Introduce matching patterns for vbroadcast AVX instruction. The idea is to

match splats in the form (splat (scalar_to_vector (load ...))) whenever
the load can be folded. All the logic and instruction emission is
working but because of PR8156, there are no ways to match loads, cause
they can never be folded for splats. Thus, the tests are XFAILed, but
I've tested and exercised all the logic using a relaxed version for
checking the foldable loads, as if the bug was already fixed. This
should work out of the box once PR8156 gets fixed since MayFoldLoad will
work as expected.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@137810 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 84 insertions, 0 deletions

diff --git a/test/CodeGen/X86/avx-vbroadcast.ll b/test/CodeGen/X86/avx-vbroadcast.ll
new file mode 100644
index 0000000..ffa9710
--- /dev/null
+++ b/test/CodeGen/X86/avx-vbroadcast.ll
@@ -0,0 +1,84 @@
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7-avx -mattr=+avx | FileCheck %s
+; XFAIL: *
+
+; xfail this file for now because of PR8156, when it gets solved merge this with avx-splat.ll
+
+; CHECK: vbroadcastsd (%
+define <4 x i64> @A(i64* %ptr) nounwind uwtable readnone ssp {
+entry:
+  %q = load i64* %ptr, align 8
+  %vecinit.i = insertelement <4 x i64> undef, i64 %q, i32 0
+  %vecinit2.i = insertelement <4 x i64> %vecinit.i, i64 %q, i32 1
+  %vecinit4.i = insertelement <4 x i64> %vecinit2.i, i64 %q, i32 2
+  %vecinit6.i = insertelement <4 x i64> %vecinit4.i, i64 %q, i32 3
+  ret <4 x i64> %vecinit6.i
+}
+
+; CHECK: vbroadcastss (%
+define <8 x i32> @B(i32* %ptr) nounwind uwtable readnone ssp {
+entry:
+  %q = load i32* %ptr, align 4
+  %vecinit.i = insertelement <8 x i32> undef, i32 %q, i32 0
+  %vecinit2.i = insertelement <8 x i32> %vecinit.i, i32 %q, i32 1
+  %vecinit4.i = insertelement <8 x i32> %vecinit2.i, i32 %q, i32 2
+  %vecinit6.i = insertelement <8 x i32> %vecinit4.i, i32 %q, i32 3
+  ret <8 x i32> %vecinit6.i
+}
+
+; CHECK: vbroadcastsd (%
+define <4 x double> @C(double* %ptr) nounwind uwtable readnone ssp {
+entry:
+  %q = load double* %ptr, align 8
+  %vecinit.i = insertelement <4 x double> undef, double %q, i32 0
+  %vecinit2.i = insertelement <4 x double> %vecinit.i, double %q, i32 1
+  %vecinit4.i = insertelement <4 x double> %vecinit2.i, double %q, i32 2
+  %vecinit6.i = insertelement <4 x double> %vecinit4.i, double %q, i32 3
+  ret <4 x double> %vecinit6.i
+}
+
+; CHECK: vbroadcastss (%
+define <8 x float> @D(float* %ptr) nounwind uwtable readnone ssp {
+entry:
+  %q = load float* %ptr, align 4
+  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
+  %vecinit2.i = insertelement <8 x float> %vecinit.i, float %q, i32 1
+  %vecinit4.i = insertelement <8 x float> %vecinit2.i, float %q, i32 2
+  %vecinit6.i = insertelement <8 x float> %vecinit4.i, float %q, i32 3
+  ret <8 x float> %vecinit6.i
+}
+
+;;;; 128-bit versions
+
+; CHECK: vbroadcastss (%
+define <4 x float> @E(float* %ptr) nounwind uwtable readnone ssp {
+entry:
+  %q = load float* %ptr, align 4
+  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
+  %vecinit2.i = insertelement <4 x float> %vecinit.i, float %q, i32 1
+  %vecinit4.i = insertelement <4 x float> %vecinit2.i, float %q, i32 2
+  %vecinit6.i = insertelement <4 x float> %vecinit4.i, float %q, i32 3
+  ret <4 x float> %vecinit6.i
+}
+
+; CHECK: vbroadcastss (%
+define <4 x i32> @F(i32* %ptr) nounwind uwtable readnone ssp {
+entry:
+  %q = load i32* %ptr, align 4
+  %vecinit.i = insertelement <4 x i32> undef, i32 %q, i32 0
+  %vecinit2.i = insertelement <4 x i32> %vecinit.i, i32 %q, i32 1
+  %vecinit4.i = insertelement <4 x i32> %vecinit2.i, i32 %q, i32 2
+  %vecinit6.i = insertelement <4 x i32> %vecinit4.i, i32 %q, i32 3
+  ret <4 x i32> %vecinit6.i
+}
+
+; Unsupported vbroadcasts
+
+; CHECK: _G
+; CHECK-NOT: vbroadcastsd (%
+define <2 x i64> @G(i64* %ptr) nounwind uwtable readnone ssp {
+entry:
+  %q = load i64* %ptr, align 8
+  %vecinit.i = insertelement <2 x i64> undef, i64 %q, i32 0
+  %vecinit2.i = insertelement <2 x i64> %vecinit.i, i64 %q, i32 1
+  ret <2 x i64> %vecinit2.i
+}