runtime: use clock_gettime to get ns resolution for time.now & runtime.nanotime

For Linux/{386,arm}, FreeBSD/{386,amd64,arm}, NetBSD/{386,amd64}, OpenBSD/{386,amd64}.
Note: our Darwin implementation already has ns resolution.

Linux/386 (Core i7-2600 @ 3.40GHz, kernel 3.5.2-gentoo)
benchmark       old ns/op    new ns/op    delta
BenchmarkNow          110          118   +7.27%

Linux/ARM (ARM Cortex-A8 @ 800MHz, kernel 2.6.32.28 android)
benchmark       old ns/op    new ns/op    delta
BenchmarkNow          625          542  -13.28%

Linux/ARM (ARM Cortex-A9 @ 1GHz, Pandaboard)
benchmark       old ns/op    new ns/op    delta
BenchmarkNow          992          909   -8.37%

FreeBSD 9-REL-p1/amd64 (Dell R610 Server with Xeon X5650 @ 2.67GHz)
benchmark       old ns/op    new ns/op    delta
BenchmarkNow          699          695   -0.57%

FreeBSD 9-REL-p1/amd64 (Atom D525 @ 1.80GHz)
benchmark       old ns/op    new ns/op    delta
BenchmarkNow         1553         1658   +6.76%

OpenBSD/amd64 (Dell E6410 with i5 CPU M 540 @ 2.53GHz)
benchmark       old ns/op    new ns/op    delta
BenchmarkNow         1262         1236   -2.06%

OpenBSD/i386 (Asus eeePC 701 with Intel Celeron M 900MHz - locked to 631MHz)
benchmark       old ns/op    new ns/op    delta
BenchmarkNow         5089         5043   -0.90%

NetBSD/i386 (VMware VM with Core i5 CPU @ 2.7GHz)
benchmark       old ns/op    new ns/op    delta
BenchmarkNow          277          278   +0.36%

NetBSD/amd64 (VMware VM with Core i5 CPU @ 2.7Ghz)
benchmark       old ns/op    new ns/op    delta
BenchmarkNow          103          105   +1.94%

Thanks Maxim Khitrov, Joel Sing, and Dave Cheney for providing benchmark data.

R=jsing, dave, rsc
CC=golang-dev
https://golang.org/cl/6820120

9 files changed, 95 insertions, 115 deletions

diff --git a/src/pkg/runtime/sys_freebsd_386.s b/src/pkg/runtime/sys_freebsd_386.s
index 9049f1e75b..a30c7ba01f 100644
--- a/src/pkg/runtime/sys_freebsd_386.s
+++ b/src/pkg/runtime/sys_freebsd_386.s
@@ -116,40 +116,38 @@ TEXT runtime·setitimer(SB), 7, $-4
 
 // func now() (sec int64, nsec int32)
 TEXT time·now(SB), 7, $32
-	MOVL	$116, AX
+	MOVL	$232, AX
 	LEAL	12(SP), BX
-	MOVL	BX, 4(SP)
-	MOVL	$0, 8(SP)
+	MOVL	$0, 4(SP)
+	MOVL	BX, 8(SP)
 	INT	$0x80
 	MOVL	12(SP), AX	// sec
-	MOVL	16(SP), BX	// usec
+	MOVL	16(SP), BX	// nsec
 
-	// sec is in AX, usec in BX
+	// sec is in AX, nsec in BX
 	MOVL	AX, sec+0(FP)
 	MOVL	$0, sec+4(FP)
-	IMULL	$1000, BX
 	MOVL	BX, nsec+8(FP)
 	RET
 
 // int64 nanotime(void) so really
 // void nanotime(int64 *nsec)
 TEXT runtime·nanotime(SB), 7, $32
-	MOVL	$116, AX
+	MOVL	$232, AX
 	LEAL	12(SP), BX
-	MOVL	BX, 4(SP)
-	MOVL	$0, 8(SP)
+	MOVL	$0, 4(SP)
+	MOVL	BX, 8(SP)
 	INT	$0x80
 	MOVL	12(SP), AX	// sec
-	MOVL	16(SP), BX	// usec
+	MOVL	16(SP), BX	// nsec
 
-	// sec is in AX, usec in BX
+	// sec is in AX, nsec in BX
 	// convert to DX:AX nsec
 	MOVL	$1000000000, CX
 	MULL	CX
-	IMULL	$1000, BX
 	ADDL	BX, AX
 	ADCL	$0, DX
-	
+
 	MOVL	ret+0(FP), DI
 	MOVL	AX, 0(DI)
 	MOVL	DX, 4(DI)
diff --git a/src/pkg/runtime/sys_freebsd_amd64.s b/src/pkg/runtime/sys_freebsd_amd64.s
index 6c52dfa15e..7b1359167b 100644
--- a/src/pkg/runtime/sys_freebsd_amd64.s
+++ b/src/pkg/runtime/sys_freebsd_amd64.s
@@ -94,31 +94,29 @@ TEXT runtime·setitimer(SB), 7, $-8
 
 // func now() (sec int64, nsec int32)
 TEXT time·now(SB), 7, $32
-	MOVL	$116, AX
-	LEAQ	8(SP), DI
-	MOVQ	$0, SI
+	MOVL	$232, AX
+	MOVQ	$0, DI
+	LEAQ	8(SP), SI
 	SYSCALL
 	MOVQ	8(SP), AX	// sec
-	MOVL	16(SP), DX	// usec
+	MOVQ	16(SP), DX	// nsec
 
-	// sec is in AX, usec in DX
+	// sec is in AX, nsec in DX
 	MOVQ	AX, sec+0(FP)
-	IMULQ	$1000, DX
 	MOVL	DX, nsec+8(FP)
 	RET
 
 TEXT runtime·nanotime(SB), 7, $32
-	MOVL	$116, AX
-	LEAQ	8(SP), DI
-	MOVQ	$0, SI
+	MOVL	$232, AX
+	MOVQ	$0, DI
+	LEAQ	8(SP), SI
 	SYSCALL
 	MOVQ	8(SP), AX	// sec
-	MOVL	16(SP), DX	// usec
+	MOVQ	16(SP), DX	// nsec
 
-	// sec is in AX, usec in DX
+	// sec is in AX, nsec in DX
 	// return nsec in AX
 	IMULQ	$1000000000, AX
-	IMULQ	$1000, DX
 	ADDQ	DX, AX
 	RET
 
diff --git a/src/pkg/runtime/sys_freebsd_arm.s b/src/pkg/runtime/sys_freebsd_arm.s
index 1edf8071bb..ec8134e4c5 100644
--- a/src/pkg/runtime/sys_freebsd_arm.s
+++ b/src/pkg/runtime/sys_freebsd_arm.s
@@ -86,39 +86,36 @@ TEXT runtime·setitimer(SB), 7, $-8
 
 // func now() (sec int64, nsec int32)
 TEXT time·now(SB), 7, $32
-	MOVW $8(R13), R0
-	MOVW $0, R1
-	SWI $116 // gettimeofday
+	MOVW $0, R0 // CLOCK_REALTIME
+	MOVW $8(R13), R1
+	SWI $232 // clock_gettime
 
 	MOVW 8(R13), R0 // sec.low
-	MOVW 16(R13), R2 // usec
+	MOVW 12(R13), R1 // sec.high
+	MOVW 16(R13), R2 // nsec
 
 	MOVW R0, 0(FP)
-	MOVW $0, R1
 	MOVW R1, 4(FP)
-	MOVW $1000, R3
-	MUL R3, R2
 	MOVW R2, 8(FP)
 	RET
 
 // int64 nanotime(void) so really
 // void nanotime(int64 *nsec)
 TEXT runtime·nanotime(SB), 7, $32
-	MOVW $8(R13), R0
-	MOVW $0, R1
-	SWI $116 // gettimeofday
+	MOVW $0, R0 // CLOCK_REALTIME
+	MOVW $8(R13), R1
+	SWI $232 // clock_gettime
 
 	MOVW 8(R13), R0 // sec.low
-	MOVW 16(R13), R2 // usec
+	MOVW 12(R13), R4 // sec.high
+	MOVW 16(R13), R2 // nsec
 
 	MOVW $1000000000, R3
 	MULLU R0, R3, (R1, R0)
-	MOVW $1000, R3
-	MOVW $0, R4
-	MUL R3, R2
+	MUL R3, R4
 	ADD.S R2, R0
 	ADC R4, R1
-	
+
 	MOVW 0(FP), R3
 	MOVW R0, 0(R3)
 	MOVW R1, 4(R3)
diff --git a/src/pkg/runtime/sys_linux_386.s b/src/pkg/runtime/sys_linux_386.s
index 28ae37b8d9..b212f806c1 100644
--- a/src/pkg/runtime/sys_linux_386.s
+++ b/src/pkg/runtime/sys_linux_386.s
@@ -104,40 +104,38 @@ TEXT runtime·mincore(SB),7,$0-24
 
 // func now() (sec int64, nsec int32)
 TEXT time·now(SB), 7, $32
-	MOVL	$78, AX			// syscall - gettimeofday
-	LEAL	8(SP), BX
-	MOVL	$0, CX
+	MOVL	$265, AX			// syscall - clock_gettime
+	MOVL	$0, BX
+	LEAL	8(SP), CX
 	MOVL	$0, DX
 	CALL	*runtime·_vdso(SB)
 	MOVL	8(SP), AX	// sec
-	MOVL	12(SP), BX	// usec
+	MOVL	12(SP), BX	// nsec
 
-	// sec is in AX, usec in BX
+	// sec is in AX, nsec in BX
 	MOVL	AX, sec+0(FP)
 	MOVL	$0, sec+4(FP)
-	IMULL	$1000, BX
 	MOVL	BX, nsec+8(FP)
 	RET
 
 // int64 nanotime(void) so really
 // void nanotime(int64 *nsec)
 TEXT runtime·nanotime(SB), 7, $32
-	MOVL	$78, AX			// syscall - gettimeofday
-	LEAL	8(SP), BX
-	MOVL	$0, CX
+	MOVL	$265, AX			// syscall - clock_gettime
+	MOVL	$0, BX
+	LEAL	8(SP), CX
 	MOVL	$0, DX
 	CALL	*runtime·_vdso(SB)
 	MOVL	8(SP), AX	// sec
-	MOVL	12(SP), BX	// usec
+	MOVL	12(SP), BX	// nsec
 
-	// sec is in AX, usec in BX
+	// sec is in AX, nsec in BX
 	// convert to DX:AX nsec
 	MOVL	$1000000000, CX
 	MULL	CX
-	IMULL	$1000, BX
 	ADDL	BX, AX
 	ADCL	$0, DX
-	
+
 	MOVL	ret+0(FP), DI
 	MOVL	AX, 0(DI)
 	MOVL	DX, 4(DI)
diff --git a/src/pkg/runtime/sys_linux_arm.s b/src/pkg/runtime/sys_linux_arm.s
index 38bcebfa1a..ec5b0ca135 100644
--- a/src/pkg/runtime/sys_linux_arm.s
+++ b/src/pkg/runtime/sys_linux_arm.s
@@ -35,6 +35,7 @@
 #define SYS_select (SYS_BASE + 142) // newselect
 #define SYS_ugetrlimit (SYS_BASE + 191)
 #define SYS_sched_getaffinity (SYS_BASE + 242)
+#define SYS_clock_gettime (SYS_BASE + 263)
 
 #define ARM_BASE (SYS_BASE + 0x0f0000)
 #define SYS_ARM_cacheflush (ARM_BASE + 2)
@@ -155,41 +156,37 @@ TEXT runtime·mincore(SB),7,$0
 	RET
 
 TEXT time·now(SB), 7, $32
-	MOVW	$8(R13), R0  // timeval
-	MOVW	$0, R1  // zone
-	MOVW	$SYS_gettimeofday, R7
+	MOVW	$0, R0  // CLOCK_REALTIME
+	MOVW	$8(R13), R1  // timespec
+	MOVW	$SYS_clock_gettime, R7
 	SWI	$0
 	
 	MOVW	8(R13), R0  // sec
-	MOVW	12(R13), R2  // usec
+	MOVW	12(R13), R2  // nsec
 	
 	MOVW	R0, 0(FP)
 	MOVW	$0, R1
 	MOVW	R1, 4(FP)
-	MOVW	$1000, R3
-	MUL	R3, R2
 	MOVW	R2, 8(FP)
 	RET	
 
 // int64 nanotime(void) so really
 // void nanotime(int64 *nsec)
 TEXT runtime·nanotime(SB),7,$32
-	MOVW	$8(R13), R0  // timeval
-	MOVW	$0, R1  // zone
-	MOVW	$SYS_gettimeofday, R7
+	MOVW	$0, R0  // CLOCK_REALTIME
+	MOVW	$8(R13), R1  // timespec
+	MOVW	$SYS_clock_gettime, R7
 	SWI	$0
 	
 	MOVW	8(R13), R0  // sec
-	MOVW	12(R13), R2  // usec
+	MOVW	12(R13), R2  // nsec
 	
 	MOVW	$1000000000, R3
 	MULLU	R0, R3, (R1, R0)
-	MOVW	$1000, R3
 	MOVW	$0, R4
-	MUL	R3, R2
 	ADD.S	R2, R0
 	ADC	R4, R1
-	
+
 	MOVW	0(FP), R3
 	MOVW	R0, 0(R3)
 	MOVW	R1, 4(R3)
diff --git a/src/pkg/runtime/sys_netbsd_386.s b/src/pkg/runtime/sys_netbsd_386.s
index 9c88b14487..1a4ff95b6d 100644
--- a/src/pkg/runtime/sys_netbsd_386.s
+++ b/src/pkg/runtime/sys_netbsd_386.s
@@ -98,9 +98,9 @@ TEXT runtime·setitimer(SB),7,$-4
 // func now() (sec int64, nsec int32)
 TEXT time·now(SB), 7, $32
 	LEAL	12(SP), BX
-	MOVL	BX, 4(SP)		// arg 1 - tp
-	MOVL	$0, 8(SP)		// arg 2 - tzp
-	MOVL	$418, AX		// sys_gettimeofday
+	MOVL	$0, 4(SP)		// arg 1 - clock_id
+	MOVL	BX, 8(SP)		// arg 2 - tp
+	MOVL	$427, AX		// sys_clock_gettime
 	INT	$0x80
 
 	MOVL	12(SP), AX		// sec - l32
@@ -108,8 +108,7 @@ TEXT time·now(SB), 7, $32
 	MOVL	16(SP), AX		// sec - h32
 	MOVL	AX, sec+4(FP)
 
-	MOVL	20(SP), BX		// usec - should not exceed 999999
-	IMULL	$1000, BX
+	MOVL	20(SP), BX		// nsec
 	MOVL	BX, nsec+8(FP)
 	RET
 
@@ -117,9 +116,9 @@ TEXT time·now(SB), 7, $32
 // void nanotime(int64 *nsec)
 TEXT runtime·nanotime(SB),7,$32
 	LEAL	12(SP), BX
-	MOVL	BX, 4(SP)		// arg 1 - tp
-	MOVL	$0, 8(SP)		// arg 2 - tzp
-	MOVL	$418, AX		// sys_gettimeofday
+	MOVL	$0, 4(SP)		// arg 1 - clock_id
+	MOVL	BX, 8(SP)		// arg 2 - tp
+	MOVL	$427, AX		// sys_clock_gettime
 	INT	$0x80
 
 	MOVL	16(SP), CX		// sec - h32
@@ -129,8 +128,7 @@ TEXT runtime·nanotime(SB),7,$32
 	MOVL	$1000000000, BX
 	MULL	BX			// result in dx:ax
 
-	MOVL	20(SP), BX		// usec
-	IMULL	$1000, BX
+	MOVL	20(SP), BX		// nsec
 	ADDL	BX, AX
 	ADCL	CX, DX			// add high bits with carry
 
diff --git a/src/pkg/runtime/sys_netbsd_amd64.s b/src/pkg/runtime/sys_netbsd_amd64.s
index e00c727ccb..fc5f28ca79 100644
--- a/src/pkg/runtime/sys_netbsd_amd64.s
+++ b/src/pkg/runtime/sys_netbsd_amd64.s
@@ -122,31 +122,29 @@ TEXT runtime·setitimer(SB),7,$-8
 
 // func now() (sec int64, nsec int32)
 TEXT time·now(SB), 7, $32
-	LEAQ	8(SP), DI		// arg 1 - tp
-	MOVQ	$0, SI			// arg 2 - tzp
-	MOVL	$418, AX		// sys_gettimeofday
+	MOVQ	$0, DI			// arg 1 - clock_id
+	LEAQ	8(SP), SI		// arg 2 - tp
+	MOVL	$427, AX		// sys_clock_gettime
 	SYSCALL
 	MOVQ	8(SP), AX		// sec
-	MOVL	16(SP), DX		// usec
+	MOVL	16(SP), DX		// nsec
 
-	// sec is in AX, usec in DX
+	// sec is in AX, nsec in DX
 	MOVQ	AX, sec+0(FP)
-	IMULQ	$1000, DX
 	MOVL	DX, nsec+8(FP)
 	RET
 
 TEXT runtime·nanotime(SB),7,$32
-	LEAQ	8(SP), DI		// arg 1 - tp
-	MOVQ	$0, SI			// arg 2 - tzp
-	MOVL	$418, AX		// sys_gettimeofday
+	MOVQ	$0, DI			// arg 1 - clock_id
+	LEAQ	8(SP), SI		// arg 2 - tp
+	MOVL	$427, AX		// sys_clock_gettime
 	SYSCALL
 	MOVQ	8(SP), AX		// sec
-	MOVL	16(SP), DX		// usec
+	MOVL	16(SP), DX		// nsec
 
-	// sec is in AX, usec in DX
+	// sec is in AX, nsec in DX
 	// return nsec in AX
 	IMULQ	$1000000000, AX
-	IMULQ	$1000, DX
 	ADDQ	DX, AX
 	RET
 
diff --git a/src/pkg/runtime/sys_openbsd_386.s b/src/pkg/runtime/sys_openbsd_386.s
index f002f8eafb..f7921552cf 100644
--- a/src/pkg/runtime/sys_openbsd_386.s
+++ b/src/pkg/runtime/sys_openbsd_386.s
@@ -98,40 +98,38 @@ TEXT runtime·setitimer(SB),7,$-4
 
 // func now() (sec int64, nsec int32)
 TEXT time·now(SB), 7, $32
-	MOVL	$116, AX
+	MOVL	$232, AX
 	LEAL	12(SP), BX
-	MOVL	BX, 4(SP)
-	MOVL	$0, 8(SP)
+	MOVL	$0, 4(SP)
+	MOVL	BX, 8(SP)
 	INT	$0x80
 	MOVL	12(SP), AX		// sec
-	MOVL	16(SP), BX		// usec
+	MOVL	16(SP), BX		// nsec
 
-	// sec is in AX, usec in BX
+	// sec is in AX, nsec in BX
 	MOVL	AX, sec+0(FP)
 	MOVL	$0, sec+4(FP)
-	IMULL	$1000, BX
 	MOVL	BX, nsec+8(FP)
 	RET
 
 // int64 nanotime(void) so really
 // void nanotime(int64 *nsec)
 TEXT runtime·nanotime(SB),7,$32
-	MOVL	$116, AX
+	MOVL	$232, AX
 	LEAL	12(SP), BX
-	MOVL	BX, 4(SP)
-	MOVL	$0, 8(SP)
+	MOVL	$0, 4(SP)
+	MOVL	BX, 8(SP)
 	INT	$0x80
 	MOVL	12(SP), AX		// sec
-	MOVL	16(SP), BX		// usec
+	MOVL	16(SP), BX		// nsec
 
-	// sec is in AX, usec in BX
+	// sec is in AX, nsec in BX
 	// convert to DX:AX nsec
 	MOVL	$1000000000, CX
 	MULL	CX
-	IMULL	$1000, BX
 	ADDL	BX, AX
 	ADCL	$0, DX
-	
+
 	MOVL	ret+0(FP), DI
 	MOVL	AX, 0(DI)
 	MOVL	DX, 4(DI)
diff --git a/src/pkg/runtime/sys_openbsd_amd64.s b/src/pkg/runtime/sys_openbsd_amd64.s
index 6b674a51ac..e569becfa1 100644
--- a/src/pkg/runtime/sys_openbsd_amd64.s
+++ b/src/pkg/runtime/sys_openbsd_amd64.s
@@ -130,31 +130,29 @@ TEXT runtime·setitimer(SB),7,$-8
 
 // func now() (sec int64, nsec int32)
 TEXT time·now(SB), 7, $32
-	LEAQ	8(SP), DI		// arg 1 - tp
-	MOVQ	$0, SI			// arg 2 - tzp
-	MOVL	$116, AX		// sys_gettimeofday
+	MOVQ	$0, DI			// arg 1 - clock_id
+	LEAQ	8(SP), SI		// arg 2 - tp
+	MOVL	$232, AX		// sys_clock_gettime
 	SYSCALL
-	MOVQ	8(SP), AX		// sec
-	MOVL	16(SP), DX		// usec
+	MOVL	8(SP), AX		// sec
+	MOVQ	16(SP), DX		// nsec
 
-	// sec is in AX, usec in DX
+	// sec is in AX, nsec in DX
 	MOVQ	AX, sec+0(FP)
-	IMULQ	$1000, DX
 	MOVL	DX, nsec+8(FP)
 	RET
 
 TEXT runtime·nanotime(SB),7,$32
-	LEAQ	8(SP), DI		// arg 1 - tp
-	MOVQ	$0, SI			// arg 2 - tzp
-	MOVL	$116, AX		// sys_gettimeofday
+	MOVQ	$0, DI			// arg 1 - clock_id
+	LEAQ	8(SP), SI		// arg 2 - tp
+	MOVL	$232, AX		// sys_clock_gettime
 	SYSCALL
-	MOVQ	8(SP), AX		// sec
-	MOVL	16(SP), DX		// usec
+	MOVL	8(SP), AX		// sec
+	MOVQ	16(SP), DX		// nsec
 
-	// sec is in AX, usec in DX
+	// sec is in AX, nsec in DX
 	// return nsec in AX
 	IMULQ	$1000000000, AX
-	IMULQ	$1000, DX
 	ADDQ	DX, AX
 	RET