From e2543ef62c98b2ce07a54698e6a8293f380a1128 Mon Sep 17 00:00:00 2001
From: Dave Cheney <dave@cheney.net>
Date: Tue, 24 Mar 2015 09:37:10 +0000
Subject: [PATCH] runtime: add runtime.cmpstring and bytes.Compare

Update #10007

Implement runtime.cmpstring and bytes.Compare in asm for arm.

benchmark                                old ns/op     new ns/op     delta
BenchmarkCompareBytesEqual               254           91.4          -64.02%
BenchmarkCompareBytesToNil               41.5          37.6          -9.40%
BenchmarkCompareBytesEmpty               40.7          37.6          -7.62%
BenchmarkCompareBytesIdentical           255           96.3          -62.24%
BenchmarkCompareBytesSameLength          125           60.9          -51.28%
BenchmarkCompareBytesDifferentLength     133           60.9          -54.21%
BenchmarkCompareBytesBigUnaligned        17985879      5669706       -68.48%
BenchmarkCompareBytesBig                 17097634      4926798       -71.18%
BenchmarkCompareBytesBigIdentical        16861941      4389206       -73.97%

benchmark                             old MB/s     new MB/s     speedup
BenchmarkCompareBytesBigUnaligned     58.30        184.95       3.17x
BenchmarkCompareBytesBig              61.33        212.83       3.47x
BenchmarkCompareBytesBigIdentical     62.19        238.90       3.84x

This is a collaboration between Josh Bleecher Snyder and myself.

Change-Id: Ib3944b8c410d0e12135c2ba9459bfe131df48edd
Reviewed-on: https://go-review.googlesource.com/8010
Reviewed-by: Keith Randall <khr@golang.org>
---
 src/bytes/compare_test.go |  3 +++
 src/runtime/asm_arm.s     | 51 +++++++++++++++++++++++++++++++++++++++
 src/runtime/noasm.go      |  4 +--
 3 files changed, 56 insertions(+), 2 deletions(-)

diff --git a/src/bytes/compare_test.go b/src/bytes/compare_test.go
index 63522374ee0..f2d81d5310c 100644
--- a/src/bytes/compare_test.go
+++ b/src/bytes/compare_test.go
@@ -17,6 +17,8 @@ var compareTests = []struct {
 	{[]byte("a"), []byte(""), 1},
 	{[]byte(""), []byte("a"), -1},
 	{[]byte("abc"), []byte("abc"), 0},
+	{[]byte("abd"), []byte("abc"), 1},
+	{[]byte("abc"), []byte("abd"), -1},
 	{[]byte("ab"), []byte("abc"), -1},
 	{[]byte("abc"), []byte("ab"), 1},
 	{[]byte("x"), []byte("ab"), 1},
@@ -27,6 +29,7 @@ var compareTests = []struct {
 	{[]byte("abcdefgh"), []byte("abcdefgh"), 0},
 	{[]byte("abcdefghi"), []byte("abcdefghi"), 0},
 	{[]byte("abcdefghi"), []byte("abcdefghj"), -1},
+	{[]byte("abcdefghj"), []byte("abcdefghi"), 1},
 	// nil tests
 	{nil, nil, 0},
 	{[]byte(""), nil, 0},
diff --git a/src/runtime/asm_arm.s b/src/runtime/asm_arm.s
index cf1c9d13025..4074e503cc5 100644
--- a/src/runtime/asm_arm.s
+++ b/src/runtime/asm_arm.s
@@ -782,6 +782,57 @@ eq:
 	MOVB	R0, ret+8(FP)
 	RET
 
+TEXT runtime·cmpstring(SB),NOSPLIT,$0-20
+	MOVW	s1_base+0(FP), R2
+	MOVW	s1_len+4(FP), R0
+	MOVW	s2_base+8(FP), R3
+	MOVW	s2_len+12(FP), R1
+	BL	runtime·cmpbody(SB)
+	MOVW	R8, ret+16(FP)
+	RET
+
+TEXT bytes·Compare(SB),NOSPLIT,$0-28
+	MOVW	s1+0(FP), R2
+	MOVW	s1+4(FP), R0
+	MOVW	s2+12(FP), R3
+	MOVW	s2+16(FP), R1
+	BL	runtime·cmpbody(SB)
+	MOVW	R8, ret+24(FP)
+	RET
+
+// On entry:
+// R0 is the length of s1
+// R1 is the length of s2
+// R2 points to the start of s1
+// R3 points to the start of s2
+//
+// On exit:
+// R8 is -1/0/+1
+// R5, R4, and R6 are clobbered
+TEXT runtime·cmpbody(SB),NOSPLIT,$-4-0
+	CMP 	R0, R1
+	MOVW 	R0, R6
+	MOVW.LT	R1, R6	// R6 is min(R0, R1)
+
+	ADD	R2, R6	// R2 is current byte in s1, R6 is last byte in s1 to compare
+loop:
+	CMP	R2, R6
+	BEQ	samebytes // all compared bytes were the same; compare lengths
+	MOVBU.P	1(R2), R4
+	MOVBU.P	1(R3), R5
+	CMP	R4, R5
+	BEQ	loop
+	// bytes differed
+	MOVW.LT	$1, R8
+	MOVW.GT	$-1, R8
+	RET
+samebytes:
+	CMP	R0, R1
+	MOVW.LT	$1, R8
+	MOVW.GT	$-1, R8
+	MOVW.EQ	$0, R8
+	RET
+
 // eqstring tests whether two strings are equal.
 // The compiler guarantees that strings passed
 // to eqstring have equal length.
diff --git a/src/runtime/noasm.go b/src/runtime/noasm.go
index 4b3c577a21c..ab9c7447891 100644
--- a/src/runtime/noasm.go
+++ b/src/runtime/noasm.go
@@ -2,9 +2,9 @@
 // Use of this source code is governed by a BSD-style
 // license that can be found in the LICENSE file.
 
-// Routines that are implemented in assembly in asm_{amd64,386}.s
+// Routines that are implemented in assembly in asm_{amd64,386,arm}.s
 
-// +build arm arm64 ppc64 ppc64le
+// +build arm64 ppc64 ppc64le
 
 package runtime