[PPC64] Add vector pack/unpack support from ISA 2.07

This patch adds support for the following new instructions in the
Power ISA 2.07:

  vpksdss
  vpksdus
  vpkudus
  vpkudum
  vupkhsw
  vupklsw

These instructions are available through the vec_packs, vec_packsu,
vec_unpackh, and vec_unpackl built-in interfaces.  These are
lane-sensitive instructions, so the built-ins have different
implementations for big- and little-endian, and the instructions must
be marked as killing the vector swap optimization for now.

The first three instructions perform saturating pack operations.  The
fourth performs a modulo pack operation, which means it can be
represented with a vector shuffle, and conversely the appropriate
vector shuffles may cause this instruction to be generated.  The other
instructions are only generated via built-in support for now.

Appropriate tests have been added.

There is a companion patch to clang for the rest of this support.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@237499 91177308-0d34-0410-b5e6-96231b3b80d8

4 files changed, 131 insertions, 0 deletions

diff --git a/test/CodeGen/PowerPC/vec_shuffle_p8vector.ll b/test/CodeGen/PowerPC/vec_shuffle_p8vector.ll
new file mode 100644
index 00000000000..d81aa729f8e
--- /dev/null
+++ b/test/CodeGen/PowerPC/vec_shuffle_p8vector.ll
@@ -0,0 +1,43 @@
+; RUN: llc -mcpu=pwr8 -mtriple=powerpc64-unknown-linux-gnu -mattr=+power8-vector < %s | FileCheck %s
+
+define void @VPKUDUM_unary(<2 x i64>* %A) {
+entry:
+        %tmp = load <2 x i64>, <2 x i64>* %A
+        %tmp2 = bitcast <2 x i64> %tmp to <4 x i32>
+        %tmp3 = extractelement <4 x i32> %tmp2, i32 1
+        %tmp4 = extractelement <4 x i32> %tmp2, i32 3
+        %tmp5 = insertelement <4 x i32> undef, i32 %tmp3, i32 0
+        %tmp6 = insertelement <4 x i32> %tmp5, i32 %tmp4, i32 1
+        %tmp7 = insertelement <4 x i32> %tmp6, i32 %tmp3, i32 2
+        %tmp8 = insertelement <4 x i32> %tmp7, i32 %tmp4, i32 3
+        %tmp9 = bitcast <4 x i32> %tmp8 to <2 x i64>
+        store <2 x i64> %tmp9, <2 x i64>* %A
+        ret void
+}
+
+; CHECK-LABEL: @VPKUDUM_unary
+; CHECK-NOT:   vperm
+; CHECK:       vpkudum
+
+define void @VPKUDUM(<2 x i64>* %A, <2 x i64>* %B) {
+entry:
+        %tmp = load <2 x i64>, <2 x i64>* %A
+        %tmp2 = bitcast <2 x i64> %tmp to <4 x i32>
+        %tmp3 = load <2 x i64>, <2 x i64>* %B
+        %tmp4 = bitcast <2 x i64> %tmp3 to <4 x i32>
+        %tmp5 = extractelement <4 x i32> %tmp2, i32 1
+        %tmp6 = extractelement <4 x i32> %tmp2, i32 3
+        %tmp7 = extractelement <4 x i32> %tmp4, i32 1
+        %tmp8 = extractelement <4 x i32> %tmp4, i32 3
+        %tmp9 = insertelement <4 x i32> undef, i32 %tmp5, i32 0
+        %tmp10 = insertelement <4 x i32> %tmp9, i32 %tmp6, i32 1
+        %tmp11 = insertelement <4 x i32> %tmp10, i32 %tmp7, i32 2
+        %tmp12 = insertelement <4 x i32> %tmp11, i32 %tmp8, i32 3
+        %tmp13 = bitcast <4 x i32> %tmp12 to <2 x i64>
+        store <2 x i64> %tmp13, <2 x i64>* %A
+        ret void
+}
+
+; CHECK-LABEL: @VPKUDUM
+; CHECK-NOT:   vperm
+; CHECK:       vpkudum
diff --git a/test/CodeGen/PowerPC/vec_shuffle_p8vector_le.ll b/test/CodeGen/PowerPC/vec_shuffle_p8vector_le.ll
new file mode 100644
index 00000000000..709388675f6
--- /dev/null
+++ b/test/CodeGen/PowerPC/vec_shuffle_p8vector_le.ll
@@ -0,0 +1,43 @@
+; RUN: llc -mcpu=pwr8 -mtriple=powerpc64le-unknown-linux-gnu -mattr=+power8-vector < %s | FileCheck %s
+
+define void @VPKUDUM_unary(<2 x i64>* %A) {
+entry:
+	%tmp = load <2 x i64>, <2 x i64>* %A
+	%tmp2 = bitcast <2 x i64> %tmp to <4 x i32>
+	%tmp3 = extractelement <4 x i32> %tmp2, i32 0
+	%tmp4 = extractelement <4 x i32> %tmp2, i32 2
+	%tmp5 = insertelement <4 x i32> undef, i32 %tmp3, i32 0
+	%tmp6 = insertelement <4 x i32> %tmp5, i32 %tmp4, i32 1
+	%tmp7 = insertelement <4 x i32> %tmp6, i32 %tmp3, i32 2
+	%tmp8 = insertelement <4 x i32> %tmp7, i32 %tmp4, i32 3
+	%tmp9 = bitcast <4 x i32> %tmp8 to <2 x i64>
+	store <2 x i64> %tmp9, <2 x i64>* %A
+	ret void
+}
+
+; CHECK-LABEL: @VPKUDUM_unary
+; CHECK-NOT:   vperm
+; CHECK:       vpkudum
+
+define void @VPKUDUM(<2 x i64>* %A, <2 x i64>* %B) {
+entry:
+	%tmp = load <2 x i64>, <2 x i64>* %A
+	%tmp2 = bitcast <2 x i64> %tmp to <4 x i32>
+        %tmp3 = load <2 x i64>, <2 x i64>* %B
+        %tmp4 = bitcast <2 x i64> %tmp3 to <4 x i32>
+	%tmp5 = extractelement <4 x i32> %tmp2, i32 0
+	%tmp6 = extractelement <4 x i32> %tmp2, i32 2
+	%tmp7 = extractelement <4 x i32> %tmp4, i32 0
+	%tmp8 = extractelement <4 x i32> %tmp4, i32 2
+	%tmp9 = insertelement <4 x i32> undef, i32 %tmp5, i32 0
+	%tmp10 = insertelement <4 x i32> %tmp9, i32 %tmp6, i32 1
+	%tmp11 = insertelement <4 x i32> %tmp10, i32 %tmp7, i32 2
+	%tmp12 = insertelement <4 x i32> %tmp11, i32 %tmp8, i32 3
+	%tmp13 = bitcast <4 x i32> %tmp12 to <2 x i64>
+	store <2 x i64> %tmp13, <2 x i64>* %A
+	ret void
+}
+
+; CHECK-LABEL: @VPKUDUM
+; CHECK-NOT:   vperm
+; CHECK:       vpkudum
diff --git a/test/MC/Disassembler/PowerPC/ppc64-encoding-p8vector.txt b/test/MC/Disassembler/PowerPC/ppc64-encoding-p8vector.txt
new file mode 100644
index 00000000000..eeea2db673f
--- /dev/null
+++ b/test/MC/Disassembler/PowerPC/ppc64-encoding-p8vector.txt
@@ -0,0 +1,19 @@
+# RUN: llvm-mc --disassemble %s -triple powerpc64-unknown-unknown -mcpu=pwr8 | FileCheck %s
+
+# CHECK: vpksdss 2, 3, 4
+0x10 0x43 0x25 0xce
+
+# CHECK: vpksdus 2, 3, 4
+0x10 0x43 0x25 0x4e
+
+# CHECK: vpkudus 2, 3, 4
+0x10 0x43 0x24 0xce
+
+# CHECK: vpkudum 2, 3, 4
+0x10 0x43 0x24 0x4e
+
+# CHECK: vupkhsw 2, 3
+0x10 0x40 0x1e 0x4e
+
+# CHECK: vupklsw 2, 3
+0x10 0x40 0x1e 0xce
diff --git a/test/MC/PowerPC/ppc64-encoding-p8vector.s b/test/MC/PowerPC/ppc64-encoding-p8vector.s
new file mode 100644
index 00000000000..657b7378fb4
--- /dev/null
+++ b/test/MC/PowerPC/ppc64-encoding-p8vector.s
@@ -0,0 +1,26 @@
+# RUN: llvm-mc -triple powerpc64-unknown-unknown --show-encoding %s | FileCheck -check-prefix=CHECK-BE %s
+# RUN: llvm-mc -triple powerpc64le-unknown-unknown --show-encoding %s | FileCheck -check-prefix=CHECK-LE %s
+
+# CHECK-BE: vpksdss 2, 3, 4                 # encoding: [0x10,0x43,0x25,0xce]
+# CHECK-LE: vpksdss 2, 3, 4                 # encoding: [0xce,0x25,0x43,0x10]
+            vpksdss 2, 3, 4
+
+# CHECK-BE: vpksdus 2, 3, 4                 # encoding: [0x10,0x43,0x25,0x4e]
+# CHECK-LE: vpksdus 2, 3, 4                 # encoding: [0x4e,0x25,0x43,0x10]
+            vpksdus 2, 3, 4
+
+# CHECK-BE: vpkudus 2, 3, 4                 # encoding: [0x10,0x43,0x24,0xce]
+# CHECK-LE: vpkudus 2, 3, 4                 # encoding: [0xce,0x24,0x43,0x10]
+            vpkudus 2, 3, 4
+
+# CHECK-BE: vpkudum 2, 3, 4                 # encoding: [0x10,0x43,0x24,0x4e]
+# CHECK-LE: vpkudum 2, 3, 4                 # encoding: [0x4e,0x24,0x43,0x10]
+            vpkudum 2, 3, 4
+
+# CHECK-BE: vupkhsw 2, 3                    # encoding: [0x10,0x40,0x1e,0x4e]
+# CHECK-LE: vupkhsw 2, 3                    # encoding: [0x4e,0x1e,0x40,0x10]
+            vupkhsw 2, 3
+
+# CHECK-BE: vupklsw 2, 3                    # encoding: [0x10,0x40,0x1e,0xce]
+# CHECK-LE: vupklsw 2, 3                    # encoding: [0xce,0x1e,0x40,0x10]
+            vupklsw 2, 3