The PMOVZXWD family of functions had patterns extends narrow vector types to wide vector types.

It had patterns for zext-loading and extending. This commit adds patterns for loading a wide type, performing a bitcast,
and extending. This is an odd pattern, but it is commonly used when writing code with intrinsics.

rdar://11897677

llvm-svn: 163995

llvm/lib/Target/X86/X86InstrSSE.td

@@ -5576,31 +5576,43 @@ let Predicates = [HasAVX] in {
             (VPMOVSXBWrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovsxbw (vzload_v2i64 addr:$src)),
             (VPMOVSXBWrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovsxbw (bc_v16i8 (loadv2i64 addr:$src))),
+            (VPMOVSXBWrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovsxwd (vzmovl_v2i64 addr:$src)),
             (VPMOVSXWDrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovsxwd (vzload_v2i64 addr:$src)),
             (VPMOVSXWDrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovsxwd (bc_v8i16 (loadv2i64 addr:$src))),
+            (VPMOVSXWDrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovsxdq (vzmovl_v2i64 addr:$src)),
             (VPMOVSXDQrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovsxdq (vzload_v2i64 addr:$src)),
             (VPMOVSXDQrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovsxdq (bc_v4i32 (loadv2i64 addr:$src))),
+            (VPMOVSXDQrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovzxbw (vzmovl_v2i64 addr:$src)),
             (VPMOVZXBWrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovzxbw (vzload_v2i64 addr:$src)),
             (VPMOVZXBWrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovzxbw (bc_v16i8 (loadv2i64 addr:$src))),
+            (VPMOVZXBWrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovzxwd (vzmovl_v2i64 addr:$src)),
             (VPMOVZXWDrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovzxwd (vzload_v2i64 addr:$src)),
             (VPMOVZXWDrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovzxwd (bc_v8i16 (loadv2i64 addr:$src))),
+            (VPMOVZXWDrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovzxdq (vzmovl_v2i64 addr:$src)),
             (VPMOVZXDQrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovzxdq (vzload_v2i64 addr:$src)),
             (VPMOVZXDQrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovzxdq (bc_v4i32 (loadv2i64 addr:$src))),
+            (VPMOVZXDQrm addr:$src)>;
 }
 
 let Predicates = [UseSSE41] in {
@@ -5609,31 +5621,43 @@ let Predicates = [UseSSE41] in {
             (PMOVSXBWrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovsxbw (vzload_v2i64 addr:$src)),
             (PMOVSXBWrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovsxbw (bc_v16i8 (loadv2i64 addr:$src))),
+            (PMOVSXBWrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovsxwd (vzmovl_v2i64 addr:$src)),
             (PMOVSXWDrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovsxwd (vzload_v2i64 addr:$src)),
             (PMOVSXWDrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovsxwd (bc_v8i16 (loadv2i64 addr:$src))),
+            (PMOVSXWDrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovsxdq (vzmovl_v2i64 addr:$src)),
             (PMOVSXDQrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovsxdq (vzload_v2i64 addr:$src)),
             (PMOVSXDQrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovsxdq (bc_v4i32 (loadv2i64 addr:$src))),
+            (PMOVSXDQrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovzxbw (vzmovl_v2i64 addr:$src)),
             (PMOVZXBWrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovzxbw (vzload_v2i64 addr:$src)),
             (PMOVZXBWrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovzxbw (bc_v16i8 (loadv2i64 addr:$src))),
+            (PMOVZXBWrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovzxwd (vzmovl_v2i64 addr:$src)),
             (PMOVZXWDrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovzxwd (vzload_v2i64 addr:$src)),
             (PMOVZXWDrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovzxwd (bc_v8i16 (loadv2i64 addr:$src))),
+            (PMOVZXWDrm addr:$src)>;
 
   def : Pat<(int_x86_sse41_pmovzxdq (vzmovl_v2i64 addr:$src)),
             (PMOVZXDQrm addr:$src)>;
   def : Pat<(int_x86_sse41_pmovzxdq (vzload_v2i64 addr:$src)),
             (PMOVZXDQrm addr:$src)>;
+  def : Pat<(int_x86_sse41_pmovzxdq (bc_v4i32 (loadv2i64 addr:$src))),
+            (PMOVZXDQrm addr:$src)>;
 }
 
 let Predicates = [HasAVX2] in {

llvm/test/CodeGen/X86/pmovext.ll

+; RUN: llc < %s -march=x86-64 -mcpu=corei7 | FileCheck %s
+
+; rdar://11897677
+
+;CHECK: intrin_pmov
+;CHECK: pmovzxbw  (%rsi), %xmm0
+;CHECK-NEXT: movdqu
+;CHECK-NEXT: ret
+define void @intrin_pmov(i16* noalias %dest, i8* noalias %src) nounwind uwtable ssp {
+  %1 = bitcast i8* %src to <2 x i64>*
+  %2 = load <2 x i64>* %1, align 16
+  %3 = bitcast <2 x i64> %2 to <16 x i8>
+  %4 = tail call <8 x i16> @llvm.x86.sse41.pmovzxbw(<16 x i8> %3) nounwind
+  %5 = bitcast i16* %dest to i8*
+  %6 = bitcast <8 x i16> %4 to <16 x i8>
+  tail call void @llvm.x86.sse2.storeu.dq(i8* %5, <16 x i8> %6) nounwind
+  ret void
+}
+
+declare <8 x i16> @llvm.x86.sse41.pmovzxbw(<16 x i8>) nounwind readnone
+
+declare void @llvm.x86.sse2.storeu.dq(i8*, <16 x i8>) nounwind