1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-linux | FileCheck %s --check-prefix=X86
; RUN: llc < %s -mtriple=x86_64-linux | FileCheck %s --check-prefix=X64
; RUN: llc < %s -mtriple=x86_64-linux-gnux32 | FileCheck %s --check-prefix=X64
; RUN: llc < %s -mtriple=x86_64-nacl | FileCheck %s --check-prefix=X64
; The computation of %t4 should match a single lea, without using actual add instructions.
define i32 @test1(i32 %A, i32 %B) {
; X86-LABEL: test1:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
; X86-NEXT: leal -5(%ecx,%eax,4), %eax
; X86-NEXT: retl
;
; X64-LABEL: test1:
; X64: # %bb.0:
; X64-NEXT: # kill: def $esi killed $esi def $rsi
; X64-NEXT: # kill: def $edi killed $edi def $rdi
; X64-NEXT: leal -5(%rsi,%rdi,4), %eax
; X64-NEXT: retq
%t1 = shl i32 %A, 2
%t3 = add i32 %B, -5
%t4 = add i32 %t3, %t1
ret i32 %t4
}
; The addlike OR instruction should fold into the LEA.
define i64 @test2(i32 %a0, i64 %a1) {
; X86-LABEL: test2:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
; X86-NEXT: movl %edx, %eax
; X86-NEXT: andl $2147483640, %eax # imm = 0x7FFFFFF8
; X86-NEXT: shrl $31, %edx
; X86-NEXT: leal 4(%eax,%eax), %eax
; X86-NEXT: addl {{[0-9]+}}(%esp), %eax
; X86-NEXT: adcl {{[0-9]+}}(%esp), %edx
; X86-NEXT: retl
;
; X64-LABEL: test2:
; X64: # %bb.0:
; X64-NEXT: # kill: def $edi killed $edi def $rdi
; X64-NEXT: andl $-8, %edi
; X64-NEXT: leaq 4(%rsi,%rdi,2), %rax
; X64-NEXT: retq
%x1 = and i32 %a0, -8
%x2 = or i32 %x1, 2
%x3 = zext i32 %x2 to i64
%x4 = shl i64 %x3, 1
%x5 = add i64 %a1, %x4
ret i64 %x5
}
|
