match_dup 0) - (match_operand:SI 2 "memory_operand")])) + (match_operand:SWI 2 "memory_operand")])) (clobber (reg:CC FLAGS_REG))])] "REGNO (operands[0]) != REGNO (operands[1]) - && GENERAL_REGNO_P (REGNO (operands[0])) - && GENERAL_REGNO_P (REGNO (operands[1]))" + && (<MODE>mode != QImode + || any_QIreg_operand (operands[1], QImode))" [(set (match_dup 0) (match_dup 4)) (parallel [(set (match_dup 0) (match_op_dup 3 [(match_dup 0) (match_dup 1)])) @@ -17926,21 +17926,29 @@ "operands[4] = replace_rtx (operands[2], operands[0], operands[1], true);") (define_peephole2 - [(set (match_operand 0 "register_operand") - (match_operand 1 "register_operand")) + [(set (match_operand 0 "mmx_reg_operand") + (match_operand 1 "mmx_reg_operand")) (set (match_dup 0) (match_operator 3 "commutative_operator" [(match_dup 0) (match_operand 2 "memory_operand")]))] - "REGNO (operands[0]) != REGNO (operands[1]) - && ((MMX_REGNO_P (REGNO (operands[0])) - && MMX_REGNO_P (REGNO (operands[1]))) - || (SSE_REGNO_P (REGNO (operands[0])) - && SSE_REGNO_P (REGNO (operands[1]))))" + "REGNO (operands[0]) != REGNO (operands[1])" [(set (match_dup 0) (match_dup 2)) (set (match_dup 0) (match_op_dup 3 [(match_dup 0) (match_dup 1)]))]) +(define_peephole2 + [(set (match_operand 0 "sse_reg_operand") + (match_operand 1 "sse_reg_operand")) + (set (match_dup 0) + (match_operator 3 "commutative_operator" + [(match_dup 0) + (match_operand 2 "memory_operand")]))] + "REGNO (operands[0]) != REGNO (operands[1])" + [(set (match_dup 0) (match_dup 2)) + (set (match_dup 0) + (match_op_dup 3 [(match_dup 0) (match_dup 1)]))]) + ; Don't do logical operations with memory outputs ; ; These two don't make sense for PPro/PII -- we're expanding a 4-uop