backport greatroar arm64 xxhash (#464)

see cespare/xxhash#51

```
benchmark                                                    old ns/op      new ns/op      delta
BenchmarkDecoder_DecoderSmall/kppkn.gtb.zst-16               14740529       14093294       -4.39%
BenchmarkDecoder_DecoderSmall/geo.protodata.zst-16           3003737        3008035        +0.14%
BenchmarkDecoder_DecoderSmall/plrabn12.txt.zst-16            70052218       70885931       +1.19%
BenchmarkDecoder_DecoderSmall/lcet10.txt.zst-16              33424884       34714144       +3.86%
BenchmarkDecoder_DecoderSmall/asyoulik.txt.zst-16            9694846        9563735        -1.35%
BenchmarkDecoder_DecoderSmall/alice29.txt.zst-16             12322865       12782681       +3.73%
BenchmarkDecoder_DecoderSmall/html_x_4.zst-16                7290319        7134523        -2.14%
BenchmarkDecoder_DecoderSmall/paper-100k.pdf.zst-16          825251         816683         -1.04%
BenchmarkDecoder_DecoderSmall/fireworks.jpeg.zst-16          462903         486823         +5.17%
BenchmarkDecoder_DecoderSmall/urls.10K.zst-16                1333159120     1114784390     -16.38%
BenchmarkDecoder_DecoderSmall/html.zst-16                    3117950        3095118        -0.73%
BenchmarkDecoder_DecoderSmall/comp-data.bin.zst-16           272085         271508         -0.21%
BenchmarkDecoder_DecodeAll/kppkn.gtb.zst-16                  1411294        1407632        -0.26%
BenchmarkDecoder_DecodeAll/geo.protodata.zst-16              370870         367499         -0.91%
BenchmarkDecoder_DecodeAll/plrabn12.txt.zst-16               4721330        4718339        -0.06%
BenchmarkDecoder_DecodeAll/lcet10.txt.zst-16                 3517766        3487756        -0.85%
BenchmarkDecoder_DecodeAll/asyoulik.txt.zst-16               1186672        1180367        -0.53%
BenchmarkDecoder_DecodeAll/alice29.txt.zst-16                1498383        1502922        +0.30%
BenchmarkDecoder_DecodeAll/html_x_4.zst-16                   748113         742537         -0.75%
BenchmarkDecoder_DecodeAll/paper-100k.pdf.zst-16             99132          98206          -0.93%
BenchmarkDecoder_DecodeAll/fireworks.jpeg.zst-16             53805          53209          -1.11%
BenchmarkDecoder_DecodeAll/urls.10K.zst-16                   4041818        4028347        -0.33%
BenchmarkDecoder_DecodeAll/html.zst-16                       387794         383309         -1.16%
BenchmarkDecoder_DecodeAll/comp-data.bin.zst-16              34390          34296          -0.27%
BenchmarkDecoder_DecodeAllParallel/kppkn.gtb.zst-16          89510          88785          -0.81%
BenchmarkDecoder_DecodeAllParallel/geo.protodata.zst-16      23315          23128          -0.80%
BenchmarkDecoder_DecodeAllParallel/plrabn12.txt.zst-16       306437         325176         +6.12%
BenchmarkDecoder_DecodeAllParallel/lcet10.txt.zst-16         226433         222589         -1.70%
BenchmarkDecoder_DecodeAllParallel/asyoulik.txt.zst-16       74612          74182          -0.58%
BenchmarkDecoder_DecodeAllParallel/alice29.txt.zst-16        95066          94304          -0.80%
BenchmarkDecoder_DecodeAllParallel/html_x_4.zst-16           47498          46946          -1.16%
BenchmarkDecoder_DecodeAllParallel/paper-100k.pdf.zst-16     6291           6237           -0.86%
BenchmarkDecoder_DecodeAllParallel/fireworks.jpeg.zst-16     3498           3453           -1.29%
BenchmarkDecoder_DecodeAllParallel/urls.10K.zst-16           295850         318916         +7.80%
BenchmarkDecoder_DecodeAllParallel/html.zst-16               24340          24196          -0.59%
BenchmarkDecoder_DecodeAllParallel/comp-data.bin.zst-16      2220           2199           -0.95%

benchmark                                                    old MB/s     new MB/s     speedup
BenchmarkDecoder_DecoderSmall/kppkn.gtb.zst-16               100.03       104.63       1.05x
BenchmarkDecoder_DecoderSmall/geo.protodata.zst-16           315.84       315.39       1.00x
BenchmarkDecoder_DecoderSmall/plrabn12.txt.zst-16            55.03        54.38        0.99x
BenchmarkDecoder_DecoderSmall/lcet10.txt.zst-16              102.14       98.35        0.96x
BenchmarkDecoder_DecoderSmall/asyoulik.txt.zst-16            103.30       104.71       1.01x
BenchmarkDecoder_DecoderSmall/alice29.txt.zst-16             98.74        95.18        0.96x
BenchmarkDecoder_DecoderSmall/html_x_4.zst-16                449.47       459.29       1.02x
BenchmarkDecoder_DecoderSmall/paper-100k.pdf.zst-16          992.67       1003.08      1.01x
BenchmarkDecoder_DecoderSmall/fireworks.jpeg.zst-16          2127.32      2022.80      0.95x
BenchmarkDecoder_DecoderSmall/urls.10K.zst-16                4.21         5.04         1.20x
BenchmarkDecoder_DecoderSmall/html.zst-16                    262.74       264.67       1.01x
BenchmarkDecoder_DecoderSmall/comp-data.bin.zst-16           119.84       120.10       1.00x
BenchmarkDecoder_DecodeAll/kppkn.gtb.zst-16                  130.60       130.94       1.00x
BenchmarkDecoder_DecodeAll/geo.protodata.zst-16              319.76       322.69       1.01x
BenchmarkDecoder_DecodeAll/plrabn12.txt.zst-16               102.06       102.13       1.00x
BenchmarkDecoder_DecodeAll/lcet10.txt.zst-16                 121.31       122.36       1.01x
BenchmarkDecoder_DecodeAll/asyoulik.txt.zst-16               105.49       106.05       1.01x
BenchmarkDecoder_DecodeAll/alice29.txt.zst-16                101.50       101.20       1.00x
BenchmarkDecoder_DecodeAll/html_x_4.zst-16                   547.51       551.62       1.01x
BenchmarkDecoder_DecodeAll/paper-100k.pdf.zst-16             1032.97      1042.70      1.01x
BenchmarkDecoder_DecodeAll/fireworks.jpeg.zst-16             2287.77      2313.39      1.01x
BenchmarkDecoder_DecodeAll/urls.10K.zst-16                   173.71       174.29       1.00x
BenchmarkDecoder_DecodeAll/html.zst-16                       264.06       267.15       1.01x
BenchmarkDecoder_DecodeAll/comp-data.bin.zst-16              118.52       118.85       1.00x
BenchmarkDecoder_DecodeAllParallel/kppkn.gtb.zst-16          2059.22      2076.03      1.01x
BenchmarkDecoder_DecodeAllParallel/geo.protodata.zst-16      5086.37      5127.46      1.01x
BenchmarkDecoder_DecodeAllParallel/plrabn12.txt.zst-16       1572.47      1481.85      0.94x
BenchmarkDecoder_DecodeAllParallel/lcet10.txt.zst-16         1884.68      1917.22      1.02x
BenchmarkDecoder_DecodeAllParallel/asyoulik.txt.zst-16       1677.73      1687.46      1.01x
BenchmarkDecoder_DecodeAllParallel/alice29.txt.zst-16        1599.82      1612.75      1.01x
BenchmarkDecoder_DecodeAllParallel/html_x_4.zst-16           8623.46      8724.94      1.01x
BenchmarkDecoder_DecodeAllParallel/paper-100k.pdf.zst-16     16276.14     16418.01     1.01x
BenchmarkDecoder_DecodeAllParallel/fireworks.jpeg.zst-16     35194.18     35648.95     1.01x
BenchmarkDecoder_DecodeAllParallel/urls.10K.zst-16           2373.12      2201.48      0.93x
BenchmarkDecoder_DecodeAllParallel/html.zst-16               4207.06      4232.02      1.01x
BenchmarkDecoder_DecodeAllParallel/comp-data.bin.zst-16      1835.78      1853.29      1.01x

benchmark                                                    old allocs     new allocs     delta
BenchmarkDecoder_DecoderSmall/kppkn.gtb.zst-16               1              3              +200.00%
BenchmarkDecoder_DecoderSmall/geo.protodata.zst-16           1              1              +0.00%
BenchmarkDecoder_DecoderSmall/plrabn12.txt.zst-16            4              5              +25.00%
BenchmarkDecoder_DecoderSmall/lcet10.txt.zst-16              1              1              +0.00%
BenchmarkDecoder_DecoderSmall/asyoulik.txt.zst-16            1              1              +0.00%
BenchmarkDecoder_DecoderSmall/alice29.txt.zst-16             1              1              +0.00%
BenchmarkDecoder_DecoderSmall/html_x_4.zst-16                1              1              +0.00%
BenchmarkDecoder_DecoderSmall/paper-100k.pdf.zst-16          1              1              +0.00%
BenchmarkDecoder_DecoderSmall/fireworks.jpeg.zst-16          1              1              +0.00%
BenchmarkDecoder_DecoderSmall/urls.10K.zst-16                52             50             -3.85%
BenchmarkDecoder_DecoderSmall/html.zst-16                    1              1              +0.00%
BenchmarkDecoder_DecoderSmall/comp-data.bin.zst-16           1              1              +0.00%
BenchmarkDecoder_DecodeAll/kppkn.gtb.zst-16                  0              0              +0.00%
BenchmarkDecoder_DecodeAll/geo.protodata.zst-16              0              0              +0.00%
BenchmarkDecoder_DecodeAll/plrabn12.txt.zst-16               0              0              +0.00%
BenchmarkDecoder_DecodeAll/lcet10.txt.zst-16                 0              0              +0.00%
BenchmarkDecoder_DecodeAll/asyoulik.txt.zst-16               0              0              +0.00%
BenchmarkDecoder_DecodeAll/alice29.txt.zst-16                0              0              +0.00%
BenchmarkDecoder_DecodeAll/html_x_4.zst-16                   0              0              +0.00%
BenchmarkDecoder_DecodeAll/paper-100k.pdf.zst-16             0              0              +0.00%
BenchmarkDecoder_DecodeAll/fireworks.jpeg.zst-16             0              0              +0.00%
BenchmarkDecoder_DecodeAll/urls.10K.zst-16                   0              0              +0.00%
BenchmarkDecoder_DecodeAll/html.zst-16                       0              0              +0.00%
BenchmarkDecoder_DecodeAll/comp-data.bin.zst-16              0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/kppkn.gtb.zst-16          0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/geo.protodata.zst-16      0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/plrabn12.txt.zst-16       0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/lcet10.txt.zst-16         0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/asyoulik.txt.zst-16       0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/alice29.txt.zst-16        0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/html_x_4.zst-16           0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/paper-100k.pdf.zst-16     0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/fireworks.jpeg.zst-16     0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/urls.10K.zst-16           0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/html.zst-16               0              0              +0.00%
BenchmarkDecoder_DecodeAllParallel/comp-data.bin.zst-16      0              0              +0.00%

benchmark                                                    old bytes     new bytes     delta
BenchmarkDecoder_DecoderSmall/kppkn.gtb.zst-16               89288         113942        +27.61%
BenchmarkDecoder_DecoderSmall/geo.protodata.zst-16           10888         12311         +13.07%
BenchmarkDecoder_DecoderSmall/plrabn12.txt.zst-16            981385        1492955       +52.13%
BenchmarkDecoder_DecoderSmall/lcet10.txt.zst-16              385552        367000        -4.81%
BenchmarkDecoder_DecoderSmall/asyoulik.txt.zst-16            30752         30503         -0.81%
BenchmarkDecoder_DecoderSmall/alice29.txt.zst-16             108837        56868         -47.75%
BenchmarkDecoder_DecoderSmall/html_x_4.zst-16                75923         97336         +28.20%
BenchmarkDecoder_DecoderSmall/paper-100k.pdf.zst-16          48            48            +0.00%
BenchmarkDecoder_DecoderSmall/fireworks.jpeg.zst-16          1964          1909          -2.80%
BenchmarkDecoder_DecoderSmall/urls.10K.zst-16                25456880      25432640      -0.10%
BenchmarkDecoder_DecoderSmall/html.zst-16                    48            48            +0.00%
BenchmarkDecoder_DecoderSmall/comp-data.bin.zst-16           51            48            -5.88%
BenchmarkDecoder_DecodeAll/kppkn.gtb.zst-16                  0             0             +0.00%
BenchmarkDecoder_DecodeAll/geo.protodata.zst-16              5             0             -100.00%
BenchmarkDecoder_DecodeAll/plrabn12.txt.zst-16               2             0             -100.00%
BenchmarkDecoder_DecodeAll/lcet10.txt.zst-16                 26            0             -100.00%
BenchmarkDecoder_DecodeAll/asyoulik.txt.zst-16               0             0             +0.00%
BenchmarkDecoder_DecodeAll/alice29.txt.zst-16                10            0             -100.00%
BenchmarkDecoder_DecodeAll/html_x_4.zst-16                   0             0             +0.00%
BenchmarkDecoder_DecodeAll/paper-100k.pdf.zst-16             0             0             +0.00%
BenchmarkDecoder_DecodeAll/fireworks.jpeg.zst-16             1             0             -100.00%
BenchmarkDecoder_DecodeAll/urls.10K.zst-16                   0             0             +0.00%
BenchmarkDecoder_DecodeAll/html.zst-16                       0             0             +0.00%
BenchmarkDecoder_DecodeAll/comp-data.bin.zst-16              0             0             +0.00%
BenchmarkDecoder_DecodeAllParallel/kppkn.gtb.zst-16          234           231           -1.28%
BenchmarkDecoder_DecodeAllParallel/geo.protodata.zst-16      38            38            +0.00%
BenchmarkDecoder_DecodeAllParallel/plrabn12.txt.zst-16       1991          2287          +14.87%
BenchmarkDecoder_DecodeAllParallel/lcet10.txt.zst-16         1501          1298          -13.52%
BenchmarkDecoder_DecodeAllParallel/asyoulik.txt.zst-16       132           130           -1.52%
BenchmarkDecoder_DecodeAllParallel/alice29.txt.zst-16        214           220           +2.80%
BenchmarkDecoder_DecodeAllParallel/html_x_4.zst-16           259           258           -0.39%
BenchmarkDecoder_DecodeAllParallel/paper-100k.pdf.zst-16     9             9             +0.00%
BenchmarkDecoder_DecodeAllParallel/fireworks.jpeg.zst-16     6             6             +0.00%
BenchmarkDecoder_DecodeAllParallel/urls.10K.zst-16           2868          3358          +17.09%
BenchmarkDecoder_DecodeAllParallel/html.zst-16               34            34            +0.00%
BenchmarkDecoder_DecodeAllParallel/comp-data.bin.zst-16      0             0             +0.00%
```

zstd/internal/xxhash/xxhash_arm64.s

@@ -0,0 +1,189 @@
+// +build gc,!purego
+
+#include "textflag.h"
+
+// Register allocation.
+#define digest	R1
+#define h	R2	// Return value.
+#define p	R3	// Input pointer.
+#define len	R4
+#define nblocks	R5	// len / 32.
+#define prime1	R7
+#define prime2	R8
+#define prime3	R9
+#define prime4	R10
+#define prime5	R11
+#define v1	R12
+#define v2	R13
+#define v3	R14
+#define v4	R15
+#define x1	R20
+#define x2	R21
+#define x3	R22
+#define x4	R23
+
+#define round(acc, x) 			\
+	MADD prime2, acc, x, acc	\
+	ROR  $64-31, acc		\
+	MUL  prime1, acc		\
+
+// x = round(0, x).
+#define round0(x)	\
+	MUL prime2, x	\
+	ROR $64-31, x	\
+	MUL prime1, x	\
+
+#define mergeRound(x)			\
+	round0(x)			\
+	EOR  x, h			\
+	MADD h, prime4, prime1, h	\
+
+// Update v[1-4] with 32-byte blocks. Assumes len >= 32.
+#define blocksLoop()		\
+	LSR  $5, len, nblocks	\
+	PCALIGN $16		\
+loop:				\
+	LDP.P 32(p), (x1, x2)	\
+	round(v1, x1)		\
+	LDP  -16(p), (x3, x4)	\
+	round(v2, x2)		\
+	SUB  $1, nblocks	\
+	round(v3, x3)		\
+	round(v4, x4)		\
+	CBNZ nblocks, loop	\
+
+
+// The primes are repeated here to ensure that they're stored
+// in a contiguous array, so we can load them with LDP.
+DATA  primes<> +0(SB)/8, $11400714785074694791
+DATA  primes<> +8(SB)/8, $14029467366897019727
+DATA  primes<>+16(SB)/8, $1609587929392839161
+DATA  primes<>+24(SB)/8, $9650029242287828579
+DATA  primes<>+32(SB)/8, $2870177450012600261
+GLOBL primes<>(SB), NOPTR+RODATA, $40
+
+
+// func Sum64(b []byte) uint64
+TEXT ·Sum64(SB), NOFRAME+NOSPLIT, $0-32
+	LDP  b_base+0(FP), (p, len)
+
+	LDP  primes<> +0(SB), (prime1, prime2)
+	LDP  primes<>+16(SB), (prime3, prime4)
+	MOVD primes<>+32(SB), prime5
+
+	CMP  $32, len
+	CSEL LO, prime5, ZR, h // if len < 32 { h = prime5 } else { h = 0 }
+	BLO  afterLoop
+
+	ADD  prime1, prime2, v1
+	MOVD prime2, v2
+	MOVD $0, v3
+	NEG  prime1, v4
+
+	blocksLoop()
+
+	ROR $64-1, v1, x1
+	ROR $64-7, v2, x2
+	ADD x1, x2
+	ROR $64-12, v3, x3
+	ROR $64-18, v4, x4
+	ADD x3, x4
+	ADD x2, x4, h
+
+	mergeRound(v1)
+	mergeRound(v2)
+	mergeRound(v3)
+	mergeRound(v4)
+
+afterLoop:
+	ADD len, h
+
+	TBZ   $4, len, try8
+	LDP.P 16(p), (x1, x2)
+
+	round0(x1)
+	ROR  $64-27, h
+	EOR  x1 @> 64-27, h, h
+	MADD h, prime4, prime1, h
+
+	round0(x2)
+	ROR  $64-27, h
+	EOR  x2 @> 64-27, h
+	MADD h, prime4, prime1, h
+
+try8:
+	TBZ    $3, len, try4
+	MOVD.P 8(p), x1
+
+	round0(x1)
+	ROR  $64-27, h
+	EOR  x1 @> 64-27, h
+	MADD h, prime4, prime1, h
+
+try4:
+	TBZ     $2, len, try2
+	MOVWU.P 4(p), x2
+
+	MUL  prime1, x2
+	ROR  $64-23, h
+	EOR  x2 @> 64-23, h
+	MADD h, prime3, prime2, h
+
+try2:
+	TBZ     $1, len, try1
+	MOVHU.P 2(p), x3
+	AND     $255, x3, x1
+	LSR     $8, x3, x2
+
+	MUL prime5, x1
+	ROR $64-11, h
+	EOR x1 @> 64-11, h
+	MUL prime1, h
+
+	MUL prime5, x2
+	ROR $64-11, h
+	EOR x2 @> 64-11, h
+	MUL prime1, h
+
+try1:
+	TBZ   $0, len, end
+	MOVBU (p), x4
+
+	MUL prime5, x4
+	ROR $64-11, h
+	EOR x4 @> 64-11, h
+	MUL prime1, h
+
+end:
+	EOR h >> 33, h
+	MUL prime2,  h
+	EOR h >> 29, h
+	MUL prime3,  h
+	EOR h >> 32, h
+
+	MOVD h, ret+24(FP)
+	RET
+
+
+// func writeBlocks(d *Digest, b []byte) int
+//
+// Assumes len(b) >= 32.
+TEXT ·writeBlocks(SB), NOFRAME+NOSPLIT, $0-40
+	LDP  primes<>(SB), (prime1, prime2)
+
+	// Load state. Assume v[1-4] are stored contiguously.
+	MOVD d+0(FP), digest
+	LDP   0(digest), (v1, v2)
+	LDP  16(digest), (v3, v4)
+
+	LDP b_base+8(FP), (p, len)
+
+	blocksLoop()
+
+	// Store updated state.
+	STP (v1, v2),  0(digest)
+	STP (v3, v4), 16(digest)
+
+	BIC  $31, len
+	MOVD len, ret+32(FP)
+	RET

zstd/internal/xxhash/xxhash_amd64.go → zstd/internal/xxhash/xxhash_asm.go

@@ -1,5 +1,8 @@
-//go:build !appengine && gc && !purego
-// +build !appengine,gc,!purego
+//go:build (amd64 || arm64) && !appengine && gc && !purego
+// +build amd64 arm64
+// +build !appengine
+// +build gc
+// +build !purego
 
 package xxhash
 

zstd/internal/xxhash/xxhash_other.go

@@ -1,5 +1,5 @@
-//go:build !amd64 || appengine || !gc || purego
-// +build !amd64 appengine !gc purego
+//go:build (!amd64 && !arm64) || appengine || !gc || purego
+// +build !amd64,!arm64 appengine !gc purego
 
 package xxhash