py: Make asm_arm_less_op take destination register as first arg.

This gets ARM native emitter working againg and addresses issue #858.

py/asmarm.c

@@ -287,10 +287,10 @@ void asm_arm_cmp_reg_reg(asm_arm_t *as, uint rd, uint rn) {
     emit_al(as, 0x1500000 | (rd << 16) | rn);
 }
 
-void asm_arm_less_op(asm_arm_t *as, uint rd, uint rn) {
-    asm_arm_cmp_reg_reg(as, rd, rn); // cmp rd, rn
-    emit(as, asm_arm_op_mov_imm(REG_RET, 1) | ARM_CC_LT); // movlt REG_RET, #1
-    emit(as, asm_arm_op_mov_imm(REG_RET, 0) | ARM_CC_GE); // movge REG_RET, #0
+void asm_arm_less_op(asm_arm_t *as, uint rd, uint rn, uint rm) {
+    asm_arm_cmp_reg_reg(as, rn, rm); // cmp rn, rm
+    emit(as, asm_arm_op_mov_imm(rd, 1) | ARM_CC_LT); // movlt rd, #1
+    emit(as, asm_arm_op_mov_imm(rd, 0) | ARM_CC_GE); // movge rd, #0
 }
 
 void asm_arm_add_reg(asm_arm_t *as, uint rd, uint rn, uint rm) {

py/asmarm.h

@@ -88,7 +88,7 @@ void asm_arm_mov_reg_local(asm_arm_t *as, uint rd, int local_num);
 
 void asm_arm_cmp_reg_i8(asm_arm_t *as, uint rd, int imm);
 void asm_arm_cmp_reg_reg(asm_arm_t *as, uint rd, uint rn);
-void asm_arm_less_op(asm_arm_t *as, uint rd, uint rn);
+void asm_arm_less_op(asm_arm_t *as, uint rd, uint rn, uint rm);
 void asm_arm_add_reg(asm_arm_t *as, uint rd, uint rn, uint rm);
 void asm_arm_mov_reg_local_addr(asm_arm_t *as, uint rd, int local_num);
 

py/emitnative.c

@@ -1514,7 +1514,7 @@ STATIC void emit_native_binary_op(emit_t *emit, mp_binary_op_t op) {
             asm_thumb_mov_rlo_i8(emit->as, REG_RET, 0); // if r0 >= r1
             asm_thumb_mov_rlo_i8(emit->as, REG_RET, 1); // if r0 < r1
 #elif N_ARM
-            asm_arm_less_op(emit->as, REG_ARG_2, REG_ARG_3);
+            asm_arm_less_op(emit->as, REG_RET, REG_ARG_2, REG_ARG_3);
 #else
     #error not implemented
 #endif