summaryrefslogtreecommitdiffstats
path: root/vendor/github.com/klauspost/compress/zstd/internal
diff options
context:
space:
mode:
Diffstat (limited to 'vendor/github.com/klauspost/compress/zstd/internal')
-rw-r--r--vendor/github.com/klauspost/compress/zstd/internal/xxhash/LICENSE.txt22
-rw-r--r--vendor/github.com/klauspost/compress/zstd/internal/xxhash/README.md58
-rw-r--r--vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash.go237
-rw-r--r--vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_amd64.s215
-rw-r--r--vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_arm64.s189
-rw-r--r--vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_asm.go15
-rw-r--r--vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_other.go77
-rw-r--r--vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_safe.go11
8 files changed, 824 insertions, 0 deletions
diff --git a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/LICENSE.txt b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/LICENSE.txt
new file mode 100644
index 00000000..24b53065
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/LICENSE.txt
@@ -0,0 +1,22 @@
+Copyright (c) 2016 Caleb Spare
+
+MIT License
+
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sublicense, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
+LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
diff --git a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/README.md b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/README.md
new file mode 100644
index 00000000..69aa3bb5
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/README.md
@@ -0,0 +1,58 @@
+# xxhash
+
+VENDORED: Go to [github.com/cespare/xxhash](https://github.com/cespare/xxhash) for original package.
+
+
+[![GoDoc](https://godoc.org/github.com/cespare/xxhash?status.svg)](https://godoc.org/github.com/cespare/xxhash)
+[![Build Status](https://travis-ci.org/cespare/xxhash.svg?branch=master)](https://travis-ci.org/cespare/xxhash)
+
+xxhash is a Go implementation of the 64-bit
+[xxHash](http://cyan4973.github.io/xxHash/) algorithm, XXH64. This is a
+high-quality hashing algorithm that is much faster than anything in the Go
+standard library.
+
+This package provides a straightforward API:
+
+```
+func Sum64(b []byte) uint64
+func Sum64String(s string) uint64
+type Digest struct{ ... }
+ func New() *Digest
+```
+
+The `Digest` type implements hash.Hash64. Its key methods are:
+
+```
+func (*Digest) Write([]byte) (int, error)
+func (*Digest) WriteString(string) (int, error)
+func (*Digest) Sum64() uint64
+```
+
+This implementation provides a fast pure-Go implementation and an even faster
+assembly implementation for amd64.
+
+## Benchmarks
+
+Here are some quick benchmarks comparing the pure-Go and assembly
+implementations of Sum64.
+
+| input size | purego | asm |
+| --- | --- | --- |
+| 5 B | 979.66 MB/s | 1291.17 MB/s |
+| 100 B | 7475.26 MB/s | 7973.40 MB/s |
+| 4 KB | 17573.46 MB/s | 17602.65 MB/s |
+| 10 MB | 17131.46 MB/s | 17142.16 MB/s |
+
+These numbers were generated on Ubuntu 18.04 with an Intel i7-8700K CPU using
+the following commands under Go 1.11.2:
+
+```
+$ go test -tags purego -benchtime 10s -bench '/xxhash,direct,bytes'
+$ go test -benchtime 10s -bench '/xxhash,direct,bytes'
+```
+
+## Projects using this package
+
+- [InfluxDB](https://github.com/influxdata/influxdb)
+- [Prometheus](https://github.com/prometheus/prometheus)
+- [FreeCache](https://github.com/coocood/freecache)
diff --git a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash.go b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash.go
new file mode 100644
index 00000000..2c112a0a
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash.go
@@ -0,0 +1,237 @@
+// Package xxhash implements the 64-bit variant of xxHash (XXH64) as described
+// at http://cyan4973.github.io/xxHash/.
+// THIS IS VENDORED: Go to github.com/cespare/xxhash for original package.
+
+package xxhash
+
+import (
+ "encoding/binary"
+ "errors"
+ "math/bits"
+)
+
+const (
+ prime1 uint64 = 11400714785074694791
+ prime2 uint64 = 14029467366897019727
+ prime3 uint64 = 1609587929392839161
+ prime4 uint64 = 9650029242287828579
+ prime5 uint64 = 2870177450012600261
+)
+
+// NOTE(caleb): I'm using both consts and vars of the primes. Using consts where
+// possible in the Go code is worth a small (but measurable) performance boost
+// by avoiding some MOVQs. Vars are needed for the asm and also are useful for
+// convenience in the Go code in a few places where we need to intentionally
+// avoid constant arithmetic (e.g., v1 := prime1 + prime2 fails because the
+// result overflows a uint64).
+var (
+ prime1v = prime1
+ prime2v = prime2
+ prime3v = prime3
+ prime4v = prime4
+ prime5v = prime5
+)
+
+// Digest implements hash.Hash64.
+type Digest struct {
+ v1 uint64
+ v2 uint64
+ v3 uint64
+ v4 uint64
+ total uint64
+ mem [32]byte
+ n int // how much of mem is used
+}
+
+// New creates a new Digest that computes the 64-bit xxHash algorithm.
+func New() *Digest {
+ var d Digest
+ d.Reset()
+ return &d
+}
+
+// Reset clears the Digest's state so that it can be reused.
+func (d *Digest) Reset() {
+ d.v1 = prime1v + prime2
+ d.v2 = prime2
+ d.v3 = 0
+ d.v4 = -prime1v
+ d.total = 0
+ d.n = 0
+}
+
+// Size always returns 8 bytes.
+func (d *Digest) Size() int { return 8 }
+
+// BlockSize always returns 32 bytes.
+func (d *Digest) BlockSize() int { return 32 }
+
+// Write adds more data to d. It always returns len(b), nil.
+func (d *Digest) Write(b []byte) (n int, err error) {
+ n = len(b)
+ d.total += uint64(n)
+
+ if d.n+n < 32 {
+ // This new data doesn't even fill the current block.
+ copy(d.mem[d.n:], b)
+ d.n += n
+ return
+ }
+
+ if d.n > 0 {
+ // Finish off the partial block.
+ copy(d.mem[d.n:], b)
+ d.v1 = round(d.v1, u64(d.mem[0:8]))
+ d.v2 = round(d.v2, u64(d.mem[8:16]))
+ d.v3 = round(d.v3, u64(d.mem[16:24]))
+ d.v4 = round(d.v4, u64(d.mem[24:32]))
+ b = b[32-d.n:]
+ d.n = 0
+ }
+
+ if len(b) >= 32 {
+ // One or more full blocks left.
+ nw := writeBlocks(d, b)
+ b = b[nw:]
+ }
+
+ // Store any remaining partial block.
+ copy(d.mem[:], b)
+ d.n = len(b)
+
+ return
+}
+
+// Sum appends the current hash to b and returns the resulting slice.
+func (d *Digest) Sum(b []byte) []byte {
+ s := d.Sum64()
+ return append(
+ b,
+ byte(s>>56),
+ byte(s>>48),
+ byte(s>>40),
+ byte(s>>32),
+ byte(s>>24),
+ byte(s>>16),
+ byte(s>>8),
+ byte(s),
+ )
+}
+
+// Sum64 returns the current hash.
+func (d *Digest) Sum64() uint64 {
+ var h uint64
+
+ if d.total >= 32 {
+ v1, v2, v3, v4 := d.v1, d.v2, d.v3, d.v4
+ h = rol1(v1) + rol7(v2) + rol12(v3) + rol18(v4)
+ h = mergeRound(h, v1)
+ h = mergeRound(h, v2)
+ h = mergeRound(h, v3)
+ h = mergeRound(h, v4)
+ } else {
+ h = d.v3 + prime5
+ }
+
+ h += d.total
+
+ i, end := 0, d.n
+ for ; i+8 <= end; i += 8 {
+ k1 := round(0, u64(d.mem[i:i+8]))
+ h ^= k1
+ h = rol27(h)*prime1 + prime4
+ }
+ if i+4 <= end {
+ h ^= uint64(u32(d.mem[i:i+4])) * prime1
+ h = rol23(h)*prime2 + prime3
+ i += 4
+ }
+ for i < end {
+ h ^= uint64(d.mem[i]) * prime5
+ h = rol11(h) * prime1
+ i++
+ }
+
+ h ^= h >> 33
+ h *= prime2
+ h ^= h >> 29
+ h *= prime3
+ h ^= h >> 32
+
+ return h
+}
+
+const (
+ magic = "xxh\x06"
+ marshaledSize = len(magic) + 8*5 + 32
+)
+
+// MarshalBinary implements the encoding.BinaryMarshaler interface.
+func (d *Digest) MarshalBinary() ([]byte, error) {
+ b := make([]byte, 0, marshaledSize)
+ b = append(b, magic...)
+ b = appendUint64(b, d.v1)
+ b = appendUint64(b, d.v2)
+ b = appendUint64(b, d.v3)
+ b = appendUint64(b, d.v4)
+ b = appendUint64(b, d.total)
+ b = append(b, d.mem[:d.n]...)
+ b = b[:len(b)+len(d.mem)-d.n]
+ return b, nil
+}
+
+// UnmarshalBinary implements the encoding.BinaryUnmarshaler interface.
+func (d *Digest) UnmarshalBinary(b []byte) error {
+ if len(b) < len(magic) || string(b[:len(magic)]) != magic {
+ return errors.New("xxhash: invalid hash state identifier")
+ }
+ if len(b) != marshaledSize {
+ return errors.New("xxhash: invalid hash state size")
+ }
+ b = b[len(magic):]
+ b, d.v1 = consumeUint64(b)
+ b, d.v2 = consumeUint64(b)
+ b, d.v3 = consumeUint64(b)
+ b, d.v4 = consumeUint64(b)
+ b, d.total = consumeUint64(b)
+ copy(d.mem[:], b)
+ d.n = int(d.total % uint64(len(d.mem)))
+ return nil
+}
+
+func appendUint64(b []byte, x uint64) []byte {
+ var a [8]byte
+ binary.LittleEndian.PutUint64(a[:], x)
+ return append(b, a[:]...)
+}
+
+func consumeUint64(b []byte) ([]byte, uint64) {
+ x := u64(b)
+ return b[8:], x
+}
+
+func u64(b []byte) uint64 { return binary.LittleEndian.Uint64(b) }
+func u32(b []byte) uint32 { return binary.LittleEndian.Uint32(b) }
+
+func round(acc, input uint64) uint64 {
+ acc += input * prime2
+ acc = rol31(acc)
+ acc *= prime1
+ return acc
+}
+
+func mergeRound(acc, val uint64) uint64 {
+ val = round(0, val)
+ acc ^= val
+ acc = acc*prime1 + prime4
+ return acc
+}
+
+func rol1(x uint64) uint64 { return bits.RotateLeft64(x, 1) }
+func rol7(x uint64) uint64 { return bits.RotateLeft64(x, 7) }
+func rol11(x uint64) uint64 { return bits.RotateLeft64(x, 11) }
+func rol12(x uint64) uint64 { return bits.RotateLeft64(x, 12) }
+func rol18(x uint64) uint64 { return bits.RotateLeft64(x, 18) }
+func rol23(x uint64) uint64 { return bits.RotateLeft64(x, 23) }
+func rol27(x uint64) uint64 { return bits.RotateLeft64(x, 27) }
+func rol31(x uint64) uint64 { return bits.RotateLeft64(x, 31) }
diff --git a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_amd64.s b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_amd64.s
new file mode 100644
index 00000000..be8db5bf
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_amd64.s
@@ -0,0 +1,215 @@
+// +build !appengine
+// +build gc
+// +build !purego
+
+#include "textflag.h"
+
+// Register allocation:
+// AX h
+// SI pointer to advance through b
+// DX n
+// BX loop end
+// R8 v1, k1
+// R9 v2
+// R10 v3
+// R11 v4
+// R12 tmp
+// R13 prime1v
+// R14 prime2v
+// DI prime4v
+
+// round reads from and advances the buffer pointer in SI.
+// It assumes that R13 has prime1v and R14 has prime2v.
+#define round(r) \
+ MOVQ (SI), R12 \
+ ADDQ $8, SI \
+ IMULQ R14, R12 \
+ ADDQ R12, r \
+ ROLQ $31, r \
+ IMULQ R13, r
+
+// mergeRound applies a merge round on the two registers acc and val.
+// It assumes that R13 has prime1v, R14 has prime2v, and DI has prime4v.
+#define mergeRound(acc, val) \
+ IMULQ R14, val \
+ ROLQ $31, val \
+ IMULQ R13, val \
+ XORQ val, acc \
+ IMULQ R13, acc \
+ ADDQ DI, acc
+
+// func Sum64(b []byte) uint64
+TEXT ·Sum64(SB), NOSPLIT, $0-32
+ // Load fixed primes.
+ MOVQ ·prime1v(SB), R13
+ MOVQ ·prime2v(SB), R14
+ MOVQ ·prime4v(SB), DI
+
+ // Load slice.
+ MOVQ b_base+0(FP), SI
+ MOVQ b_len+8(FP), DX
+ LEAQ (SI)(DX*1), BX
+
+ // The first loop limit will be len(b)-32.
+ SUBQ $32, BX
+
+ // Check whether we have at least one block.
+ CMPQ DX, $32
+ JLT noBlocks
+
+ // Set up initial state (v1, v2, v3, v4).
+ MOVQ R13, R8
+ ADDQ R14, R8
+ MOVQ R14, R9
+ XORQ R10, R10
+ XORQ R11, R11
+ SUBQ R13, R11
+
+ // Loop until SI > BX.
+blockLoop:
+ round(R8)
+ round(R9)
+ round(R10)
+ round(R11)
+
+ CMPQ SI, BX
+ JLE blockLoop
+
+ MOVQ R8, AX
+ ROLQ $1, AX
+ MOVQ R9, R12
+ ROLQ $7, R12
+ ADDQ R12, AX
+ MOVQ R10, R12
+ ROLQ $12, R12
+ ADDQ R12, AX
+ MOVQ R11, R12
+ ROLQ $18, R12
+ ADDQ R12, AX
+
+ mergeRound(AX, R8)
+ mergeRound(AX, R9)
+ mergeRound(AX, R10)
+ mergeRound(AX, R11)
+
+ JMP afterBlocks
+
+noBlocks:
+ MOVQ ·prime5v(SB), AX
+
+afterBlocks:
+ ADDQ DX, AX
+
+ // Right now BX has len(b)-32, and we want to loop until SI > len(b)-8.
+ ADDQ $24, BX
+
+ CMPQ SI, BX
+ JG fourByte
+
+wordLoop:
+ // Calculate k1.
+ MOVQ (SI), R8
+ ADDQ $8, SI
+ IMULQ R14, R8
+ ROLQ $31, R8
+ IMULQ R13, R8
+
+ XORQ R8, AX
+ ROLQ $27, AX
+ IMULQ R13, AX
+ ADDQ DI, AX
+
+ CMPQ SI, BX
+ JLE wordLoop
+
+fourByte:
+ ADDQ $4, BX
+ CMPQ SI, BX
+ JG singles
+
+ MOVL (SI), R8
+ ADDQ $4, SI
+ IMULQ R13, R8
+ XORQ R8, AX
+
+ ROLQ $23, AX
+ IMULQ R14, AX
+ ADDQ ·prime3v(SB), AX
+
+singles:
+ ADDQ $4, BX
+ CMPQ SI, BX
+ JGE finalize
+
+singlesLoop:
+ MOVBQZX (SI), R12
+ ADDQ $1, SI
+ IMULQ ·prime5v(SB), R12
+ XORQ R12, AX
+
+ ROLQ $11, AX
+ IMULQ R13, AX
+
+ CMPQ SI, BX
+ JL singlesLoop
+
+finalize:
+ MOVQ AX, R12
+ SHRQ $33, R12
+ XORQ R12, AX
+ IMULQ R14, AX
+ MOVQ AX, R12
+ SHRQ $29, R12
+ XORQ R12, AX
+ IMULQ ·prime3v(SB), AX
+ MOVQ AX, R12
+ SHRQ $32, R12
+ XORQ R12, AX
+
+ MOVQ AX, ret+24(FP)
+ RET
+
+// writeBlocks uses the same registers as above except that it uses AX to store
+// the d pointer.
+
+// func writeBlocks(d *Digest, b []byte) int
+TEXT ·writeBlocks(SB), NOSPLIT, $0-40
+ // Load fixed primes needed for round.
+ MOVQ ·prime1v(SB), R13
+ MOVQ ·prime2v(SB), R14
+
+ // Load slice.
+ MOVQ b_base+8(FP), SI
+ MOVQ b_len+16(FP), DX
+ LEAQ (SI)(DX*1), BX
+ SUBQ $32, BX
+
+ // Load vN from d.
+ MOVQ d+0(FP), AX
+ MOVQ 0(AX), R8 // v1
+ MOVQ 8(AX), R9 // v2
+ MOVQ 16(AX), R10 // v3
+ MOVQ 24(AX), R11 // v4
+
+ // We don't need to check the loop condition here; this function is
+ // always called with at least one block of data to process.
+blockLoop:
+ round(R8)
+ round(R9)
+ round(R10)
+ round(R11)
+
+ CMPQ SI, BX
+ JLE blockLoop
+
+ // Copy vN back to d.
+ MOVQ R8, 0(AX)
+ MOVQ R9, 8(AX)
+ MOVQ R10, 16(AX)
+ MOVQ R11, 24(AX)
+
+ // The number of bytes written is SI minus the old base pointer.
+ SUBQ b_base+8(FP), SI
+ MOVQ SI, ret+32(FP)
+
+ RET
diff --git a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_arm64.s b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_arm64.s
new file mode 100644
index 00000000..66260958
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/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
diff --git a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_asm.go b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_asm.go
new file mode 100644
index 00000000..9216e0a4
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_asm.go
@@ -0,0 +1,15 @@
+//go:build (amd64 || arm64) && !appengine && gc && !purego
+// +build amd64 arm64
+// +build !appengine
+// +build gc
+// +build !purego
+
+package xxhash
+
+// Sum64 computes the 64-bit xxHash digest of b.
+//
+//go:noescape
+func Sum64(b []byte) uint64
+
+//go:noescape
+func writeBlocks(d *Digest, b []byte) int
diff --git a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_other.go b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_other.go
new file mode 100644
index 00000000..2deb1ca7
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_other.go
@@ -0,0 +1,77 @@
+//go:build (!amd64 && !arm64) || appengine || !gc || purego
+// +build !amd64,!arm64 appengine !gc purego
+
+package xxhash
+
+// Sum64 computes the 64-bit xxHash digest of b.
+func Sum64(b []byte) uint64 {
+ // A simpler version would be
+ // d := New()
+ // d.Write(b)
+ // return d.Sum64()
+ // but this is faster, particularly for small inputs.
+
+ n := len(b)
+ var h uint64
+
+ if n >= 32 {
+ v1 := prime1v + prime2
+ v2 := prime2
+ v3 := uint64(0)
+ v4 := -prime1v
+ for len(b) >= 32 {
+ v1 = round(v1, u64(b[0:8:len(b)]))
+ v2 = round(v2, u64(b[8:16:len(b)]))
+ v3 = round(v3, u64(b[16:24:len(b)]))
+ v4 = round(v4, u64(b[24:32:len(b)]))
+ b = b[32:len(b):len(b)]
+ }
+ h = rol1(v1) + rol7(v2) + rol12(v3) + rol18(v4)
+ h = mergeRound(h, v1)
+ h = mergeRound(h, v2)
+ h = mergeRound(h, v3)
+ h = mergeRound(h, v4)
+ } else {
+ h = prime5
+ }
+
+ h += uint64(n)
+
+ i, end := 0, len(b)
+ for ; i+8 <= end; i += 8 {
+ k1 := round(0, u64(b[i:i+8:len(b)]))
+ h ^= k1
+ h = rol27(h)*prime1 + prime4
+ }
+ if i+4 <= end {
+ h ^= uint64(u32(b[i:i+4:len(b)])) * prime1
+ h = rol23(h)*prime2 + prime3
+ i += 4
+ }
+ for ; i < end; i++ {
+ h ^= uint64(b[i]) * prime5
+ h = rol11(h) * prime1
+ }
+
+ h ^= h >> 33
+ h *= prime2
+ h ^= h >> 29
+ h *= prime3
+ h ^= h >> 32
+
+ return h
+}
+
+func writeBlocks(d *Digest, b []byte) int {
+ v1, v2, v3, v4 := d.v1, d.v2, d.v3, d.v4
+ n := len(b)
+ for len(b) >= 32 {
+ v1 = round(v1, u64(b[0:8:len(b)]))
+ v2 = round(v2, u64(b[8:16:len(b)]))
+ v3 = round(v3, u64(b[16:24:len(b)]))
+ v4 = round(v4, u64(b[24:32:len(b)]))
+ b = b[32:len(b):len(b)]
+ }
+ d.v1, d.v2, d.v3, d.v4 = v1, v2, v3, v4
+ return n - len(b)
+}
diff --git a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_safe.go b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_safe.go
new file mode 100644
index 00000000..6f3b0cb1
--- /dev/null
+++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_safe.go
@@ -0,0 +1,11 @@
+package xxhash
+
+// Sum64String computes the 64-bit xxHash digest of s.
+func Sum64String(s string) uint64 {
+ return Sum64([]byte(s))
+}
+
+// WriteString adds more data to d. It always returns len(s), nil.
+func (d *Digest) WriteString(s string) (n int, err error) {
+ return d.Write([]byte(s))
+}