optimize bitcasts from large integers to vector into vector

element insertion from the pieces that feed into the vector.
This handles a pattern that occurs frequently due to code
generated for the x86-64 abi.  We now compile something like
this:

struct S { float A, B, C, D; };
struct S g;
struct S bar() { 
  struct S A = g;
  ++A.A;
  ++A.C;
  return A;
}

into all nice vector operations:

_bar:                                   ## @bar
## BB#0:                                ## %entry
	movq	_g@GOTPCREL(%rip), %rax
	movss	LCPI1_0(%rip), %xmm1
	movss	(%rax), %xmm0
	addss	%xmm1, %xmm0
	pshufd	$16, %xmm0, %xmm0
	movss	4(%rax), %xmm2
	movss	12(%rax), %xmm3
	pshufd	$16, %xmm2, %xmm2
	unpcklps	%xmm2, %xmm0
	addss	8(%rax), %xmm1
	pshufd	$16, %xmm1, %xmm1
	pshufd	$16, %xmm3, %xmm2
	unpcklps	%xmm2, %xmm1
	ret

instead of icky integer operations:

_bar:                                   ## @bar
	movq	_g@GOTPCREL(%rip), %rax
	movss	LCPI1_0(%rip), %xmm1
	movss	(%rax), %xmm0
	addss	%xmm1, %xmm0
	movd	%xmm0, %ecx
	movl	4(%rax), %edx
	movl	12(%rax), %esi
	shlq	$32, %rdx
	addq	%rcx, %rdx
	movd	%rdx, %xmm0
	addss	8(%rax), %xmm1
	movd	%xmm1, %eax
	shlq	$32, %rsi
	addq	%rax, %rsi
	movd	%rsi, %xmm1
	ret

This resolves rdar://8360454

llvm-svn: 112343

1 files changed, 31 insertions, 0 deletions

diff --git a/llvm/test/Transforms/InstCombine/bitcast.ll b/llvm/test/Transforms/InstCombine/bitcast.ll
index 10898397b98..87e413ea27f 100644
--- a/llvm/test/Transforms/InstCombine/bitcast.ll
+++ b/llvm/test/Transforms/InstCombine/bitcast.ll
@@ -60,3 +60,34 @@ define float @test3(<2 x float> %A, <2 x i64> %B) {
 ; CHECK-NEXT:  %add = fadd float %tmp24, %tmp4
 ; CHECK-NEXT:  ret float %add
 }
+
+
+define <2 x i32> @test4(i32 %A, i32 %B){
+  %tmp38 = zext i32 %A to i64
+  %tmp32 = zext i32 %B to i64
+  %tmp33 = shl i64 %tmp32, 32
+  %ins35 = or i64 %tmp33, %tmp38
+  %tmp43 = bitcast i64 %ins35 to <2 x i32>
+  ret <2 x i32> %tmp43
+  ; CHECK: @test4
+  ; CHECK-NEXT: insertelement <2 x i32> undef, i32 %A, i32 0
+  ; CHECK-NEXT: insertelement <2 x i32> {{.*}}, i32 %B, i32 1
+  ; CHECK-NEXT: ret <2 x i32> 
+
+}
+
+; rdar://8360454
+define <2 x float> @test5(float %A, float %B) {
+  %tmp37 = bitcast float %A to i32
+  %tmp38 = zext i32 %tmp37 to i64
+  %tmp31 = bitcast float %B to i32
+  %tmp32 = zext i32 %tmp31 to i64
+  %tmp33 = shl i64 %tmp32, 32
+  %ins35 = or i64 %tmp33, %tmp38
+  %tmp43 = bitcast i64 %ins35 to <2 x float>
+  ret <2 x float> %tmp43
+  ; CHECK: @test5
+  ; CHECK-NEXT: insertelement <2 x float> undef, float %A, i32 0
+  ; CHECK-NEXT: insertelement <2 x float> {{.*}}, float %B, i32 1
+  ; CHECK-NEXT: ret <2 x float> 
+}