Avoid introducing a move for struct return (#6245)

* Move sret_reg handling to Lower:new, from gen_arg_setup

* Update exp files

* Review feedback

* Review feedback

cranelift/filetests/filetests/isa/aarch64/call.clif

@@ -850,16 +850,14 @@ block0(v0: i64):
 
 ; VCode:
 ; block0:
-;   mov x5, x8
-;   movz x4, #42
-;   str x4, [x8]
+;   movz x2, #42
+;   str x2, [x8]
 ;   ret
 ; 
 ; Disassembled:
 ; block0: ; offset 0x0
-;   mov x5, x8
-;   mov x4, #0x2a
-;   str x4, [x8]
+;   mov x2, #0x2a
+;   str x2, [x8]
 ;   ret
 
 function %f18(i64) -> i64 {
@@ -905,13 +903,14 @@ block0(v0: i64):
 ; VCode:
 ;   stp fp, lr, [sp, #-16]!
 ;   mov fp, sp
-;   str x24, [sp, #-16]!
+;   str x23, [sp, #-16]!
 ; block0:
-;   mov x24, x8
-;   load_ext_name x4, TestCase(%g)+0
-;   blr x4
-;   mov x8, x24
-;   ldr x24, [sp], #16
+;   mov x23, x8
+;   load_ext_name x2, TestCase(%g)+0
+;   mov x8, x23
+;   blr x2
+;   mov x8, x23
+;   ldr x23, [sp], #16
 ;   ldp fp, lr, [sp], #16
 ;   ret
 ; 
@@ -919,16 +918,17 @@ block0(v0: i64):
 ; block0: ; offset 0x0
 ;   stp x29, x30, [sp, #-0x10]!
 ;   mov x29, sp
-;   str x24, [sp, #-0x10]!
+;   str x23, [sp, #-0x10]!
 ; block1: ; offset 0xc
-;   mov x24, x8
-;   ldr x4, #0x18
+;   mov x23, x8
+;   ldr x2, #0x18
 ;   b #0x20
 ;   .byte 0x00, 0x00, 0x00, 0x00 ; reloc_external Abs8 %g 0
 ;   .byte 0x00, 0x00, 0x00, 0x00
-;   blr x4
-;   mov x8, x24
-;   ldr x24, [sp], #0x10
+;   mov x8, x23
+;   blr x2
+;   mov x8, x23
+;   ldr x23, [sp], #0x10
 ;   ldp x29, x30, [sp], #0x10
 ;   ret
 

cranelift/filetests/filetests/isa/x64/struct-ret.clif

@@ -12,9 +12,9 @@ block0(v0: i64):
 ;   pushq   %rbp
 ;   movq    %rsp, %rbp
 ; block0:
+;   movl    $42, %eax
+;   movq    %rax, 0(%rdi)
 ;   movq    %rdi, %rax
-;   movl    $42, %edx
-;   movq    %rdx, 0(%rdi)
 ;   movq    %rbp, %rsp
 ;   popq    %rbp
 ;   ret
@@ -24,9 +24,9 @@ block0(v0: i64):
 ;   pushq %rbp
 ;   movq %rsp, %rbp
 ; block1: ; offset 0x4
+;   movl $0x2a, %eax
+;   movq %rax, (%rdi) ; trap: heap_oob
 ;   movq %rdi, %rax
-;   movl $0x2a, %edx
-;   movq %rdx, (%rdi) ; trap: heap_oob
 ;   movq %rbp, %rsp
 ;   popq %rbp
 ;   retq
@@ -79,8 +79,9 @@ block0(v0: i64):
 ;   movq    %r15, 0(%rsp)
 ; block0:
 ;   movq    %rdi, %r15
-;   load_ext_name %f4+0, %rdx
-;   call    *%rdx
+;   load_ext_name %f4+0, %rax
+;   movq    %r15, %rdi
+;   call    *%rax
 ;   movq    %r15, %rax
 ;   movq    0(%rsp), %r15
 ;   addq    %rsp, $16, %rsp
@@ -96,8 +97,9 @@ block0(v0: i64):
 ;   movq %r15, (%rsp)
 ; block1: ; offset 0xc
 ;   movq %rdi, %r15
-;   movabsq $0, %rdx ; reloc_external Abs8 %f4 0
-;   callq *%rdx
+;   movabsq $0, %rax ; reloc_external Abs8 %f4 0
+;   movq %r15, %rdi
+;   callq *%rax
 ;   movq %r15, %rax
 ;   movq (%rsp), %r15
 ;   addq $0x10, %rsp