runtime: remove VZEROUPPER in asyncPreempt on darwin/amd64

Updates golang#37174
Updates golang#49233
Fixes golang#41152

Change-Id: I35b148c8bc132f02dd6a5a6bb48b711fb5c5df9e
Reviewed-on: https://go-review.googlesource.com/c/go/+/560955
Reviewed-by: Keith Randall <[email protected]>
Auto-Submit: Dmitri Shuralyov <[email protected]>
LUCI-TryBot-Result: Go LUCI <[email protected]>
Reviewed-by: Cherry Mui <[email protected]>

src/runtime/mkpreempt.go

@@ -264,19 +264,6 @@ func genAMD64() {
 
 	l.save()
 
-	// Apparently, the signal handling code path in darwin kernel leaves
-	// the upper bits of Y registers in a dirty state, which causes
-	// many SSE operations (128-bit and narrower) become much slower.
-	// Clear the upper bits to get to a clean state. See issue #37174.
-	// It is safe here as Go code don't use the upper bits of Y registers.
-	p("#ifdef GOOS_darwin")
-	p("#ifndef hasAVX")
-	p("CMPB internal∕cpu·X86+const_offsetX86HasAVX(SB), $0")
-	p("JE 2(PC)")
-	p("#endif")
-	p("VZEROUPPER")
-	p("#endif")
-
 	lSSE.save()
 	p("CALL ·asyncPreempt2(SB)")
 	lSSE.restore()

src/runtime/preempt_amd64.s

@@ -27,13 +27,6 @@ TEXT ·asyncPreempt(SB),NOSPLIT|NOFRAME,$0-0
 	MOVQ R13, 88(SP)
 	MOVQ R14, 96(SP)
 	MOVQ R15, 104(SP)
-	#ifdef GOOS_darwin
-	#ifndef hasAVX
-	CMPB internal∕cpu·X86+const_offsetX86HasAVX(SB), $0
-	JE 2(PC)
-	#endif
-	VZEROUPPER
-	#endif
 	MOVUPS X0, 112(SP)
 	MOVUPS X1, 128(SP)
 	MOVUPS X2, 144(SP)