Add patterns for integer forms of SHUFPD/VSHUFPD with a memory load.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@147393 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index 911150347f0aa69584dc0edb1f42a36636f47521..8b094686f5fbed8d05304c6e2e2739d2a8948c13 100644 (file)
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -2348,6 +2348,9 @@ let Predicates = [HasSSE2] in {
             (SHUFPDrri VR128:$src1, VR128:$src2,
                        (SHUFFLE_get_shuf_imm VR128:$src3))>;
   // Generic SHUFPD patterns
+  def : Pat<(v2i64 (X86Shufpd VR128:$src1,
+                       (memopv2i64 addr:$src2), (i8 imm:$imm))),
+            (SHUFPDrmi VR128:$src1, addr:$src2, imm:$imm)>;
   def : Pat<(v2f64 (X86Shufpd VR128:$src1,
                        (memopv2f64 addr:$src2), (i8 imm:$imm))),
             (SHUFPDrmi VR128:$src1, addr:$src2, imm:$imm)>;
@@ -2397,6 +2400,9 @@ let Predicates = [HasAVX] in {
             (VSHUFPDrri VR128:$src1, VR128:$src2,
                         (SHUFFLE_get_shuf_imm VR128:$src3))>;
 
+  def : Pat<(v2i64 (X86Shufpd VR128:$src1,
+                       (memopv2i64 addr:$src2), (i8 imm:$imm))),
+            (VSHUFPDrmi VR128:$src1, addr:$src2, imm:$imm)>;
   def : Pat<(v2f64 (X86Shufpd VR128:$src1,
                        (memopv2f64 addr:$src2), (i8 imm:$imm))),
             (VSHUFPDrmi VR128:$src1, addr:$src2, imm:$imm)>;

diff --git a/test/CodeGen/X86/avx-vshufp.ll b/test/CodeGen/X86/avx-vshufp.ll
index a60d8c25dc5a91455081c0734259c4f1edce762e..45883b717380d6222259d7edf49d0537341893b2 100644 (file)
--- a/test/CodeGen/X86/avx-vshufp.ll
+++ b/test/CodeGen/X86/avx-vshufp.ll
@@ -16,6 +16,22 @@ entry:
   ret <8 x float> %shuffle
 }
 
+; CHECK: vshufps  $-53, %ymm
+define <8 x i32> @A3(<8 x i32> %a, <8 x i32> %b) nounwind uwtable readnone ssp {
+entry:
+  %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 3, i32 2, i32 8, i32 11, i32 7, i32 6, i32 12, i32 15>
+  ret <8 x i32> %shuffle
+}
+
+; CHECK: vshufps  $-53, (%{{.*}}), %ymm
+define <8 x i32> @A4(<8 x i32>* %a, <8 x i32>* %b) nounwind uwtable readnone ssp {
+entry:
+  %a2 = load <8 x i32>* %a
+  %b2 = load <8 x i32>* %b
+  %shuffle = shufflevector <8 x i32> %a2, <8 x i32> %b2, <8 x i32> <i32 3, i32 2, i32 8, i32 11, i32 7, i32 6, i32 12, i32 15>
+  ret <8 x i32> %shuffle
+}
+
 ; CHECK: vshufpd  $10, %ymm
 define <4 x double> @B(<4 x double> %a, <4 x double> %b) nounwind uwtable readnone ssp {
 entry:
@@ -32,6 +48,22 @@ entry:
   ret <4 x double> %shuffle
 }
 
+; CHECK: vshufpd  $10, %ymm
+define <4 x i64> @B3(<4 x i64> %a, <4 x i64> %b) nounwind uwtable readnone ssp {
+entry:
+  %shuffle = shufflevector <4 x i64> %a, <4 x i64> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
+  ret <4 x i64> %shuffle
+}
+
+; CHECK: vshufpd  $10, (%{{.*}}), %ymm
+define <4 x i64> @B4(<4 x i64>* %a, <4 x i64>* %b) nounwind uwtable readnone ssp {
+entry:
+  %a2 = load <4 x i64>* %a
+  %b2 = load <4 x i64>* %b
+  %shuffle = shufflevector <4 x i64> %a2, <4 x i64> %b2, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
+  ret <4 x i64> %shuffle
+}
+
 ; CHECK: vshufps  $-53, %ymm
 define <8 x float> @C(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
 entry:
@@ -76,6 +108,22 @@ entry:
   ret <4 x float> %shuffle
 }
 
+; CHECK: vshufps  $-53, %xmm
+define <4 x i32> @A3128(<4 x i32> %a, <4 x i32> %b) nounwind uwtable readnone ssp {
+entry:
+  %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 3, i32 2, i32 4, i32 7>
+  ret <4 x i32> %shuffle
+}
+
+; CHECK: vshufps  $-53, (%{{.*}}), %xmm
+define <4 x i32> @A4128(<4 x i32>* %a, <4 x i32>* %b) nounwind uwtable readnone ssp {
+entry:
+  %a2 = load <4 x i32>* %a
+  %b2 = load <4 x i32>* %b
+  %shuffle = shufflevector <4 x i32> %a2, <4 x i32> %b2, <4 x i32> <i32 3, i32 2, i32 4, i32 7>
+  ret <4 x i32> %shuffle
+}
+
 ; CHECK: vshufpd  $1, %xmm
 define <2 x double> @B128(<2 x double> %a, <2 x double> %b) nounwind uwtable readnone ssp {
 entry:
@@ -91,3 +139,19 @@ entry:
   %shuffle = shufflevector <2 x double> %a2, <2 x double> %b2, <2 x i32> <i32 1, i32 2>
   ret <2 x double> %shuffle
 }
+
+; CHECK: vshufpd  $1, %xmm
+define <2 x i64> @B3128(<2 x i64> %a, <2 x i64> %b) nounwind uwtable readnone ssp {
+entry:
+  %shuffle = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 1, i32 2>
+  ret <2 x i64> %shuffle
+}
+
+; CHECK: vshufpd  $1, (%{{.*}}), %xmm
+define <2 x i64> @B4128(<2 x i64>* %a, <2 x i64>* %b) nounwind uwtable readnone ssp {
+entry:
+  %a2 = load <2 x i64>* %a
+  %b2 = load <2 x i64>* %b
+  %shuffle = shufflevector <2 x i64> %a2, <2 x i64> %b2, <2 x i32> <i32 1, i32 2>
+  ret <2 x i64> %shuffle
+}