Add AVX version of CLMUL instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@109248 91177308-0d34-0410-b5e6-96231b3b80d8

5 files changed, 142 insertions, 0 deletions

diff --git a/lib/Target/X86/AsmParser/X86AsmParser.cpp b/lib/Target/X86/AsmParser/X86AsmParser.cpp
index 4e05961..3095bdb 100644
--- a/lib/Target/X86/AsmParser/X86AsmParser.cpp
+++ b/lib/Target/X86/AsmParser/X86AsmParser.cpp
@@ -750,6 +750,23 @@ ParseInstruction(StringRef Name, SMLoc NameLoc,
       }
     }
   }
+
+  // FIXME: Hack to recognize vpclmul<src1_quadword, src2_quadword>dq
+  if (PatchedName.startswith("vpclmul")) {
+    unsigned CLMULQuadWordSelect = StringSwitch<unsigned>(
+      PatchedName.slice(7, PatchedName.size() - 2))
+      .Case("lqlq", 0x00) // src1[63:0],   src2[63:0]
+      .Case("hqlq", 0x01) // src1[127:64], src2[63:0]
+      .Case("lqhq", 0x10) // src1[63:0],   src2[127:64]
+      .Case("hqhq", 0x11) // src1[127:64], src2[127:64]
+      .Default(~0U);
+    if (CLMULQuadWordSelect != ~0U) {
+      ExtraImmOp = MCConstantExpr::Create(CLMULQuadWordSelect,
+                                          getParser().getContext());
+      assert(PatchedName.endswith("dq") && "Unexpected mnemonic!");
+      PatchedName = "vpclmulqdq";
+    }
+  }
   Operands.push_back(X86Operand::CreateToken(PatchedName, NameLoc));
 
   if (ExtraImmOp)
diff --git a/lib/Target/X86/X86InstrFormats.td b/lib/Target/X86/X86InstrFormats.td
index 380426f..cc4dd5b 100644
--- a/lib/Target/X86/X86InstrFormats.td
+++ b/lib/Target/X86/X86InstrFormats.td
@@ -439,6 +439,12 @@ class AESAI<bits<8> o, Format F, dag outs, dag ins, string asm,
       : Ii8<o, F, outs, ins, asm, pattern, SSEPackedInt>, TA,
         Requires<[HasAES]>;
 
+// CLMUL Instruction Templates
+class CLMULIi8<bits<8> o, Format F, dag outs, dag ins, string asm,
+               list<dag>pattern>
+      : Ii8<o, F, outs, ins, asm, pattern, SSEPackedInt>, TA,
+        OpSize, VEX_4V, Requires<[HasAVX, HasCLMUL]>;
+
 // FMA3 Instruction Templates
 class FMA3<bits<8> o, Format F, dag outs, dag ins, string asm,
            list<dag>pattern>
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index 0e69105..741e841 100644
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -5118,6 +5118,41 @@ def AESKEYGENASSIST128rm : AESAI<0xDF, MRMSrcMem, (outs VR128:$dst),
   OpSize;
 
 //===----------------------------------------------------------------------===//
+// CLMUL Instructions
+//===----------------------------------------------------------------------===//
+
+// Only the AVX version of CLMUL instructions are described here.
+
+// Carry-less Multiplication instructions
+let isAsmParserOnly = 1 in {
+def VPCLMULQDQrr : CLMULIi8<0x44, MRMSrcReg, (outs VR128:$dst),
+           (ins VR128:$src1, VR128:$src2, i8imm:$src3),
+           "vpclmulqdq\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
+           []>;
+
+def VPCLMULQDQrm : CLMULIi8<0x44, MRMSrcMem, (outs VR128:$dst),
+           (ins VR128:$src1, i128mem:$src2, i8imm:$src3),
+           "vpclmulqdq\t{$src3, $src2, $src1, $dst|$dst, $src1, $src2, $src3}",
+           []>;
+
+// Assembler Only
+multiclass avx_vpclmul<string asm> {
+  def rr : I<0, Pseudo, (outs VR128:$dst), (ins VR128:$src1, VR128:$src2),
+             !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
+             []>;
+
+  def rm : I<0, Pseudo, (outs VR128:$dst), (ins VR128:$src1, i128mem:$src2),
+             !strconcat(asm, "\t{$src2, $src1, $dst|$dst, $src1, $src2}"),
+             []>;
+}
+defm VPCLMULHQHQDQ : avx_vpclmul<"vpclmulhqhqdq">;
+defm VPCLMULHQLQDQ : avx_vpclmul<"vpclmulhqlqdq">;
+defm VPCLMULLQHQDQ : avx_vpclmul<"vpclmullqhqdq">;
+defm VPCLMULLQLQDQ : avx_vpclmul<"vpclmullqlqdq">;
+
+} // isAsmParserOnly
+
+//===----------------------------------------------------------------------===//
 // AVX Instructions
 //===----------------------------------------------------------------------===//
 
diff --git a/test/MC/AsmParser/X86/x86_32-avx-clmul-encoding.s b/test/MC/AsmParser/X86/x86_32-avx-clmul-encoding.s
new file mode 100644
index 0000000..47bf980
--- /dev/null
+++ b/test/MC/AsmParser/X86/x86_32-avx-clmul-encoding.s
@@ -0,0 +1,42 @@
+// RUN: llvm-mc -triple i386-unknown-unknown --show-encoding %s | FileCheck %s
+
+// CHECK: vpclmulqdq  $17, %xmm2, %xmm5, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0xca,0x11]
+          vpclmulhqhqdq %xmm2, %xmm5, %xmm1
+
+// CHECK: vpclmulqdq  $17, (%eax), %xmm5, %xmm3
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0x18,0x11]
+          vpclmulhqhqdq (%eax), %xmm5, %xmm3
+
+// CHECK: vpclmulqdq  $1, %xmm2, %xmm5, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0xca,0x01]
+          vpclmulhqlqdq %xmm2, %xmm5, %xmm1
+
+// CHECK: vpclmulqdq  $1, (%eax), %xmm5, %xmm3
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0x18,0x01]
+          vpclmulhqlqdq (%eax), %xmm5, %xmm3
+
+// CHECK: vpclmulqdq  $16, %xmm2, %xmm5, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0xca,0x10]
+          vpclmullqhqdq %xmm2, %xmm5, %xmm1
+
+// CHECK: vpclmulqdq  $16, (%eax), %xmm5, %xmm3
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0x18,0x10]
+          vpclmullqhqdq (%eax), %xmm5, %xmm3
+
+// CHECK: vpclmulqdq  $0, %xmm2, %xmm5, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0xca,0x00]
+          vpclmullqlqdq %xmm2, %xmm5, %xmm1
+
+// CHECK: vpclmulqdq  $0, (%eax), %xmm5, %xmm3
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0x18,0x00]
+          vpclmullqlqdq (%eax), %xmm5, %xmm3
+
+// CHECK: vpclmulqdq  $17, %xmm2, %xmm5, %xmm1
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0xca,0x11]
+          vpclmulqdq  $17, %xmm2, %xmm5, %xmm1
+
+// CHECK: vpclmulqdq  $17, (%eax), %xmm5, %xmm3
+// CHECK: encoding: [0xc4,0xe3,0x51,0x44,0x18,0x11]
+          vpclmulqdq  $17, (%eax), %xmm5, %xmm3
+
diff --git a/test/MC/AsmParser/X86/x86_64-avx-clmul-encoding.s b/test/MC/AsmParser/X86/x86_64-avx-clmul-encoding.s
new file mode 100644
index 0000000..67e82c6
--- /dev/null
+++ b/test/MC/AsmParser/X86/x86_64-avx-clmul-encoding.s
@@ -0,0 +1,42 @@
+// RUN: llvm-mc -triple x86_64-unknown-unknown --show-encoding %s | FileCheck %s
+
+// CHECK: vpclmulqdq  $17, %xmm12, %xmm10, %xmm11
+// CHECK: encoding: [0xc4,0x43,0x29,0x44,0xdc,0x11]
+          vpclmulhqhqdq %xmm12, %xmm10, %xmm11
+
+// CHECK: vpclmulqdq  $17, (%rax), %xmm10, %xmm13
+// CHECK: encoding: [0xc4,0x63,0x29,0x44,0x28,0x11]
+          vpclmulhqhqdq (%rax), %xmm10, %xmm13
+
+// CHECK: vpclmulqdq  $1, %xmm12, %xmm10, %xmm11
+// CHECK: encoding: [0xc4,0x43,0x29,0x44,0xdc,0x01]
+          vpclmulhqlqdq %xmm12, %xmm10, %xmm11
+
+// CHECK: vpclmulqdq  $1, (%rax), %xmm10, %xmm13
+// CHECK: encoding: [0xc4,0x63,0x29,0x44,0x28,0x01]
+          vpclmulhqlqdq (%rax), %xmm10, %xmm13
+
+// CHECK: vpclmulqdq  $16, %xmm12, %xmm10, %xmm11
+// CHECK: encoding: [0xc4,0x43,0x29,0x44,0xdc,0x10]
+          vpclmullqhqdq %xmm12, %xmm10, %xmm11
+
+// CHECK: vpclmulqdq  $16, (%rax), %xmm10, %xmm13
+// CHECK: encoding: [0xc4,0x63,0x29,0x44,0x28,0x10]
+          vpclmullqhqdq (%rax), %xmm10, %xmm13
+
+// CHECK: vpclmulqdq  $0, %xmm12, %xmm10, %xmm11
+// CHECK: encoding: [0xc4,0x43,0x29,0x44,0xdc,0x00]
+          vpclmullqlqdq %xmm12, %xmm10, %xmm11
+
+// CHECK: vpclmulqdq  $0, (%rax), %xmm10, %xmm13
+// CHECK: encoding: [0xc4,0x63,0x29,0x44,0x28,0x00]
+          vpclmullqlqdq (%rax), %xmm10, %xmm13
+
+// CHECK: vpclmulqdq  $17, %xmm12, %xmm10, %xmm11
+// CHECK: encoding: [0xc4,0x43,0x29,0x44,0xdc,0x11]
+          vpclmulqdq  $17, %xmm12, %xmm10, %xmm11
+
+// CHECK: vpclmulqdq  $17, (%rax), %xmm10, %xmm13
+// CHECK: encoding: [0xc4,0x63,0x29,0x44,0x28,0x11]
+          vpclmulqdq  $17, (%rax), %xmm10, %xmm13
+