runtime: use internal/cpu variables in assembler code

Using internal/cpu variables has the benefit of avoiding false sharing
(as those are padded) and allows memory and cache usage for these variables
to be shared by multiple packages.

Change-Id: I2bf68d03091bf52b466cf689230d5d25d5950037
Reviewed-on: https://go-review.googlesource.com/126599
Run-TryBot: Martin Möhrmann <moehrmann@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

src/runtime/asm_386.s

@@ -881,7 +881,7 @@ TEXT runtime·stackcheck(SB), NOSPLIT, $0-0
 
 // func cputicks() int64
 TEXT runtime·cputicks(SB),NOSPLIT,$0-8
-	CMPB	runtime·support_sse2(SB), $1
+	CMPB	internal∕cpu·X86+const_offset_x86_HasSSE2(SB), $1
 	JNE	done
 	CMPB	runtime·lfenceBeforeRdtsc(SB), $1
 	JNE	mfence

src/runtime/cpuflags.go

+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package runtime
+
+import (
+	"internal/cpu"
+	"unsafe"
+)
+
+// Offsets into internal/cpu records for use in assembly.
+const (
+	offset_x86_HasAVX2 = unsafe.Offsetof(cpu.X86.HasAVX2)
+	offset_x86_HasERMS = unsafe.Offsetof(cpu.X86.HasERMS)
+	offset_x86_HasSSE2 = unsafe.Offsetof(cpu.X86.HasSSE2)
+)

src/runtime/cpuflags_amd64.go

@@ -6,12 +6,6 @@ package runtime
 
 import (
 	"internal/cpu"
-	"unsafe"
-)
-
-// Offsets into internal/cpu records for use in assembly.
-const (
-	offsetX86HasAVX2 = unsafe.Offsetof(cpu.X86.HasAVX2)
 )
 
 var useAVXmemmove bool

src/runtime/memclr_386.s

@@ -4,6 +4,7 @@
 
 // +build !plan9
 
+#include "go_asm.h"
 #include "textflag.h"
 
 // NOTE: Windows externalthreadhandler expects memclr to preserve DX.
@@ -28,7 +29,7 @@ tail:
 	JBE	_5through8
 	CMPL	BX, $16
 	JBE	_9through16
-	CMPB	runtime·support_sse2(SB), $1
+	CMPB	internal∕cpu·X86+const_offset_x86_HasSSE2(SB), $1
 	JNE	nosse2
 	PXOR	X0, X0
 	CMPL	BX, $32

src/runtime/memclr_amd64.s

@@ -38,7 +38,7 @@ tail:
 	JBE	_65through128
 	CMPQ	BX, $256
 	JBE	_129through256
-	CMPB	internal∕cpu·X86+const_offsetX86HasAVX2(SB), $1
+	CMPB	internal∕cpu·X86+const_offset_x86_HasAVX2(SB), $1
 	JE loop_preheader_avx2
 	// TODO: for really big clears, use MOVNTDQ, even without AVX2.
 

src/runtime/memmove_386.s

@@ -25,6 +25,7 @@
 
 // +build !plan9
 
+#include "go_asm.h"
 #include "textflag.h"
 
 // func memmove(to, from unsafe.Pointer, n uintptr)
@@ -51,7 +52,7 @@ tail:
 	JBE	move_5through8
 	CMPL	BX, $16
 	JBE	move_9through16
-	CMPB	runtime·support_sse2(SB), $1
+	CMPB	internal∕cpu·X86+const_offset_x86_HasSSE2(SB), $1
 	JNE	nosse2
 	CMPL	BX, $32
 	JBE	move_17through32
@@ -72,7 +73,7 @@ nosse2:
  */
 forward:
 	// If REP MOVSB isn't fast, don't use it
-	CMPB	runtime·support_erms(SB), $1 // enhanced REP MOVSB/STOSB
+	CMPB	internal∕cpu·X86+const_offset_x86_HasERMS(SB), $1 // enhanced REP MOVSB/STOSB
 	JNE	fwdBy4
 
 	// Check alignment

src/runtime/memmove_amd64.s

@@ -25,6 +25,7 @@
 
 // +build !plan9
 
+#include "go_asm.h"
 #include "textflag.h"
 
 // func memmove(to, from unsafe.Pointer, n uintptr)
@@ -83,7 +84,7 @@ forward:
 	JLS	move_256through2048
 
 	// If REP MOVSB isn't fast, don't use it
-	CMPB	runtime·support_erms(SB), $1 // enhanced REP MOVSB/STOSB
+	CMPB	internal∕cpu·X86+const_offset_x86_HasERMS(SB), $1 // enhanced REP MOVSB/STOSB
 	JNE	fwdBy8
 
 	// Check alignment

src/runtime/proc.go

@@ -507,7 +507,8 @@ func cpuinit() {
 
 	cpu.Initialize(env)
 
-	support_erms = cpu.X86.HasERMS
+	// Support cpu feature variables are used in code generated by the compiler
+	// to guard execution of instructions that can not be assumed to be always supported.
 	support_popcnt = cpu.X86.HasPOPCNT
 	support_sse2 = cpu.X86.HasSSE2
 	support_sse41 = cpu.X86.HasSSE41

src/runtime/runtime2.go

@@ -836,16 +836,15 @@ var (
 	newprocs   int32
 
 	// Information about what cpu features are available.
-	// Set on startup in runtime.cpuinit.
 	// Packages outside the runtime should not use these
 	// as they are not an external api.
-	// TODO: deprecate these; use internal/cpu directly.
+	// Set on startup in asm_{386,amd64,amd64p32}.s
 	processorVersionInfo uint32
 	isIntel              bool
 	lfenceBeforeRdtsc    bool
 
 	// Set in runtime.cpuinit.
-	support_erms          bool
+	// TODO: deprecate these; use internal/cpu directly.
 	support_popcnt        bool
 	support_sse2          bool
 	support_sse41         bool