cmd/compile, simd: capture VAES instructions and fix AVX512VAES feature

The code previously filters out VAES-only instructions, this CL added
them back.

This CL added the VAES feature check following the Intel xed data:

  XED_ISA_SET_VAES:              vaes.7.0.ecx.9 # avx.1.0.ecx.28

This CL also found out that the old AVX512VAES feature check is not
checking the correct bits, it also fixes it:

  XED_ISA_SET_AVX512_VAES_128:    vaes.7.0.ecx.9  aes.1.0.ecx.25  avx512f.7.0.ebx.16 avx512vl.7.0.ebx.31
  XED_ISA_SET_AVX512_VAES_256:    vaes.7.0.ecx.9  aes.1.0.ecx.25  avx512f.7.0.ebx.16 avx512vl.7.0.ebx.31
  XED_ISA_SET_AVX512_VAES_512:    vaes.7.0.ecx.9  aes.1.0.ecx.25  avx512f.7.0.ebx.16

It restricts to the most strict common set - includes avx512vl for even
512-bits although it doesn't requires it.

Change-Id: I4e2f72b312fd2411589fbc12f9ee5c63c09c2e9a
Reviewed-on: https://go-review.googlesource.com/c/go/+/738500
Reviewed-by: Cherry Mui <cherryyz@google.com>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>

3 files changed, 18 insertions, 5 deletions

diff --git a/src/simd/archsimd/_gen/simdgen/xed.go b/src/simd/archsimd/_gen/simdgen/xed.go
index 5d6fac64d0..2175fa4a8d 100644
--- a/src/simd/archsimd/_gen/simdgen/xed.go
+++ b/src/simd/archsimd/_gen/simdgen/xed.go
@@ -77,7 +77,8 @@ func loadXED(xedPath string) []*unify.Value {
 		switch {
 		case inst.RealOpcode == "N":
 			return // Skip unstable instructions
-		case !(strings.HasPrefix(inst.Extension, "AVX") || strings.HasPrefix(inst.Extension, "SHA") || inst.Extension == "FMA"):
+		case !(strings.HasPrefix(inst.Extension, "AVX") || strings.HasPrefix(inst.Extension, "SHA") ||
+			inst.Extension == "FMA" || inst.Extension == "VAES"):
 			// We're only interested in AVX and SHA instructions.
 			return
 		}
@@ -796,6 +797,7 @@ var cpuFeatureMap = map[string]string{
 	"AVXAES":   "AVXAES",
 	"SHA":      "SHA",
 	"FMA":      "FMA",
+	"VAES":     "VAES",
 
 	// AVX-512 foundational features. We combine all of these into one "AVX512" feature.
 	"AVX512F":  "AVX512",
@@ -829,6 +831,7 @@ func init() {
 
 		"AVXAES": {Virtual: true, Implies: []string{"AVX", "AES"}},
 		"FMA":    {Implies: []string{"AVX"}},
+		"VAES":   {Implies: []string{"AVX"}},
 
 		// AVX-512 subfeatures.
 		"AVX512BITALG":    {Implies: []string{"AVX512"}},
diff --git a/src/simd/archsimd/cpu.go b/src/simd/archsimd/cpu.go
index 8069ee7f26..a25556ae65 100644
--- a/src/simd/archsimd/cpu.go
+++ b/src/simd/archsimd/cpu.go
@@ -158,3 +158,13 @@ func (X86Features) FMA() bool {
 func (X86Features) SHA() bool {
 	return cpu.X86.HasSHA
 }
+
+// VAES returns whether the CPU supports the VAES feature.
+//
+// If it returns true, then the CPU also supports AVX.
+//
+// VAES is defined on all GOARCHes, but will only return true on
+// GOARCH amd64.
+func (X86Features) VAES() bool {
+	return cpu.X86.HasVAES
+}
diff --git a/src/simd/archsimd/ops_amd64.go b/src/simd/archsimd/ops_amd64.go
index ec50cc72c5..9c5bb22eee 100644
--- a/src/simd/archsimd/ops_amd64.go
+++ b/src/simd/archsimd/ops_amd64.go
@@ -19,7 +19,7 @@ func (x Uint8x16) AESDecryptLastRound(y Uint32x4) Uint8x16
 // y is the chunk of dw array in use.
 // result = AddRoundKey(InvShiftRows(InvSubBytes(x)), y)
 //
-// Asm: VAESDECLAST, CPU Feature: AVX512VAES
+// Asm: VAESDECLAST, CPU Feature: VAES
 func (x Uint8x32) AESDecryptLastRound(y Uint32x8) Uint8x32
 
 // AESDecryptLastRound performs a series of operations in AES cipher algorithm defined in FIPS 197.
@@ -45,7 +45,7 @@ func (x Uint8x16) AESDecryptOneRound(y Uint32x4) Uint8x16
 // y is the chunk of dw array in use.
 // result = AddRoundKey(InvMixColumns(InvShiftRows(InvSubBytes(x))), y)
 //
-// Asm: VAESDEC, CPU Feature: AVX512VAES
+// Asm: VAESDEC, CPU Feature: VAES
 func (x Uint8x32) AESDecryptOneRound(y Uint32x8) Uint8x32
 
 // AESDecryptOneRound performs a series of operations in AES cipher algorithm defined in FIPS 197.
@@ -71,7 +71,7 @@ func (x Uint8x16) AESEncryptLastRound(y Uint32x4) Uint8x16
 // y is the chunk of w array in use.
 // result = AddRoundKey((ShiftRows(SubBytes(x))), y)
 //
-// Asm: VAESENCLAST, CPU Feature: AVX512VAES
+// Asm: VAESENCLAST, CPU Feature: VAES
 func (x Uint8x32) AESEncryptLastRound(y Uint32x8) Uint8x32
 
 // AESEncryptLastRound performs a series of operations in AES cipher algorithm defined in FIPS 197.
@@ -97,7 +97,7 @@ func (x Uint8x16) AESEncryptOneRound(y Uint32x4) Uint8x16
 // y is the chunk of w array in use.
 // result = AddRoundKey(MixColumns(ShiftRows(SubBytes(x))), y)
 //
-// Asm: VAESENC, CPU Feature: AVX512VAES
+// Asm: VAESENC, CPU Feature: VAES
 func (x Uint8x32) AESEncryptOneRound(y Uint32x8) Uint8x32
 
 // AESEncryptOneRound performs a series of operations in AES cipher algorithm defined in FIPS 197.