When widening a vector load, use the correct chain. This fixes PR4891.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@81343 91177308-0d34-0410-b5e6-96231b3b80d8

2 files changed, 66 insertions, 0 deletions

diff --git a/test/CodeGen/X86/widen_load-0.ll b/test/CodeGen/X86/widen_load-0.ll
new file mode 100644
index 0000000..f6c4af0
--- /dev/null
+++ b/test/CodeGen/X86/widen_load-0.ll
@@ -0,0 +1,21 @@
+; RUN: llc < %s -o - -march=x86-64 | FileCheck %s
+; PR4891
+
+; Both loads should happen before either store.
+
+; CHECK: movl  (%rdi), %eax
+; CHECK: movl  (%rsi), %ecx
+; CHECK: movl  %ecx, (%rdi)
+; CHECK: movl  %eax, (%rsi)
+
+define void @short2_int_swap(<2 x i16>* nocapture %b, i32* nocapture %c) nounwind {
+entry:
+  %0 = load <2 x i16>* %b, align 2                ; <<2 x i16>> [#uses=1]
+  %1 = load i32* %c, align 4                      ; <i32> [#uses=1]
+  %tmp1 = bitcast i32 %1 to <2 x i16>             ; <<2 x i16>> [#uses=1]
+  store <2 x i16> %tmp1, <2 x i16>* %b, align 2
+  %tmp5 = bitcast <2 x i16> %0 to <1 x i32>       ; <<1 x i32>> [#uses=1]
+  %tmp3 = extractelement <1 x i32> %tmp5, i32 0   ; <i32> [#uses=1]
+  store i32 %tmp3, i32* %c, align 4
+  ret void
+}
diff --git a/test/CodeGen/X86/widen_load-1.ll b/test/CodeGen/X86/widen_load-1.ll
new file mode 100644
index 0000000..2d34b31
--- /dev/null
+++ b/test/CodeGen/X86/widen_load-1.ll
@@ -0,0 +1,45 @@
+; RUN: llc %s -o - -march=x86-64 -mtriple=x86_64-unknown-linux-gnu -disable-mmx | FileCheck %s
+; PR4891
+
+; This load should be before the call, not after.
+
+; CHECK: movq    compl+128(%rip), %xmm0
+; CHECK: movaps  %xmm0, (%rsp)
+; CHECK: call    killcommon
+
+@compl = linkonce global [20 x i64] zeroinitializer, align 64 ; <[20 x i64]*> [#uses=1]
+
+declare void @killcommon(i32* noalias)
+
+define void @reset(<2 x float>* noalias %garbage1) {
+"file complex.c, line 27, bb1":
+  %changed = alloca i32, align 4                  ; <i32*> [#uses=3]
+  br label %"file complex.c, line 27, bb13"
+
+"file complex.c, line 27, bb13":                  ; preds = %"file complex.c, line 27, bb1"
+  store i32 0, i32* %changed, align 4
+  %r2 = getelementptr float* bitcast ([20 x i64]* @compl to float*), i64 32 ; <float*> [#uses=1]
+  %r3 = bitcast float* %r2 to <2 x float>*        ; <<2 x float>*> [#uses=1]
+  %r4 = load <2 x float>* %r3, align 4            ; <<2 x float>> [#uses=1]
+  call void @killcommon(i32* %changed)
+  br label %"file complex.c, line 34, bb4"
+
+"file complex.c, line 34, bb4":                   ; preds = %"file complex.c, line 27, bb13"
+  %r5 = load i32* %changed, align 4               ; <i32> [#uses=1]
+  %r6 = icmp eq i32 %r5, 0                        ; <i1> [#uses=1]
+  %r7 = zext i1 %r6 to i32                        ; <i32> [#uses=1]
+  %r8 = icmp ne i32 %r7, 0                        ; <i1> [#uses=1]
+  br i1 %r8, label %"file complex.c, line 34, bb7", label %"file complex.c, line 27, bb5"
+
+"file complex.c, line 27, bb5":                   ; preds = %"file complex.c, line 34, bb4"
+  br label %"file complex.c, line 35, bb6"
+
+"file complex.c, line 35, bb6":                   ; preds = %"file complex.c, line 27, bb5"
+  %r11 = ptrtoint <2 x float>* %garbage1 to i64   ; <i64> [#uses=1]
+  %r12 = inttoptr i64 %r11 to <2 x float>*        ; <<2 x float>*> [#uses=1]
+  store <2 x float> %r4, <2 x float>* %r12, align 4
+  br label %"file complex.c, line 34, bb7"
+
+"file complex.c, line 34, bb7":                   ; preds = %"file complex.c, line 35, bb6", %"file complex.c, line 34, bb4"
+  ret void
+}