runtime: add support for VDSO on ppc64x for use in walltime/nanotime

This change adds support for VDSO on ppc64x, making it possible to
avoid a syscall in walltime and nanotime.

BenchmarkClockVDSOAndFallbackPaths/vDSO-192             20000000    66.0 ns/op
BenchmarkClockVDSOAndFallbackPaths/Fallback-192          1000000    1456 ns/op

Change-Id: I3373bd804b6f122961de3ae9d034e6ccf35748e6
Reviewed-on: https://go-review.googlesource.com/131135
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>

src/runtime/os_linux_novdso.go

@@ -3,7 +3,7 @@
 // license that can be found in the LICENSE file.
 
 // +build linux
-// +build !386,!amd64,!arm,!arm64
+// +build !386,!amd64,!arm,!arm64,!ppc64,!ppc64le
 
 package runtime
 

src/runtime/sys_linux_ppc64x.s

@@ -154,21 +154,87 @@ TEXT runtime·mincore(SB),NOSPLIT|NOFRAME,$0-28
 
 // func walltime() (sec int64, nsec int32)
 TEXT runtime·walltime(SB),NOSPLIT,$16
-	MOVD	$0, R3 // CLOCK_REALTIME
-	MOVD	$0(R1), R4
-	SYSCALL	$SYS_clock_gettime
-	MOVD	0(R1), R3	// sec
-	MOVD	8(R1), R5	// nsec
+	MOVD	R1, R15		// R15 is unchanged by C code
+	MOVD	g_m(g), R21	// R21 = m
+
+	MOVD	$0, R3		// CLOCK_REALTIME
+
+	MOVD	runtime·vdsoClockgettimeSym(SB), R12	// Check for VDSO availability
+	CMP	R12, R0
+	BEQ	fallback
+
+	// Set vdsoPC and vdsoSP for SIGPROF traceback.
+	MOVD	LR, R14
+	MOVD	R14, m_vdsoPC(R21)
+	MOVD	R15, m_vdsoSP(R21)
+
+	MOVD	m_curg(R21), R6
+	CMP	g, R6
+	BNE	noswitch
+
+	MOVD	m_g0(R21), R7
+	MOVD	(g_sched+gobuf_sp)(R7), R1	// Set SP to g0 stack
+
+noswitch:
+	SUB	$16, R1			// Space for results
+	RLDICR	$0, R1, $59, R1		// Align for C code
+	MOVD	R12, CTR
+	MOVD	R1, R4
+	BL	(CTR)			// Call from VDSO
+	MOVD	$0, R0			// Restore R0
+	MOVD	R0, m_vdsoSP(R21)	// Clear vdsoSP
+	MOVD	0(R1), R3		// sec
+	MOVD	8(R1), R5		// nsec
+	MOVD	R15, R1			// Restore SP
+
+finish:
 	MOVD	R3, sec+0(FP)
 	MOVW	R5, nsec+8(FP)
 	RET
 
+	// Syscall fallback
+fallback:
+	ADD	$32, R1, R4
+	SYSCALL $SYS_clock_gettime
+	MOVD	32(R1), R3
+	MOVD	40(R1), R5
+	JMP	finish
+
 TEXT runtime·nanotime(SB),NOSPLIT,$16
-	MOVW	$1, R3 // CLOCK_MONOTONIC
-	MOVD	$0(R1), R4
-	SYSCALL	$SYS_clock_gettime
-	MOVD	0(R1), R3	// sec
-	MOVD	8(R1), R5	// nsec
+	MOVD	$1, R3		// CLOCK_MONOTONIC
+
+	MOVD	R1, R15		// R15 is unchanged by C code
+	MOVD	g_m(g), R21	// R21 = m
+
+	MOVD	runtime·vdsoClockgettimeSym(SB), R12	// Check for VDSO availability
+	CMP	R12, R0
+	BEQ	fallback
+
+	// Set vdsoPC and vdsoSP for SIGPROF traceback.
+	MOVD	LR, R14		// R14 is unchanged by C code
+	MOVD	R14, m_vdsoPC(R21)
+	MOVD	R15, m_vdsoSP(R21)
+
+	MOVD	m_curg(R21), R6
+	CMP	g, R6
+	BNE	noswitch
+
+	MOVD	m_g0(R21), R7
+	MOVD	(g_sched+gobuf_sp)(R7), R1	// Set SP to g0 stack
+
+noswitch:
+	SUB	$16, R1			// Space for results
+	RLDICR	$0, R1, $59, R1		// Align for C code
+	MOVD	R12, CTR
+	MOVD	R1, R4
+	BL	(CTR)			// Call from VDSO
+	MOVD	$0, R0			// Restore R0
+	MOVD	$0, m_vdsoSP(R21)	// Clear vdsoSP
+	MOVD	0(R1), R3		// sec
+	MOVD	8(R1), R5		// nsec
+	MOVD	R15, R1			// Restore SP
+
+finish:
 	// sec is in R3, nsec in R5
 	// return nsec in R3
 	MOVD	$1000000000, R4
@@ -177,6 +243,14 @@ TEXT runtime·nanotime(SB),NOSPLIT,$16
 	MOVD	R3, ret+0(FP)
 	RET
 
+	// Syscall fallback
+fallback:
+	ADD	$32, R1, R4
+	SYSCALL $SYS_clock_gettime
+	MOVD	32(R1), R3
+	MOVD	48(R1), R5
+	JMP	finish
+
 TEXT runtime·rtsigprocmask(SB),NOSPLIT|NOFRAME,$0-28
 	MOVW	how+0(FP), R3
 	MOVD	new+8(FP), R4

src/runtime/vdso_elf64.go

@@ -3,7 +3,7 @@
 // license that can be found in the LICENSE file.
 
 // +build linux
-// +build amd64 arm64
+// +build amd64 arm64 ppc64 ppc64le
 
 package runtime
 

src/runtime/vdso_in_none.go

@@ -2,7 +2,7 @@
 // Use of this source code is governed by a BSD-style
 // license that can be found in the LICENSE file.
 
-// +build linux,!386,!amd64,!arm,!arm64 !linux
+// +build linux,!386,!amd64,!arm,!arm64,!ppc64,!ppc64le !linux
 
 package runtime
 

src/runtime/vdso_linux.go

@@ -3,7 +3,7 @@
 // license that can be found in the LICENSE file.
 
 // +build linux
-// +build 386 amd64 arm arm64
+// +build 386 amd64 arm arm64 ppc64 ppc64le
 
 package runtime
 
@@ -42,6 +42,8 @@ const (
 
 	_STT_FUNC = 2 /* Symbol is a code object */
 
+	_STT_NOTYPE = 0 /* Symbol type is not specified */
+
 	_STB_GLOBAL = 1 /* Global symbol */
 	_STB_WEAK   = 2 /* Weak symbol */
 
@@ -212,7 +214,8 @@ func vdsoParseSymbols(info *vdsoInfo, version int32) {
 		sym := &info.symtab[symIndex]
 		typ := _ELF_ST_TYPE(sym.st_info)
 		bind := _ELF_ST_BIND(sym.st_info)
-		if typ != _STT_FUNC || bind != _STB_GLOBAL && bind != _STB_WEAK || sym.st_shndx == _SHN_UNDEF {
+		// On ppc64x, VDSO functions are of type _STT_NOTYPE.
+		if typ != _STT_FUNC && typ != _STT_NOTYPE || bind != _STB_GLOBAL && bind != _STB_WEAK || sym.st_shndx == _SHN_UNDEF {
 			return false
 		}
 		if k.name != gostringnocopy(&info.symstrings[sym.st_name]) {

src/runtime/vdso_linux_ppc64x.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.
+
+// +build linux
+// +build ppc64 ppc64le
+
+package runtime
+
+const (
+	// vdsoArrayMax is the byte-size of a maximally sized array on this architecture.
+	// See cmd/compile/internal/ppc64/galign.go arch.MAXWIDTH initialization.
+	vdsoArrayMax = 1<<50 - 1
+)
+
+var vdsoLinuxVersion = vdsoVersionKey{"LINUX_2.6.15", 0x75fcba5}
+
+var vdsoSymbolKeys = []vdsoSymbolKey{
+	{"__kernel_clock_gettime", 0xb0cd725, 0xdfa941fd, &vdsoClockgettimeSym},
+}
+
+// initialize with vsyscall fallbacks
+var (
+	vdsoClockgettimeSym uintptr = 0
+)

src/runtime/vdso_linux_test.go

@@ -3,7 +3,7 @@
 // license that can be found in the LICENSE file.
 
 // +build linux
-// +build 386 amd64 arm arm64
+// +build 386 amd64 arm arm64 ppc64 ppc64le
 
 package runtime_test