diff --git a/go.mod b/go.mod index 35ec2a5627..d2b8853321 100644 --- a/go.mod +++ b/go.mod @@ -20,8 +20,8 @@ require ( github.com/openshift/client-go v0.0.0-20260723174158-ae2315de9d73 github.com/operator-framework/api v0.46.0 github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.94.1 - github.com/prometheus/client_golang v1.24.1 - github.com/prometheus/client_model v0.6.2 + github.com/prometheus/client_golang v1.25.0 + github.com/prometheus/client_model v0.6.3 github.com/regclient/regclient v0.11.5 github.com/sirupsen/logrus v1.10.2 github.com/stretchr/testify v1.12.1 @@ -79,7 +79,7 @@ require ( github.com/huandu/xstrings v1.5.0 // indirect github.com/inconshreveable/mousetrap v1.1.0 // indirect github.com/json-iterator/go v1.1.12 // indirect - github.com/klauspost/compress v1.19.1 // indirect + github.com/klauspost/compress v1.20.1 // indirect github.com/liggitt/tabwriter v0.0.0-20181228230101-89fcab3d43de // indirect github.com/mitchellh/copystructure v1.2.0 // indirect github.com/mitchellh/go-wordwrap v1.0.1 // indirect @@ -95,8 +95,8 @@ require ( github.com/opencontainers/go-digest v1.0.0 // indirect github.com/peterbourgon/diskv v2.0.1+incompatible // indirect github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect - github.com/prometheus/common v0.70.1 // indirect - github.com/prometheus/procfs v0.21.1 // indirect + github.com/prometheus/common v0.72.0 // indirect + github.com/prometheus/procfs v0.22.0 // indirect github.com/russross/blackfriday/v2 v2.1.0 // indirect github.com/shopspring/decimal v1.4.0 // indirect github.com/spf13/cast v1.7.0 // indirect diff --git a/go.sum b/go.sum index dcca81fa76..7cc01a9eb0 100644 --- a/go.sum +++ b/go.sum @@ -130,8 +130,8 @@ github.com/joshdk/go-junit v1.0.0 h1:S86cUKIdwBHWwA6xCmFlf3RTLfVXYQfvanM5Uh+K6GE github.com/joshdk/go-junit v1.0.0/go.mod h1:TiiV0PqkaNfFXjEiyjWM3XXrhVyCa1K4Zfga6W52ung= github.com/json-iterator/go v1.1.12 h1:PV8peI4a0ysnczrg+LtxykD8LfKY9ML6u2jnxaEnrnM= github.com/json-iterator/go v1.1.12/go.mod h1:e30LSqwooZae/UwlEbR2852Gd8hjQvJoHmT4TnhNGBo= -github.com/klauspost/compress v1.19.1 h1:VsB4HPswih7mmZ8WleSFQ75c/Ui1M4trX5oAsJnhSlk= -github.com/klauspost/compress v1.19.1/go.mod h1:cwPg85FWrGar70rWktvGQj8/hthj3wpl0PGDogxkrSQ= +github.com/klauspost/compress v1.20.1 h1:T7kKElXUMXrUJ2E9QhQhxFtcK5rPyLdsGZvdbLMPdiQ= +github.com/klauspost/compress v1.20.1/go.mod h1:LUdAzn7YLVvxLpc7y3V1m40wESHTgc1422pwwBSKYuI= github.com/kr/pretty v0.3.1 h1:flRD4NNwYAUpkphVc1HcthR4KEIFJ65n8Mw5qdRn3LE= github.com/kr/pretty v0.3.1/go.mod h1:hoEshYVHaxMs3cyo3Yncou5ZscifuDolrwPKZanG3xk= github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY= @@ -189,14 +189,14 @@ github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 h1:Jamvg5psRI github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4= github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.94.1 h1:TWZqERWd1r20mHm6rCahhWuCaDq4PQFHgzQifOpFsc0= github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.94.1/go.mod h1:n4uH73YCwqQomLLIKEthHFwh4F9PneNaP6eaIIscpVg= -github.com/prometheus/client_golang v1.24.1 h1:JnJkREXzWxUdCuPFpIWZiPispT9xVV59uiuyR2bPlnU= -github.com/prometheus/client_golang v1.24.1/go.mod h1:F+oSRECHg4sse5ucfYpYDeIv/hu68Zo0uoHKetWnzcE= -github.com/prometheus/client_model v0.6.2 h1:oBsgwpGs7iVziMvrGhE53c/GrLUsZdHnqNwqPLxwZyk= -github.com/prometheus/client_model v0.6.2/go.mod h1:y3m2F6Gdpfy6Ut/GBsUqTWZqCUvMVzSfMLjcu6wAwpE= -github.com/prometheus/common v0.70.1 h1:1HvjP4D5oL3t8RsPlwxA9onvvStjtIHYE5XuuwOi/PY= -github.com/prometheus/common v0.70.1/go.mod h1:VdFUQDMZK3VLkurFUVhia6uys/0suUp86TJz5qbJRhc= -github.com/prometheus/procfs v0.21.1 h1:GljZCt+zSTS+NZq88cyQ1LjZ+RCHp3uVuabBWA5+OJI= -github.com/prometheus/procfs v0.21.1/go.mod h1:aB55Cww9pdSJVHk0hUf0inxWyyjPogFIjmHKYgMKmtY= +github.com/prometheus/client_golang v1.25.0 h1:cDtW1TcFsMi1RJc8khwCKB4VhzwV04aKqbYtNOXuiK0= +github.com/prometheus/client_golang v1.25.0/go.mod h1:j2X/4/kaQHLPnmyEU8zjmkndo1T317VpCIQEml8LTKk= +github.com/prometheus/client_model v0.6.3 h1:O0jaTVAYNxTHYInEPFJt5I3+sN8zqBtVMPTB1qyxiEo= +github.com/prometheus/client_model v0.6.3/go.mod h1:gpN5P9S7Rr6Yr92PiQ+Ixvhf6JZEkF1dnxsYL2aPBEM= +github.com/prometheus/common v0.72.0 h1:tAYsE+sPJxIncDAobm4H5aQjmox9ZxEIIqPbiffa8G4= +github.com/prometheus/common v0.72.0/go.mod h1:77NWqAQ2tXT7BIK40qjJdw5Acrsrg1TlHAnsQi3i6mk= +github.com/prometheus/procfs v0.22.0 h1:6q9+/JL9IKAPbCmBrv9n5O5Ty3NKnciV5X7YGw0oics= +github.com/prometheus/procfs v0.22.0/go.mod h1:CvmFr/GVhIjIvWJZW3tgkODBQMRIf0EyWMQLHCHab58= github.com/regclient/regclient v0.11.5 h1:OHRsXO0F3qHGfa4HEUv+EkMH9NXNcCTBKjNzyC/UhIA= github.com/regclient/regclient v0.11.5/go.mod h1:DZUOfIT14WFTK2Pj4vjd93avy9O4Fdpjrf9ir23TbRE= github.com/rogpeppe/go-internal v1.14.1 h1:UQB4HGPB6osV0SQTLymcB4TgvyWu6ZyliaW0tI/otEQ= diff --git a/vendor/github.com/klauspost/compress/README.md b/vendor/github.com/klauspost/compress/README.md index 0e9f170d01..bf31b905f8 100644 --- a/vendor/github.com/klauspost/compress/README.md +++ b/vendor/github.com/klauspost/compress/README.md @@ -6,7 +6,9 @@ This package provides various compression algorithms. * [S2](https://github.com/klauspost/compress/tree/master/s2#s2-compression) is a high performance replacement for Snappy. * Optimized [deflate](https://godoc.org/github.com/klauspost/compress/flate) packages which can be used as a dropin replacement for [gzip](https://godoc.org/github.com/klauspost/compress/gzip), [zip](https://godoc.org/github.com/klauspost/compress/zip) and [zlib](https://godoc.org/github.com/klauspost/compress/zlib). * [snappy](https://github.com/klauspost/compress/tree/master/snappy) is a drop-in replacement for `github.com/golang/snappy` offering better compression and concurrent streams. +* [lzw](https://github.com/klauspost/compress/tree/master/lzw) is a drop-in replacement for `compress/lzw` with 1.4-4x faster decompression and 1.1-2.7x faster compression, depending on the data. * [huff0](https://github.com/klauspost/compress/tree/master/huff0) and [FSE](https://github.com/klauspost/compress/tree/master/fse) implementations for raw entropy encoding. +* [Xpress](https://github.com/klauspost/compress/tree/master/xpress) decompression of the Microsoft XPRESS (MS-XCA) plain LZ77 and LZ77+Huffman formats (the LZ77+Huffman variant is the one used in WIM images and Windows Compact OS / WOF data). * [gzhttp](https://github.com/klauspost/compress/tree/master/gzhttp) Provides client and server wrappers for handling gzipped/zstd HTTP requests efficiently. * [pgzip](https://github.com/klauspost/pgzip) is a separate package that provides a very fast parallel gzip implementation. diff --git a/vendor/github.com/klauspost/compress/huff0/bitreader.go b/vendor/github.com/klauspost/compress/huff0/bitreader.go index bfc7a523de..faaa418a75 100644 --- a/vendor/github.com/klauspost/compress/huff0/bitreader.go +++ b/vendor/github.com/klauspost/compress/huff0/bitreader.go @@ -9,6 +9,7 @@ import ( "errors" "fmt" "io" + "math/bits" "github.com/klauspost/compress/internal/le" ) @@ -210,6 +211,68 @@ func (b *bitReaderShifted) remaining() uint { return b.off*8 + uint(64-b.bitsRead) } +// canUseAsm reports whether the reader has a full 8-byte window ahead of +// its read pointer, which the Decompress4X asm loops need to take over. +func (b *bitReaderShifted) canUseAsm() bool { + return b.off >= 8 +} + +// prepareForAsm establishes the invariant the Decompress4X asm loops rely +// on: value == load64(in[off:off+8]) << bitsRead with bitsRead <= 7, so +// that a full group of symbols can never shift their sentinel bit out of +// the container. init leaves bitsRead == 8 when the final byte of the +// stream is exactly 0x01; that whole byte is consumed, so the same position +// is the window one byte lower with nothing consumed. Requires canUseAsm. +func (b *bitReaderShifted) prepareForAsm() { + if b.bitsRead >= 8 { + b.off-- + b.value = le.Load64(b.in, b.off) + b.bitsRead -= 8 + } +} + +// restoreFromAsm converts the state left behind by the Decompress4X asm +// loops back to the invariant the Go code relies on: value holds the 8 +// bytes at in[off:off+8] shifted left by bitsRead, with the low bitsRead +// bits zero. +// +// The asm keeps a sentinel bit in value just below the unread bits, so its +// trailing zero count is the number of consumed bits. The sentinel is ORed +// over the lowest bit of the window, which the loop never consumes before +// re-reading memory, so the window is re-read here too rather than taken +// from value. The asm reports off as the signed distance from the start of +// the stream, and it can be negative: a reload always reads a whole 8-byte +// window, so a stream that is nearly drained ends with up to 7 bytes of the +// previous stream (or the jump table) below its start inside the window. +// Those bytes sit below the stream's own bits and count as consumed. +// Anything further below, or more bits consumed than the stream holds, is +// corruption. +func (b *bitReaderShifted) restoreFromAsm() error { + off := int(b.off) + consumed := uint(bits.TrailingZeros64(b.value)) + if off < 0 { + if off < -7 { + return errors.New("corruption detected: stream underrun") + } + consumed += uint(-off) * 8 + if consumed > 64 { + return errors.New("corruption detected: stream underrun") + } + off = 0 + } + if off+8 > len(b.in) { + return errors.New("corruption detected: stream overrun") + } + b.off = uint(off) + b.bitsRead = uint8(consumed) + if consumed >= 64 { + b.value = 0 + } else { + b.value = le.Load64(b.in, b.off) << consumed + } + return nil +} + // close the bitstream and returns an error if out-of-buffer reads occurred. func (b *bitReaderShifted) close() error { // Release reference. diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.go b/vendor/github.com/klauspost/compress/huff0/decompress_amd64.go index 2d6ef64be1..e651f30ecb 100644 --- a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.go +++ b/vendor/github.com/klauspost/compress/huff0/decompress_amd64.go @@ -1,222 +1,128 @@ //go:build amd64 && !appengine && !noasm && gc -// This file contains the specialisation of Decoder.Decompress4X -// and Decoder.Decompress1X that use an asm implementation of thir main loops. +// amd64 stubs and dispatch for the asm loops used by decompress_asm.go. package huff0 import ( - "errors" - "fmt" - "github.com/klauspost/compress/internal/cpuinfo" ) -// decompress4x_main_loop_x86 is an x86 assembler implementation -// of Decompress4X when tablelog > 8. +// decompress4x_main_loop_amd64 is an x86 assembler implementation +// of Decompress4X when tablelog > 8, decoding fastSymbols symbols per +// stream between bit container reloads. // //go:noescape func decompress4x_main_loop_amd64(ctx *decompress4xContext) -// decompress4x_8b_loop_x86 is an x86 assembler implementation -// of Decompress4X when tablelog <= 8 which decodes 4 entries -// per loop. +// decompress4x_8b_main_loop_amd64 is an x86 assembler implementation +// of Decompress4X when tablelog <= 8, decoding fast8bSymbols symbols +// per stream between bit container reloads. // //go:noescape func decompress4x_8b_main_loop_amd64(ctx *decompress4xContext) -// fallback8BitSize is the size where using Go version is faster. -const fallback8BitSize = 800 - -type decompress4xContext struct { - pbr *[4]bitReaderShifted - peekBits uint8 - out *byte - dstEvery int - tbl *dEntrySingle - decoded int - limit *byte -} +// decompress4x_4b_main_loop_amd64 is an x86 assembler implementation +// of Decompress4X when tablelog <= 4, decoding fast4bSymbols symbols +// per stream between bit container reloads. +// +//go:noescape +func decompress4x_4b_main_loop_amd64(ctx *decompress4xContext) -// Decompress4X will decompress a 4X encoded stream. -// The length of the supplied input must match the end of a block exactly. -// The *capacity* of the dst slice must match the destination size of -// the uncompressed data exactly. -func (d *Decoder) Decompress4X(dst, src []byte) ([]byte, error) { - if len(d.dt.single) == 0 { - return nil, errors.New("no table loaded") - } - if len(src) < 6+(4*1) { - return nil, errors.New("input too small") - } +// decompress4x_4b_main_loop_bmi2 is the BMI2 twin of decompress4x_4b_main_loop_amd64. +// +//go:noescape +func decompress4x_4b_main_loop_bmi2(ctx *decompress4xContext) - use8BitTables := d.actualTableLog <= 8 - if cap(dst) < fallback8BitSize && use8BitTables { - return d.decompress4X8bit(dst, src) - } +// decompress4x_main_loop_bmi2 is the BMI2 twin of decompress4x_main_loop_amd64. +// +//go:noescape +func decompress4x_main_loop_bmi2(ctx *decompress4xContext) - var br [4]bitReaderShifted - // Decode "jump table" - start := 6 - for i := range 3 { - length := int(src[i*2]) | (int(src[i*2+1]) << 8) - if start+length >= len(src) { - return nil, errors.New("truncated input (or invalid offset)") - } - err := br[i].init(src[start : start+length]) - if err != nil { - return nil, err - } - start += length - } - err := br[3].init(src[start:]) - if err != nil { - return nil, err - } +// decompress4x_8b_main_loop_bmi2 is the BMI2 twin of decompress4x_8b_main_loop_amd64. +// +//go:noescape +func decompress4x_8b_main_loop_bmi2(ctx *decompress4xContext) - // destination, offset to match first output - dstSize := cap(dst) - dst = dst[:dstSize] - out := dst - dstEvery := (dstSize + 3) / 4 - - const tlSize = 1 << tableLogMax - const tlMask = tlSize - 1 - single := d.dt.single[:tlSize] - - var decoded int - - if len(out) > 4*4 && !(br[0].off < 4 || br[1].off < 4 || br[2].off < 4 || br[3].off < 4) { - ctx := decompress4xContext{ - pbr: &br, - peekBits: uint8((64 - d.actualTableLog) & 63), // see: bitReaderShifted.peekBitsFast() - out: &out[0], - dstEvery: dstEvery, - tbl: &single[0], - limit: &out[dstEvery-4], // Always stop decoding when first buffer gets here to avoid writing OOB on last. - } - if use8BitTables { - decompress4x_8b_main_loop_amd64(&ctx) - } else { - decompress4x_main_loop_amd64(&ctx) - } - - decoded = ctx.decoded - out = out[decoded/4:] - } +// decompress1x_main_loop_amd64 is an x86 assembler implementation +// of Decompress1X when tablelog > 8, decoding fastSymbols symbols +// between bit container reloads. +// +//go:noescape +func decompress1x_main_loop_amd64(ctx *decompress1xContext) - // Decode remaining. - remainBytes := dstEvery - (decoded / 4) - for i := range br { - offset := dstEvery * i - endsAt := min(offset+remainBytes, len(out)) - br := &br[i] - bitsLeft := br.remaining() - for bitsLeft > 0 { - br.fill() - if offset >= endsAt { - return nil, errors.New("corruption detected: stream overrun 4") - } - - // Read value and increment offset. - val := br.peekBitsFast(d.actualTableLog) - v := single[val&tlMask].entry - nBits := uint8(v) - br.advance(nBits) - bitsLeft -= uint(nBits) - out[offset] = uint8(v >> 8) - offset++ - } - if offset != endsAt { - return nil, fmt.Errorf("corruption detected: short output block %d, end %d != %d", i, offset, endsAt) - } - decoded += offset - dstEvery*i - err = br.close() - if err != nil { - return nil, err - } - } - if dstSize != decoded { - return nil, errors.New("corruption detected: short output block") - } - return dst, nil -} +// decompress1x_8b_main_loop_amd64 is an x86 assembler implementation +// of Decompress1X when tablelog <= 8, decoding fast8bSymbols symbols +// between bit container reloads. +// +//go:noescape +func decompress1x_8b_main_loop_amd64(ctx *decompress1xContext) -// decompress4x_main_loop_x86 is an x86 assembler implementation -// of Decompress1X when tablelog > 8. +// decompress1x_4b_main_loop_amd64 is an x86 assembler implementation +// of Decompress1X when tablelog <= 4, decoding fast4bSymbols symbols +// between bit container reloads. // //go:noescape -func decompress1x_main_loop_amd64(ctx *decompress1xContext) +func decompress1x_4b_main_loop_amd64(ctx *decompress1xContext) -// decompress4x_main_loop_x86 is an x86 with BMI2 assembler implementation -// of Decompress1X when tablelog > 8. +// decompress1x_main_loop_bmi2 is the BMI2 twin of decompress1x_main_loop_amd64. // //go:noescape func decompress1x_main_loop_bmi2(ctx *decompress1xContext) -type decompress1xContext struct { - pbr *bitReaderShifted - peekBits uint8 - out *byte - outCap int - tbl *dEntrySingle - decoded int +// decompress1x_8b_main_loop_bmi2 is the BMI2 twin of decompress1x_8b_main_loop_amd64. +// +//go:noescape +func decompress1x_8b_main_loop_bmi2(ctx *decompress1xContext) + +// decompress1x_4b_main_loop_bmi2 is the BMI2 twin of decompress1x_4b_main_loop_amd64. +// +//go:noescape +func decompress1x_4b_main_loop_bmi2(ctx *decompress1xContext) + +func decompress4x_main_loop_asm(ctx *decompress4xContext) { + if cpuinfo.HasBMI2() { + decompress4x_main_loop_bmi2(ctx) + } else { + decompress4x_main_loop_amd64(ctx) + } } -// Error reported by asm implementations -const error_max_decoded_size_exeeded = -1 +func decompress4x_8b_main_loop_asm(ctx *decompress4xContext) { + if cpuinfo.HasBMI2() { + decompress4x_8b_main_loop_bmi2(ctx) + } else { + decompress4x_8b_main_loop_amd64(ctx) + } +} -// Decompress1X will decompress a 1X encoded stream. -// The cap of the output buffer will be the maximum decompressed size. -// The length of the supplied input must match the end of a block exactly. -func (d *Decoder) Decompress1X(dst, src []byte) ([]byte, error) { - if len(d.dt.single) == 0 { - return nil, errors.New("no table loaded") +func decompress4x_4b_main_loop_asm(ctx *decompress4xContext) { + if cpuinfo.HasBMI2() { + decompress4x_4b_main_loop_bmi2(ctx) + } else { + decompress4x_4b_main_loop_amd64(ctx) } - var br bitReaderShifted - err := br.init(src) - if err != nil { - return dst, err +} + +func decompress1x_main_loop_asm(ctx *decompress1xContext) { + if cpuinfo.HasBMI2() { + decompress1x_main_loop_bmi2(ctx) + } else { + decompress1x_main_loop_amd64(ctx) } - maxDecodedSize := cap(dst) - dst = dst[:maxDecodedSize] - - const tlSize = 1 << tableLogMax - const tlMask = tlSize - 1 - - if maxDecodedSize >= 4 { - ctx := decompress1xContext{ - pbr: &br, - out: &dst[0], - outCap: maxDecodedSize, - peekBits: uint8((64 - d.actualTableLog) & 63), // see: bitReaderShifted.peekBitsFast() - tbl: &d.dt.single[0], - } - - if cpuinfo.HasBMI2() { - decompress1x_main_loop_bmi2(&ctx) - } else { - decompress1x_main_loop_amd64(&ctx) - } - if ctx.decoded == error_max_decoded_size_exeeded { - return nil, ErrMaxDecodedSizeExceeded - } - - dst = dst[:ctx.decoded] +} + +func decompress1x_8b_main_loop_asm(ctx *decompress1xContext) { + if cpuinfo.HasBMI2() { + decompress1x_8b_main_loop_bmi2(ctx) + } else { + decompress1x_8b_main_loop_amd64(ctx) } +} - // br < 8, so uint8 is fine - bitsLeft := uint8(br.off)*8 + 64 - br.bitsRead - for bitsLeft > 0 { - br.fill() - if len(dst) >= maxDecodedSize { - br.close() - return nil, ErrMaxDecodedSizeExceeded - } - v := d.dt.single[br.peekBitsFast(d.actualTableLog)&tlMask] - nBits := uint8(v.entry) - br.advance(nBits) - bitsLeft -= nBits - dst = append(dst, uint8(v.entry>>8)) +func decompress1x_4b_main_loop_asm(ctx *decompress1xContext) { + if cpuinfo.HasBMI2() { + decompress1x_4b_main_loop_bmi2(ctx) + } else { + decompress1x_4b_main_loop_amd64(ctx) } - return dst, br.close() } diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s b/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s index c4c7ab2d1f..f072c18f04 100644 --- a/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s +++ b/vendor/github.com/klauspost/compress/huff0/decompress_amd64.s @@ -1,830 +1,3386 @@ -// Code generated by command: go run gen.go -out ../decompress_amd64.s -pkg=huff0. DO NOT EDIT. +// Code generated by command: go run gen.go -out ../decompress.s -arch amd64,arm64 -pkg=huff0. DO NOT EDIT. -//go:build amd64 && !appengine && !noasm && gc +//go:build !appengine && !noasm && gc // func decompress4x_main_loop_amd64(ctx *decompress4xContext) +// Requires: BMI, CMOV TEXT ·decompress4x_main_loop_amd64(SB), $0-8 - // Preload values - MOVQ ctx+0(FP), AX - MOVBQZX 8(AX), DI - MOVQ 16(AX), BX - MOVQ 48(AX), SI - MOVQ 24(AX), R8 - MOVQ 32(AX), R9 - MOVQ (AX), R10 - - // Main loop -main_loop: - XORL DX, DX - CMPQ BX, SI - SETGE DL - - // br0.fillFast32() - MOVQ 32(R10), R11 - MOVBQZX 40(R10), R12 - CMPQ R12, $0x20 - JBE skip_fill0 - MOVQ 24(R10), AX - SUBQ $0x20, R12 - SUBQ $0x04, AX - MOVQ (R10), R13 - - // b.value |= uint64(low) << (b.bitsRead & 63) - MOVL (AX)(R13*1), R13 - MOVQ R12, CX - SHLQ CL, R13 - MOVQ AX, 24(R10) - ORQ R13, R11 - - // exhausted += (br0.off < 4) - CMPQ AX, $0x04 - ADCB $+0, DL - -skip_fill0: - // val0 := br0.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v0 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br0.advance(uint8(v0.entry) - MOVB CH, AL - SHLQ CL, R11 - ADDB CL, R12 - - // val1 := br0.peekTopBits(peekBits) - MOVQ DI, CX - MOVQ R11, R13 - SHRQ CL, R13 - - // v1 := table[val1&mask] - MOVW (R9)(R13*2), CX - - // br0.advance(uint8(v1.entry)) - MOVB CH, AH - SHLQ CL, R11 - ADDB CL, R12 - - // these two writes get coalesced - // out[id * dstEvery + 0] = uint8(v0.entry >> 8) - // out[id * dstEvery + 1] = uint8(v1.entry >> 8) - MOVW AX, (BX) - - // update the bitreader structure - MOVQ R11, 32(R10) - MOVB R12, 40(R10) - - // br1.fillFast32() - MOVQ 80(R10), R11 - MOVBQZX 88(R10), R12 - CMPQ R12, $0x20 - JBE skip_fill1 - MOVQ 72(R10), AX - SUBQ $0x20, R12 - SUBQ $0x04, AX - MOVQ 48(R10), R13 - - // b.value |= uint64(low) << (b.bitsRead & 63) - MOVL (AX)(R13*1), R13 - MOVQ R12, CX - SHLQ CL, R13 - MOVQ AX, 72(R10) - ORQ R13, R11 - - // exhausted += (br1.off < 4) - CMPQ AX, $0x04 - ADCB $+0, DL - -skip_fill1: - // val0 := br1.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v0 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br1.advance(uint8(v0.entry) - MOVB CH, AL - SHLQ CL, R11 - ADDB CL, R12 - - // val1 := br1.peekTopBits(peekBits) - MOVQ DI, CX - MOVQ R11, R13 - SHRQ CL, R13 - - // v1 := table[val1&mask] - MOVW (R9)(R13*2), CX - - // br1.advance(uint8(v1.entry)) - MOVB CH, AH - SHLQ CL, R11 - ADDB CL, R12 - - // these two writes get coalesced - // out[id * dstEvery + 0] = uint8(v0.entry >> 8) - // out[id * dstEvery + 1] = uint8(v1.entry >> 8) - MOVW AX, (BX)(R8*1) - - // update the bitreader structure - MOVQ R11, 80(R10) - MOVB R12, 88(R10) - - // br2.fillFast32() - MOVQ 128(R10), R11 - MOVBQZX 136(R10), R12 - CMPQ R12, $0x20 - JBE skip_fill2 - MOVQ 120(R10), AX - SUBQ $0x20, R12 - SUBQ $0x04, AX - MOVQ 96(R10), R13 - - // b.value |= uint64(low) << (b.bitsRead & 63) - MOVL (AX)(R13*1), R13 - MOVQ R12, CX - SHLQ CL, R13 - MOVQ AX, 120(R10) - ORQ R13, R11 - - // exhausted += (br2.off < 4) - CMPQ AX, $0x04 - ADCB $+0, DL - -skip_fill2: - // val0 := br2.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v0 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br2.advance(uint8(v0.entry) - MOVB CH, AL - SHLQ CL, R11 - ADDB CL, R12 - - // val1 := br2.peekTopBits(peekBits) - MOVQ DI, CX - MOVQ R11, R13 - SHRQ CL, R13 - - // v1 := table[val1&mask] - MOVW (R9)(R13*2), CX - - // br2.advance(uint8(v1.entry)) - MOVB CH, AH - SHLQ CL, R11 - ADDB CL, R12 - - // these two writes get coalesced - // out[id * dstEvery + 0] = uint8(v0.entry >> 8) - // out[id * dstEvery + 1] = uint8(v1.entry >> 8) - MOVW AX, (BX)(R8*2) - - // update the bitreader structure - MOVQ R11, 128(R10) - MOVB R12, 136(R10) - - // br3.fillFast32() - MOVQ 176(R10), R11 - MOVBQZX 184(R10), R12 - CMPQ R12, $0x20 - JBE skip_fill3 - MOVQ 168(R10), AX - SUBQ $0x20, R12 - SUBQ $0x04, AX - MOVQ 144(R10), R13 - - // b.value |= uint64(low) << (b.bitsRead & 63) - MOVL (AX)(R13*1), R13 - MOVQ R12, CX - SHLQ CL, R13 - MOVQ AX, 168(R10) - ORQ R13, R11 - - // exhausted += (br3.off < 4) - CMPQ AX, $0x04 - ADCB $+0, DL - -skip_fill3: - // val0 := br3.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v0 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br3.advance(uint8(v0.entry) - MOVB CH, AL - SHLQ CL, R11 - ADDB CL, R12 + MOVQ ctx+0(FP), AX - // val1 := br3.peekTopBits(peekBits) - MOVQ DI, CX - MOVQ R11, R13 - SHRQ CL, R13 - - // v1 := table[val1&mask] - MOVW (R9)(R13*2), CX - - // br3.advance(uint8(v1.entry)) - MOVB CH, AH - SHLQ CL, R11 - ADDB CL, R12 - - // these two writes get coalesced - // out[id * dstEvery + 0] = uint8(v0.entry >> 8) - // out[id * dstEvery + 1] = uint8(v1.entry >> 8) - LEAQ (R8)(R8*2), CX - MOVW AX, (BX)(CX*1) - - // update the bitreader structure - MOVQ R11, 176(R10) - MOVB R12, 184(R10) - ADDQ $0x02, BX - TESTB DL, DL - JZ main_loop - MOVQ ctx+0(FP), AX - SUBQ 16(AX), BX - SHLQ $0x02, BX - MOVQ BX, 40(AX) + // Preload values + MOVBQZX 8(AX), DX + MOVQ 32(AX), BX + MOVQ 16(AX), SI + MOVQ 24(AX), CX + LEAQ (SI)(CX*1), R8 + LEAQ (SI)(CX*2), R10 + LEAQ (CX)(CX*2), R12 + ADDQ SI, R12 + + // Convert each bit reader to sentinel form: bits = value | 1<> 8) - // out[id * dstEvery + 1] = uint8(v1.entry >> 8) - // out[id * dstEvery + 3] = uint8(v2.entry >> 8) - // out[id * dstEvery + 4] = uint8(v3.entry >> 8) - MOVL AX, (BX) - - // update the bitreader structure - MOVQ R11, 32(R10) - MOVB R12, 40(R10) - - // br1.fillFast32() - MOVQ 80(R10), R11 - MOVBQZX 88(R10), R12 - CMPQ R12, $0x20 - JBE skip_fill1 - MOVQ 72(R10), R13 - SUBQ $0x20, R12 - SUBQ $0x04, R13 - MOVQ 48(R10), R14 - - // b.value |= uint64(low) << (b.bitsRead & 63) - MOVL (R13)(R14*1), R14 - MOVQ R12, CX - SHLQ CL, R14 - MOVQ R13, 72(R10) - ORQ R14, R11 + TZCNTQ R13, R14 + MOVQ R14, CX + ANDQ $0x07, CX + SHRQ $0x03, R14 + MOVQ 88(AX), R13 + SUBQ R14, R13 + MOVQ R13, 88(AX) + MOVQ (R13), R13 + ORQ $0x01, R13 + SHLQ CL, R13 + ADDQ $0x0e, SI + ADDQ $0x0e, R8 + ADDQ $0x0e, R10 + ADDQ $0x0e, R12 + CMPQ SI, 96(AX) + JB inner_loop + JMP outer_loop + +done: + // Hand the state back: off = ip - in (negative when the window reached below the stream start), + // value = the sentinel-form container. bitReaderShifted.restoreFromAsm normalizes both. + MOVQ (AX), CX + MOVQ 64(AX), DX + SUBQ (CX), DX + MOVQ DX, 24(CX) + MOVQ DI, 32(CX) + MOVQ 72(AX), DX + SUBQ 48(CX), DX + MOVQ DX, 72(CX) + MOVQ R9, 80(CX) + MOVQ 80(AX), DX + SUBQ 96(CX), DX + MOVQ DX, 120(CX) + MOVQ R11, 128(CX) + MOVQ 88(AX), DX + SUBQ 144(CX), DX + MOVQ DX, 168(CX) + MOVQ R13, 176(CX) + SUBQ 16(AX), SI + SHLQ $0x02, SI + MOVQ SI, 40(AX) + RET - // exhausted += (br1.off < 4) - CMPQ R13, $0x04 - ADCB $+0, DL +// func decompress4x_main_loop_bmi2(ctx *decompress4xContext) +// Requires: BMI, BMI2, CMOV +TEXT ·decompress4x_main_loop_bmi2(SB), $0-8 + MOVQ ctx+0(FP), AX -skip_fill1: - // val0 := br1.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 + // Preload values + MOVBQZX 8(AX), CX + MOVQ 32(AX), DX + MOVQ 16(AX), BX + MOVQ 24(AX), SI + LEAQ (BX)(SI*1), DI + LEAQ (BX)(SI*2), R9 + LEAQ (SI)(SI*2), R11 + ADDQ BX, R11 + + // Convert each bit reader to sentinel form: bits = value | 1<> 8) - // out[id * dstEvery + 1] = uint8(v1.entry >> 8) - // out[id * dstEvery + 3] = uint8(v2.entry >> 8) - // out[id * dstEvery + 4] = uint8(v3.entry >> 8) - MOVL AX, (BX)(R8*1) - - // update the bitreader structure - MOVQ R11, 80(R10) - MOVB R12, 88(R10) - - // br2.fillFast32() - MOVQ 128(R10), R11 - MOVBQZX 136(R10), R12 - CMPQ R12, $0x20 - JBE skip_fill2 - MOVQ 120(R10), R13 - SUBQ $0x20, R12 - SUBQ $0x04, R13 - MOVQ 96(R10), R14 - - // b.value |= uint64(low) << (b.bitsRead & 63) - MOVL (R13)(R14*1), R14 - MOVQ R12, CX + ANDQ $0x38, R13 + MOVQ -8(R8), R14 + SHRQ $0x01, R14 + ANDQ $0x07, R11 + MOVQ R13, CX + XORQ $0x3f, CX + SHRQ CL, R14 + MOVQ R11, CX SHLQ CL, R14 - MOVQ R13, 120(R10) - ORQ R14, R11 - - // exhausted += (br2.off < 4) - CMPQ R13, $0x04 - ADCB $+0, DL - -skip_fill2: - // val0 := br2.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v0 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br2.advance(uint8(v0.entry) - MOVB CH, AL - SHLQ CL, R11 - ADDB CL, R12 - - // val1 := br2.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v1 := table[val0&mask] - MOVW (R9)(R13*2), CX + ORQ R14, R10 + SHRQ $0x03, R13 + SUBQ R13, R8 + ADDQ $0x05, SI + CMPQ SI, R12 + JB inner_loop + JMP outer_loop + +done: + // Hand the state back in the Go bit reader's own form: off = ip - in, value, bitsRead. + MOVQ (AX), CX + SUBQ (CX), R8 + MOVQ R8, 24(CX) + MOVQ R10, 32(CX) + MOVB R11, 40(CX) + SUBQ 16(AX), SI + MOVQ SI, 40(AX) + RET - // br2.advance(uint8(v1.entry) - MOVB CH, AH - SHLQ CL, R11 - ADDB CL, R12 - BSWAPL AX +// func decompress1x_8b_main_loop_amd64(ctx *decompress1xContext) +// Requires: CMOV +TEXT ·decompress1x_8b_main_loop_amd64(SB), $0-8 + MOVQ ctx+0(FP), AX - // val2 := br2.peekTopBits(peekBits) - MOVQ R11, R13 + // Preload values + MOVBQZX 8(AX), DX + MOVQ 32(AX), BX + MOVQ 16(AX), SI + MOVQ 24(AX), DI + ADDQ SI, DI + MOVQ 56(AX), R8 + MOVQ 48(AX), R9 + MOVQ (AX), CX + MOVQ 32(CX), R10 + MOVBQZX 40(CX), R11 + +outer_loop: + // Iterations allowed by the output: (limit - op) / 8 MOVQ DI, CX - SHRQ CL, R13 - - // v2 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br2.advance(uint8(v2.entry) - MOVB CH, AH - SHLQ CL, R11 - ADDB CL, R12 + SUBQ SI, CX + JLE done + SHRQ $0x03, CX + + // Iterations allowed by the input: a refill reads the 8 bytes below the window and + // moves it down by at most 7, so every read stays inside the stream while ip stays + // at least 8 above ilowest. + MOVQ R8, R12 + SUBQ R9, R12 + SHRQ $0x03, R12 + CMPQ R12, CX + CMOVQCS R12, CX + TESTQ CX, CX + JZ done + IMUL3Q $0x07, CX, R12 + ADDQ SI, R12 + +inner_loop: + // symbol 0 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, (SI) + + // symbol 1 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 1(SI) + + // symbol 2 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 2(SI) + + // symbol 3 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 3(SI) + + // symbol 4 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 4(SI) + + // symbol 5 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 5(SI) + + // symbol 6 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 6(SI) - // val3 := br2.peekTopBits(peekBits) + // Refill the whole bytes consumed from below the window MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v3 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br2.advance(uint8(v3.entry) - MOVB CH, AL - SHLQ CL, R11 - ADDB CL, R12 - BSWAPL AX - - // these four writes get coalesced - // out[id * dstEvery + 0] = uint8(v0.entry >> 8) - // out[id * dstEvery + 1] = uint8(v1.entry >> 8) - // out[id * dstEvery + 3] = uint8(v2.entry >> 8) - // out[id * dstEvery + 4] = uint8(v3.entry >> 8) - MOVL AX, (BX)(R8*2) - - // update the bitreader structure - MOVQ R11, 128(R10) - MOVB R12, 136(R10) - - // br3.fillFast32() - MOVQ 176(R10), R11 - MOVBQZX 184(R10), R12 - CMPQ R12, $0x20 - JBE skip_fill3 - MOVQ 168(R10), R13 - SUBQ $0x20, R12 - SUBQ $0x04, R13 - MOVQ 144(R10), R14 - - // b.value |= uint64(low) << (b.bitsRead & 63) - MOVL (R13)(R14*1), R14 - MOVQ R12, CX + ANDQ $0x38, R13 + MOVQ -8(R8), R14 + SHRQ $0x01, R14 + ANDQ $0x07, R11 + MOVQ R13, CX + XORQ $0x3f, CX + SHRQ CL, R14 + MOVQ R11, CX SHLQ CL, R14 - MOVQ R13, 168(R10) - ORQ R14, R11 - - // exhausted += (br3.off < 4) - CMPQ R13, $0x04 - ADCB $+0, DL - -skip_fill3: - // val0 := br3.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v0 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br3.advance(uint8(v0.entry) - MOVB CH, AL - SHLQ CL, R11 - ADDB CL, R12 - - // val1 := br3.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v1 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br3.advance(uint8(v1.entry) - MOVB CH, AH - SHLQ CL, R11 - ADDB CL, R12 - BSWAPL AX - - // val2 := br3.peekTopBits(peekBits) - MOVQ R11, R13 - MOVQ DI, CX - SHRQ CL, R13 - - // v2 := table[val0&mask] - MOVW (R9)(R13*2), CX + ORQ R14, R10 + SHRQ $0x03, R13 + SUBQ R13, R8 + ADDQ $0x07, SI + CMPQ SI, R12 + JB inner_loop + JMP outer_loop + +done: + // Hand the state back in the Go bit reader's own form: off = ip - in, value, bitsRead. + MOVQ (AX), CX + SUBQ (CX), R8 + MOVQ R8, 24(CX) + MOVQ R10, 32(CX) + MOVB R11, 40(CX) + SUBQ 16(AX), SI + MOVQ SI, 40(AX) + RET - // br3.advance(uint8(v2.entry) - MOVB CH, AH - SHLQ CL, R11 - ADDB CL, R12 +// func decompress1x_4b_main_loop_amd64(ctx *decompress1xContext) +// Requires: CMOV +TEXT ·decompress1x_4b_main_loop_amd64(SB), $0-8 + MOVQ ctx+0(FP), AX - // val3 := br3.peekTopBits(peekBits) - MOVQ R11, R13 + // Preload values + MOVBQZX 8(AX), DX + MOVQ 32(AX), BX + MOVQ 16(AX), SI + MOVQ 24(AX), DI + ADDQ SI, DI + MOVQ 56(AX), R8 + MOVQ 48(AX), R9 + MOVQ (AX), CX + MOVQ 32(CX), R10 + MOVBQZX 40(CX), R11 + +outer_loop: + // Iterations allowed by the output: (limit - op) / 16 MOVQ DI, CX - SHRQ CL, R13 - - // v3 := table[val0&mask] - MOVW (R9)(R13*2), CX - - // br3.advance(uint8(v3.entry) - MOVB CH, AL - SHLQ CL, R11 - ADDB CL, R12 - BSWAPL AX - - // these four writes get coalesced - // out[id * dstEvery + 0] = uint8(v0.entry >> 8) - // out[id * dstEvery + 1] = uint8(v1.entry >> 8) - // out[id * dstEvery + 3] = uint8(v2.entry >> 8) - // out[id * dstEvery + 4] = uint8(v3.entry >> 8) - LEAQ (R8)(R8*2), CX - MOVL AX, (BX)(CX*1) - - // update the bitreader structure - MOVQ R11, 176(R10) - MOVB R12, 184(R10) - ADDQ $0x04, BX - TESTB DL, DL - JZ main_loop - MOVQ ctx+0(FP), AX - SUBQ 16(AX), BX - SHLQ $0x02, BX - MOVQ BX, 40(AX) - RET - -// func decompress1x_main_loop_amd64(ctx *decompress1xContext) -TEXT ·decompress1x_main_loop_amd64(SB), $0-8 - MOVQ ctx+0(FP), CX - MOVQ 16(CX), DX - MOVQ 24(CX), BX - CMPQ BX, $0x04 - JB error_max_decoded_size_exceeded - LEAQ (DX)(BX*1), BX - MOVQ (CX), SI - MOVQ (SI), R8 - MOVQ 24(SI), R9 - MOVQ 32(SI), R10 - MOVBQZX 40(SI), R11 - MOVQ 32(CX), SI - MOVBQZX 8(CX), DI - JMP loop_condition - -main_loop: - // Check if we have room for 4 bytes in the output buffer - LEAQ 4(DX), CX - CMPQ CX, BX - JGE error_max_decoded_size_exceeded - - // Decode 4 values - CMPQ R11, $0x20 - JL bitReader_fillFast_1_end - SUBQ $0x20, R11 - SUBQ $0x04, R9 - MOVL (R8)(R9*1), R12 - MOVQ R11, CX - SHLQ CL, R12 - ORQ R12, R10 - -bitReader_fillFast_1_end: - MOVQ DI, CX - MOVQ R10, R12 - SHRQ CL, R12 - MOVW (SI)(R12*2), CX - MOVB CH, AL - MOVBQZX CL, CX + SUBQ SI, CX + JLE done + SHRQ $0x04, CX + + // Iterations allowed by the input: a refill reads the 8 bytes below the window and + // moves it down by at most 7, so every read stays inside the stream while ip stays + // at least 8 above ilowest. + MOVQ R8, R12 + SUBQ R9, R12 + SHRQ $0x03, R12 + CMPQ R12, CX + CMOVQCS R12, CX + TESTQ CX, CX + JZ done + IMUL3Q $0x0e, CX, R12 + ADDQ SI, R12 + +inner_loop: + // symbol 0 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, (SI) + + // symbol 1 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX SHLQ CL, R10 - MOVQ DI, CX - MOVQ R10, R12 - SHRQ CL, R12 - MOVW (SI)(R12*2), CX - MOVB CH, AH - MOVBQZX CL, CX ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 1(SI) + + // symbol 2 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX SHLQ CL, R10 - BSWAPL AX - CMPQ R11, $0x20 - JL bitReader_fillFast_2_end - SUBQ $0x20, R11 - SUBQ $0x04, R9 - MOVL (R8)(R9*1), R12 - MOVQ R11, CX - SHLQ CL, R12 - ORQ R12, R10 - -bitReader_fillFast_2_end: - MOVQ DI, CX - MOVQ R10, R12 - SHRQ CL, R12 - MOVW (SI)(R12*2), CX - MOVB CH, AH - MOVBQZX CL, CX ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 2(SI) + + // symbol 3 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX SHLQ CL, R10 - MOVQ DI, CX - MOVQ R10, R12 - SHRQ CL, R12 - MOVW (SI)(R12*2), CX - MOVB CH, AL - MOVBQZX CL, CX ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 3(SI) + + // symbol 4 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 4(SI) + + // symbol 5 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 5(SI) + + // symbol 6 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 6(SI) + + // symbol 7 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 7(SI) + + // symbol 8 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 8(SI) + + // symbol 9 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX SHLQ CL, R10 - BSWAPL AX - - // Store the decoded values - MOVL AX, (DX) - ADDQ $0x04, DX - -loop_condition: - CMPQ R9, $0x08 - JGE main_loop - - // Update ctx structure - MOVQ ctx+0(FP), AX - SUBQ 16(AX), DX - MOVQ DX, 40(AX) - MOVQ (AX), AX - MOVQ R9, 24(AX) - MOVQ R10, 32(AX) - MOVB R11, 40(AX) - RET - - // Report error -error_max_decoded_size_exceeded: - MOVQ ctx+0(FP), AX - MOVQ $-1, CX - MOVQ CX, 40(AX) - RET - -// func decompress1x_main_loop_bmi2(ctx *decompress1xContext) -// Requires: BMI2 -TEXT ·decompress1x_main_loop_bmi2(SB), $0-8 - MOVQ ctx+0(FP), CX - MOVQ 16(CX), DX - MOVQ 24(CX), BX - CMPQ BX, $0x04 - JB error_max_decoded_size_exceeded - LEAQ (DX)(BX*1), BX - MOVQ (CX), SI - MOVQ (SI), R8 - MOVQ 24(SI), R9 - MOVQ 32(SI), R10 - MOVBQZX 40(SI), R11 - MOVQ 32(CX), SI - MOVBQZX 8(CX), DI - JMP loop_condition - -main_loop: - // Check if we have room for 4 bytes in the output buffer - LEAQ 4(DX), CX - CMPQ CX, BX - JGE error_max_decoded_size_exceeded - - // Decode 4 values - CMPQ R11, $0x20 - JL bitReader_fillFast_1_end - SUBQ $0x20, R11 - SUBQ $0x04, R9 - MOVL (R8)(R9*1), CX - SHLXQ R11, CX, CX - ORQ CX, R10 - -bitReader_fillFast_1_end: - SHRXQ DI, R10, CX - MOVW (SI)(CX*2), CX - MOVB CH, AL - MOVBQZX CL, CX ADDQ CX, R11 - SHLXQ CX, R10, R10 - SHRXQ DI, R10, CX - MOVW (SI)(CX*2), CX - MOVB CH, AH - MOVBQZX CL, CX + SHRQ $0x08, CX + MOVB CL, 9(SI) + + // symbol 10 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 ADDQ CX, R11 - SHLXQ CX, R10, R10 - BSWAPL AX - CMPQ R11, $0x20 - JL bitReader_fillFast_2_end - SUBQ $0x20, R11 - SUBQ $0x04, R9 - MOVL (R8)(R9*1), CX - SHLXQ R11, CX, CX - ORQ CX, R10 - -bitReader_fillFast_2_end: - SHRXQ DI, R10, CX - MOVW (SI)(CX*2), CX - MOVB CH, AH - MOVBQZX CL, CX + SHRQ $0x08, CX + MOVB CL, 10(SI) + + // symbol 11 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 ADDQ CX, R11 - SHLXQ CX, R10, R10 - SHRXQ DI, R10, CX - MOVW (SI)(CX*2), CX - MOVB CH, AL - MOVBQZX CL, CX + SHRQ $0x08, CX + MOVB CL, 11(SI) + + // symbol 12 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 ADDQ CX, R11 - SHLXQ CX, R10, R10 - BSWAPL AX + SHRQ $0x08, CX + MOVB CL, 12(SI) + + // symbol 13 + MOVQ DX, CX + MOVQ R10, R13 + SHRQ CL, R13 + MOVWQZX (BX)(R13*2), CX + SHLQ CL, R10 + ADDQ CX, R11 + SHRQ $0x08, CX + MOVB CL, 13(SI) - // Store the decoded values - MOVL AX, (DX) - ADDQ $0x04, DX + // Refill the whole bytes consumed from below the window + MOVQ R11, R13 + ANDQ $0x38, R13 + MOVQ -8(R8), R14 + SHRQ $0x01, R14 + ANDQ $0x07, R11 + MOVQ R13, CX + XORQ $0x3f, CX + SHRQ CL, R14 + MOVQ R11, CX + SHLQ CL, R14 + ORQ R14, R10 + SHRQ $0x03, R13 + SUBQ R13, R8 + ADDQ $0x0e, SI + CMPQ SI, R12 + JB inner_loop + JMP outer_loop + +done: + // Hand the state back in the Go bit reader's own form: off = ip - in, value, bitsRead. + MOVQ (AX), CX + SUBQ (CX), R8 + MOVQ R8, 24(CX) + MOVQ R10, 32(CX) + MOVB R11, 40(CX) + SUBQ 16(AX), SI + MOVQ SI, 40(AX) + RET + +// func decompress1x_main_loop_bmi2(ctx *decompress1xContext) +// Requires: BMI2, CMOV +TEXT ·decompress1x_main_loop_bmi2(SB), $0-8 + MOVQ ctx+0(FP), AX -loop_condition: - CMPQ R9, $0x08 - JGE main_loop + // Preload values + MOVBQZX 8(AX), CX + MOVQ 32(AX), DX + MOVQ 16(AX), BX + MOVQ 24(AX), SI + ADDQ BX, SI + MOVQ 56(AX), DI + MOVQ 48(AX), R8 + MOVQ (AX), R10 + MOVQ 32(R10), R9 + MOVBQZX 40(R10), R10 + +outer_loop: + // Iterations allowed by the output: (limit - op) / 8 + MOVQ SI, R11 + SUBQ BX, R11 + JLE done + SHRQ $0x03, R11 + + // Iterations allowed by the input: a refill reads the 8 bytes below the window and + // moves it down by at most 7, so every read stays inside the stream while ip stays + // at least 8 above ilowest. + MOVQ DI, R12 + SUBQ R8, R12 + SHRQ $0x03, R12 + CMPQ R12, R11 + CMOVQCS R12, R11 + TESTQ R11, R11 + JZ done + IMUL3Q $0x05, R11, R11 + ADDQ BX, R11 + +inner_loop: + // symbol 0 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, (BX) + + // symbol 1 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 1(BX) + + // symbol 2 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 2(BX) + + // symbol 3 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 3(BX) + + // symbol 4 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 4(BX) + + // Refill the whole bytes consumed from below the window + MOVQ R10, R12 + ANDQ $0x38, R12 + MOVQ -8(DI), R13 + SHRQ $0x01, R13 + ANDQ $0x07, R10 + MOVQ R12, R14 + XORQ $0x3f, R14 + SHRXQ R14, R13, R13 + SHLXQ R10, R13, R13 + ORQ R13, R9 + SHRQ $0x03, R12 + SUBQ R12, DI + ADDQ $0x05, BX + CMPQ BX, R11 + JB inner_loop + JMP outer_loop + +done: + // Hand the state back in the Go bit reader's own form: off = ip - in, value, bitsRead. + MOVQ (AX), CX + SUBQ (CX), DI + MOVQ DI, 24(CX) + MOVQ R9, 32(CX) + MOVB R10, 40(CX) + SUBQ 16(AX), BX + MOVQ BX, 40(AX) + RET - // Update ctx structure +// func decompress1x_8b_main_loop_bmi2(ctx *decompress1xContext) +// Requires: BMI2, CMOV +TEXT ·decompress1x_8b_main_loop_bmi2(SB), $0-8 MOVQ ctx+0(FP), AX - SUBQ 16(AX), DX - MOVQ DX, 40(AX) - MOVQ (AX), AX - MOVQ R9, 24(AX) - MOVQ R10, 32(AX) - MOVB R11, 40(AX) + + // Preload values + MOVBQZX 8(AX), CX + MOVQ 32(AX), DX + MOVQ 16(AX), BX + MOVQ 24(AX), SI + ADDQ BX, SI + MOVQ 56(AX), DI + MOVQ 48(AX), R8 + MOVQ (AX), R10 + MOVQ 32(R10), R9 + MOVBQZX 40(R10), R10 + +outer_loop: + // Iterations allowed by the output: (limit - op) / 8 + MOVQ SI, R11 + SUBQ BX, R11 + JLE done + SHRQ $0x03, R11 + + // Iterations allowed by the input: a refill reads the 8 bytes below the window and + // moves it down by at most 7, so every read stays inside the stream while ip stays + // at least 8 above ilowest. + MOVQ DI, R12 + SUBQ R8, R12 + SHRQ $0x03, R12 + CMPQ R12, R11 + CMOVQCS R12, R11 + TESTQ R11, R11 + JZ done + IMUL3Q $0x07, R11, R11 + ADDQ BX, R11 + +inner_loop: + // symbol 0 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, (BX) + + // symbol 1 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 1(BX) + + // symbol 2 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 2(BX) + + // symbol 3 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 3(BX) + + // symbol 4 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 4(BX) + + // symbol 5 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 5(BX) + + // symbol 6 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 6(BX) + + // Refill the whole bytes consumed from below the window + MOVQ R10, R12 + ANDQ $0x38, R12 + MOVQ -8(DI), R13 + SHRQ $0x01, R13 + ANDQ $0x07, R10 + MOVQ R12, R14 + XORQ $0x3f, R14 + SHRXQ R14, R13, R13 + SHLXQ R10, R13, R13 + ORQ R13, R9 + SHRQ $0x03, R12 + SUBQ R12, DI + ADDQ $0x07, BX + CMPQ BX, R11 + JB inner_loop + JMP outer_loop + +done: + // Hand the state back in the Go bit reader's own form: off = ip - in, value, bitsRead. + MOVQ (AX), CX + SUBQ (CX), DI + MOVQ DI, 24(CX) + MOVQ R9, 32(CX) + MOVB R10, 40(CX) + SUBQ 16(AX), BX + MOVQ BX, 40(AX) RET - // Report error -error_max_decoded_size_exceeded: +// func decompress1x_4b_main_loop_bmi2(ctx *decompress1xContext) +// Requires: BMI2, CMOV +TEXT ·decompress1x_4b_main_loop_bmi2(SB), $0-8 MOVQ ctx+0(FP), AX - MOVQ $-1, CX - MOVQ CX, 40(AX) + + // Preload values + MOVBQZX 8(AX), CX + MOVQ 32(AX), DX + MOVQ 16(AX), BX + MOVQ 24(AX), SI + ADDQ BX, SI + MOVQ 56(AX), DI + MOVQ 48(AX), R8 + MOVQ (AX), R10 + MOVQ 32(R10), R9 + MOVBQZX 40(R10), R10 + +outer_loop: + // Iterations allowed by the output: (limit - op) / 16 + MOVQ SI, R11 + SUBQ BX, R11 + JLE done + SHRQ $0x04, R11 + + // Iterations allowed by the input: a refill reads the 8 bytes below the window and + // moves it down by at most 7, so every read stays inside the stream while ip stays + // at least 8 above ilowest. + MOVQ DI, R12 + SUBQ R8, R12 + SHRQ $0x03, R12 + CMPQ R12, R11 + CMOVQCS R12, R11 + TESTQ R11, R11 + JZ done + IMUL3Q $0x0e, R11, R11 + ADDQ BX, R11 + +inner_loop: + // symbol 0 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, (BX) + + // symbol 1 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 1(BX) + + // symbol 2 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 2(BX) + + // symbol 3 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 3(BX) + + // symbol 4 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 4(BX) + + // symbol 5 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 5(BX) + + // symbol 6 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 6(BX) + + // symbol 7 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 7(BX) + + // symbol 8 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 8(BX) + + // symbol 9 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 9(BX) + + // symbol 10 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 10(BX) + + // symbol 11 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 11(BX) + + // symbol 12 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 12(BX) + + // symbol 13 + SHRXQ CX, R9, R12 + MOVWQZX (DX)(R12*2), R12 + SHLXQ R12, R9, R9 + ADDQ R12, R10 + SHRQ $0x08, R12 + MOVB R12, 13(BX) + + // Refill the whole bytes consumed from below the window + MOVQ R10, R12 + ANDQ $0x38, R12 + MOVQ -8(DI), R13 + SHRQ $0x01, R13 + ANDQ $0x07, R10 + MOVQ R12, R14 + XORQ $0x3f, R14 + SHRXQ R14, R13, R13 + SHLXQ R10, R13, R13 + ORQ R13, R9 + SHRQ $0x03, R12 + SUBQ R12, DI + ADDQ $0x0e, BX + CMPQ BX, R11 + JB inner_loop + JMP outer_loop + +done: + // Hand the state back in the Go bit reader's own form: off = ip - in, value, bitsRead. + MOVQ (AX), CX + SUBQ (CX), DI + MOVQ DI, 24(CX) + MOVQ R9, 32(CX) + MOVB R10, 40(CX) + SUBQ 16(AX), BX + MOVQ BX, 40(AX) RET diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_arm64.go b/vendor/github.com/klauspost/compress/huff0/decompress_arm64.go new file mode 100644 index 0000000000..d1988ad54d --- /dev/null +++ b/vendor/github.com/klauspost/compress/huff0/decompress_arm64.go @@ -0,0 +1,72 @@ +//go:build arm64 && !appengine && !noasm && gc + +// arm64 stubs and dispatch for the asm loops used by decompress_asm.go. +// The asm (decompress_arm64.s) is generated by the avo arm64 lowering +// printer from the same source as the amd64 asm; see _generate/gen.go. +package huff0 + +// decompress4x_main_loop_arm64 is an arm64 assembler implementation +// of Decompress4X when tablelog > 8, decoding fastSymbols symbols per +// stream between bit container reloads. +// +//go:noescape +func decompress4x_main_loop_arm64(ctx *decompress4xContext) + +// decompress4x_8b_main_loop_arm64 is an arm64 assembler implementation +// of Decompress4X when tablelog <= 8, decoding fast8bSymbols symbols +// per stream between bit container reloads. +// +//go:noescape +func decompress4x_8b_main_loop_arm64(ctx *decompress4xContext) + +// decompress4x_4b_main_loop_arm64 is an arm64 assembler implementation +// of Decompress4X when tablelog <= 4, decoding fast4bSymbols symbols +// per stream between bit container reloads. +// +//go:noescape +func decompress4x_4b_main_loop_arm64(ctx *decompress4xContext) + +// decompress1x_main_loop_arm64 is an arm64 assembler implementation +// of Decompress1X when tablelog > 8, decoding fastSymbols symbols +// between bit container reloads. +// +//go:noescape +func decompress1x_main_loop_arm64(ctx *decompress1xContext) + +// decompress1x_8b_main_loop_arm64 is an arm64 assembler implementation +// of Decompress1X when tablelog <= 8, decoding fast8bSymbols symbols +// between bit container reloads. +// +//go:noescape +func decompress1x_8b_main_loop_arm64(ctx *decompress1xContext) + +// decompress1x_4b_main_loop_arm64 is an arm64 assembler implementation +// of Decompress1X when tablelog <= 4, decoding fast4bSymbols symbols +// between bit container reloads. +// +//go:noescape +func decompress1x_4b_main_loop_arm64(ctx *decompress1xContext) + +func decompress4x_main_loop_asm(ctx *decompress4xContext) { + decompress4x_main_loop_arm64(ctx) +} + +func decompress4x_8b_main_loop_asm(ctx *decompress4xContext) { + decompress4x_8b_main_loop_arm64(ctx) +} + +func decompress4x_4b_main_loop_asm(ctx *decompress4xContext) { + decompress4x_4b_main_loop_arm64(ctx) +} + +func decompress1x_main_loop_asm(ctx *decompress1xContext) { + decompress1x_main_loop_arm64(ctx) +} + +func decompress1x_8b_main_loop_asm(ctx *decompress1xContext) { + decompress1x_8b_main_loop_arm64(ctx) +} + +func decompress1x_4b_main_loop_asm(ctx *decompress1xContext) { + decompress1x_4b_main_loop_arm64(ctx) +} diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_arm64.s b/vendor/github.com/klauspost/compress/huff0/decompress_arm64.s new file mode 100644 index 0000000000..971b8eb8bd --- /dev/null +++ b/vendor/github.com/klauspost/compress/huff0/decompress_arm64.s @@ -0,0 +1,1657 @@ +// Code generated by command: go run gen.go -out ../decompress.s -arch amd64,arm64 -pkg=huff0. DO NOT EDIT. +// EXPERIMENTAL arm64 output lowered from an amd64 avo program. + +//go:build arm64 && (!appengine && !noasm && gc) + +// func decompress4x_main_loop_amd64(ctx *decompress4xContext) +// Requires: BMI, CMOV +TEXT ·decompress4x_main_loop_arm64(SB), $0-8 + MOVD ctx+0(FP), R0 + + // Preload values + MOVBU 8(R0), R2 + MOVD 32(R0), R3 + MOVD 16(R0), R5 + MOVD 24(R0), R1 + ADD R1, R5, R7 + ADD R1<<1, R5, R9 + ADD R1<<1, R1, R11 + ADD R5, R11, R11 + + // Convert each bit reader to sentinel form: bits = value | 1<= len(src) { + return nil, errors.New("truncated input (or invalid offset)") + } + err := br[i].init(src[start : start+length]) + if err != nil { + return nil, err + } + start += length + } + err := br[3].init(src[start:]) + if err != nil { + return nil, err + } + + // destination, offset to match first output + dstSize := cap(dst) + dst = dst[:dstSize] + out := dst + dstEvery := (dstSize + 3) / 4 + + const tlSize = 1 << tableLogMax + const tlMask = tlSize - 1 + single := d.dt.single[:tlSize] + + var decoded int + + nSyms := fastSymbols + if d.actualTableLog <= 4 { + nSyms = fast4bSymbols + } else if use8BitTables { + nSyms = fast8bSymbols + } + // The asm writes nSyms bytes per stream per iteration and only re-checks + // its bounds between batches of iterations (the batch size is derived + // from limit in _generate/gen.go), so stream 0 must stop early enough + // that stream 3 (which may be up to 3 bytes shorter than dstEvery) + // never writes past the end of out. Every stream needs a full 8-byte + // window ahead of its read pointer to enter the loop. + if limit := dstEvery - nSyms - 2; limit > 0 && br[0].canUseAsm() && br[1].canUseAsm() && br[2].canUseAsm() && br[3].canUseAsm() { + for i := range br { + br[i].prepareForAsm() + } + ctx := decompress4xContext{ + pbr: &br, + peekBits: uint8((64 - d.actualTableLog) & 63), // see: bitReaderShifted.peekBitsFast() + out: &out[0], + dstEvery: dstEvery, + tbl: &single[0], + limit: &out[limit], + // The 6-byte jump table sits below stream 0 inside src, so the + // lowest window may reach into it; restoreFromAsm accounts for + // bytes below a stream's start. Bounding by src rather than by + // stream 0 keeps the asm running about six bytes longer, which + // matters for streams with very short codes. + ilowest: &src[0], + } + for i := range br { + ctx.ip[i] = &br[i].in[br[i].off] + } + switch nSyms { + case fast4bSymbols: + decompress4x_4b_main_loop_asm(&ctx) + case fast8bSymbols: + decompress4x_8b_main_loop_asm(&ctx) + default: + decompress4x_main_loop_asm(&ctx) + } + + decoded = ctx.decoded + out = out[decoded/4:] + for i := range br { + if err := br[i].restoreFromAsm(); err != nil { + return nil, err + } + } + } + + // Decode remaining. + remainBytes := dstEvery - (decoded / 4) + for i := range br { + offset := dstEvery * i + endsAt := min(offset+remainBytes, len(out)) + br := &br[i] + bitsLeft := br.remaining() + for bitsLeft > 0 { + br.fill() + if offset >= endsAt { + return nil, errors.New("corruption detected: stream overrun 4") + } + + // Read value and increment offset. + val := br.peekBitsFast(d.actualTableLog) + v := single[val&tlMask].entry + nBits := uint8(v) + br.advance(nBits) + bitsLeft -= uint(nBits) + out[offset] = uint8(v >> 8) + offset++ + } + if offset != endsAt { + return nil, fmt.Errorf("corruption detected: short output block %d, end %d != %d", i, offset, endsAt) + } + decoded += offset - dstEvery*i + err = br.close() + if err != nil { + return nil, err + } + } + if dstSize != decoded { + return nil, errors.New("corruption detected: short output block") + } + return dst, nil +} + +// decompress1xContext is the argument block of the Decompress1X asm loops. +// Go fills every field but decoded; the asm advances ip, and on return +// leaves the bit reader in the form bitReaderShifted.restoreFromAsm expects. +type decompress1xContext struct { + pbr *bitReaderShifted + peekBits uint8 + out *byte + outCap int // no write reaches out[outCap] + tbl *dEntrySingle + decoded int + ilowest *byte // start of the stream; no read goes below it + ip *byte // the 8-byte input window, advanced by the asm +} + +// Decompress1X will decompress a 1X encoded stream. +// The cap of the output buffer will be the maximum decompressed size. +// The length of the supplied input must match the end of a block exactly. +func (d *Decoder) Decompress1X(dst, src []byte) ([]byte, error) { + if len(d.dt.single) == 0 { + return nil, errors.New("no table loaded") + } + var br bitReaderShifted + err := br.init(src) + if err != nil { + return dst, err + } + maxDecodedSize := cap(dst) + dst = dst[:maxDecodedSize] + + const tlSize = 1 << tableLogMax + const tlMask = tlSize - 1 + + // The asm decodes whole batches of symbols and only re-checks its + // bounds between them, so it needs at least one batch of room (the + // output bound rounds nSyms up to 16, see _generate/gen.go) and a full + // 8-byte window ahead of the read pointer to enter the loop. It also + // needs at most 7 bits consumed on entry so that a batch cannot run + // the container dry, which prepareForAsm establishes. + decoded := 0 + if maxDecodedSize >= 16 && br.canUseAsm() { + br.prepareForAsm() + ctx := decompress1xContext{ + pbr: &br, + out: &dst[0], + outCap: maxDecodedSize, + peekBits: uint8((64 - d.actualTableLog) & 63), // see: bitReaderShifted.peekBitsFast() + tbl: &d.dt.single[0], + ilowest: &src[0], + ip: &br.in[br.off], + } + if d.actualTableLog <= 4 { + decompress1x_4b_main_loop_asm(&ctx) + } else if d.actualTableLog <= 8 { + decompress1x_8b_main_loop_asm(&ctx) + } else { + decompress1x_main_loop_asm(&ctx) + } + decoded = ctx.decoded + } + dst = dst[:decoded] + + bitsLeft := br.remaining() + for bitsLeft > 0 { + br.fill() + if len(dst) >= maxDecodedSize { + br.close() + return nil, ErrMaxDecodedSizeExceeded + } + v := d.dt.single[br.peekBitsFast(d.actualTableLog)&tlMask] + nBits := uint8(v.entry) + br.advance(nBits) + bitsLeft -= uint(nBits) + dst = append(dst, uint8(v.entry>>8)) + } + return dst, br.close() +} diff --git a/vendor/github.com/klauspost/compress/huff0/decompress_generic.go b/vendor/github.com/klauspost/compress/huff0/decompress_generic.go index 6103923222..e1bca5c686 100644 --- a/vendor/github.com/klauspost/compress/huff0/decompress_generic.go +++ b/vendor/github.com/klauspost/compress/huff0/decompress_generic.go @@ -1,4 +1,4 @@ -//go:build !amd64 || appengine || !gc || noasm +//go:build (!amd64 && !arm64) || appengine || !gc || noasm // This file contains a generic implementation of Decoder.Decompress4X. package huff0 diff --git a/vendor/github.com/klauspost/compress/zstd/blockdec.go b/vendor/github.com/klauspost/compress/zstd/blockdec.go index 2329e996f8..89b20d6a84 100644 --- a/vendor/github.com/klauspost/compress/zstd/blockdec.go +++ b/vendor/github.com/klauspost/compress/zstd/blockdec.go @@ -240,9 +240,11 @@ func (b *blockDec) decodeBuf(hist *history) error { b.dst[i] = v } hist.appendKeep(b.dst) + hist.decoders.consumeSyncLen(len(b.dst)) return nil case blockTypeRaw: hist.appendKeep(b.data) + hist.decoders.consumeSyncLen(len(b.data)) return nil case blockTypeCompressed: saved := b.dst @@ -400,8 +402,9 @@ func (b *blockDec) decodeLiterals(in []byte, hist *history) (remain []byte, err } } var err error - // Use our out buffer. - huff.MaxDecodedSize = litRegenSize + // Decoder.Decompress* uses cap(dst) for the size limit. Do not write + // MaxDecodedSize on hist.huffTree: with a trained dictionary that + // pointer aliases the shared dict.litEnc and concurrent DecodeAll races. if fourStreams { literals, err = huff.Decoder().Decompress4X(b.literalBuf[:0:litRegenSize], literals) } else { @@ -486,6 +489,7 @@ func (b *blockDec) decodeCompressed(hist *history) error { } if hist.decoders.nSeqs == 0 { b.dst = append(b.dst, hist.decoders.literals...) + hist.decoders.consumeSyncLen(len(hist.decoders.literals)) return nil } before := len(hist.decoders.out) @@ -673,10 +677,6 @@ func (b *blockDec) executeSequences(hist *history) error { hbytes := hist.b if len(hbytes) > hist.windowSize { hbytes = hbytes[len(hbytes)-hist.windowSize:] - // We do not need history anymore. - if hist.dict != nil { - hist.dict.content = nil - } } hist.decoders.windowSize = hist.windowSize hist.decoders.out = b.dst[:0] diff --git a/vendor/github.com/klauspost/compress/zstd/decoder.go b/vendor/github.com/klauspost/compress/zstd/decoder.go index c7e500f02a..90ef3bb03f 100644 --- a/vendor/github.com/klauspost/compress/zstd/decoder.go +++ b/vendor/github.com/klauspost/compress/zstd/decoder.go @@ -7,8 +7,11 @@ package zstd import ( "context" "encoding/binary" + "errors" "io" + "maps" "sync" + "sync/atomic" "github.com/klauspost/compress/zstd/internal/xxhash" ) @@ -20,7 +23,8 @@ import ( // A decoder can safely be re-used even if the previous stream failed. // To release the resources, you must call the Close() function on a decoder. type Decoder struct { - o decoderOptions + // publishedOptions contains the current immutable decoder options. + publishedOptions atomic.Pointer[decoderOptions] // Unreferenced decoders, ready for use. decoders chan *blockDec @@ -84,9 +88,10 @@ var ( func NewReader(r io.Reader, opts ...DOption) (*Decoder, error) { initPredefined() var d Decoder - d.o.setDefault() - for _, o := range opts { - err := o(&d.o) + var o decoderOptions + o.setDefault() + for _, opt := range opts { + err := opt(&o) if err != nil { return nil, err } @@ -99,17 +104,19 @@ func NewReader(r io.Reader, opts ...DOption) (*Decoder, error) { } // Initialize dict map if needed. - if d.o.dicts == nil { - d.o.dicts = make(map[uint32]*dict) + if o.dicts == nil { + o.dicts = make(map[uint32]*dict) } // Create decoders - d.decoders = make(chan *blockDec, d.o.concurrent) - for i := 0; i < d.o.concurrent; i++ { - dec := newBlockDec(d.o.lowMem) - dec.localFrame = newFrameDec(d.o) + d.decoders = make(chan *blockDec, o.concurrent) + for i := 0; i < o.concurrent; i++ { + dec := newBlockDec(o.lowMem) + dec.localFrame = newFrameDec(o) + dec.localFrame.o.dicts = nil d.decoders <- dec } + d.publishedOptions.Store(&o) if r == nil { return &d, nil @@ -171,6 +178,9 @@ func (d *Decoder) Reset(r io.Reader) error { d.drainOutput() d.syncStream.br.r = nil + if d.frame != nil { + d.frame.o.dicts = nil + } if r == nil { d.current.err = ErrDecoderNilInput if len(d.current.b) > 0 { @@ -180,8 +190,10 @@ func (d *Decoder) Reset(r io.Reader) error { return nil } + o := *d.publishedOptions.Load() + // If bytes buffer and < 5MB, do sync decoding anyway. - if bb, ok := r.(byter); ok && bb.Len() < d.o.decodeBufsBelow && !d.o.limitToCap { + if bb, ok := r.(byter); ok && bb.Len() < o.decodeBufsBelow && !o.limitToCap { bb2 := bb if debugDecoder { println("*bytes.Buffer detected, doing sync decode, len:", bb.Len()) @@ -193,18 +205,30 @@ func (d *Decoder) Reset(r io.Reader) error { } dst, err := d.DecodeAll(b, dst) - if err == nil { - err = io.EOF + if !errors.Is(err, ErrDecoderSizeExceeded) { + if err == nil { + err = io.EOF + } + // Save output buffer + d.syncStream.dstBuf = dst + d.current.b = dst + d.current.err = err + d.current.flushed = true + if debugDecoder { + println("sync decode to", len(dst), "bytes, err:", err) + } + return nil } - // Save output buffer - d.syncStream.dstBuf = dst - d.current.b = dst - d.current.err = err - d.current.flushed = true + // The output does not fit WithDecoderMaxMemory, which caps the total + // decoded size when decoding in memory. A Reader follows the + // streaming contract, where the option caps the window instead, so + // decode this input as a stream after all; the shortcut has not + // consumed the reader. Drop the oversized buffer rather than keep it + // for reuse. if debugDecoder { - println("sync decode to", len(dst), "bytes, err:", err) + println("sync decode exceeded max memory after", len(dst), "bytes; streaming instead") } - return nil + d.syncStream.dstBuf = nil } // Remove current block. d.stashDecoder() @@ -217,14 +241,16 @@ func (d *Decoder) Reset(r io.Reader) error { // Ensure no-one else is still running... d.streamWg.Wait() if d.frame == nil { - d.frame = newFrameDec(d.o) + d.frame = newFrameDec(o) + } else { + d.frame.setOptions(o) } - if d.o.concurrent == 1 { + if o.concurrent == 1 { return d.startSyncDecoder(r) } - d.current.output = make(chan decodeOutput, d.o.concurrent) + d.current.output = make(chan decodeOutput, o.concurrent) ctx, cancel := context.WithCancel(context.Background()) d.current.cancel = cancel d.streamWg.Add(1) @@ -238,13 +264,21 @@ func (d *Decoder) Reset(r io.Reader) error { // Options are applied on top of the existing options. // Some options cannot be changed on reset and will return an error. func (d *Decoder) ResetWithOptions(r io.Reader, opts ...DOption) error { - d.o.resetOpt = true - defer func() { d.o.resetOpt = false }() + if d.current.err == ErrDecoderClosed { + return d.current.err + } + next := *d.publishedOptions.Load() + next.dicts = maps.Clone(next.dicts) + next.resetOpt = true for _, o := range opts { - if err := o(&d.o); err != nil { + if err := o(&next); err != nil { + next.resetOpt = false + d.publishedOptions.Store(&next) return err } } + next.resetOpt = false + d.publishedOptions.Store(&next) return d.Reset(r) } @@ -324,6 +358,8 @@ func (d *Decoder) DecodeAll(input, dst []byte) ([]byte, error) { // Grab a block decoder and frame decoder. block := <-d.decoders frame := block.localFrame + o := *d.publishedOptions.Load() + frame.setOptions(o) initialSize := len(dst) defer func() { if debugDecoder { @@ -335,6 +371,7 @@ func (d *Decoder) DecodeAll(input, dst []byte) ([]byte, error) { frame.history.decoders.br.in = nil frame.history.decoders.br.cursor = 0 } + frame.o.dicts = nil d.decoders <- block }() frame.bBuf = input @@ -354,20 +391,20 @@ func (d *Decoder) DecodeAll(input, dst []byte) ([]byte, error) { if err = d.setDict(frame); err != nil { return nil, err } - if frame.WindowSize > d.o.maxWindowSize { + if frame.WindowSize > frame.o.maxWindowSize { if debugDecoder { - println("window size exceeded:", frame.WindowSize, ">", d.o.maxWindowSize) + println("window size exceeded:", frame.WindowSize, ">", frame.o.maxWindowSize) } return dst, ErrWindowSizeExceeded } if frame.FrameContentSize != fcsUnknown { - if frame.FrameContentSize > d.o.maxDecodedSize-uint64(len(dst)-initialSize) { + if frame.FrameContentSize > frame.o.maxDecodedSize-uint64(len(dst)-initialSize) { if debugDecoder { - println("decoder size exceeded; fcs:", frame.FrameContentSize, "> mcs:", d.o.maxDecodedSize-uint64(len(dst)-initialSize), "len:", len(dst)) + println("decoder size exceeded; fcs:", frame.FrameContentSize, "> mcs:", frame.o.maxDecodedSize-uint64(len(dst)-initialSize), "len:", len(dst)) } return dst, ErrDecoderSizeExceeded } - if d.o.limitToCap && frame.FrameContentSize > uint64(cap(dst)-len(dst)) { + if frame.o.limitToCap && frame.FrameContentSize > uint64(cap(dst)-len(dst)) { if debugDecoder { println("decoder size exceeded; fcs:", frame.FrameContentSize, "> (cap-len)", cap(dst)-len(dst)) } @@ -380,14 +417,14 @@ func (d *Decoder) DecodeAll(input, dst []byte) ([]byte, error) { } } - if cap(dst) == 0 && !d.o.limitToCap { + if cap(dst) == 0 && !frame.o.limitToCap { // Allocate len(input) * 2 by default if nothing is provided // and we didn't get frame content size. size := min( // Cap to 1 MB. len(input)*2, 1<<20) - if uint64(size) > d.o.maxDecodedSize { - size = int(d.o.maxDecodedSize) + if uint64(size) > frame.o.maxDecodedSize { + size = int(frame.o.maxDecodedSize) } dst = make([]byte, 0, size) } @@ -396,7 +433,7 @@ func (d *Decoder) DecodeAll(input, dst []byte) ([]byte, error) { if err != nil { return dst, err } - if uint64(len(dst)-initialSize) > d.o.maxDecodedSize { + if uint64(len(dst)-initialSize) > frame.o.maxDecodedSize { return dst, ErrDecoderSizeExceeded } if len(frame.bBuf) == 0 { @@ -459,7 +496,7 @@ func (d *Decoder) nextBlock(blocking bool) (ok bool) { println("got", len(d.current.b), "bytes, error:", d.current.err, "data crc:", tmp) } - if d.o.ignoreChecksum { + if d.frame.o.ignoreChecksum { return true } @@ -497,7 +534,7 @@ func (d *Decoder) nextBlockSync() (ok bool) { if d.current.err != nil { return false } - if d.frame.WindowSize > d.o.maxDecodedSize || d.frame.WindowSize > d.o.maxWindowSize { + if d.frame.WindowSize > d.frame.o.maxDecodedSize || d.frame.WindowSize > d.frame.o.maxWindowSize { d.current.err = ErrDecoderSizeExceeded return false } @@ -546,11 +583,11 @@ func (d *Decoder) nextBlockSync() (ok bool) { // Update/Check CRC if d.frame.HasCheckSum { - if !d.o.ignoreChecksum { + if !d.frame.o.ignoreChecksum { d.frame.crc.Write(d.current.b) } if d.current.d.Last { - if !d.o.ignoreChecksum { + if !d.frame.o.ignoreChecksum { d.current.err = d.frame.checkCRC() } else { d.current.err = d.frame.consumeCRC() @@ -656,8 +693,8 @@ func (d *Decoder) startStreamDecoder(ctx context.Context, r io.Reader, output ch defer d.streamWg.Done() br := readerWrapper{r: r} - var seqDecode = make(chan *blockDec, d.o.concurrent) - var seqExecute = make(chan *blockDec, d.o.concurrent) + var seqDecode = make(chan *blockDec, d.frame.o.concurrent) + var seqExecute = make(chan *blockDec, d.frame.o.concurrent) // Async 1: Decode sequences... go func() { @@ -858,9 +895,9 @@ decodeStream: if err == nil { err = d.setDict(frame) } - if err == nil && d.frame.WindowSize > d.o.maxWindowSize { + if err == nil && d.frame.WindowSize > d.frame.o.maxWindowSize { if debugDecoder { - println("decoder size exceeded, fws:", d.frame.WindowSize, "> mws:", d.o.maxWindowSize) + println("decoder size exceeded, fws:", d.frame.WindowSize, "> mws:", d.frame.o.maxWindowSize) } err = ErrDecoderSizeExceeded @@ -940,7 +977,7 @@ decodeStream: } func (d *Decoder) setDict(frame *frameDec) (err error) { - dict, ok := d.o.dicts[frame.DictionaryID] + dict, ok := frame.o.dicts[frame.DictionaryID] if ok { if debugDecoder { println("setting dict", frame.DictionaryID) diff --git a/vendor/github.com/klauspost/compress/zstd/dict.go b/vendor/github.com/klauspost/compress/zstd/dict.go index 4f1c4938cd..67f40e7914 100644 --- a/vendor/github.com/klauspost/compress/zstd/dict.go +++ b/vendor/github.com/klauspost/compress/zstd/dict.go @@ -296,40 +296,81 @@ func BuildDict(o BuildDictOptions) ([]byte, error) { if offset > 3 { newOffsets[offset-3]++ } else { - newOffsets[uint32(o.Offsets[offset-1])]++ + // Repeat codes reference the training Offsets. Skip unset + // (zero) entries so they are not ranked as real offsets. + prev := o.Offsets[offset-1] + if prev > 0 { + newOffsets[uint32(prev)]++ + } } } } // Find most used offsets. var sortedOffsets []uint32 for k := range newOffsets { + if k == 0 { + continue + } sortedOffsets = append(sortedOffsets, k) } sort.Slice(sortedOffsets, func(i, j int) bool { a, b := sortedOffsets[i], sortedOffsets[j] - if a == b { + ca, cb := newOffsets[a], newOffsets[b] + if ca == cb { // Prefer the longer offset - return sortedOffsets[i] > sortedOffsets[j] + return a > b } - return newOffsets[sortedOffsets[i]] > newOffsets[sortedOffsets[j]] + return ca > cb }) - if len(sortedOffsets) > 3 { - if debug { - print("Offsets:") - for i, v := range sortedOffsets { - if i > 20 { - break - } - printf("[%d: %d],", v, newOffsets[v]) + if debug { + print("Offsets:") + for i, v := range sortedOffsets { + if i > 20 { + break } - println("") + printf("[%d: %d],", v, newOffsets[v]) + } + println("") + } + // Dictionary recent-offsets must be three positive values within the + // history. Ranked matches may be fewer (or empty when only unset + // repeat codes were seen), so fill remaining slots with defaults. + used := make(map[int]bool, 3) + var finalOffsets [3]int + nOff := 0 + for _, v := range sortedOffsets { + iv := int(v) + if iv <= 0 || iv > len(hist) || used[iv] { + continue + } + finalOffsets[nOff] = iv + used[iv] = true + nOff++ + if nOff == 3 { + break } - - sortedOffsets = sortedOffsets[:3] } - for i, v := range sortedOffsets { - o.Offsets[i] = int(v) + for _, def := range []int{1, 4, 8} { + if nOff == 3 { + break + } + if def <= len(hist) && !used[def] { + finalOffsets[nOff] = def + used[def] = true + nOff++ + } + } + for def := 1; nOff < 3 && def <= len(hist); def++ { + if !used[def] { + finalOffsets[nOff] = def + used[def] = true + nOff++ + } + } + if nOff < 3 { + return nil, fmt.Errorf("could not determine 3 valid dictionary offsets (history size %d)", len(hist)) } + o.Offsets = finalOffsets if debug { println("New repeat offsets", o.Offsets) } @@ -337,6 +378,9 @@ func BuildDict(o BuildDictOptions) ([]byte, error) { if nUsed == 0 || seqs == 0 { return nil, fmt.Errorf("%d blocks, %d sequences found", nUsed, seqs) } + if litTotal == 0 { + return nil, errors.New("0 literals found") + } if debug { println("Sequences:", seqs, "Blocks:", nUsed, "Literals:", litTotal) } @@ -517,11 +561,10 @@ func BuildDict(o BuildDictOptions) ([]byte, error) { out.Write(binary.LittleEndian.AppendUint32(nil, uint32(o.Offsets[1]))) out.Write(binary.LittleEndian.AppendUint32(nil, uint32(o.Offsets[2]))) out.Write(hist) + if _, err := loadDict(out.Bytes()); err != nil { + return nil, fmt.Errorf("built dictionary failed validation: %w", err) + } if debug { - _, err := loadDict(out.Bytes()) - if err != nil { - panic(err) - } i, err := InspectDictionary(out.Bytes()) if err != nil { panic(err) diff --git a/vendor/github.com/klauspost/compress/zstd/framedec.go b/vendor/github.com/klauspost/compress/zstd/framedec.go index d88f067e5c..81d8b9c922 100644 --- a/vendor/github.com/klauspost/compress/zstd/framedec.go +++ b/vendor/github.com/klauspost/compress/zstd/framedec.go @@ -49,13 +49,16 @@ const ( ) func newFrameDec(o decoderOptions) *frameDec { + d := frameDec{} + d.setOptions(o) + return &d +} + +func (d *frameDec) setOptions(o decoderOptions) { if o.maxWindowSize > o.maxDecodedSize { o.maxWindowSize = o.maxDecodedSize } - d := frameDec{ - o: o, - } - return &d + d.o = o } // reset will read the frame header and prepare for block decoding. diff --git a/vendor/github.com/klauspost/compress/zstd/fse_decoder_arm64.s b/vendor/github.com/klauspost/compress/zstd/fse_decoder_arm64.s index 77ee3913f0..8ca913eaec 100644 --- a/vendor/github.com/klauspost/compress/zstd/fse_decoder_arm64.s +++ b/vendor/github.com/klauspost/compress/zstd/fse_decoder_arm64.s @@ -1,7 +1,7 @@ // Code generated by command: go run gen_fse.go -out ../fse_decoder.s -arch amd64,arm64 -pkg=zstd. DO NOT EDIT. // EXPERIMENTAL arm64 output lowered from an amd64 avo program. -//go:build arm64 && !appengine && !noasm && gc && !noasm +//go:build arm64 && (!appengine && !noasm && gc && !noasm) // func buildDtable_asm(s *fseDecoder, ctx *buildDtableAsmContext) int TEXT ·buildDtable_asm(SB), $0-24 @@ -26,11 +26,10 @@ TEXT ·buildDtable_asm(SB), $0-24 JMP init_main_loop_condition init_main_loop: - ADD R8<<1, R1, R15 - MOVH (R15), R9 - AND $0xffff, R9, R15 + MOVH (R1)(R8<<1), R9 MOVD $-1, R16 AND $0xffff, R16, R16 + AND $0xffff, R9, R15 CMP R16, R15 BNE do_not_update_high_threshold ADD R7<<3, R5, R15 @@ -39,8 +38,7 @@ init_main_loop: MOVD $0x0000000000000001, R9 do_not_update_high_threshold: - ADD R8<<1, R3, R15 - MOVH R9, (R15) + MOVH R9, (R3)(R8<<1) ADD $1, R8, R8 init_main_loop_condition: @@ -49,12 +47,10 @@ init_main_loop_condition: // Spread symbols // Calculate table step - MOVD R0, R8 - LSR $0x01, R8, R8 - MOVD R0, R9 - LSR $0x03, R9, R9 - ADD R9, R8, R8 - ADD $3, R8, R8 + LSR $0x01, R0, R8 + LSR $0x03, R0, R9 + ADD R9, R8, R8 + ADD $3, R8, R8 // Fill add bits values SUB $1, R0, R9 @@ -64,8 +60,7 @@ init_main_loop_condition: spread_main_loop: MOVD $0, R12 - ADD R11<<1, R1, R15 - MOVH (R15), R13 + MOVH (R1)(R11<<1), R13 JMP spread_inner_loop_condition spread_inner_loop: @@ -102,11 +97,9 @@ spread_check_ok: build_table_main_table: ADD R6<<3, R5, R15 MOVBU 1(R15), R1 - ADD R1<<1, R3, R15 - MOVHU (R15), R7 + MOVHU (R3)(R1<<1), R7 ADD $1, R7, R8 - ADD R1<<1, R3, R15 - MOVH R8, (R15) + MOVH R8, (R3)(R1<<1) MOVD R7, R8 CLZ R8, R16 MOVD $63, R8 @@ -129,12 +122,10 @@ build_table_main_table: RET build_table_check1_ok: - AND $0xff, R1, R15 - AND $0xff, R1, R16 - TST R16, R15 + TST $0xff, R1 BNE build_table_check2_ok - AND $0xffff, R7, R15 AND $0xffff, R6, R16 + AND $0xffff, R7, R15 CMP R16, R15 BNE build_table_check2_ok MOVD ctx+8(FP), R0 diff --git a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash.go b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash.go index fc40c82001..9c0cbc1681 100644 --- a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash.go +++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash.go @@ -52,6 +52,11 @@ func (d *Digest) Reset() { d.n = 0 } +// maxAsmSize bounds the input handed to a single writeBlocks call. Assembly is +// never preemptible, so an unbounded call holds every P in stop-the-world for +// its duration. Whole 32-byte blocks, so Write can feed it in pieces. +const maxAsmSize = 128 << 10 // 4096 whole 32-byte blocks + // Size always returns 8 bytes. func (d *Digest) Size() int { return 8 } @@ -83,6 +88,13 @@ func (d *Digest) Write(b []byte) (n int, err error) { d.n = 0 } + for len(b) >= maxAsmSize { + // Assembly is not preemptible, so hashing a large buffer in one call + // blocks every stop-the-world for the whole call. Feed it in chunks. + writeBlocks(d, b[:maxAsmSize]) + b = b[maxAsmSize:] + } + if len(b) >= 32 { // One or more full blocks left. nw := writeBlocks(d, b) 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 index ddb63aa91b..e4c688efd3 100644 --- a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_amd64.s +++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_amd64.s @@ -175,7 +175,11 @@ finalize: RET // func writeBlocks(d *Digest, b []byte) int -TEXT ·writeBlocks(SB), NOSPLIT|NOFRAME, $0-40 +// +// Deliberately not NOSPLIT: the stack check is the preemption point that lets a +// stop-the-world proceed between chunks. The frame must be >= abi.StackSmall +// (128), or the assembler marks this leaf NOSPLIT anyway and drops that check. +TEXT ·writeBlocks(SB), $128-40 // Load fixed primes needed for round. MOVQ ·primes+0(SB), prime1 MOVQ ·primes+8(SB), prime2 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 index ae7d4d3295..e88cc2fc91 100644 --- a/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_arm64.s +++ b/vendor/github.com/klauspost/compress/zstd/internal/xxhash/xxhash_arm64.s @@ -163,7 +163,11 @@ finalize: RET // func writeBlocks(s *Digest, b []byte) int -TEXT ·writeBlocks(SB), NOSPLIT|NOFRAME, $0-40 +// +// Deliberately not NOSPLIT: the stack check is the preemption point that lets a +// stop-the-world proceed between chunks. The frame must be >= abi.StackSmall +// (128), or the assembler marks this leaf NOSPLIT anyway and drops that check. +TEXT ·writeBlocks(SB), $128-40 LDP ·primes+0(SB), (prime1, prime2) // Load state. Assume v[1-4] are stored contiguously. diff --git a/vendor/github.com/klauspost/compress/zstd/seqdec.go b/vendor/github.com/klauspost/compress/zstd/seqdec.go index 0bfb0e43c7..d1a7732fcc 100644 --- a/vendor/github.com/klauspost/compress/zstd/seqdec.go +++ b/vendor/github.com/klauspost/compress/zstd/seqdec.go @@ -99,6 +99,29 @@ func (s *sequenceDecs) initialize(br *bitReader, hist *history, out []byte) erro return nil } +// consumeSyncLen accounts for n bytes of frame output produced outside +// decodeSync. maxSyncLen bounds how much the frame may still produce and +// decodeSyncSimple compares it with the output buffer's slack to decide +// whether the extended 16-byte copies are safe. Blocks with sequences +// subtracted their output; raw blocks, RLE blocks and compressed blocks +// holding only literals did not, so a frame that opened with a raw block +// overstated its remaining size for every block after it and ran the +// bounds-exact copies for the rest of the frame. +// +// Zero means "no bound" and selects the conservative paths, so a block that +// meets or exceeds the bound lands there; the frame-size checks in +// runDecoder reject the excess afterwards. +func (s *sequenceDecs) consumeSyncLen(n int) { + if s.maxSyncLen == 0 { + return + } + if uint64(n) >= s.maxSyncLen { + s.maxSyncLen = 0 + return + } + s.maxSyncLen -= uint64(n) +} + func (s *sequenceDecs) freeDecoders() { if f := s.litLengths.fse; f != nil && !f.preDefined { fseDecoderPool.Put(f) diff --git a/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.s b/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.s index 3fc381c7a7..798d940231 100644 --- a/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.s +++ b/vendor/github.com/klauspost/compress/zstd/seqdec_amd64.s @@ -4,7 +4,7 @@ // func sequenceDecs_decode_amd64(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int // Requires: CMOV -TEXT ·sequenceDecs_decode_amd64(SB), $8-32 +TEXT ·sequenceDecs_decode_amd64(SB), $40-32 MOVQ br+8(FP), CX MOVQ 24(CX), DX MOVBQZX 40(CX), BX @@ -16,6 +16,14 @@ TEXT ·sequenceDecs_decode_amd64(SB), $8-32 MOVQ 72(AX), DI MOVQ 80(AX), R8 MOVQ 88(AX), R9 + MOVQ (AX), CX + MOVQ CX, 8(SP) + MOVQ 24(AX), CX + MOVQ CX, 16(SP) + MOVQ 48(AX), CX + MOVQ CX, 24(SP) + MOVQ 96(AX), CX + MOVQ CX, 32(SP) MOVQ 104(AX), R10 MOVQ s+0(FP), AX MOVQ 144(AX), R11 @@ -55,46 +63,32 @@ sequenceDecs_decode_amd64_fill_check_overread: sequenceDecs_decode_amd64_fill_end: // Update offset - MOVQ R9, AX - MOVQ BX, CX - MOVQ DX, R15 - SHLQ CL, R15 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decode_amd64_of_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decode_amd64_of_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decode_amd64_of_update_zero - NEGQ CX - SHRQ CL, R15 - ADDQ R15, AX - -sequenceDecs_decode_amd64_of_update_zero: - MOVQ AX, 16(R10) + MOVQ R9, AX + MOVQ BX, CX + MOVQ DX, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R15 + ADDQ R15, AX + MOVQ AX, 16(R10) // Update match length - MOVQ R8, AX - MOVQ BX, CX - MOVQ DX, R15 - SHLQ CL, R15 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decode_amd64_ml_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decode_amd64_ml_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decode_amd64_ml_update_zero - NEGQ CX - SHRQ CL, R15 - ADDQ R15, AX - -sequenceDecs_decode_amd64_ml_update_zero: - MOVQ AX, 8(R10) + MOVQ R8, AX + MOVQ BX, CX + MOVQ DX, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R15 + ADDQ R15, AX + MOVQ AX, 8(R10) // Fill bitreader to have enough for the remaining CMPQ SI, $0x08 @@ -126,90 +120,76 @@ sequenceDecs_decode_amd64_fill_2_check_overread: sequenceDecs_decode_amd64_fill_2_end: // Update literal length - MOVQ DI, AX - MOVQ BX, CX - MOVQ DX, R15 - SHLQ CL, R15 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decode_amd64_ll_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decode_amd64_ll_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decode_amd64_ll_update_zero - NEGQ CX - SHRQ CL, R15 - ADDQ R15, AX - -sequenceDecs_decode_amd64_ll_update_zero: - MOVQ AX, (R10) + MOVQ DI, AX + MOVQ BX, CX + MOVQ DX, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R15 + ADDQ R15, AX + MOVQ AX, (R10) // Fill bitreader for state updates MOVQ R14, (SP) MOVQ R9, AX SHRQ $0x08, AX MOVBQZX AL, AX - MOVQ ctx+16(FP), CX - CMPQ 96(CX), $0x00 + CMPQ 32(SP), $0x00 JZ sequenceDecs_decode_amd64_skip_update // Update Literal Length State MOVBQZX DI, R14 SHRL $0x10, DI - LEAQ (BX)(R14*1), CX + MOVQ BX, CX MOVQ DX, R15 - MOVQ CX, BX - ROLQ CL, R15 - MOVL $0x00000001, BP - MOVB R14, CL - SHLL CL, BP - DECL BP - ANDQ BP, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + ADDQ R14, BX + XORQ $0x3f, R14 + MOVQ R14, CX + SHRQ CL, R15 ADDQ R15, DI // Load ctx.llTable - MOVQ ctx+16(FP), CX - MOVQ (CX), CX + MOVQ 8(SP), CX MOVQ (CX)(DI*8), DI // Update Match Length State MOVBQZX R8, R14 SHRL $0x10, R8 - LEAQ (BX)(R14*1), CX + MOVQ BX, CX MOVQ DX, R15 - MOVQ CX, BX - ROLQ CL, R15 - MOVL $0x00000001, BP - MOVB R14, CL - SHLL CL, BP - DECL BP - ANDQ BP, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + ADDQ R14, BX + XORQ $0x3f, R14 + MOVQ R14, CX + SHRQ CL, R15 ADDQ R15, R8 // Load ctx.mlTable - MOVQ ctx+16(FP), CX - MOVQ 24(CX), CX + MOVQ 16(SP), CX MOVQ (CX)(R8*8), R8 // Update Offset State MOVBQZX R9, R14 SHRL $0x10, R9 - LEAQ (BX)(R14*1), CX + MOVQ BX, CX MOVQ DX, R15 - MOVQ CX, BX - ROLQ CL, R15 - MOVL $0x00000001, BP - MOVB R14, CL - SHLL CL, BP - DECL BP - ANDQ BP, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + ADDQ R14, BX + XORQ $0x3f, R14 + MOVQ R14, CX + SHRQ CL, R15 ADDQ R15, R9 // Load ctx.ofTable - MOVQ ctx+16(FP), CX - MOVQ 48(CX), CX + MOVQ 24(SP), CX MOVQ (CX)(R9*8), R9 sequenceDecs_decode_amd64_skip_update: @@ -290,8 +270,7 @@ sequenceDecs_decode_amd64_after_adjust: sequenceDecs_decode_amd64_match_len_ofs_ok: ADDQ $0x18, R10 - MOVQ ctx+16(FP), AX - DECQ 96(AX) + DECQ 32(SP) JNS sequenceDecs_decode_amd64_main_loop MOVQ s+0(FP), AX MOVQ R11, 144(AX) @@ -303,36 +282,54 @@ sequenceDecs_decode_amd64_match_len_ofs_ok: MOVQ SI, 32(AX) // Return success + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000000, ret+24(FP) RET // Return with match length error sequenceDecs_decode_amd64_error_match_len_ofs_mismatch: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000001, ret+24(FP) RET // Return with match too long error sequenceDecs_decode_amd64_error_match_len_too_big: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000002, ret+24(FP) RET // Return with match offset too long error + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000003, ret+24(FP) RET // Return with not enough literals error error_not_enough_literals: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000004, ret+24(FP) RET // Return with overread error error_overread: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000006, ret+24(FP) RET // func sequenceDecs_decode_56_amd64(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int // Requires: CMOV -TEXT ·sequenceDecs_decode_56_amd64(SB), $8-32 +TEXT ·sequenceDecs_decode_56_amd64(SB), $40-32 MOVQ br+8(FP), CX MOVQ 24(CX), DX MOVBQZX 40(CX), BX @@ -344,6 +341,14 @@ TEXT ·sequenceDecs_decode_56_amd64(SB), $8-32 MOVQ 72(AX), DI MOVQ 80(AX), R8 MOVQ 88(AX), R9 + MOVQ (AX), CX + MOVQ CX, 8(SP) + MOVQ 24(AX), CX + MOVQ CX, 16(SP) + MOVQ 48(AX), CX + MOVQ CX, 24(SP) + MOVQ 96(AX), CX + MOVQ CX, 32(SP) MOVQ 104(AX), R10 MOVQ s+0(FP), AX MOVQ 144(AX), R11 @@ -383,132 +388,104 @@ sequenceDecs_decode_56_amd64_fill_check_overread: sequenceDecs_decode_56_amd64_fill_end: // Update offset - MOVQ R9, AX - MOVQ BX, CX - MOVQ DX, R15 - SHLQ CL, R15 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decode_56_amd64_of_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decode_56_amd64_of_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decode_56_amd64_of_update_zero - NEGQ CX - SHRQ CL, R15 - ADDQ R15, AX - -sequenceDecs_decode_56_amd64_of_update_zero: - MOVQ AX, 16(R10) + MOVQ R9, AX + MOVQ BX, CX + MOVQ DX, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R15 + ADDQ R15, AX + MOVQ AX, 16(R10) // Update match length - MOVQ R8, AX - MOVQ BX, CX - MOVQ DX, R15 - SHLQ CL, R15 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decode_56_amd64_ml_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decode_56_amd64_ml_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decode_56_amd64_ml_update_zero - NEGQ CX - SHRQ CL, R15 - ADDQ R15, AX - -sequenceDecs_decode_56_amd64_ml_update_zero: - MOVQ AX, 8(R10) + MOVQ R8, AX + MOVQ BX, CX + MOVQ DX, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R15 + ADDQ R15, AX + MOVQ AX, 8(R10) // Update literal length - MOVQ DI, AX - MOVQ BX, CX - MOVQ DX, R15 - SHLQ CL, R15 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decode_56_amd64_ll_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decode_56_amd64_ll_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decode_56_amd64_ll_update_zero - NEGQ CX - SHRQ CL, R15 - ADDQ R15, AX - -sequenceDecs_decode_56_amd64_ll_update_zero: - MOVQ AX, (R10) + MOVQ DI, AX + MOVQ BX, CX + MOVQ DX, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R15 + ADDQ R15, AX + MOVQ AX, (R10) // Fill bitreader for state updates MOVQ R14, (SP) MOVQ R9, AX SHRQ $0x08, AX MOVBQZX AL, AX - MOVQ ctx+16(FP), CX - CMPQ 96(CX), $0x00 + CMPQ 32(SP), $0x00 JZ sequenceDecs_decode_56_amd64_skip_update // Update Literal Length State MOVBQZX DI, R14 SHRL $0x10, DI - LEAQ (BX)(R14*1), CX + MOVQ BX, CX MOVQ DX, R15 - MOVQ CX, BX - ROLQ CL, R15 - MOVL $0x00000001, BP - MOVB R14, CL - SHLL CL, BP - DECL BP - ANDQ BP, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + ADDQ R14, BX + XORQ $0x3f, R14 + MOVQ R14, CX + SHRQ CL, R15 ADDQ R15, DI // Load ctx.llTable - MOVQ ctx+16(FP), CX - MOVQ (CX), CX + MOVQ 8(SP), CX MOVQ (CX)(DI*8), DI // Update Match Length State MOVBQZX R8, R14 SHRL $0x10, R8 - LEAQ (BX)(R14*1), CX + MOVQ BX, CX MOVQ DX, R15 - MOVQ CX, BX - ROLQ CL, R15 - MOVL $0x00000001, BP - MOVB R14, CL - SHLL CL, BP - DECL BP - ANDQ BP, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + ADDQ R14, BX + XORQ $0x3f, R14 + MOVQ R14, CX + SHRQ CL, R15 ADDQ R15, R8 // Load ctx.mlTable - MOVQ ctx+16(FP), CX - MOVQ 24(CX), CX + MOVQ 16(SP), CX MOVQ (CX)(R8*8), R8 // Update Offset State MOVBQZX R9, R14 SHRL $0x10, R9 - LEAQ (BX)(R14*1), CX + MOVQ BX, CX MOVQ DX, R15 - MOVQ CX, BX - ROLQ CL, R15 - MOVL $0x00000001, BP - MOVB R14, CL - SHLL CL, BP - DECL BP - ANDQ BP, R15 + SHLQ CL, R15 + SHRQ $0x01, R15 + ADDQ R14, BX + XORQ $0x3f, R14 + MOVQ R14, CX + SHRQ CL, R15 ADDQ R15, R9 // Load ctx.ofTable - MOVQ ctx+16(FP), CX - MOVQ 48(CX), CX + MOVQ 24(SP), CX MOVQ (CX)(R9*8), R9 sequenceDecs_decode_56_amd64_skip_update: @@ -589,8 +566,7 @@ sequenceDecs_decode_56_amd64_after_adjust: sequenceDecs_decode_56_amd64_match_len_ofs_ok: ADDQ $0x18, R10 - MOVQ ctx+16(FP), AX - DECQ 96(AX) + DECQ 32(SP) JNS sequenceDecs_decode_56_amd64_main_loop MOVQ s+0(FP), AX MOVQ R11, 144(AX) @@ -602,36 +578,54 @@ sequenceDecs_decode_56_amd64_match_len_ofs_ok: MOVQ SI, 32(AX) // Return success + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000000, ret+24(FP) RET // Return with match length error sequenceDecs_decode_56_amd64_error_match_len_ofs_mismatch: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000001, ret+24(FP) RET // Return with match too long error sequenceDecs_decode_56_amd64_error_match_len_too_big: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000002, ret+24(FP) RET // Return with match offset too long error + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000003, ret+24(FP) RET // Return with not enough literals error error_not_enough_literals: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000004, ret+24(FP) RET // Return with overread error error_overread: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000006, ret+24(FP) RET // func sequenceDecs_decode_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int // Requires: BMI, BMI2, CMOV -TEXT ·sequenceDecs_decode_bmi2(SB), $8-32 +TEXT ·sequenceDecs_decode_bmi2(SB), $40-32 MOVQ br+8(FP), BX MOVQ 24(BX), AX MOVBQZX 40(BX), DX @@ -643,6 +637,14 @@ TEXT ·sequenceDecs_decode_bmi2(SB), $8-32 MOVQ 72(CX), SI MOVQ 80(CX), DI MOVQ 88(CX), R8 + MOVQ (CX), R9 + MOVQ R9, 8(SP) + MOVQ 24(CX), R9 + MOVQ R9, 16(SP) + MOVQ 48(CX), R9 + MOVQ R9, 24(SP) + MOVQ 96(CX), R9 + MOVQ R9, 32(SP) MOVQ 104(CX), R9 MOVQ s+0(FP), CX MOVQ 144(CX), R10 @@ -753,8 +755,7 @@ sequenceDecs_decode_bmi2_fill_2_end: MOVQ R13, (SP) MOVQ $0x00000808, CX BEXTRQ CX, R8, R13 - MOVQ ctx+16(FP), CX - CMPQ 96(CX), $0x00 + CMPQ 32(SP), $0x00 JZ sequenceDecs_decode_bmi2_skip_update LEAQ (SI)(DI*1), R14 ADDQ R8, R14 @@ -772,8 +773,7 @@ sequenceDecs_decode_bmi2_fill_2_end: ADDQ CX, R8 // Load ctx.ofTable - MOVQ ctx+16(FP), CX - MOVQ 48(CX), CX + MOVQ 24(SP), CX MOVQ (CX)(R8*8), R8 // Update Match Length State @@ -783,8 +783,7 @@ sequenceDecs_decode_bmi2_fill_2_end: ADDQ CX, DI // Load ctx.mlTable - MOVQ ctx+16(FP), CX - MOVQ 24(CX), CX + MOVQ 16(SP), CX MOVQ (CX)(DI*8), DI // Update Literal Length State @@ -793,8 +792,7 @@ sequenceDecs_decode_bmi2_fill_2_end: ADDQ CX, SI // Load ctx.llTable - MOVQ ctx+16(FP), CX - MOVQ (CX), CX + MOVQ 8(SP), CX MOVQ (CX)(SI*8), SI sequenceDecs_decode_bmi2_skip_update: @@ -875,8 +873,7 @@ sequenceDecs_decode_bmi2_after_adjust: sequenceDecs_decode_bmi2_match_len_ofs_ok: ADDQ $0x18, R9 - MOVQ ctx+16(FP), CX - DECQ 96(CX) + DECQ 32(SP) JNS sequenceDecs_decode_bmi2_main_loop MOVQ s+0(FP), CX MOVQ R10, 144(CX) @@ -888,36 +885,54 @@ sequenceDecs_decode_bmi2_match_len_ofs_ok: MOVQ BX, 32(CX) // Return success + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000000, ret+24(FP) RET // Return with match length error sequenceDecs_decode_bmi2_error_match_len_ofs_mismatch: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000001, ret+24(FP) RET // Return with match too long error sequenceDecs_decode_bmi2_error_match_len_too_big: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000002, ret+24(FP) RET // Return with match offset too long error + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000003, ret+24(FP) RET // Return with not enough literals error error_not_enough_literals: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000004, ret+24(FP) RET // Return with overread error error_overread: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000006, ret+24(FP) RET // func sequenceDecs_decode_56_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int // Requires: BMI, BMI2, CMOV -TEXT ·sequenceDecs_decode_56_bmi2(SB), $8-32 +TEXT ·sequenceDecs_decode_56_bmi2(SB), $40-32 MOVQ br+8(FP), BX MOVQ 24(BX), AX MOVBQZX 40(BX), DX @@ -929,6 +944,14 @@ TEXT ·sequenceDecs_decode_56_bmi2(SB), $8-32 MOVQ 72(CX), SI MOVQ 80(CX), DI MOVQ 88(CX), R8 + MOVQ (CX), R9 + MOVQ R9, 8(SP) + MOVQ 24(CX), R9 + MOVQ R9, 16(SP) + MOVQ 48(CX), R9 + MOVQ R9, 24(SP) + MOVQ 96(CX), R9 + MOVQ R9, 32(SP) MOVQ 104(CX), R9 MOVQ s+0(FP), CX MOVQ 144(CX), R10 @@ -1010,8 +1033,7 @@ sequenceDecs_decode_56_bmi2_fill_end: MOVQ R13, (SP) MOVQ $0x00000808, CX BEXTRQ CX, R8, R13 - MOVQ ctx+16(FP), CX - CMPQ 96(CX), $0x00 + CMPQ 32(SP), $0x00 JZ sequenceDecs_decode_56_bmi2_skip_update LEAQ (SI)(DI*1), R14 ADDQ R8, R14 @@ -1029,8 +1051,7 @@ sequenceDecs_decode_56_bmi2_fill_end: ADDQ CX, R8 // Load ctx.ofTable - MOVQ ctx+16(FP), CX - MOVQ 48(CX), CX + MOVQ 24(SP), CX MOVQ (CX)(R8*8), R8 // Update Match Length State @@ -1040,8 +1061,7 @@ sequenceDecs_decode_56_bmi2_fill_end: ADDQ CX, DI // Load ctx.mlTable - MOVQ ctx+16(FP), CX - MOVQ 24(CX), CX + MOVQ 16(SP), CX MOVQ (CX)(DI*8), DI // Update Literal Length State @@ -1050,8 +1070,7 @@ sequenceDecs_decode_56_bmi2_fill_end: ADDQ CX, SI // Load ctx.llTable - MOVQ ctx+16(FP), CX - MOVQ (CX), CX + MOVQ 8(SP), CX MOVQ (CX)(SI*8), SI sequenceDecs_decode_56_bmi2_skip_update: @@ -1132,8 +1151,7 @@ sequenceDecs_decode_56_bmi2_after_adjust: sequenceDecs_decode_56_bmi2_match_len_ofs_ok: ADDQ $0x18, R9 - MOVQ ctx+16(FP), CX - DECQ 96(CX) + DECQ 32(SP) JNS sequenceDecs_decode_56_bmi2_main_loop MOVQ s+0(FP), CX MOVQ R10, 144(CX) @@ -1145,30 +1163,48 @@ sequenceDecs_decode_56_bmi2_match_len_ofs_ok: MOVQ BX, 32(CX) // Return success + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000000, ret+24(FP) RET // Return with match length error sequenceDecs_decode_56_bmi2_error_match_len_ofs_mismatch: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000001, ret+24(FP) RET // Return with match too long error sequenceDecs_decode_56_bmi2_error_match_len_too_big: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000002, ret+24(FP) RET // Return with match offset too long error + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000003, ret+24(FP) RET // Return with not enough literals error error_not_enough_literals: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000004, ret+24(FP) RET // Return with overread error error_overread: + MOVQ 32(SP), AX + MOVQ ctx+16(FP), CX + MOVQ AX, 96(CX) MOVQ $0x00000006, ret+24(FP) RET @@ -1203,9 +1239,11 @@ main_loop: MOVQ 8(AX), R13 // Copy literals - TESTQ R11, R11 - JZ check_offset - XORQ R14, R14 + MOVUPS (SI), X0 + MOVUPS X0, (BX) + CMPQ R11, $0x10 + JBE copy_1_end + MOVQ $0x00000010, R14 copy_1: MOVUPS (SI)(R14*1), X0 @@ -1213,12 +1251,13 @@ copy_1: ADDQ $0x10, R14 CMPQ R14, R11 JB copy_1 - ADDQ R11, SI - ADDQ R11, BX - ADDQ R11, DI + +copy_1_end: + ADDQ R11, SI + ADDQ R11, BX + ADDQ R11, DI // Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize) -check_offset: LEAQ (DI)(R10*1), R11 CMPQ R12, R11 JG error_match_off_too_big @@ -1258,13 +1297,13 @@ copy_4_small: JMP copy_4_move_8through16 copy_4_move_3: - MOVW (R14), R11 - MOVB 2(R14), R12 - MOVW R11, (BX) - MOVB R12, 2(BX) - ADDQ R13, R14 - ADDQ R13, BX - JMP copy_4_end + MOVWQZX (R14), R11 + MOVB 2(R14), R12 + MOVW R11, (BX) + MOVB R12, 2(BX) + ADDQ R13, R14 + ADDQ R13, BX + JMP copy_4_end copy_4_move_4through7: MOVL (R14), R11 @@ -1327,13 +1366,13 @@ copy_5_move_1or2: JMP copy_5_end copy_5_move_3: - MOVW (R14), R15 - MOVB 2(R14), BP - MOVW R15, (BX) - MOVB BP, 2(BX) - ADDQ R11, R14 - ADDQ R11, BX - JMP copy_5_end + MOVWQZX (R14), R15 + MOVB 2(R14), BP + MOVW R15, (BX) + MOVB BP, 2(BX) + ADDQ R11, R14 + ADDQ R11, BX + JMP copy_5_end copy_5_move_4through7: MOVL (R14), R15 @@ -1366,9 +1405,19 @@ copy_match: JA copy_overlapping_match // Copy non-overlapping match - ADDQ R13, DI - MOVQ BX, R12 + ADDQ R13, DI + MOVUPS (R11), X0 + MOVUPS X0, (BX) + CMPQ R13, $0x10 + JA copy_2_long + ADDQ R13, BX + JMP handle_loop + +copy_2_long: + LEAQ 16(BX), R12 ADDQ R13, BX + ADDQ $0x10, R11 + SUBQ $0x10, R13 copy_2: MOVUPS (R11), X0 @@ -1384,12 +1433,12 @@ copy_overlapping_match: ADDQ R13, DI copy_slow_3: - MOVB (R11), R12 - MOVB R12, (BX) - INCQ R11 - INCQ BX - DECQ R13 - JNZ copy_slow_3 + MOVBQZX (R11), R12 + MOVB R12, (BX) + INCQ R11 + INCQ BX + DECQ R13 + JNZ copy_slow_3 handle_loop: ADDQ $0x18, AX @@ -1494,13 +1543,13 @@ copy_1_move_1or2: JMP copy_1_end copy_1_move_3: - MOVW (SI), R14 - MOVB 2(SI), R15 - MOVW R14, (BX) - MOVB R15, 2(BX) - ADDQ R11, SI - ADDQ R11, BX - JMP copy_1_end + MOVWQZX (SI), R14 + MOVB 2(SI), R15 + MOVW R14, (BX) + MOVB R15, 2(BX) + ADDQ R11, SI + ADDQ R11, BX + JMP copy_1_end copy_1_move_4through7: MOVL (SI), R14 @@ -1563,13 +1612,13 @@ copy_4_small: JMP copy_4_move_8through16 copy_4_move_3: - MOVW (R14), R11 - MOVB 2(R14), R12 - MOVW R11, (BX) - MOVB R12, 2(BX) - ADDQ R13, R14 - ADDQ R13, BX - JMP copy_4_end + MOVWQZX (R14), R11 + MOVB 2(R14), R12 + MOVW R11, (BX) + MOVB R12, 2(BX) + ADDQ R13, R14 + ADDQ R13, BX + JMP copy_4_end copy_4_move_4through7: MOVL (R14), R11 @@ -1632,13 +1681,13 @@ copy_5_move_1or2: JMP copy_5_end copy_5_move_3: - MOVW (R14), R15 - MOVB 2(R14), BP - MOVW R15, (BX) - MOVB BP, 2(BX) - ADDQ R11, R14 - ADDQ R11, BX - JMP copy_5_end + MOVWQZX (R14), R15 + MOVB 2(R14), BP + MOVW R15, (BX) + MOVB BP, 2(BX) + ADDQ R11, R14 + ADDQ R11, BX + JMP copy_5_end copy_5_move_4through7: MOVL (R14), R15 @@ -1707,13 +1756,13 @@ copy_2_move_1or2: JMP copy_2_end copy_2_move_3: - MOVW (R11), R12 - MOVB 2(R11), R14 - MOVW R12, (BX) - MOVB R14, 2(BX) - ADDQ R13, R11 - ADDQ R13, BX - JMP copy_2_end + MOVWQZX (R11), R12 + MOVB 2(R11), R14 + MOVW R12, (BX) + MOVB R14, 2(BX) + ADDQ R13, R11 + ADDQ R13, BX + JMP copy_2_end copy_2_move_4through7: MOVL (R11), R12 @@ -1740,12 +1789,12 @@ copy_overlapping_match: ADDQ R13, DI copy_slow_3: - MOVB (R11), R12 - MOVB R12, (BX) - INCQ R11 - INCQ BX - DECQ R13 - JNZ copy_slow_3 + MOVBQZX (R11), R12 + MOVB R12, (BX) + INCQ R11 + INCQ BX + DECQ R13 + JNZ copy_slow_3 handle_loop: ADDQ $0x18, AX @@ -1784,7 +1833,7 @@ empty_seqs: // func sequenceDecs_decodeSync_amd64(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int // Requires: CMOV, SSE -TEXT ·sequenceDecs_decodeSync_amd64(SB), $64-32 +TEXT ·sequenceDecs_decodeSync_amd64(SB), $96-32 MOVQ br+8(FP), CX MOVQ 24(CX), DX MOVBQZX 40(CX), BX @@ -1796,26 +1845,34 @@ TEXT ·sequenceDecs_decodeSync_amd64(SB), $64-32 MOVQ 72(AX), DI MOVQ 80(AX), R8 MOVQ 88(AX), R9 - XORQ CX, CX + MOVQ (AX), CX MOVQ CX, 8(SP) + MOVQ 24(AX), CX MOVQ CX, 16(SP) + MOVQ 48(AX), CX MOVQ CX, 24(SP) + MOVQ 96(AX), CX + MOVQ CX, 32(SP) + XORQ CX, CX + MOVQ CX, 40(SP) + MOVQ CX, 48(SP) + MOVQ CX, 56(SP) MOVQ 112(AX), R10 MOVQ 128(AX), CX - MOVQ CX, 32(SP) + MOVQ CX, 64(SP) MOVQ 144(AX), R11 MOVQ 136(AX), R12 MOVQ 200(AX), CX - MOVQ CX, 56(SP) + MOVQ CX, 88(SP) MOVQ 176(AX), CX - MOVQ CX, 48(SP) + MOVQ CX, 80(SP) MOVQ 184(AX), AX - MOVQ AX, 40(SP) - MOVQ 40(SP), AX - ADDQ AX, 48(SP) + MOVQ AX, 72(SP) + MOVQ 72(SP), AX + ADDQ AX, 80(SP) // Calculate pointer to s.out[cap(s.out)] (a past-end pointer) - ADDQ R10, 32(SP) + ADDQ R10, 64(SP) // outBase += outPosition ADDQ R12, R10 @@ -1853,46 +1910,32 @@ sequenceDecs_decodeSync_amd64_fill_check_overread: sequenceDecs_decodeSync_amd64_fill_end: // Update offset - MOVQ R9, AX - MOVQ BX, CX - MOVQ DX, R14 - SHLQ CL, R14 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decodeSync_amd64_of_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decodeSync_amd64_of_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decodeSync_amd64_of_update_zero - NEGQ CX - SHRQ CL, R14 - ADDQ R14, AX - -sequenceDecs_decodeSync_amd64_of_update_zero: - MOVQ AX, 8(SP) + MOVQ R9, AX + MOVQ BX, CX + MOVQ DX, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R14 + ADDQ R14, AX + MOVQ AX, 40(SP) // Update match length - MOVQ R8, AX - MOVQ BX, CX - MOVQ DX, R14 - SHLQ CL, R14 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decodeSync_amd64_ml_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decodeSync_amd64_ml_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decodeSync_amd64_ml_update_zero - NEGQ CX - SHRQ CL, R14 - ADDQ R14, AX - -sequenceDecs_decodeSync_amd64_ml_update_zero: - MOVQ AX, 16(SP) + MOVQ R8, AX + MOVQ BX, CX + MOVQ DX, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R14 + ADDQ R14, AX + MOVQ AX, 48(SP) // Fill bitreader to have enough for the remaining CMPQ SI, $0x08 @@ -1924,96 +1967,82 @@ sequenceDecs_decodeSync_amd64_fill_2_check_overread: sequenceDecs_decodeSync_amd64_fill_2_end: // Update literal length - MOVQ DI, AX - MOVQ BX, CX - MOVQ DX, R14 - SHLQ CL, R14 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decodeSync_amd64_ll_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decodeSync_amd64_ll_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decodeSync_amd64_ll_update_zero - NEGQ CX - SHRQ CL, R14 - ADDQ R14, AX - -sequenceDecs_decodeSync_amd64_ll_update_zero: - MOVQ AX, 24(SP) + MOVQ DI, AX + MOVQ BX, CX + MOVQ DX, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R14 + ADDQ R14, AX + MOVQ AX, 56(SP) // Fill bitreader for state updates MOVQ R13, (SP) MOVQ R9, AX SHRQ $0x08, AX MOVBQZX AL, AX - MOVQ ctx+16(FP), CX - CMPQ 96(CX), $0x00 + CMPQ 32(SP), $0x00 JZ sequenceDecs_decodeSync_amd64_skip_update // Update Literal Length State MOVBQZX DI, R13 SHRL $0x10, DI - LEAQ (BX)(R13*1), CX + MOVQ BX, CX MOVQ DX, R14 - MOVQ CX, BX - ROLQ CL, R14 - MOVL $0x00000001, R15 - MOVB R13, CL - SHLL CL, R15 - DECL R15 - ANDQ R15, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + ADDQ R13, BX + XORQ $0x3f, R13 + MOVQ R13, CX + SHRQ CL, R14 ADDQ R14, DI // Load ctx.llTable - MOVQ ctx+16(FP), CX - MOVQ (CX), CX + MOVQ 8(SP), CX MOVQ (CX)(DI*8), DI // Update Match Length State MOVBQZX R8, R13 SHRL $0x10, R8 - LEAQ (BX)(R13*1), CX + MOVQ BX, CX MOVQ DX, R14 - MOVQ CX, BX - ROLQ CL, R14 - MOVL $0x00000001, R15 - MOVB R13, CL - SHLL CL, R15 - DECL R15 - ANDQ R15, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + ADDQ R13, BX + XORQ $0x3f, R13 + MOVQ R13, CX + SHRQ CL, R14 ADDQ R14, R8 // Load ctx.mlTable - MOVQ ctx+16(FP), CX - MOVQ 24(CX), CX + MOVQ 16(SP), CX MOVQ (CX)(R8*8), R8 // Update Offset State MOVBQZX R9, R13 SHRL $0x10, R9 - LEAQ (BX)(R13*1), CX + MOVQ BX, CX MOVQ DX, R14 - MOVQ CX, BX - ROLQ CL, R14 - MOVL $0x00000001, R15 - MOVB R13, CL - SHLL CL, R15 - DECL R15 - ANDQ R15, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + ADDQ R13, BX + XORQ $0x3f, R13 + MOVQ R13, CX + SHRQ CL, R14 ADDQ R14, R9 // Load ctx.ofTable - MOVQ ctx+16(FP), CX - MOVQ 48(CX), CX + MOVQ 24(SP), CX MOVQ (CX)(R9*8), R9 sequenceDecs_decodeSync_amd64_skip_update: // Adjust offset MOVQ s+0(FP), CX - MOVQ 8(SP), R13 + MOVQ 40(SP), R13 CMPQ AX, $0x01 JBE sequenceDecs_decodeSync_amd64_adjust_offsetB_1_or_0 MOVUPS 144(CX), X0 @@ -2022,7 +2051,7 @@ sequenceDecs_decodeSync_amd64_skip_update: JMP sequenceDecs_decodeSync_amd64_after_adjust sequenceDecs_decodeSync_amd64_adjust_offsetB_1_or_0: - CMPQ 24(SP), $0x00000000 + CMPQ 56(SP), $0x00000000 JNE sequenceDecs_decodeSync_amd64_adjust_offset_maybezero INCQ R13 JMP sequenceDecs_decodeSync_amd64_adjust_offset_nonzero @@ -2057,14 +2086,11 @@ sequenceDecs_decodeSync_amd64_adjust_skip: MOVQ R14, R13 sequenceDecs_decodeSync_amd64_after_adjust: - MOVQ R13, 8(SP) + MOVQ R13, 40(SP) // Check values - MOVQ 16(SP), AX - MOVQ 24(SP), CX - LEAQ (AX)(CX*1), R14 - MOVQ s+0(FP), R15 - ADDQ R14, 256(R15) + MOVQ 48(SP), AX + MOVQ 56(SP), CX MOVQ ctx+16(FP), R14 SUBQ CX, 104(R14) JS error_not_enough_literals @@ -2076,50 +2102,49 @@ sequenceDecs_decodeSync_amd64_after_adjust: JNZ sequenceDecs_decodeSync_amd64_error_match_len_ofs_mismatch sequenceDecs_decodeSync_amd64_match_len_ofs_ok: - MOVQ 24(SP), AX - MOVQ 8(SP), CX - MOVQ 16(SP), R13 - // Check if we have enough space in s.out - LEAQ (AX)(R13*1), R14 + LEAQ 16(CX)(AX*1), R14 ADDQ R10, R14 - CMPQ R14, 32(SP) + CMPQ R14, 64(SP) JA error_not_enough_space // Copy literals - TESTQ AX, AX - JZ check_offset - XORQ R14, R14 + MOVUPS (R11), X0 + MOVUPS X0, (R10) + CMPQ CX, $0x10 + JBE copy_1_end + MOVQ $0x00000010, R14 copy_1: MOVUPS (R11)(R14*1), X0 MOVUPS X0, (R10)(R14*1) ADDQ $0x10, R14 - CMPQ R14, AX + CMPQ R14, CX JB copy_1 - ADDQ AX, R11 - ADDQ AX, R10 - ADDQ AX, R12 + +copy_1_end: + ADDQ CX, R11 + ADDQ CX, R10 + ADDQ CX, R12 // Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize) -check_offset: - MOVQ R12, AX - ADDQ 40(SP), AX - CMPQ CX, AX + MOVQ R12, CX + ADDQ 72(SP), CX + CMPQ R13, CX JG error_match_off_too_big - CMPQ CX, 56(SP) + CMPQ R13, 88(SP) JG error_match_off_too_big // Copy match from history - MOVQ CX, AX - SUBQ R12, AX + MOVQ R13, CX + SUBQ R12, CX JLS copy_match - MOVQ 48(SP), R14 - SUBQ AX, R14 - CMPQ R13, AX + MOVQ 80(SP), R14 + SUBQ CX, R14 + CMPQ AX, CX JG copy_all_from_history - MOVQ R13, AX - SUBQ $0x10, AX + MOVQ AX, CX + SUBQ $0x10, CX JB copy_4_small copy_4_loop: @@ -2127,54 +2152,54 @@ copy_4_loop: MOVUPS X0, (R10) ADDQ $0x10, R14 ADDQ $0x10, R10 - SUBQ $0x10, AX + SUBQ $0x10, CX JAE copy_4_loop - LEAQ 16(R14)(AX*1), R14 - LEAQ 16(R10)(AX*1), R10 + LEAQ 16(R14)(CX*1), R14 + LEAQ 16(R10)(CX*1), R10 MOVUPS -16(R14), X0 MOVUPS X0, -16(R10) JMP copy_4_end copy_4_small: - CMPQ R13, $0x03 + CMPQ AX, $0x03 JE copy_4_move_3 - CMPQ R13, $0x08 + CMPQ AX, $0x08 JB copy_4_move_4through7 JMP copy_4_move_8through16 copy_4_move_3: - MOVW (R14), AX - MOVB 2(R14), CL - MOVW AX, (R10) - MOVB CL, 2(R10) - ADDQ R13, R14 - ADDQ R13, R10 - JMP copy_4_end + MOVWQZX (R14), CX + MOVB 2(R14), R13 + MOVW CX, (R10) + MOVB R13, 2(R10) + ADDQ AX, R14 + ADDQ AX, R10 + JMP copy_4_end copy_4_move_4through7: - MOVL (R14), AX - MOVL -4(R14)(R13*1), CX - MOVL AX, (R10) - MOVL CX, -4(R10)(R13*1) - ADDQ R13, R14 - ADDQ R13, R10 + MOVL (R14), CX + MOVL -4(R14)(AX*1), R13 + MOVL CX, (R10) + MOVL R13, -4(R10)(AX*1) + ADDQ AX, R14 + ADDQ AX, R10 JMP copy_4_end copy_4_move_8through16: - MOVQ (R14), AX - MOVQ -8(R14)(R13*1), CX - MOVQ AX, (R10) - MOVQ CX, -8(R10)(R13*1) - ADDQ R13, R14 - ADDQ R13, R10 + MOVQ (R14), CX + MOVQ -8(R14)(AX*1), R13 + MOVQ CX, (R10) + MOVQ R13, -8(R10)(AX*1) + ADDQ AX, R14 + ADDQ AX, R10 copy_4_end: - ADDQ R13, R12 + ADDQ AX, R12 JMP handle_loop JMP loop_finished copy_all_from_history: - MOVQ AX, R15 + MOVQ CX, R15 SUBQ $0x10, R15 JB copy_5_small @@ -2192,90 +2217,99 @@ copy_5_loop: JMP copy_5_end copy_5_small: - CMPQ AX, $0x03 + CMPQ CX, $0x03 JE copy_5_move_3 JB copy_5_move_1or2 - CMPQ AX, $0x08 + CMPQ CX, $0x08 JB copy_5_move_4through7 JMP copy_5_move_8through16 copy_5_move_1or2: MOVB (R14), R15 - MOVB -1(R14)(AX*1), BP + MOVB -1(R14)(CX*1), BP MOVB R15, (R10) - MOVB BP, -1(R10)(AX*1) - ADDQ AX, R14 - ADDQ AX, R10 + MOVB BP, -1(R10)(CX*1) + ADDQ CX, R14 + ADDQ CX, R10 JMP copy_5_end copy_5_move_3: - MOVW (R14), R15 - MOVB 2(R14), BP - MOVW R15, (R10) - MOVB BP, 2(R10) - ADDQ AX, R14 - ADDQ AX, R10 - JMP copy_5_end + MOVWQZX (R14), R15 + MOVB 2(R14), BP + MOVW R15, (R10) + MOVB BP, 2(R10) + ADDQ CX, R14 + ADDQ CX, R10 + JMP copy_5_end copy_5_move_4through7: MOVL (R14), R15 - MOVL -4(R14)(AX*1), BP + MOVL -4(R14)(CX*1), BP MOVL R15, (R10) - MOVL BP, -4(R10)(AX*1) - ADDQ AX, R14 - ADDQ AX, R10 + MOVL BP, -4(R10)(CX*1) + ADDQ CX, R14 + ADDQ CX, R10 JMP copy_5_end copy_5_move_8through16: MOVQ (R14), R15 - MOVQ -8(R14)(AX*1), BP + MOVQ -8(R14)(CX*1), BP MOVQ R15, (R10) - MOVQ BP, -8(R10)(AX*1) - ADDQ AX, R14 - ADDQ AX, R10 + MOVQ BP, -8(R10)(CX*1) + ADDQ CX, R14 + ADDQ CX, R10 copy_5_end: - ADDQ AX, R12 - SUBQ AX, R13 + ADDQ CX, R12 + SUBQ CX, AX // Copy match from the current buffer copy_match: - MOVQ R10, AX - SUBQ CX, AX + MOVQ R10, CX + SUBQ R13, CX // ml <= mo - CMPQ R13, CX + CMPQ AX, R13 JA copy_overlapping_match // Copy non-overlapping match - ADDQ R13, R12 - MOVQ R10, CX - ADDQ R13, R10 + ADDQ AX, R12 + MOVUPS (CX), X0 + MOVUPS X0, (R10) + CMPQ AX, $0x10 + JA copy_2_long + ADDQ AX, R10 + JMP handle_loop + +copy_2_long: + LEAQ 16(R10), R13 + ADDQ AX, R10 + ADDQ $0x10, CX + SUBQ $0x10, AX copy_2: - MOVUPS (AX), X0 - MOVUPS X0, (CX) - ADDQ $0x10, AX + MOVUPS (CX), X0 + MOVUPS X0, (R13) ADDQ $0x10, CX - SUBQ $0x10, R13 + ADDQ $0x10, R13 + SUBQ $0x10, AX JHI copy_2 JMP handle_loop // Copy overlapping match copy_overlapping_match: - ADDQ R13, R12 + ADDQ AX, R12 copy_slow_3: - MOVB (AX), CL - MOVB CL, (R10) - INCQ AX - INCQ R10 - DECQ R13 - JNZ copy_slow_3 + MOVBQZX (CX), R13 + MOVB R13, (R10) + INCQ CX + INCQ R10 + DECQ AX + JNZ copy_slow_3 handle_loop: - MOVQ ctx+16(FP), AX - DECQ 96(AX) + DECQ 32(SP) JNS sequenceDecs_decodeSync_amd64_main_loop loop_finished: @@ -2284,6 +2318,13 @@ loop_finished: MOVB BL, 40(AX) MOVQ SI, 32(AX) + // s.seqSize += outPosition - ctx.outPosition + MOVQ ctx+16(FP), AX + MOVQ s+0(FP), CX + MOVQ R12, DX + SUBQ 136(AX), DX + ADDQ DX, 256(CX) + // Update the context MOVQ ctx+16(FP), AX MOVQ R12, 136(AX) @@ -2297,7 +2338,7 @@ loop_finished: // Return with match length error sequenceDecs_decodeSync_amd64_error_match_len_ofs_mismatch: - MOVQ 16(SP), AX + MOVQ 48(SP), AX MOVQ ctx+16(FP), CX MOVQ AX, 216(CX) MOVQ $0x00000001, ret+24(FP) @@ -2306,7 +2347,7 @@ sequenceDecs_decodeSync_amd64_error_match_len_ofs_mismatch: // Return with match too long error sequenceDecs_decodeSync_amd64_error_match_len_too_big: MOVQ ctx+16(FP), AX - MOVQ 16(SP), CX + MOVQ 48(SP), CX MOVQ CX, 216(AX) MOVQ $0x00000002, ret+24(FP) RET @@ -2314,7 +2355,7 @@ sequenceDecs_decodeSync_amd64_error_match_len_too_big: // Return with match offset too long error error_match_off_too_big: MOVQ ctx+16(FP), AX - MOVQ 8(SP), CX + MOVQ 40(SP), CX MOVQ CX, 224(AX) MOVQ R12, 136(AX) MOVQ $0x00000003, ret+24(FP) @@ -2323,7 +2364,7 @@ error_match_off_too_big: // Return with not enough literals error error_not_enough_literals: MOVQ ctx+16(FP), AX - MOVQ 24(SP), CX + MOVQ 56(SP), CX MOVQ CX, 208(AX) MOVQ $0x00000004, ret+24(FP) RET @@ -2336,9 +2377,9 @@ error_overread: // Return with not enough output space error error_not_enough_space: MOVQ ctx+16(FP), AX - MOVQ 24(SP), CX + MOVQ 56(SP), CX MOVQ CX, 208(AX) - MOVQ 16(SP), CX + MOVQ 48(SP), CX MOVQ CX, 216(AX) MOVQ R12, 136(AX) MOVQ $0x00000005, ret+24(FP) @@ -2346,7 +2387,7 @@ error_not_enough_space: // func sequenceDecs_decodeSync_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int // Requires: BMI, BMI2, CMOV, SSE -TEXT ·sequenceDecs_decodeSync_bmi2(SB), $64-32 +TEXT ·sequenceDecs_decodeSync_bmi2(SB), $96-32 MOVQ br+8(FP), BX MOVQ 24(BX), AX MOVBQZX 40(BX), DX @@ -2358,26 +2399,34 @@ TEXT ·sequenceDecs_decodeSync_bmi2(SB), $64-32 MOVQ 72(CX), SI MOVQ 80(CX), DI MOVQ 88(CX), R8 - XORQ R9, R9 + MOVQ (CX), R9 MOVQ R9, 8(SP) + MOVQ 24(CX), R9 MOVQ R9, 16(SP) + MOVQ 48(CX), R9 MOVQ R9, 24(SP) + MOVQ 96(CX), R9 + MOVQ R9, 32(SP) + XORQ R9, R9 + MOVQ R9, 40(SP) + MOVQ R9, 48(SP) + MOVQ R9, 56(SP) MOVQ 112(CX), R9 MOVQ 128(CX), R10 - MOVQ R10, 32(SP) + MOVQ R10, 64(SP) MOVQ 144(CX), R10 MOVQ 136(CX), R11 MOVQ 200(CX), R12 - MOVQ R12, 56(SP) + MOVQ R12, 88(SP) MOVQ 176(CX), R12 - MOVQ R12, 48(SP) + MOVQ R12, 80(SP) MOVQ 184(CX), CX - MOVQ CX, 40(SP) - MOVQ 40(SP), CX - ADDQ CX, 48(SP) + MOVQ CX, 72(SP) + MOVQ 72(SP), CX + ADDQ CX, 80(SP) // Calculate pointer to s.out[cap(s.out)] (a past-end pointer) - ADDQ R9, 32(SP) + ADDQ R9, 64(SP) // outBase += outPosition ADDQ R11, R9 @@ -2425,7 +2474,7 @@ sequenceDecs_decodeSync_bmi2_fill_end: MOVQ R8, CX SHRQ $0x20, CX ADDQ R14, CX - MOVQ CX, 8(SP) + MOVQ CX, 40(SP) // Update match length MOVQ $0x00000808, CX @@ -2438,7 +2487,7 @@ sequenceDecs_decodeSync_bmi2_fill_end: MOVQ DI, CX SHRQ $0x20, CX ADDQ R14, CX - MOVQ CX, 16(SP) + MOVQ CX, 48(SP) // Fill bitreader to have enough for the remaining CMPQ BX, $0x08 @@ -2480,14 +2529,13 @@ sequenceDecs_decodeSync_bmi2_fill_2_end: MOVQ SI, CX SHRQ $0x20, CX ADDQ R14, CX - MOVQ CX, 24(SP) + MOVQ CX, 56(SP) // Fill bitreader for state updates MOVQ R12, (SP) MOVQ $0x00000808, CX BEXTRQ CX, R8, R12 - MOVQ ctx+16(FP), CX - CMPQ 96(CX), $0x00 + CMPQ 32(SP), $0x00 JZ sequenceDecs_decodeSync_bmi2_skip_update LEAQ (SI)(DI*1), R13 ADDQ R8, R13 @@ -2505,8 +2553,7 @@ sequenceDecs_decodeSync_bmi2_fill_2_end: ADDQ CX, R8 // Load ctx.ofTable - MOVQ ctx+16(FP), CX - MOVQ 48(CX), CX + MOVQ 24(SP), CX MOVQ (CX)(R8*8), R8 // Update Match Length State @@ -2516,8 +2563,7 @@ sequenceDecs_decodeSync_bmi2_fill_2_end: ADDQ CX, DI // Load ctx.mlTable - MOVQ ctx+16(FP), CX - MOVQ 24(CX), CX + MOVQ 16(SP), CX MOVQ (CX)(DI*8), DI // Update Literal Length State @@ -2526,14 +2572,13 @@ sequenceDecs_decodeSync_bmi2_fill_2_end: ADDQ CX, SI // Load ctx.llTable - MOVQ ctx+16(FP), CX - MOVQ (CX), CX + MOVQ 8(SP), CX MOVQ (CX)(SI*8), SI sequenceDecs_decodeSync_bmi2_skip_update: // Adjust offset MOVQ s+0(FP), CX - MOVQ 8(SP), R13 + MOVQ 40(SP), R13 CMPQ R12, $0x01 JBE sequenceDecs_decodeSync_bmi2_adjust_offsetB_1_or_0 MOVUPS 144(CX), X0 @@ -2542,7 +2587,7 @@ sequenceDecs_decodeSync_bmi2_skip_update: JMP sequenceDecs_decodeSync_bmi2_after_adjust sequenceDecs_decodeSync_bmi2_adjust_offsetB_1_or_0: - CMPQ 24(SP), $0x00000000 + CMPQ 56(SP), $0x00000000 JNE sequenceDecs_decodeSync_bmi2_adjust_offset_maybezero INCQ R13 JMP sequenceDecs_decodeSync_bmi2_adjust_offset_nonzero @@ -2577,14 +2622,11 @@ sequenceDecs_decodeSync_bmi2_adjust_skip: MOVQ R14, R13 sequenceDecs_decodeSync_bmi2_after_adjust: - MOVQ R13, 8(SP) + MOVQ R13, 40(SP) // Check values - MOVQ 16(SP), CX - MOVQ 24(SP), R12 - LEAQ (CX)(R12*1), R14 - MOVQ s+0(FP), R15 - ADDQ R14, 256(R15) + MOVQ 48(SP), CX + MOVQ 56(SP), R12 MOVQ ctx+16(FP), R14 SUBQ R12, 104(R14) JS error_not_enough_literals @@ -2596,50 +2638,49 @@ sequenceDecs_decodeSync_bmi2_after_adjust: JNZ sequenceDecs_decodeSync_bmi2_error_match_len_ofs_mismatch sequenceDecs_decodeSync_bmi2_match_len_ofs_ok: - MOVQ 24(SP), CX - MOVQ 8(SP), R12 - MOVQ 16(SP), R13 - // Check if we have enough space in s.out - LEAQ (CX)(R13*1), R14 + LEAQ 16(R12)(CX*1), R14 ADDQ R9, R14 - CMPQ R14, 32(SP) + CMPQ R14, 64(SP) JA error_not_enough_space // Copy literals - TESTQ CX, CX - JZ check_offset - XORQ R14, R14 + MOVUPS (R10), X0 + MOVUPS X0, (R9) + CMPQ R12, $0x10 + JBE copy_1_end + MOVQ $0x00000010, R14 copy_1: MOVUPS (R10)(R14*1), X0 MOVUPS X0, (R9)(R14*1) ADDQ $0x10, R14 - CMPQ R14, CX + CMPQ R14, R12 JB copy_1 - ADDQ CX, R10 - ADDQ CX, R9 - ADDQ CX, R11 + +copy_1_end: + ADDQ R12, R10 + ADDQ R12, R9 + ADDQ R12, R11 // Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize) -check_offset: - MOVQ R11, CX - ADDQ 40(SP), CX - CMPQ R12, CX + MOVQ R11, R12 + ADDQ 72(SP), R12 + CMPQ R13, R12 JG error_match_off_too_big - CMPQ R12, 56(SP) + CMPQ R13, 88(SP) JG error_match_off_too_big // Copy match from history - MOVQ R12, CX - SUBQ R11, CX + MOVQ R13, R12 + SUBQ R11, R12 JLS copy_match - MOVQ 48(SP), R14 - SUBQ CX, R14 - CMPQ R13, CX + MOVQ 80(SP), R14 + SUBQ R12, R14 + CMPQ CX, R12 JG copy_all_from_history - MOVQ R13, CX - SUBQ $0x10, CX + MOVQ CX, R12 + SUBQ $0x10, R12 JB copy_4_small copy_4_loop: @@ -2647,54 +2688,54 @@ copy_4_loop: MOVUPS X0, (R9) ADDQ $0x10, R14 ADDQ $0x10, R9 - SUBQ $0x10, CX + SUBQ $0x10, R12 JAE copy_4_loop - LEAQ 16(R14)(CX*1), R14 - LEAQ 16(R9)(CX*1), R9 + LEAQ 16(R14)(R12*1), R14 + LEAQ 16(R9)(R12*1), R9 MOVUPS -16(R14), X0 MOVUPS X0, -16(R9) JMP copy_4_end copy_4_small: - CMPQ R13, $0x03 + CMPQ CX, $0x03 JE copy_4_move_3 - CMPQ R13, $0x08 + CMPQ CX, $0x08 JB copy_4_move_4through7 JMP copy_4_move_8through16 copy_4_move_3: - MOVW (R14), CX - MOVB 2(R14), R12 - MOVW CX, (R9) - MOVB R12, 2(R9) - ADDQ R13, R14 - ADDQ R13, R9 - JMP copy_4_end + MOVWQZX (R14), R12 + MOVB 2(R14), R13 + MOVW R12, (R9) + MOVB R13, 2(R9) + ADDQ CX, R14 + ADDQ CX, R9 + JMP copy_4_end copy_4_move_4through7: - MOVL (R14), CX - MOVL -4(R14)(R13*1), R12 - MOVL CX, (R9) - MOVL R12, -4(R9)(R13*1) - ADDQ R13, R14 - ADDQ R13, R9 + MOVL (R14), R12 + MOVL -4(R14)(CX*1), R13 + MOVL R12, (R9) + MOVL R13, -4(R9)(CX*1) + ADDQ CX, R14 + ADDQ CX, R9 JMP copy_4_end copy_4_move_8through16: - MOVQ (R14), CX - MOVQ -8(R14)(R13*1), R12 - MOVQ CX, (R9) - MOVQ R12, -8(R9)(R13*1) - ADDQ R13, R14 - ADDQ R13, R9 + MOVQ (R14), R12 + MOVQ -8(R14)(CX*1), R13 + MOVQ R12, (R9) + MOVQ R13, -8(R9)(CX*1) + ADDQ CX, R14 + ADDQ CX, R9 copy_4_end: - ADDQ R13, R11 + ADDQ CX, R11 JMP handle_loop JMP loop_finished copy_all_from_history: - MOVQ CX, R15 + MOVQ R12, R15 SUBQ $0x10, R15 JB copy_5_small @@ -2712,90 +2753,99 @@ copy_5_loop: JMP copy_5_end copy_5_small: - CMPQ CX, $0x03 + CMPQ R12, $0x03 JE copy_5_move_3 JB copy_5_move_1or2 - CMPQ CX, $0x08 + CMPQ R12, $0x08 JB copy_5_move_4through7 JMP copy_5_move_8through16 copy_5_move_1or2: MOVB (R14), R15 - MOVB -1(R14)(CX*1), BP + MOVB -1(R14)(R12*1), BP MOVB R15, (R9) - MOVB BP, -1(R9)(CX*1) - ADDQ CX, R14 - ADDQ CX, R9 + MOVB BP, -1(R9)(R12*1) + ADDQ R12, R14 + ADDQ R12, R9 JMP copy_5_end copy_5_move_3: - MOVW (R14), R15 - MOVB 2(R14), BP - MOVW R15, (R9) - MOVB BP, 2(R9) - ADDQ CX, R14 - ADDQ CX, R9 - JMP copy_5_end + MOVWQZX (R14), R15 + MOVB 2(R14), BP + MOVW R15, (R9) + MOVB BP, 2(R9) + ADDQ R12, R14 + ADDQ R12, R9 + JMP copy_5_end copy_5_move_4through7: MOVL (R14), R15 - MOVL -4(R14)(CX*1), BP + MOVL -4(R14)(R12*1), BP MOVL R15, (R9) - MOVL BP, -4(R9)(CX*1) - ADDQ CX, R14 - ADDQ CX, R9 + MOVL BP, -4(R9)(R12*1) + ADDQ R12, R14 + ADDQ R12, R9 JMP copy_5_end copy_5_move_8through16: MOVQ (R14), R15 - MOVQ -8(R14)(CX*1), BP + MOVQ -8(R14)(R12*1), BP MOVQ R15, (R9) - MOVQ BP, -8(R9)(CX*1) - ADDQ CX, R14 - ADDQ CX, R9 + MOVQ BP, -8(R9)(R12*1) + ADDQ R12, R14 + ADDQ R12, R9 copy_5_end: - ADDQ CX, R11 - SUBQ CX, R13 + ADDQ R12, R11 + SUBQ R12, CX // Copy match from the current buffer copy_match: - MOVQ R9, CX - SUBQ R12, CX + MOVQ R9, R12 + SUBQ R13, R12 // ml <= mo - CMPQ R13, R12 + CMPQ CX, R13 JA copy_overlapping_match // Copy non-overlapping match - ADDQ R13, R11 - MOVQ R9, R12 - ADDQ R13, R9 + ADDQ CX, R11 + MOVUPS (R12), X0 + MOVUPS X0, (R9) + CMPQ CX, $0x10 + JA copy_2_long + ADDQ CX, R9 + JMP handle_loop + +copy_2_long: + LEAQ 16(R9), R13 + ADDQ CX, R9 + ADDQ $0x10, R12 + SUBQ $0x10, CX copy_2: - MOVUPS (CX), X0 - MOVUPS X0, (R12) - ADDQ $0x10, CX + MOVUPS (R12), X0 + MOVUPS X0, (R13) ADDQ $0x10, R12 - SUBQ $0x10, R13 + ADDQ $0x10, R13 + SUBQ $0x10, CX JHI copy_2 JMP handle_loop // Copy overlapping match copy_overlapping_match: - ADDQ R13, R11 + ADDQ CX, R11 copy_slow_3: - MOVB (CX), R12 - MOVB R12, (R9) - INCQ CX - INCQ R9 - DECQ R13 - JNZ copy_slow_3 + MOVBQZX (R12), R13 + MOVB R13, (R9) + INCQ R12 + INCQ R9 + DECQ CX + JNZ copy_slow_3 handle_loop: - MOVQ ctx+16(FP), CX - DECQ 96(CX) + DECQ 32(SP) JNS sequenceDecs_decodeSync_bmi2_main_loop loop_finished: @@ -2804,6 +2854,13 @@ loop_finished: MOVB DL, 40(CX) MOVQ BX, 32(CX) + // s.seqSize += outPosition - ctx.outPosition + MOVQ ctx+16(FP), AX + MOVQ s+0(FP), CX + MOVQ R11, DX + SUBQ 136(AX), DX + ADDQ DX, 256(CX) + // Update the context MOVQ ctx+16(FP), AX MOVQ R11, 136(AX) @@ -2817,7 +2874,7 @@ loop_finished: // Return with match length error sequenceDecs_decodeSync_bmi2_error_match_len_ofs_mismatch: - MOVQ 16(SP), AX + MOVQ 48(SP), AX MOVQ ctx+16(FP), CX MOVQ AX, 216(CX) MOVQ $0x00000001, ret+24(FP) @@ -2826,7 +2883,7 @@ sequenceDecs_decodeSync_bmi2_error_match_len_ofs_mismatch: // Return with match too long error sequenceDecs_decodeSync_bmi2_error_match_len_too_big: MOVQ ctx+16(FP), AX - MOVQ 16(SP), CX + MOVQ 48(SP), CX MOVQ CX, 216(AX) MOVQ $0x00000002, ret+24(FP) RET @@ -2834,7 +2891,7 @@ sequenceDecs_decodeSync_bmi2_error_match_len_too_big: // Return with match offset too long error error_match_off_too_big: MOVQ ctx+16(FP), AX - MOVQ 8(SP), CX + MOVQ 40(SP), CX MOVQ CX, 224(AX) MOVQ R11, 136(AX) MOVQ $0x00000003, ret+24(FP) @@ -2843,7 +2900,7 @@ error_match_off_too_big: // Return with not enough literals error error_not_enough_literals: MOVQ ctx+16(FP), AX - MOVQ 24(SP), CX + MOVQ 56(SP), CX MOVQ CX, 208(AX) MOVQ $0x00000004, ret+24(FP) RET @@ -2856,9 +2913,9 @@ error_overread: // Return with not enough output space error error_not_enough_space: MOVQ ctx+16(FP), AX - MOVQ 24(SP), CX + MOVQ 56(SP), CX MOVQ CX, 208(AX) - MOVQ 16(SP), CX + MOVQ 48(SP), CX MOVQ CX, 216(AX) MOVQ R11, 136(AX) MOVQ $0x00000005, ret+24(FP) @@ -2866,7 +2923,7 @@ error_not_enough_space: // func sequenceDecs_decodeSync_safe_amd64(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int // Requires: CMOV, SSE -TEXT ·sequenceDecs_decodeSync_safe_amd64(SB), $64-32 +TEXT ·sequenceDecs_decodeSync_safe_amd64(SB), $96-32 MOVQ br+8(FP), CX MOVQ 24(CX), DX MOVBQZX 40(CX), BX @@ -2878,26 +2935,34 @@ TEXT ·sequenceDecs_decodeSync_safe_amd64(SB), $64-32 MOVQ 72(AX), DI MOVQ 80(AX), R8 MOVQ 88(AX), R9 - XORQ CX, CX + MOVQ (AX), CX MOVQ CX, 8(SP) + MOVQ 24(AX), CX MOVQ CX, 16(SP) + MOVQ 48(AX), CX MOVQ CX, 24(SP) + MOVQ 96(AX), CX + MOVQ CX, 32(SP) + XORQ CX, CX + MOVQ CX, 40(SP) + MOVQ CX, 48(SP) + MOVQ CX, 56(SP) MOVQ 112(AX), R10 MOVQ 128(AX), CX - MOVQ CX, 32(SP) + MOVQ CX, 64(SP) MOVQ 144(AX), R11 MOVQ 136(AX), R12 MOVQ 200(AX), CX - MOVQ CX, 56(SP) + MOVQ CX, 88(SP) MOVQ 176(AX), CX - MOVQ CX, 48(SP) + MOVQ CX, 80(SP) MOVQ 184(AX), AX - MOVQ AX, 40(SP) - MOVQ 40(SP), AX - ADDQ AX, 48(SP) + MOVQ AX, 72(SP) + MOVQ 72(SP), AX + ADDQ AX, 80(SP) // Calculate pointer to s.out[cap(s.out)] (a past-end pointer) - ADDQ R10, 32(SP) + ADDQ R10, 64(SP) // outBase += outPosition ADDQ R12, R10 @@ -2935,46 +3000,32 @@ sequenceDecs_decodeSync_safe_amd64_fill_check_overread: sequenceDecs_decodeSync_safe_amd64_fill_end: // Update offset - MOVQ R9, AX - MOVQ BX, CX - MOVQ DX, R14 - SHLQ CL, R14 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decodeSync_safe_amd64_of_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decodeSync_safe_amd64_of_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decodeSync_safe_amd64_of_update_zero - NEGQ CX - SHRQ CL, R14 - ADDQ R14, AX - -sequenceDecs_decodeSync_safe_amd64_of_update_zero: - MOVQ AX, 8(SP) + MOVQ R9, AX + MOVQ BX, CX + MOVQ DX, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R14 + ADDQ R14, AX + MOVQ AX, 40(SP) // Update match length - MOVQ R8, AX - MOVQ BX, CX - MOVQ DX, R14 - SHLQ CL, R14 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decodeSync_safe_amd64_ml_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decodeSync_safe_amd64_ml_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decodeSync_safe_amd64_ml_update_zero - NEGQ CX - SHRQ CL, R14 - ADDQ R14, AX - -sequenceDecs_decodeSync_safe_amd64_ml_update_zero: - MOVQ AX, 16(SP) + MOVQ R8, AX + MOVQ BX, CX + MOVQ DX, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R14 + ADDQ R14, AX + MOVQ AX, 48(SP) // Fill bitreader to have enough for the remaining CMPQ SI, $0x08 @@ -3006,96 +3057,82 @@ sequenceDecs_decodeSync_safe_amd64_fill_2_check_overread: sequenceDecs_decodeSync_safe_amd64_fill_2_end: // Update literal length - MOVQ DI, AX - MOVQ BX, CX - MOVQ DX, R14 - SHLQ CL, R14 - MOVB AH, CL - SHRQ $0x20, AX - TESTQ CX, CX - JZ sequenceDecs_decodeSync_safe_amd64_ll_update_zero - ADDQ CX, BX - CMPQ BX, $0x40 - JA sequenceDecs_decodeSync_safe_amd64_ll_update_zero - CMPQ CX, $0x40 - JAE sequenceDecs_decodeSync_safe_amd64_ll_update_zero - NEGQ CX - SHRQ CL, R14 - ADDQ R14, AX - -sequenceDecs_decodeSync_safe_amd64_ll_update_zero: - MOVQ AX, 24(SP) + MOVQ DI, AX + MOVQ BX, CX + MOVQ DX, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + MOVBLZX AH, CX + SHRQ $0x20, AX + ADDQ CX, BX + XORQ $0x3f, CX + SHRQ CL, R14 + ADDQ R14, AX + MOVQ AX, 56(SP) // Fill bitreader for state updates MOVQ R13, (SP) MOVQ R9, AX SHRQ $0x08, AX MOVBQZX AL, AX - MOVQ ctx+16(FP), CX - CMPQ 96(CX), $0x00 + CMPQ 32(SP), $0x00 JZ sequenceDecs_decodeSync_safe_amd64_skip_update // Update Literal Length State MOVBQZX DI, R13 SHRL $0x10, DI - LEAQ (BX)(R13*1), CX + MOVQ BX, CX MOVQ DX, R14 - MOVQ CX, BX - ROLQ CL, R14 - MOVL $0x00000001, R15 - MOVB R13, CL - SHLL CL, R15 - DECL R15 - ANDQ R15, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + ADDQ R13, BX + XORQ $0x3f, R13 + MOVQ R13, CX + SHRQ CL, R14 ADDQ R14, DI // Load ctx.llTable - MOVQ ctx+16(FP), CX - MOVQ (CX), CX + MOVQ 8(SP), CX MOVQ (CX)(DI*8), DI // Update Match Length State MOVBQZX R8, R13 SHRL $0x10, R8 - LEAQ (BX)(R13*1), CX + MOVQ BX, CX MOVQ DX, R14 - MOVQ CX, BX - ROLQ CL, R14 - MOVL $0x00000001, R15 - MOVB R13, CL - SHLL CL, R15 - DECL R15 - ANDQ R15, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + ADDQ R13, BX + XORQ $0x3f, R13 + MOVQ R13, CX + SHRQ CL, R14 ADDQ R14, R8 // Load ctx.mlTable - MOVQ ctx+16(FP), CX - MOVQ 24(CX), CX + MOVQ 16(SP), CX MOVQ (CX)(R8*8), R8 // Update Offset State MOVBQZX R9, R13 SHRL $0x10, R9 - LEAQ (BX)(R13*1), CX + MOVQ BX, CX MOVQ DX, R14 - MOVQ CX, BX - ROLQ CL, R14 - MOVL $0x00000001, R15 - MOVB R13, CL - SHLL CL, R15 - DECL R15 - ANDQ R15, R14 + SHLQ CL, R14 + SHRQ $0x01, R14 + ADDQ R13, BX + XORQ $0x3f, R13 + MOVQ R13, CX + SHRQ CL, R14 ADDQ R14, R9 // Load ctx.ofTable - MOVQ ctx+16(FP), CX - MOVQ 48(CX), CX + MOVQ 24(SP), CX MOVQ (CX)(R9*8), R9 sequenceDecs_decodeSync_safe_amd64_skip_update: // Adjust offset MOVQ s+0(FP), CX - MOVQ 8(SP), R13 + MOVQ 40(SP), R13 CMPQ AX, $0x01 JBE sequenceDecs_decodeSync_safe_amd64_adjust_offsetB_1_or_0 MOVUPS 144(CX), X0 @@ -3104,7 +3141,7 @@ sequenceDecs_decodeSync_safe_amd64_skip_update: JMP sequenceDecs_decodeSync_safe_amd64_after_adjust sequenceDecs_decodeSync_safe_amd64_adjust_offsetB_1_or_0: - CMPQ 24(SP), $0x00000000 + CMPQ 56(SP), $0x00000000 JNE sequenceDecs_decodeSync_safe_amd64_adjust_offset_maybezero INCQ R13 JMP sequenceDecs_decodeSync_safe_amd64_adjust_offset_nonzero @@ -3139,14 +3176,11 @@ sequenceDecs_decodeSync_safe_amd64_adjust_skip: MOVQ R14, R13 sequenceDecs_decodeSync_safe_amd64_after_adjust: - MOVQ R13, 8(SP) + MOVQ R13, 40(SP) // Check values - MOVQ 16(SP), AX - MOVQ 24(SP), CX - LEAQ (AX)(CX*1), R14 - MOVQ s+0(FP), R15 - ADDQ R14, 256(R15) + MOVQ 48(SP), AX + MOVQ 56(SP), CX MOVQ ctx+16(FP), R14 SUBQ CX, 104(R14) JS error_not_enough_literals @@ -3158,20 +3192,16 @@ sequenceDecs_decodeSync_safe_amd64_after_adjust: JNZ sequenceDecs_decodeSync_safe_amd64_error_match_len_ofs_mismatch sequenceDecs_decodeSync_safe_amd64_match_len_ofs_ok: - MOVQ 24(SP), AX - MOVQ 8(SP), CX - MOVQ 16(SP), R13 - // Check if we have enough space in s.out - LEAQ (AX)(R13*1), R14 + LEAQ (CX)(AX*1), R14 ADDQ R10, R14 - CMPQ R14, 32(SP) + CMPQ R14, 64(SP) JA error_not_enough_space // Copy literals - TESTQ AX, AX + TESTQ CX, CX JZ check_offset - MOVQ AX, R14 + MOVQ CX, R14 SUBQ $0x10, R14 JB copy_1_small @@ -3189,70 +3219,70 @@ copy_1_loop: JMP copy_1_end copy_1_small: - CMPQ AX, $0x03 + CMPQ CX, $0x03 JE copy_1_move_3 JB copy_1_move_1or2 - CMPQ AX, $0x08 + CMPQ CX, $0x08 JB copy_1_move_4through7 JMP copy_1_move_8through16 copy_1_move_1or2: MOVB (R11), R14 - MOVB -1(R11)(AX*1), R15 + MOVB -1(R11)(CX*1), R15 MOVB R14, (R10) - MOVB R15, -1(R10)(AX*1) - ADDQ AX, R11 - ADDQ AX, R10 + MOVB R15, -1(R10)(CX*1) + ADDQ CX, R11 + ADDQ CX, R10 JMP copy_1_end copy_1_move_3: - MOVW (R11), R14 - MOVB 2(R11), R15 - MOVW R14, (R10) - MOVB R15, 2(R10) - ADDQ AX, R11 - ADDQ AX, R10 - JMP copy_1_end + MOVWQZX (R11), R14 + MOVB 2(R11), R15 + MOVW R14, (R10) + MOVB R15, 2(R10) + ADDQ CX, R11 + ADDQ CX, R10 + JMP copy_1_end copy_1_move_4through7: MOVL (R11), R14 - MOVL -4(R11)(AX*1), R15 + MOVL -4(R11)(CX*1), R15 MOVL R14, (R10) - MOVL R15, -4(R10)(AX*1) - ADDQ AX, R11 - ADDQ AX, R10 + MOVL R15, -4(R10)(CX*1) + ADDQ CX, R11 + ADDQ CX, R10 JMP copy_1_end copy_1_move_8through16: MOVQ (R11), R14 - MOVQ -8(R11)(AX*1), R15 + MOVQ -8(R11)(CX*1), R15 MOVQ R14, (R10) - MOVQ R15, -8(R10)(AX*1) - ADDQ AX, R11 - ADDQ AX, R10 + MOVQ R15, -8(R10)(CX*1) + ADDQ CX, R11 + ADDQ CX, R10 copy_1_end: - ADDQ AX, R12 + ADDQ CX, R12 // Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize) check_offset: - MOVQ R12, AX - ADDQ 40(SP), AX - CMPQ CX, AX + MOVQ R12, CX + ADDQ 72(SP), CX + CMPQ R13, CX JG error_match_off_too_big - CMPQ CX, 56(SP) + CMPQ R13, 88(SP) JG error_match_off_too_big // Copy match from history - MOVQ CX, AX - SUBQ R12, AX + MOVQ R13, CX + SUBQ R12, CX JLS copy_match - MOVQ 48(SP), R14 - SUBQ AX, R14 - CMPQ R13, AX + MOVQ 80(SP), R14 + SUBQ CX, R14 + CMPQ AX, CX JG copy_all_from_history - MOVQ R13, AX - SUBQ $0x10, AX + MOVQ AX, CX + SUBQ $0x10, CX JB copy_4_small copy_4_loop: @@ -3260,54 +3290,54 @@ copy_4_loop: MOVUPS X0, (R10) ADDQ $0x10, R14 ADDQ $0x10, R10 - SUBQ $0x10, AX + SUBQ $0x10, CX JAE copy_4_loop - LEAQ 16(R14)(AX*1), R14 - LEAQ 16(R10)(AX*1), R10 + LEAQ 16(R14)(CX*1), R14 + LEAQ 16(R10)(CX*1), R10 MOVUPS -16(R14), X0 MOVUPS X0, -16(R10) JMP copy_4_end copy_4_small: - CMPQ R13, $0x03 + CMPQ AX, $0x03 JE copy_4_move_3 - CMPQ R13, $0x08 + CMPQ AX, $0x08 JB copy_4_move_4through7 JMP copy_4_move_8through16 copy_4_move_3: - MOVW (R14), AX - MOVB 2(R14), CL - MOVW AX, (R10) - MOVB CL, 2(R10) - ADDQ R13, R14 - ADDQ R13, R10 - JMP copy_4_end + MOVWQZX (R14), CX + MOVB 2(R14), R13 + MOVW CX, (R10) + MOVB R13, 2(R10) + ADDQ AX, R14 + ADDQ AX, R10 + JMP copy_4_end copy_4_move_4through7: - MOVL (R14), AX - MOVL -4(R14)(R13*1), CX - MOVL AX, (R10) - MOVL CX, -4(R10)(R13*1) - ADDQ R13, R14 - ADDQ R13, R10 + MOVL (R14), CX + MOVL -4(R14)(AX*1), R13 + MOVL CX, (R10) + MOVL R13, -4(R10)(AX*1) + ADDQ AX, R14 + ADDQ AX, R10 JMP copy_4_end copy_4_move_8through16: - MOVQ (R14), AX - MOVQ -8(R14)(R13*1), CX - MOVQ AX, (R10) - MOVQ CX, -8(R10)(R13*1) - ADDQ R13, R14 - ADDQ R13, R10 + MOVQ (R14), CX + MOVQ -8(R14)(AX*1), R13 + MOVQ CX, (R10) + MOVQ R13, -8(R10)(AX*1) + ADDQ AX, R14 + ADDQ AX, R10 copy_4_end: - ADDQ R13, R12 + ADDQ AX, R12 JMP handle_loop JMP loop_finished copy_all_from_history: - MOVQ AX, R15 + MOVQ CX, R15 SUBQ $0x10, R15 JB copy_5_small @@ -3325,141 +3355,140 @@ copy_5_loop: JMP copy_5_end copy_5_small: - CMPQ AX, $0x03 + CMPQ CX, $0x03 JE copy_5_move_3 JB copy_5_move_1or2 - CMPQ AX, $0x08 + CMPQ CX, $0x08 JB copy_5_move_4through7 JMP copy_5_move_8through16 copy_5_move_1or2: MOVB (R14), R15 - MOVB -1(R14)(AX*1), BP + MOVB -1(R14)(CX*1), BP MOVB R15, (R10) - MOVB BP, -1(R10)(AX*1) - ADDQ AX, R14 - ADDQ AX, R10 + MOVB BP, -1(R10)(CX*1) + ADDQ CX, R14 + ADDQ CX, R10 JMP copy_5_end copy_5_move_3: - MOVW (R14), R15 - MOVB 2(R14), BP - MOVW R15, (R10) - MOVB BP, 2(R10) - ADDQ AX, R14 - ADDQ AX, R10 - JMP copy_5_end + MOVWQZX (R14), R15 + MOVB 2(R14), BP + MOVW R15, (R10) + MOVB BP, 2(R10) + ADDQ CX, R14 + ADDQ CX, R10 + JMP copy_5_end copy_5_move_4through7: MOVL (R14), R15 - MOVL -4(R14)(AX*1), BP + MOVL -4(R14)(CX*1), BP MOVL R15, (R10) - MOVL BP, -4(R10)(AX*1) - ADDQ AX, R14 - ADDQ AX, R10 + MOVL BP, -4(R10)(CX*1) + ADDQ CX, R14 + ADDQ CX, R10 JMP copy_5_end copy_5_move_8through16: MOVQ (R14), R15 - MOVQ -8(R14)(AX*1), BP + MOVQ -8(R14)(CX*1), BP MOVQ R15, (R10) - MOVQ BP, -8(R10)(AX*1) - ADDQ AX, R14 - ADDQ AX, R10 + MOVQ BP, -8(R10)(CX*1) + ADDQ CX, R14 + ADDQ CX, R10 copy_5_end: - ADDQ AX, R12 - SUBQ AX, R13 + ADDQ CX, R12 + SUBQ CX, AX // Copy match from the current buffer copy_match: - MOVQ R10, AX - SUBQ CX, AX + MOVQ R10, CX + SUBQ R13, CX // ml <= mo - CMPQ R13, CX + CMPQ AX, R13 JA copy_overlapping_match // Copy non-overlapping match - ADDQ R13, R12 - MOVQ R13, CX - SUBQ $0x10, CX + ADDQ AX, R12 + MOVQ AX, R13 + SUBQ $0x10, R13 JB copy_2_small copy_2_loop: - MOVUPS (AX), X0 + MOVUPS (CX), X0 MOVUPS X0, (R10) - ADDQ $0x10, AX + ADDQ $0x10, CX ADDQ $0x10, R10 - SUBQ $0x10, CX + SUBQ $0x10, R13 JAE copy_2_loop - LEAQ 16(AX)(CX*1), AX - LEAQ 16(R10)(CX*1), R10 - MOVUPS -16(AX), X0 + LEAQ 16(CX)(R13*1), CX + LEAQ 16(R10)(R13*1), R10 + MOVUPS -16(CX), X0 MOVUPS X0, -16(R10) JMP copy_2_end copy_2_small: - CMPQ R13, $0x03 + CMPQ AX, $0x03 JE copy_2_move_3 JB copy_2_move_1or2 - CMPQ R13, $0x08 + CMPQ AX, $0x08 JB copy_2_move_4through7 JMP copy_2_move_8through16 copy_2_move_1or2: - MOVB (AX), CL - MOVB -1(AX)(R13*1), R14 - MOVB CL, (R10) - MOVB R14, -1(R10)(R13*1) - ADDQ R13, AX - ADDQ R13, R10 + MOVB (CX), R13 + MOVB -1(CX)(AX*1), R14 + MOVB R13, (R10) + MOVB R14, -1(R10)(AX*1) + ADDQ AX, CX + ADDQ AX, R10 JMP copy_2_end copy_2_move_3: - MOVW (AX), CX - MOVB 2(AX), R14 - MOVW CX, (R10) - MOVB R14, 2(R10) - ADDQ R13, AX - ADDQ R13, R10 - JMP copy_2_end + MOVWQZX (CX), R13 + MOVB 2(CX), R14 + MOVW R13, (R10) + MOVB R14, 2(R10) + ADDQ AX, CX + ADDQ AX, R10 + JMP copy_2_end copy_2_move_4through7: - MOVL (AX), CX - MOVL -4(AX)(R13*1), R14 - MOVL CX, (R10) - MOVL R14, -4(R10)(R13*1) - ADDQ R13, AX - ADDQ R13, R10 + MOVL (CX), R13 + MOVL -4(CX)(AX*1), R14 + MOVL R13, (R10) + MOVL R14, -4(R10)(AX*1) + ADDQ AX, CX + ADDQ AX, R10 JMP copy_2_end copy_2_move_8through16: - MOVQ (AX), CX - MOVQ -8(AX)(R13*1), R14 - MOVQ CX, (R10) - MOVQ R14, -8(R10)(R13*1) - ADDQ R13, AX - ADDQ R13, R10 + MOVQ (CX), R13 + MOVQ -8(CX)(AX*1), R14 + MOVQ R13, (R10) + MOVQ R14, -8(R10)(AX*1) + ADDQ AX, CX + ADDQ AX, R10 copy_2_end: JMP handle_loop // Copy overlapping match copy_overlapping_match: - ADDQ R13, R12 + ADDQ AX, R12 copy_slow_3: - MOVB (AX), CL - MOVB CL, (R10) - INCQ AX - INCQ R10 - DECQ R13 - JNZ copy_slow_3 + MOVBQZX (CX), R13 + MOVB R13, (R10) + INCQ CX + INCQ R10 + DECQ AX + JNZ copy_slow_3 handle_loop: - MOVQ ctx+16(FP), AX - DECQ 96(AX) + DECQ 32(SP) JNS sequenceDecs_decodeSync_safe_amd64_main_loop loop_finished: @@ -3468,6 +3497,13 @@ loop_finished: MOVB BL, 40(AX) MOVQ SI, 32(AX) + // s.seqSize += outPosition - ctx.outPosition + MOVQ ctx+16(FP), AX + MOVQ s+0(FP), CX + MOVQ R12, DX + SUBQ 136(AX), DX + ADDQ DX, 256(CX) + // Update the context MOVQ ctx+16(FP), AX MOVQ R12, 136(AX) @@ -3481,7 +3517,7 @@ loop_finished: // Return with match length error sequenceDecs_decodeSync_safe_amd64_error_match_len_ofs_mismatch: - MOVQ 16(SP), AX + MOVQ 48(SP), AX MOVQ ctx+16(FP), CX MOVQ AX, 216(CX) MOVQ $0x00000001, ret+24(FP) @@ -3490,7 +3526,7 @@ sequenceDecs_decodeSync_safe_amd64_error_match_len_ofs_mismatch: // Return with match too long error sequenceDecs_decodeSync_safe_amd64_error_match_len_too_big: MOVQ ctx+16(FP), AX - MOVQ 16(SP), CX + MOVQ 48(SP), CX MOVQ CX, 216(AX) MOVQ $0x00000002, ret+24(FP) RET @@ -3498,7 +3534,7 @@ sequenceDecs_decodeSync_safe_amd64_error_match_len_too_big: // Return with match offset too long error error_match_off_too_big: MOVQ ctx+16(FP), AX - MOVQ 8(SP), CX + MOVQ 40(SP), CX MOVQ CX, 224(AX) MOVQ R12, 136(AX) MOVQ $0x00000003, ret+24(FP) @@ -3507,7 +3543,7 @@ error_match_off_too_big: // Return with not enough literals error error_not_enough_literals: MOVQ ctx+16(FP), AX - MOVQ 24(SP), CX + MOVQ 56(SP), CX MOVQ CX, 208(AX) MOVQ $0x00000004, ret+24(FP) RET @@ -3520,9 +3556,9 @@ error_overread: // Return with not enough output space error error_not_enough_space: MOVQ ctx+16(FP), AX - MOVQ 24(SP), CX + MOVQ 56(SP), CX MOVQ CX, 208(AX) - MOVQ 16(SP), CX + MOVQ 48(SP), CX MOVQ CX, 216(AX) MOVQ R12, 136(AX) MOVQ $0x00000005, ret+24(FP) @@ -3530,7 +3566,7 @@ error_not_enough_space: // func sequenceDecs_decodeSync_safe_bmi2(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int // Requires: BMI, BMI2, CMOV, SSE -TEXT ·sequenceDecs_decodeSync_safe_bmi2(SB), $64-32 +TEXT ·sequenceDecs_decodeSync_safe_bmi2(SB), $96-32 MOVQ br+8(FP), BX MOVQ 24(BX), AX MOVBQZX 40(BX), DX @@ -3542,26 +3578,34 @@ TEXT ·sequenceDecs_decodeSync_safe_bmi2(SB), $64-32 MOVQ 72(CX), SI MOVQ 80(CX), DI MOVQ 88(CX), R8 - XORQ R9, R9 + MOVQ (CX), R9 MOVQ R9, 8(SP) + MOVQ 24(CX), R9 MOVQ R9, 16(SP) + MOVQ 48(CX), R9 MOVQ R9, 24(SP) + MOVQ 96(CX), R9 + MOVQ R9, 32(SP) + XORQ R9, R9 + MOVQ R9, 40(SP) + MOVQ R9, 48(SP) + MOVQ R9, 56(SP) MOVQ 112(CX), R9 MOVQ 128(CX), R10 - MOVQ R10, 32(SP) + MOVQ R10, 64(SP) MOVQ 144(CX), R10 MOVQ 136(CX), R11 MOVQ 200(CX), R12 - MOVQ R12, 56(SP) + MOVQ R12, 88(SP) MOVQ 176(CX), R12 - MOVQ R12, 48(SP) + MOVQ R12, 80(SP) MOVQ 184(CX), CX - MOVQ CX, 40(SP) - MOVQ 40(SP), CX - ADDQ CX, 48(SP) + MOVQ CX, 72(SP) + MOVQ 72(SP), CX + ADDQ CX, 80(SP) // Calculate pointer to s.out[cap(s.out)] (a past-end pointer) - ADDQ R9, 32(SP) + ADDQ R9, 64(SP) // outBase += outPosition ADDQ R11, R9 @@ -3609,7 +3653,7 @@ sequenceDecs_decodeSync_safe_bmi2_fill_end: MOVQ R8, CX SHRQ $0x20, CX ADDQ R14, CX - MOVQ CX, 8(SP) + MOVQ CX, 40(SP) // Update match length MOVQ $0x00000808, CX @@ -3622,7 +3666,7 @@ sequenceDecs_decodeSync_safe_bmi2_fill_end: MOVQ DI, CX SHRQ $0x20, CX ADDQ R14, CX - MOVQ CX, 16(SP) + MOVQ CX, 48(SP) // Fill bitreader to have enough for the remaining CMPQ BX, $0x08 @@ -3664,14 +3708,13 @@ sequenceDecs_decodeSync_safe_bmi2_fill_2_end: MOVQ SI, CX SHRQ $0x20, CX ADDQ R14, CX - MOVQ CX, 24(SP) + MOVQ CX, 56(SP) // Fill bitreader for state updates MOVQ R12, (SP) MOVQ $0x00000808, CX BEXTRQ CX, R8, R12 - MOVQ ctx+16(FP), CX - CMPQ 96(CX), $0x00 + CMPQ 32(SP), $0x00 JZ sequenceDecs_decodeSync_safe_bmi2_skip_update LEAQ (SI)(DI*1), R13 ADDQ R8, R13 @@ -3689,8 +3732,7 @@ sequenceDecs_decodeSync_safe_bmi2_fill_2_end: ADDQ CX, R8 // Load ctx.ofTable - MOVQ ctx+16(FP), CX - MOVQ 48(CX), CX + MOVQ 24(SP), CX MOVQ (CX)(R8*8), R8 // Update Match Length State @@ -3700,8 +3742,7 @@ sequenceDecs_decodeSync_safe_bmi2_fill_2_end: ADDQ CX, DI // Load ctx.mlTable - MOVQ ctx+16(FP), CX - MOVQ 24(CX), CX + MOVQ 16(SP), CX MOVQ (CX)(DI*8), DI // Update Literal Length State @@ -3710,14 +3751,13 @@ sequenceDecs_decodeSync_safe_bmi2_fill_2_end: ADDQ CX, SI // Load ctx.llTable - MOVQ ctx+16(FP), CX - MOVQ (CX), CX + MOVQ 8(SP), CX MOVQ (CX)(SI*8), SI sequenceDecs_decodeSync_safe_bmi2_skip_update: // Adjust offset MOVQ s+0(FP), CX - MOVQ 8(SP), R13 + MOVQ 40(SP), R13 CMPQ R12, $0x01 JBE sequenceDecs_decodeSync_safe_bmi2_adjust_offsetB_1_or_0 MOVUPS 144(CX), X0 @@ -3726,7 +3766,7 @@ sequenceDecs_decodeSync_safe_bmi2_skip_update: JMP sequenceDecs_decodeSync_safe_bmi2_after_adjust sequenceDecs_decodeSync_safe_bmi2_adjust_offsetB_1_or_0: - CMPQ 24(SP), $0x00000000 + CMPQ 56(SP), $0x00000000 JNE sequenceDecs_decodeSync_safe_bmi2_adjust_offset_maybezero INCQ R13 JMP sequenceDecs_decodeSync_safe_bmi2_adjust_offset_nonzero @@ -3761,14 +3801,11 @@ sequenceDecs_decodeSync_safe_bmi2_adjust_skip: MOVQ R14, R13 sequenceDecs_decodeSync_safe_bmi2_after_adjust: - MOVQ R13, 8(SP) + MOVQ R13, 40(SP) // Check values - MOVQ 16(SP), CX - MOVQ 24(SP), R12 - LEAQ (CX)(R12*1), R14 - MOVQ s+0(FP), R15 - ADDQ R14, 256(R15) + MOVQ 48(SP), CX + MOVQ 56(SP), R12 MOVQ ctx+16(FP), R14 SUBQ R12, 104(R14) JS error_not_enough_literals @@ -3780,20 +3817,16 @@ sequenceDecs_decodeSync_safe_bmi2_after_adjust: JNZ sequenceDecs_decodeSync_safe_bmi2_error_match_len_ofs_mismatch sequenceDecs_decodeSync_safe_bmi2_match_len_ofs_ok: - MOVQ 24(SP), CX - MOVQ 8(SP), R12 - MOVQ 16(SP), R13 - // Check if we have enough space in s.out - LEAQ (CX)(R13*1), R14 + LEAQ (R12)(CX*1), R14 ADDQ R9, R14 - CMPQ R14, 32(SP) + CMPQ R14, 64(SP) JA error_not_enough_space // Copy literals - TESTQ CX, CX + TESTQ R12, R12 JZ check_offset - MOVQ CX, R14 + MOVQ R12, R14 SUBQ $0x10, R14 JB copy_1_small @@ -3811,70 +3844,70 @@ copy_1_loop: JMP copy_1_end copy_1_small: - CMPQ CX, $0x03 + CMPQ R12, $0x03 JE copy_1_move_3 JB copy_1_move_1or2 - CMPQ CX, $0x08 + CMPQ R12, $0x08 JB copy_1_move_4through7 JMP copy_1_move_8through16 copy_1_move_1or2: MOVB (R10), R14 - MOVB -1(R10)(CX*1), R15 + MOVB -1(R10)(R12*1), R15 MOVB R14, (R9) - MOVB R15, -1(R9)(CX*1) - ADDQ CX, R10 - ADDQ CX, R9 + MOVB R15, -1(R9)(R12*1) + ADDQ R12, R10 + ADDQ R12, R9 JMP copy_1_end copy_1_move_3: - MOVW (R10), R14 - MOVB 2(R10), R15 - MOVW R14, (R9) - MOVB R15, 2(R9) - ADDQ CX, R10 - ADDQ CX, R9 - JMP copy_1_end + MOVWQZX (R10), R14 + MOVB 2(R10), R15 + MOVW R14, (R9) + MOVB R15, 2(R9) + ADDQ R12, R10 + ADDQ R12, R9 + JMP copy_1_end copy_1_move_4through7: MOVL (R10), R14 - MOVL -4(R10)(CX*1), R15 + MOVL -4(R10)(R12*1), R15 MOVL R14, (R9) - MOVL R15, -4(R9)(CX*1) - ADDQ CX, R10 - ADDQ CX, R9 + MOVL R15, -4(R9)(R12*1) + ADDQ R12, R10 + ADDQ R12, R9 JMP copy_1_end copy_1_move_8through16: MOVQ (R10), R14 - MOVQ -8(R10)(CX*1), R15 + MOVQ -8(R10)(R12*1), R15 MOVQ R14, (R9) - MOVQ R15, -8(R9)(CX*1) - ADDQ CX, R10 - ADDQ CX, R9 + MOVQ R15, -8(R9)(R12*1) + ADDQ R12, R10 + ADDQ R12, R9 copy_1_end: - ADDQ CX, R11 + ADDQ R12, R11 // Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize) check_offset: - MOVQ R11, CX - ADDQ 40(SP), CX - CMPQ R12, CX + MOVQ R11, R12 + ADDQ 72(SP), R12 + CMPQ R13, R12 JG error_match_off_too_big - CMPQ R12, 56(SP) + CMPQ R13, 88(SP) JG error_match_off_too_big // Copy match from history - MOVQ R12, CX - SUBQ R11, CX + MOVQ R13, R12 + SUBQ R11, R12 JLS copy_match - MOVQ 48(SP), R14 - SUBQ CX, R14 - CMPQ R13, CX + MOVQ 80(SP), R14 + SUBQ R12, R14 + CMPQ CX, R12 JG copy_all_from_history - MOVQ R13, CX - SUBQ $0x10, CX + MOVQ CX, R12 + SUBQ $0x10, R12 JB copy_4_small copy_4_loop: @@ -3882,54 +3915,54 @@ copy_4_loop: MOVUPS X0, (R9) ADDQ $0x10, R14 ADDQ $0x10, R9 - SUBQ $0x10, CX + SUBQ $0x10, R12 JAE copy_4_loop - LEAQ 16(R14)(CX*1), R14 - LEAQ 16(R9)(CX*1), R9 + LEAQ 16(R14)(R12*1), R14 + LEAQ 16(R9)(R12*1), R9 MOVUPS -16(R14), X0 MOVUPS X0, -16(R9) JMP copy_4_end copy_4_small: - CMPQ R13, $0x03 + CMPQ CX, $0x03 JE copy_4_move_3 - CMPQ R13, $0x08 + CMPQ CX, $0x08 JB copy_4_move_4through7 JMP copy_4_move_8through16 copy_4_move_3: - MOVW (R14), CX - MOVB 2(R14), R12 - MOVW CX, (R9) - MOVB R12, 2(R9) - ADDQ R13, R14 - ADDQ R13, R9 - JMP copy_4_end + MOVWQZX (R14), R12 + MOVB 2(R14), R13 + MOVW R12, (R9) + MOVB R13, 2(R9) + ADDQ CX, R14 + ADDQ CX, R9 + JMP copy_4_end copy_4_move_4through7: - MOVL (R14), CX - MOVL -4(R14)(R13*1), R12 - MOVL CX, (R9) - MOVL R12, -4(R9)(R13*1) - ADDQ R13, R14 - ADDQ R13, R9 + MOVL (R14), R12 + MOVL -4(R14)(CX*1), R13 + MOVL R12, (R9) + MOVL R13, -4(R9)(CX*1) + ADDQ CX, R14 + ADDQ CX, R9 JMP copy_4_end copy_4_move_8through16: - MOVQ (R14), CX - MOVQ -8(R14)(R13*1), R12 - MOVQ CX, (R9) - MOVQ R12, -8(R9)(R13*1) - ADDQ R13, R14 - ADDQ R13, R9 + MOVQ (R14), R12 + MOVQ -8(R14)(CX*1), R13 + MOVQ R12, (R9) + MOVQ R13, -8(R9)(CX*1) + ADDQ CX, R14 + ADDQ CX, R9 copy_4_end: - ADDQ R13, R11 + ADDQ CX, R11 JMP handle_loop JMP loop_finished copy_all_from_history: - MOVQ CX, R15 + MOVQ R12, R15 SUBQ $0x10, R15 JB copy_5_small @@ -3947,141 +3980,140 @@ copy_5_loop: JMP copy_5_end copy_5_small: - CMPQ CX, $0x03 + CMPQ R12, $0x03 JE copy_5_move_3 JB copy_5_move_1or2 - CMPQ CX, $0x08 + CMPQ R12, $0x08 JB copy_5_move_4through7 JMP copy_5_move_8through16 copy_5_move_1or2: MOVB (R14), R15 - MOVB -1(R14)(CX*1), BP + MOVB -1(R14)(R12*1), BP MOVB R15, (R9) - MOVB BP, -1(R9)(CX*1) - ADDQ CX, R14 - ADDQ CX, R9 + MOVB BP, -1(R9)(R12*1) + ADDQ R12, R14 + ADDQ R12, R9 JMP copy_5_end copy_5_move_3: - MOVW (R14), R15 - MOVB 2(R14), BP - MOVW R15, (R9) - MOVB BP, 2(R9) - ADDQ CX, R14 - ADDQ CX, R9 - JMP copy_5_end + MOVWQZX (R14), R15 + MOVB 2(R14), BP + MOVW R15, (R9) + MOVB BP, 2(R9) + ADDQ R12, R14 + ADDQ R12, R9 + JMP copy_5_end copy_5_move_4through7: MOVL (R14), R15 - MOVL -4(R14)(CX*1), BP + MOVL -4(R14)(R12*1), BP MOVL R15, (R9) - MOVL BP, -4(R9)(CX*1) - ADDQ CX, R14 - ADDQ CX, R9 + MOVL BP, -4(R9)(R12*1) + ADDQ R12, R14 + ADDQ R12, R9 JMP copy_5_end copy_5_move_8through16: MOVQ (R14), R15 - MOVQ -8(R14)(CX*1), BP + MOVQ -8(R14)(R12*1), BP MOVQ R15, (R9) - MOVQ BP, -8(R9)(CX*1) - ADDQ CX, R14 - ADDQ CX, R9 + MOVQ BP, -8(R9)(R12*1) + ADDQ R12, R14 + ADDQ R12, R9 copy_5_end: - ADDQ CX, R11 - SUBQ CX, R13 + ADDQ R12, R11 + SUBQ R12, CX // Copy match from the current buffer copy_match: - MOVQ R9, CX - SUBQ R12, CX + MOVQ R9, R12 + SUBQ R13, R12 // ml <= mo - CMPQ R13, R12 + CMPQ CX, R13 JA copy_overlapping_match // Copy non-overlapping match - ADDQ R13, R11 - MOVQ R13, R12 - SUBQ $0x10, R12 + ADDQ CX, R11 + MOVQ CX, R13 + SUBQ $0x10, R13 JB copy_2_small copy_2_loop: - MOVUPS (CX), X0 + MOVUPS (R12), X0 MOVUPS X0, (R9) - ADDQ $0x10, CX + ADDQ $0x10, R12 ADDQ $0x10, R9 - SUBQ $0x10, R12 + SUBQ $0x10, R13 JAE copy_2_loop - LEAQ 16(CX)(R12*1), CX - LEAQ 16(R9)(R12*1), R9 - MOVUPS -16(CX), X0 + LEAQ 16(R12)(R13*1), R12 + LEAQ 16(R9)(R13*1), R9 + MOVUPS -16(R12), X0 MOVUPS X0, -16(R9) JMP copy_2_end copy_2_small: - CMPQ R13, $0x03 + CMPQ CX, $0x03 JE copy_2_move_3 JB copy_2_move_1or2 - CMPQ R13, $0x08 + CMPQ CX, $0x08 JB copy_2_move_4through7 JMP copy_2_move_8through16 copy_2_move_1or2: - MOVB (CX), R12 - MOVB -1(CX)(R13*1), R14 - MOVB R12, (R9) - MOVB R14, -1(R9)(R13*1) - ADDQ R13, CX - ADDQ R13, R9 + MOVB (R12), R13 + MOVB -1(R12)(CX*1), R14 + MOVB R13, (R9) + MOVB R14, -1(R9)(CX*1) + ADDQ CX, R12 + ADDQ CX, R9 JMP copy_2_end copy_2_move_3: - MOVW (CX), R12 - MOVB 2(CX), R14 - MOVW R12, (R9) - MOVB R14, 2(R9) - ADDQ R13, CX - ADDQ R13, R9 - JMP copy_2_end + MOVWQZX (R12), R13 + MOVB 2(R12), R14 + MOVW R13, (R9) + MOVB R14, 2(R9) + ADDQ CX, R12 + ADDQ CX, R9 + JMP copy_2_end copy_2_move_4through7: - MOVL (CX), R12 - MOVL -4(CX)(R13*1), R14 - MOVL R12, (R9) - MOVL R14, -4(R9)(R13*1) - ADDQ R13, CX - ADDQ R13, R9 + MOVL (R12), R13 + MOVL -4(R12)(CX*1), R14 + MOVL R13, (R9) + MOVL R14, -4(R9)(CX*1) + ADDQ CX, R12 + ADDQ CX, R9 JMP copy_2_end copy_2_move_8through16: - MOVQ (CX), R12 - MOVQ -8(CX)(R13*1), R14 - MOVQ R12, (R9) - MOVQ R14, -8(R9)(R13*1) - ADDQ R13, CX - ADDQ R13, R9 + MOVQ (R12), R13 + MOVQ -8(R12)(CX*1), R14 + MOVQ R13, (R9) + MOVQ R14, -8(R9)(CX*1) + ADDQ CX, R12 + ADDQ CX, R9 copy_2_end: JMP handle_loop // Copy overlapping match copy_overlapping_match: - ADDQ R13, R11 + ADDQ CX, R11 copy_slow_3: - MOVB (CX), R12 - MOVB R12, (R9) - INCQ CX - INCQ R9 - DECQ R13 - JNZ copy_slow_3 + MOVBQZX (R12), R13 + MOVB R13, (R9) + INCQ R12 + INCQ R9 + DECQ CX + JNZ copy_slow_3 handle_loop: - MOVQ ctx+16(FP), CX - DECQ 96(CX) + DECQ 32(SP) JNS sequenceDecs_decodeSync_safe_bmi2_main_loop loop_finished: @@ -4090,6 +4122,13 @@ loop_finished: MOVB DL, 40(CX) MOVQ BX, 32(CX) + // s.seqSize += outPosition - ctx.outPosition + MOVQ ctx+16(FP), AX + MOVQ s+0(FP), CX + MOVQ R11, DX + SUBQ 136(AX), DX + ADDQ DX, 256(CX) + // Update the context MOVQ ctx+16(FP), AX MOVQ R11, 136(AX) @@ -4103,7 +4142,7 @@ loop_finished: // Return with match length error sequenceDecs_decodeSync_safe_bmi2_error_match_len_ofs_mismatch: - MOVQ 16(SP), AX + MOVQ 48(SP), AX MOVQ ctx+16(FP), CX MOVQ AX, 216(CX) MOVQ $0x00000001, ret+24(FP) @@ -4112,7 +4151,7 @@ sequenceDecs_decodeSync_safe_bmi2_error_match_len_ofs_mismatch: // Return with match too long error sequenceDecs_decodeSync_safe_bmi2_error_match_len_too_big: MOVQ ctx+16(FP), AX - MOVQ 16(SP), CX + MOVQ 48(SP), CX MOVQ CX, 216(AX) MOVQ $0x00000002, ret+24(FP) RET @@ -4120,7 +4159,7 @@ sequenceDecs_decodeSync_safe_bmi2_error_match_len_too_big: // Return with match offset too long error error_match_off_too_big: MOVQ ctx+16(FP), AX - MOVQ 8(SP), CX + MOVQ 40(SP), CX MOVQ CX, 224(AX) MOVQ R11, 136(AX) MOVQ $0x00000003, ret+24(FP) @@ -4129,7 +4168,7 @@ error_match_off_too_big: // Return with not enough literals error error_not_enough_literals: MOVQ ctx+16(FP), AX - MOVQ 24(SP), CX + MOVQ 56(SP), CX MOVQ CX, 208(AX) MOVQ $0x00000004, ret+24(FP) RET @@ -4142,9 +4181,9 @@ error_overread: // Return with not enough output space error error_not_enough_space: MOVQ ctx+16(FP), AX - MOVQ 24(SP), CX + MOVQ 56(SP), CX MOVQ CX, 208(AX) - MOVQ 16(SP), CX + MOVQ 48(SP), CX MOVQ CX, 216(AX) MOVQ R11, 136(AX) MOVQ $0x00000005, ret+24(FP) diff --git a/vendor/github.com/klauspost/compress/zstd/seqdec_arm64.s b/vendor/github.com/klauspost/compress/zstd/seqdec_arm64.s index a468e5fc2c..56391bfc18 100644 --- a/vendor/github.com/klauspost/compress/zstd/seqdec_arm64.s +++ b/vendor/github.com/klauspost/compress/zstd/seqdec_arm64.s @@ -1,22 +1,30 @@ // Code generated by command: go run gen.go -out ../seqdec.s -arch amd64,arm64 -pkg=zstd. DO NOT EDIT. // EXPERIMENTAL arm64 output lowered from an amd64 avo program. -//go:build arm64 && !appengine && !noasm && gc && !noasm +//go:build arm64 && (!appengine && !noasm && gc && !noasm) // func sequenceDecs_decode_amd64(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int // Requires: CMOV -TEXT ·sequenceDecs_decode_arm64(SB), $8-32 +TEXT ·sequenceDecs_decode_arm64(SB), $40-32 MOVD br+8(FP), R1 MOVD 24(R1), R2 MOVBU 40(R1), R3 MOVD (R1), R0 MOVD 32(R1), R5 ADD R5, R0, R0 - MOVD R0, (RSP) + MOVD R0, 8(RSP) MOVD ctx+16(FP), R0 MOVD 72(R0), R6 MOVD 80(R0), R7 MOVD 88(R0), R8 + MOVD (R0), R1 + MOVD R1, 16(RSP) + MOVD 24(R0), R1 + MOVD R1, 24(RSP) + MOVD 48(R0), R1 + MOVD R1, 32(RSP) + MOVD 96(R0), R1 + MOVD R1, 40(RSP) MOVD 104(R0), R9 MOVD s+0(FP), R0 MOVD 144(R0), R10 @@ -24,13 +32,12 @@ TEXT ·sequenceDecs_decode_arm64(SB), $8-32 MOVD 160(R0), R12 sequenceDecs_decode_amd64_main_loop: - MOVD (RSP), R13 + MOVD 8(RSP), R13 // Fill bitreader to have enough for the offset and match length. CMP $0x08, R5 BLT sequenceDecs_decode_amd64_fill_byte_by_byte - MOVD R3, R0 - LSR $0x03, R0, R0 + LSR $0x03, R3, R0 SUB R0, R13, R13 MOVD (R13), R2 SUB R0, R5, R5 @@ -57,51 +64,32 @@ sequenceDecs_decode_amd64_fill_check_overread: sequenceDecs_decode_amd64_fill_end: // Update offset MOVD R8, R0 - MOVD R3, R1 - MOVD R2, R14 - LSL R1, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decode_amd64_of_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decode_amd64_of_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decode_amd64_of_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R14, R14 ADD R14, R0, R0 - -sequenceDecs_decode_amd64_of_update_zero: MOVD R0, 16(R9) // Update match length MOVD R7, R0 - MOVD R3, R1 - MOVD R2, R14 - LSL R1, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decode_amd64_ml_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decode_amd64_ml_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decode_amd64_ml_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R14, R14 ADD R14, R0, R0 - -sequenceDecs_decode_amd64_ml_update_zero: MOVD R0, 8(R9) // Fill bitreader to have enough for the remaining CMP $0x08, R5 BLT sequenceDecs_decode_amd64_fill_2_byte_by_byte - MOVD R3, R0 - LSR $0x03, R0, R0 + LSR $0x03, R3, R0 SUB R0, R13, R13 MOVD (R13), R2 SUB R0, R5, R5 @@ -128,97 +116,64 @@ sequenceDecs_decode_amd64_fill_2_check_overread: sequenceDecs_decode_amd64_fill_2_end: // Update literal length MOVD R6, R0 - MOVD R3, R1 - MOVD R2, R14 - LSL R1, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decode_amd64_ll_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decode_amd64_ll_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decode_amd64_ll_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R14, R14 ADD R14, R0, R0 - -sequenceDecs_decode_amd64_ll_update_zero: MOVD R0, (R9) // Fill bitreader for state updates - MOVD R13, (RSP) - MOVD R8, R0 - LSR $0x08, R0, R0 - MOVBU R0, R0 - MOVD ctx+16(FP), R1 - MOVD 96(R1), R16 - CMP $0x00, R16 - BEQ sequenceDecs_decode_amd64_skip_update + MOVD R13, 8(RSP) + UBFX $8, R8, $8, R0 + MOVD 40(RSP), R16 + CMP $0x00, R16 + BEQ sequenceDecs_decode_amd64_skip_update // Update Literal Length State MOVBU R6, R13 LSRW $0x10, R6, R6 - ADD R13, R3, R1 - MOVD R2, R14 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R14, R14 - MOVD $0x00000001, R4 - MOVB R13, R1 - LSLW R1, R4, R4 - SUBW $1, R4, R4 - AND R4, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 + ADD R13, R3, R3 + EOR $0x3f, R13, R13 + LSR R13, R14, R14 ADD R14, R6, R6 // Load ctx.llTable - MOVD ctx+16(FP), R1 - MOVD (R1), R1 - ADD R6<<3, R1, R15 - MOVD (R15), R6 + MOVD 16(RSP), R1 + MOVD (R1)(R6<<3), R6 // Update Match Length State MOVBU R7, R13 LSRW $0x10, R7, R7 - ADD R13, R3, R1 - MOVD R2, R14 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R14, R14 - MOVD $0x00000001, R4 - MOVB R13, R1 - LSLW R1, R4, R4 - SUBW $1, R4, R4 - AND R4, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 + ADD R13, R3, R3 + EOR $0x3f, R13, R13 + LSR R13, R14, R14 ADD R14, R7, R7 // Load ctx.mlTable - MOVD ctx+16(FP), R1 - MOVD 24(R1), R1 - ADD R7<<3, R1, R15 - MOVD (R15), R7 + MOVD 24(RSP), R1 + MOVD (R1)(R7<<3), R7 // Update Offset State MOVBU R8, R13 LSRW $0x10, R8, R8 - ADD R13, R3, R1 - MOVD R2, R14 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R14, R14 - MOVD $0x00000001, R4 - MOVB R13, R1 - LSLW R1, R4, R4 - SUBW $1, R4, R4 - AND R4, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 + ADD R13, R3, R3 + EOR $0x3f, R13, R13 + LSR R13, R14, R14 ADD R14, R8, R8 // Load ctx.ofTable - MOVD ctx+16(FP), R1 - MOVD 48(R1), R1 - ADD R8<<3, R1, R15 - MOVD (R15), R8 + MOVD 32(RSP), R1 + MOVD (R1)(R8<<3), R8 sequenceDecs_decode_amd64_skip_update: // Adjust offset @@ -303,10 +258,9 @@ sequenceDecs_decode_amd64_after_adjust: sequenceDecs_decode_amd64_match_len_ofs_ok: ADD $0x18, R9, R9 - MOVD ctx+16(FP), R0 - MOVD 96(R0), R16 + MOVD 40(RSP), R16 SUBS $1, R16, R16 - MOVD R16, 96(R0) + MOVD R16, 40(RSP) BPL sequenceDecs_decode_amd64_main_loop MOVD s+0(FP), R0 MOVD R10, 144(R0) @@ -318,53 +272,79 @@ sequenceDecs_decode_amd64_match_len_ofs_ok: MOVD R5, 32(R0) // Return success + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000000, R16 MOVD R16, ret+24(FP) RET // Return with match length error sequenceDecs_decode_amd64_error_match_len_ofs_mismatch: + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000001, R16 MOVD R16, ret+24(FP) RET // Return with match too long error sequenceDecs_decode_amd64_error_match_len_too_big: + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000002, R16 MOVD R16, ret+24(FP) RET // Return with match offset too long error + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000003, R16 MOVD R16, ret+24(FP) RET // Return with not enough literals error error_not_enough_literals: + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000004, R16 MOVD R16, ret+24(FP) RET // Return with overread error error_overread: + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000006, R16 MOVD R16, ret+24(FP) RET // func sequenceDecs_decode_56_amd64(s *sequenceDecs, br *bitReader, ctx *decodeAsmContext) int // Requires: CMOV -TEXT ·sequenceDecs_decode_56_arm64(SB), $8-32 +TEXT ·sequenceDecs_decode_56_arm64(SB), $40-32 MOVD br+8(FP), R1 MOVD 24(R1), R2 MOVBU 40(R1), R3 MOVD (R1), R0 MOVD 32(R1), R5 ADD R5, R0, R0 - MOVD R0, (RSP) + MOVD R0, 8(RSP) MOVD ctx+16(FP), R0 MOVD 72(R0), R6 MOVD 80(R0), R7 MOVD 88(R0), R8 + MOVD (R0), R1 + MOVD R1, 16(RSP) + MOVD 24(R0), R1 + MOVD R1, 24(RSP) + MOVD 48(R0), R1 + MOVD R1, 32(RSP) + MOVD 96(R0), R1 + MOVD R1, 40(RSP) MOVD 104(R0), R9 MOVD s+0(FP), R0 MOVD 144(R0), R10 @@ -372,13 +352,12 @@ TEXT ·sequenceDecs_decode_56_arm64(SB), $8-32 MOVD 160(R0), R12 sequenceDecs_decode_56_amd64_main_loop: - MOVD (RSP), R13 + MOVD 8(RSP), R13 // Fill bitreader to have enough for the offset and match length. CMP $0x08, R5 BLT sequenceDecs_decode_56_amd64_fill_byte_by_byte - MOVD R3, R0 - LSR $0x03, R0, R0 + LSR $0x03, R3, R0 SUB R0, R13, R13 MOVD (R13), R2 SUB R0, R5, R5 @@ -405,139 +384,88 @@ sequenceDecs_decode_56_amd64_fill_check_overread: sequenceDecs_decode_56_amd64_fill_end: // Update offset MOVD R8, R0 - MOVD R3, R1 - MOVD R2, R14 - LSL R1, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decode_56_amd64_of_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decode_56_amd64_of_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decode_56_amd64_of_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R14, R14 ADD R14, R0, R0 - -sequenceDecs_decode_56_amd64_of_update_zero: MOVD R0, 16(R9) // Update match length MOVD R7, R0 - MOVD R3, R1 - MOVD R2, R14 - LSL R1, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decode_56_amd64_ml_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decode_56_amd64_ml_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decode_56_amd64_ml_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R14, R14 ADD R14, R0, R0 - -sequenceDecs_decode_56_amd64_ml_update_zero: MOVD R0, 8(R9) // Update literal length MOVD R6, R0 - MOVD R3, R1 - MOVD R2, R14 - LSL R1, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decode_56_amd64_ll_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decode_56_amd64_ll_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decode_56_amd64_ll_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R14, R14 ADD R14, R0, R0 - -sequenceDecs_decode_56_amd64_ll_update_zero: MOVD R0, (R9) // Fill bitreader for state updates - MOVD R13, (RSP) - MOVD R8, R0 - LSR $0x08, R0, R0 - MOVBU R0, R0 - MOVD ctx+16(FP), R1 - MOVD 96(R1), R16 - CMP $0x00, R16 - BEQ sequenceDecs_decode_56_amd64_skip_update + MOVD R13, 8(RSP) + UBFX $8, R8, $8, R0 + MOVD 40(RSP), R16 + CMP $0x00, R16 + BEQ sequenceDecs_decode_56_amd64_skip_update // Update Literal Length State MOVBU R6, R13 LSRW $0x10, R6, R6 - ADD R13, R3, R1 - MOVD R2, R14 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R14, R14 - MOVD $0x00000001, R4 - MOVB R13, R1 - LSLW R1, R4, R4 - SUBW $1, R4, R4 - AND R4, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 + ADD R13, R3, R3 + EOR $0x3f, R13, R13 + LSR R13, R14, R14 ADD R14, R6, R6 // Load ctx.llTable - MOVD ctx+16(FP), R1 - MOVD (R1), R1 - ADD R6<<3, R1, R15 - MOVD (R15), R6 + MOVD 16(RSP), R1 + MOVD (R1)(R6<<3), R6 // Update Match Length State MOVBU R7, R13 LSRW $0x10, R7, R7 - ADD R13, R3, R1 - MOVD R2, R14 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R14, R14 - MOVD $0x00000001, R4 - MOVB R13, R1 - LSLW R1, R4, R4 - SUBW $1, R4, R4 - AND R4, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 + ADD R13, R3, R3 + EOR $0x3f, R13, R13 + LSR R13, R14, R14 ADD R14, R7, R7 // Load ctx.mlTable - MOVD ctx+16(FP), R1 - MOVD 24(R1), R1 - ADD R7<<3, R1, R15 - MOVD (R15), R7 + MOVD 24(RSP), R1 + MOVD (R1)(R7<<3), R7 // Update Offset State MOVBU R8, R13 LSRW $0x10, R8, R8 - ADD R13, R3, R1 - MOVD R2, R14 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R14, R14 - MOVD $0x00000001, R4 - MOVB R13, R1 - LSLW R1, R4, R4 - SUBW $1, R4, R4 - AND R4, R14, R14 + LSL R3, R2, R14 + LSR $0x01, R14, R14 + ADD R13, R3, R3 + EOR $0x3f, R13, R13 + LSR R13, R14, R14 ADD R14, R8, R8 // Load ctx.ofTable - MOVD ctx+16(FP), R1 - MOVD 48(R1), R1 - ADD R8<<3, R1, R15 - MOVD (R15), R8 + MOVD 32(RSP), R1 + MOVD (R1)(R8<<3), R8 sequenceDecs_decode_56_amd64_skip_update: // Adjust offset @@ -622,10 +550,9 @@ sequenceDecs_decode_56_amd64_after_adjust: sequenceDecs_decode_56_amd64_match_len_ofs_ok: ADD $0x18, R9, R9 - MOVD ctx+16(FP), R0 - MOVD 96(R0), R16 + MOVD 40(RSP), R16 SUBS $1, R16, R16 - MOVD R16, 96(R0) + MOVD R16, 40(RSP) BPL sequenceDecs_decode_56_amd64_main_loop MOVD s+0(FP), R0 MOVD R10, 144(R0) @@ -637,35 +564,53 @@ sequenceDecs_decode_56_amd64_match_len_ofs_ok: MOVD R5, 32(R0) // Return success + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000000, R16 MOVD R16, ret+24(FP) RET // Return with match length error sequenceDecs_decode_56_amd64_error_match_len_ofs_mismatch: + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000001, R16 MOVD R16, ret+24(FP) RET // Return with match too long error sequenceDecs_decode_56_amd64_error_match_len_too_big: + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000002, R16 MOVD R16, ret+24(FP) RET // Return with match offset too long error + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000003, R16 MOVD R16, ret+24(FP) RET // Return with not enough literals error error_not_enough_literals: + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000004, R16 MOVD R16, ret+24(FP) RET // Return with overread error error_overread: + MOVD 40(RSP), R0 + MOVD ctx+16(FP), R1 + MOVD R0, 96(R1) MOVD $0x00000006, R16 MOVD R16, ret+24(FP) RET @@ -705,24 +650,27 @@ main_loop: MOVD 8(R0), R12 // Copy literals - TST R10, R10 - BEQ check_offset - MOVD $0, R13 + FMOVQ (R5), F0 + FMOVQ F0, (R3) + CMP $0x10, R10 + BLS copy_1_end + MOVD $0x00000010, R13 copy_1: - ADD R13, R5, R15 - VLD1 (R15), [V0.B16] - ADD R13, R3, R15 - VST1 [V0.B16], (R15) - ADD $0x10, R13, R13 - CMP R10, R13 - BLO copy_1 - ADD R10, R5, R5 - ADD R10, R3, R3 - ADD R10, R6, R6 + ADD R13, R5, R15 + FMOVQ (R15), F0 + ADD R13, R3, R15 + FMOVQ F0, (R15) + ADD $0x10, R13, R13 + CMP R10, R13 + BLO copy_1 + +copy_1_end: + ADD R10, R5, R5 + ADD R10, R3, R3 + ADD R10, R6, R6 // Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize) -check_offset: ADD R9, R6, R10 CMP R10, R11 BGT error_match_off_too_big @@ -742,21 +690,19 @@ check_offset: BLO copy_4_small copy_4_loop: - VLD1 (R13), [V0.B16] - VST1 [V0.B16], (R3) - ADD $0x10, R13, R13 - ADD $0x10, R3, R3 - SUBS $0x10, R10, R10 - BHS copy_4_loop - ADD R10, R13, R13 - ADD $16, R13, R13 - ADD R10, R3, R3 - ADD $16, R3, R3 - ADD $-16, R13, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R3, R15 - VST1 [V0.B16], (R15) - JMP copy_4_end + FMOVQ (R13), F0 + FMOVQ F0, (R3) + ADD $0x10, R13, R13 + ADD $0x10, R3, R3 + SUBS $0x10, R10, R10 + BHS copy_4_loop + ADD R10, R13, R13 + ADD $16, R13, R13 + ADD R10, R3, R3 + ADD $16, R3, R3 + FMOVQ -16(R13), F0 + FMOVQ F0, -16(R3) + JMP copy_4_end copy_4_small: CMP $0x03, R12 @@ -766,13 +712,14 @@ copy_4_small: JMP copy_4_move_8through16 copy_4_move_3: - MOVH (R13), R10 - MOVB 2(R13), R11 - MOVH R10, (R3) - MOVB R11, 2(R3) - ADD R12, R13, R13 - ADD R12, R3, R3 - JMP copy_4_end + MOVHU (R13), R10 + MOVBU 2(R13), R16 + BFI $0, R16, $8, R11 + MOVH R10, (R3) + MOVB R11, 2(R3) + ADD R12, R13, R13 + ADD R12, R3, R3 + JMP copy_4_end copy_4_move_4through7: MOVWU (R13), R10 @@ -809,21 +756,19 @@ copy_all_from_history: BLO copy_5_small copy_5_loop: - VLD1 (R13), [V0.B16] - VST1 [V0.B16], (R3) - ADD $0x10, R13, R13 - ADD $0x10, R3, R3 - SUBS $0x10, R14, R14 - BHS copy_5_loop - ADD R14, R13, R13 - ADD $16, R13, R13 - ADD R14, R3, R3 - ADD $16, R3, R3 - ADD $-16, R13, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R3, R15 - VST1 [V0.B16], (R15) - JMP copy_5_end + FMOVQ (R13), F0 + FMOVQ F0, (R3) + ADD $0x10, R13, R13 + ADD $0x10, R3, R3 + SUBS $0x10, R14, R14 + BHS copy_5_loop + ADD R14, R13, R13 + ADD $16, R13, R13 + ADD R14, R3, R3 + ADD $16, R3, R3 + FMOVQ -16(R13), F0 + FMOVQ F0, -16(R3) + JMP copy_5_end copy_5_small: CMP $0x03, R10 @@ -834,24 +779,27 @@ copy_5_small: JMP copy_5_move_8through16 copy_5_move_1or2: - MOVB (R13), R14 - ADD R10, R13, R15 - MOVB -1(R15), R4 - MOVB R14, (R3) - ADD R10, R3, R15 - MOVB R4, -1(R15) - ADD R10, R13, R13 - ADD R10, R3, R3 - JMP copy_5_end + MOVBU (R13), R16 + BFI $0, R16, $8, R14 + ADD R10, R13, R15 + MOVBU -1(R15), R16 + BFI $0, R16, $8, R4 + MOVB R14, (R3) + ADD R10, R3, R15 + MOVB R4, -1(R15) + ADD R10, R13, R13 + ADD R10, R3, R3 + JMP copy_5_end copy_5_move_3: - MOVH (R13), R14 - MOVB 2(R13), R4 - MOVH R14, (R3) - MOVB R4, 2(R3) - ADD R10, R13, R13 - ADD R10, R3, R3 - JMP copy_5_end + MOVHU (R13), R14 + MOVBU 2(R13), R16 + BFI $0, R16, $8, R4 + MOVH R14, (R3) + MOVB R4, 2(R3) + ADD R10, R13, R13 + ADD R10, R3, R3 + JMP copy_5_end copy_5_move_4through7: MOVWU (R13), R14 @@ -888,30 +836,40 @@ copy_match: BHI copy_overlapping_match // Copy non-overlapping match - ADD R12, R6, R6 - MOVD R3, R11 - ADD R12, R3, R3 + ADD R12, R6, R6 + FMOVQ (R10), F0 + FMOVQ F0, (R3) + CMP $0x10, R12 + BHI copy_2_long + ADD R12, R3, R3 + JMP handle_loop + +copy_2_long: + ADD $16, R3, R11 + ADD R12, R3, R3 + ADD $0x10, R10, R10 + SUB $0x10, R12, R12 copy_2: - VLD1 (R10), [V0.B16] - VST1 [V0.B16], (R11) - ADD $0x10, R10, R10 - ADD $0x10, R11, R11 - SUBS $0x10, R12, R12 - BHI copy_2 - JMP handle_loop + FMOVQ (R10), F0 + FMOVQ F0, (R11) + ADD $0x10, R10, R10 + ADD $0x10, R11, R11 + SUBS $0x10, R12, R12 + BHI copy_2 + JMP handle_loop // Copy overlapping match copy_overlapping_match: ADD R12, R6, R6 copy_slow_3: - MOVB (R10), R11 - MOVB R11, (R3) - ADD $1, R10, R10 - ADD $1, R3, R3 - SUBS $1, R12, R12 - BNE copy_slow_3 + MOVBU (R10), R11 + MOVB R11, (R3) + ADD $1, R10, R10 + ADD $1, R3, R3 + SUBS $1, R12, R12 + BNE copy_slow_3 handle_loop: ADD $0x18, R0, R0 @@ -991,21 +949,19 @@ main_loop: BLO copy_1_small copy_1_loop: - VLD1 (R5), [V0.B16] - VST1 [V0.B16], (R3) - ADD $0x10, R5, R5 - ADD $0x10, R3, R3 - SUBS $0x10, R13, R13 - BHS copy_1_loop - ADD R13, R5, R5 - ADD $16, R5, R5 - ADD R13, R3, R3 - ADD $16, R3, R3 - ADD $-16, R5, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R3, R15 - VST1 [V0.B16], (R15) - JMP copy_1_end + FMOVQ (R5), F0 + FMOVQ F0, (R3) + ADD $0x10, R5, R5 + ADD $0x10, R3, R3 + SUBS $0x10, R13, R13 + BHS copy_1_loop + ADD R13, R5, R5 + ADD $16, R5, R5 + ADD R13, R3, R3 + ADD $16, R3, R3 + FMOVQ -16(R5), F0 + FMOVQ F0, -16(R3) + JMP copy_1_end copy_1_small: CMP $0x03, R10 @@ -1016,24 +972,27 @@ copy_1_small: JMP copy_1_move_8through16 copy_1_move_1or2: - MOVB (R5), R13 - ADD R10, R5, R15 - MOVB -1(R15), R14 - MOVB R13, (R3) - ADD R10, R3, R15 - MOVB R14, -1(R15) - ADD R10, R5, R5 - ADD R10, R3, R3 - JMP copy_1_end + MOVBU (R5), R16 + BFI $0, R16, $8, R13 + ADD R10, R5, R15 + MOVBU -1(R15), R16 + BFI $0, R16, $8, R14 + MOVB R13, (R3) + ADD R10, R3, R15 + MOVB R14, -1(R15) + ADD R10, R5, R5 + ADD R10, R3, R3 + JMP copy_1_end copy_1_move_3: - MOVH (R5), R13 - MOVB 2(R5), R14 - MOVH R13, (R3) - MOVB R14, 2(R3) - ADD R10, R5, R5 - ADD R10, R3, R3 - JMP copy_1_end + MOVHU (R5), R13 + MOVBU 2(R5), R16 + BFI $0, R16, $8, R14 + MOVH R13, (R3) + MOVB R14, 2(R3) + ADD R10, R5, R5 + ADD R10, R3, R3 + JMP copy_1_end copy_1_move_4through7: MOVWU (R5), R13 @@ -1080,21 +1039,19 @@ check_offset: BLO copy_4_small copy_4_loop: - VLD1 (R13), [V0.B16] - VST1 [V0.B16], (R3) - ADD $0x10, R13, R13 - ADD $0x10, R3, R3 - SUBS $0x10, R10, R10 - BHS copy_4_loop - ADD R10, R13, R13 - ADD $16, R13, R13 - ADD R10, R3, R3 - ADD $16, R3, R3 - ADD $-16, R13, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R3, R15 - VST1 [V0.B16], (R15) - JMP copy_4_end + FMOVQ (R13), F0 + FMOVQ F0, (R3) + ADD $0x10, R13, R13 + ADD $0x10, R3, R3 + SUBS $0x10, R10, R10 + BHS copy_4_loop + ADD R10, R13, R13 + ADD $16, R13, R13 + ADD R10, R3, R3 + ADD $16, R3, R3 + FMOVQ -16(R13), F0 + FMOVQ F0, -16(R3) + JMP copy_4_end copy_4_small: CMP $0x03, R12 @@ -1104,13 +1061,14 @@ copy_4_small: JMP copy_4_move_8through16 copy_4_move_3: - MOVH (R13), R10 - MOVB 2(R13), R11 - MOVH R10, (R3) - MOVB R11, 2(R3) - ADD R12, R13, R13 - ADD R12, R3, R3 - JMP copy_4_end + MOVHU (R13), R10 + MOVBU 2(R13), R16 + BFI $0, R16, $8, R11 + MOVH R10, (R3) + MOVB R11, 2(R3) + ADD R12, R13, R13 + ADD R12, R3, R3 + JMP copy_4_end copy_4_move_4through7: MOVWU (R13), R10 @@ -1147,21 +1105,19 @@ copy_all_from_history: BLO copy_5_small copy_5_loop: - VLD1 (R13), [V0.B16] - VST1 [V0.B16], (R3) - ADD $0x10, R13, R13 - ADD $0x10, R3, R3 - SUBS $0x10, R14, R14 - BHS copy_5_loop - ADD R14, R13, R13 - ADD $16, R13, R13 - ADD R14, R3, R3 - ADD $16, R3, R3 - ADD $-16, R13, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R3, R15 - VST1 [V0.B16], (R15) - JMP copy_5_end + FMOVQ (R13), F0 + FMOVQ F0, (R3) + ADD $0x10, R13, R13 + ADD $0x10, R3, R3 + SUBS $0x10, R14, R14 + BHS copy_5_loop + ADD R14, R13, R13 + ADD $16, R13, R13 + ADD R14, R3, R3 + ADD $16, R3, R3 + FMOVQ -16(R13), F0 + FMOVQ F0, -16(R3) + JMP copy_5_end copy_5_small: CMP $0x03, R10 @@ -1172,24 +1128,27 @@ copy_5_small: JMP copy_5_move_8through16 copy_5_move_1or2: - MOVB (R13), R14 - ADD R10, R13, R15 - MOVB -1(R15), R4 - MOVB R14, (R3) - ADD R10, R3, R15 - MOVB R4, -1(R15) - ADD R10, R13, R13 - ADD R10, R3, R3 - JMP copy_5_end + MOVBU (R13), R16 + BFI $0, R16, $8, R14 + ADD R10, R13, R15 + MOVBU -1(R15), R16 + BFI $0, R16, $8, R4 + MOVB R14, (R3) + ADD R10, R3, R15 + MOVB R4, -1(R15) + ADD R10, R13, R13 + ADD R10, R3, R3 + JMP copy_5_end copy_5_move_3: - MOVH (R13), R14 - MOVB 2(R13), R4 - MOVH R14, (R3) - MOVB R4, 2(R3) - ADD R10, R13, R13 - ADD R10, R3, R3 - JMP copy_5_end + MOVHU (R13), R14 + MOVBU 2(R13), R16 + BFI $0, R16, $8, R4 + MOVH R14, (R3) + MOVB R4, 2(R3) + ADD R10, R13, R13 + ADD R10, R3, R3 + JMP copy_5_end copy_5_move_4through7: MOVWU (R13), R14 @@ -1232,21 +1191,19 @@ copy_match: BLO copy_2_small copy_2_loop: - VLD1 (R10), [V0.B16] - VST1 [V0.B16], (R3) - ADD $0x10, R10, R10 - ADD $0x10, R3, R3 - SUBS $0x10, R11, R11 - BHS copy_2_loop - ADD R11, R10, R10 - ADD $16, R10, R10 - ADD R11, R3, R3 - ADD $16, R3, R3 - ADD $-16, R10, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R3, R15 - VST1 [V0.B16], (R15) - JMP copy_2_end + FMOVQ (R10), F0 + FMOVQ F0, (R3) + ADD $0x10, R10, R10 + ADD $0x10, R3, R3 + SUBS $0x10, R11, R11 + BHS copy_2_loop + ADD R11, R10, R10 + ADD $16, R10, R10 + ADD R11, R3, R3 + ADD $16, R3, R3 + FMOVQ -16(R10), F0 + FMOVQ F0, -16(R3) + JMP copy_2_end copy_2_small: CMP $0x03, R12 @@ -1257,24 +1214,27 @@ copy_2_small: JMP copy_2_move_8through16 copy_2_move_1or2: - MOVB (R10), R11 - ADD R12, R10, R15 - MOVB -1(R15), R13 - MOVB R11, (R3) - ADD R12, R3, R15 - MOVB R13, -1(R15) - ADD R12, R10, R10 - ADD R12, R3, R3 - JMP copy_2_end + MOVBU (R10), R16 + BFI $0, R16, $8, R11 + ADD R12, R10, R15 + MOVBU -1(R15), R16 + BFI $0, R16, $8, R13 + MOVB R11, (R3) + ADD R12, R3, R15 + MOVB R13, -1(R15) + ADD R12, R10, R10 + ADD R12, R3, R3 + JMP copy_2_end copy_2_move_3: - MOVH (R10), R11 - MOVB 2(R10), R13 - MOVH R11, (R3) - MOVB R13, 2(R3) - ADD R12, R10, R10 - ADD R12, R3, R3 - JMP copy_2_end + MOVHU (R10), R11 + MOVBU 2(R10), R16 + BFI $0, R16, $8, R13 + MOVH R11, (R3) + MOVB R13, 2(R3) + ADD R12, R10, R10 + ADD R12, R3, R3 + JMP copy_2_end copy_2_move_4through7: MOVWU (R10), R11 @@ -1305,12 +1265,12 @@ copy_overlapping_match: ADD R12, R6, R6 copy_slow_3: - MOVB (R10), R11 - MOVB R11, (R3) - ADD $1, R10, R10 - ADD $1, R3, R3 - SUBS $1, R12, R12 - BNE copy_slow_3 + MOVBU (R10), R11 + MOVB R11, (R3) + ADD $1, R10, R10 + ADD $1, R3, R3 + SUBS $1, R12, R12 + BNE copy_slow_3 handle_loop: ADD $0x18, R0, R0 @@ -1354,54 +1314,61 @@ empty_seqs: // func sequenceDecs_decodeSync_amd64(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int // Requires: CMOV, SSE -TEXT ·sequenceDecs_decodeSync_arm64(SB), $64-32 +TEXT ·sequenceDecs_decodeSync_arm64(SB), $96-32 MOVD br+8(FP), R1 MOVD 24(R1), R2 MOVBU 40(R1), R3 MOVD (R1), R0 MOVD 32(R1), R5 ADD R5, R0, R0 - MOVD R0, (RSP) + MOVD R0, 8(RSP) MOVD ctx+16(FP), R0 MOVD 72(R0), R6 MOVD 80(R0), R7 MOVD 88(R0), R8 - MOVD $0, R1 - MOVD R1, 8(RSP) + MOVD (R0), R1 MOVD R1, 16(RSP) + MOVD 24(R0), R1 MOVD R1, 24(RSP) + MOVD 48(R0), R1 + MOVD R1, 32(RSP) + MOVD 96(R0), R1 + MOVD R1, 40(RSP) + MOVD $0, R1 + MOVD R1, 48(RSP) + MOVD R1, 56(RSP) + MOVD R1, 64(RSP) MOVD 112(R0), R9 MOVD 128(R0), R1 - MOVD R1, 32(RSP) + MOVD R1, 72(RSP) MOVD 144(R0), R10 MOVD 136(R0), R11 MOVD 200(R0), R1 - MOVD R1, 56(RSP) + MOVD R1, 96(RSP) MOVD 176(R0), R1 - MOVD R1, 48(RSP) + MOVD R1, 88(RSP) MOVD 184(R0), R0 - MOVD R0, 40(RSP) - MOVD 40(RSP), R0 - MOVD 48(RSP), R16 + MOVD R0, 80(RSP) + MOVD 80(RSP), R0 + MOVD 88(RSP), R16 ADD R0, R16, R16 - MOVD R16, 48(RSP) + MOVD R16, 88(RSP) // Calculate pointer to s.out[cap(s.out)] (a past-end pointer) - MOVD 32(RSP), R16 + MOVD 72(RSP), R16 ADD R9, R16, R16 - MOVD R16, 32(RSP) + MOVD R16, 72(RSP) // outBase += outPosition ADD R11, R9, R9 sequenceDecs_decodeSync_amd64_main_loop: - MOVD (RSP), R12 + MOVD 8(RSP), R12 // Fill bitreader to have enough for the offset and match length. CMP $0x08, R5 BLT sequenceDecs_decodeSync_amd64_fill_byte_by_byte - MOVD R3, R0 - LSR $0x03, R0, R0 + LSR $0x03, R3, R0 SUB R0, R12, R12 MOVD (R12), R2 SUB R0, R5, R5 @@ -1428,51 +1395,32 @@ sequenceDecs_decodeSync_amd64_fill_check_overread: sequenceDecs_decodeSync_amd64_fill_end: // Update offset MOVD R8, R0 - MOVD R3, R1 - MOVD R2, R13 - LSL R1, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decodeSync_amd64_of_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decodeSync_amd64_of_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decodeSync_amd64_of_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R13, R13 ADD R13, R0, R0 - -sequenceDecs_decodeSync_amd64_of_update_zero: - MOVD R0, 8(RSP) + MOVD R0, 48(RSP) // Update match length MOVD R7, R0 - MOVD R3, R1 - MOVD R2, R13 - LSL R1, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decodeSync_amd64_ml_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decodeSync_amd64_ml_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decodeSync_amd64_ml_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R13, R13 ADD R13, R0, R0 - -sequenceDecs_decodeSync_amd64_ml_update_zero: - MOVD R0, 16(RSP) + MOVD R0, 56(RSP) // Fill bitreader to have enough for the remaining CMP $0x08, R5 BLT sequenceDecs_decodeSync_amd64_fill_2_byte_by_byte - MOVD R3, R0 - LSR $0x03, R0, R0 + LSR $0x03, R3, R0 SUB R0, R12, R12 MOVD (R12), R2 SUB R0, R5, R5 @@ -1499,113 +1447,78 @@ sequenceDecs_decodeSync_amd64_fill_2_check_overread: sequenceDecs_decodeSync_amd64_fill_2_end: // Update literal length MOVD R6, R0 - MOVD R3, R1 - MOVD R2, R13 - LSL R1, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decodeSync_amd64_ll_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decodeSync_amd64_ll_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decodeSync_amd64_ll_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R13, R13 ADD R13, R0, R0 - -sequenceDecs_decodeSync_amd64_ll_update_zero: - MOVD R0, 24(RSP) + MOVD R0, 64(RSP) // Fill bitreader for state updates - MOVD R12, (RSP) - MOVD R8, R0 - LSR $0x08, R0, R0 - MOVBU R0, R0 - MOVD ctx+16(FP), R1 - MOVD 96(R1), R16 - CMP $0x00, R16 - BEQ sequenceDecs_decodeSync_amd64_skip_update + MOVD R12, 8(RSP) + UBFX $8, R8, $8, R0 + MOVD 40(RSP), R16 + CMP $0x00, R16 + BEQ sequenceDecs_decodeSync_amd64_skip_update // Update Literal Length State MOVBU R6, R12 LSRW $0x10, R6, R6 - ADD R12, R3, R1 - MOVD R2, R13 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R13, R13 - MOVD $0x00000001, R14 - MOVB R12, R1 - LSLW R1, R14, R14 - SUBW $1, R14, R14 - AND R14, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 + ADD R12, R3, R3 + EOR $0x3f, R12, R12 + LSR R12, R13, R13 ADD R13, R6, R6 // Load ctx.llTable - MOVD ctx+16(FP), R1 - MOVD (R1), R1 - ADD R6<<3, R1, R15 - MOVD (R15), R6 + MOVD 16(RSP), R1 + MOVD (R1)(R6<<3), R6 // Update Match Length State MOVBU R7, R12 LSRW $0x10, R7, R7 - ADD R12, R3, R1 - MOVD R2, R13 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R13, R13 - MOVD $0x00000001, R14 - MOVB R12, R1 - LSLW R1, R14, R14 - SUBW $1, R14, R14 - AND R14, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 + ADD R12, R3, R3 + EOR $0x3f, R12, R12 + LSR R12, R13, R13 ADD R13, R7, R7 // Load ctx.mlTable - MOVD ctx+16(FP), R1 - MOVD 24(R1), R1 - ADD R7<<3, R1, R15 - MOVD (R15), R7 + MOVD 24(RSP), R1 + MOVD (R1)(R7<<3), R7 // Update Offset State MOVBU R8, R12 LSRW $0x10, R8, R8 - ADD R12, R3, R1 - MOVD R2, R13 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R13, R13 - MOVD $0x00000001, R14 - MOVB R12, R1 - LSLW R1, R14, R14 - SUBW $1, R14, R14 - AND R14, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 + ADD R12, R3, R3 + EOR $0x3f, R12, R12 + LSR R12, R13, R13 ADD R13, R8, R8 // Load ctx.ofTable - MOVD ctx+16(FP), R1 - MOVD 48(R1), R1 - ADD R8<<3, R1, R15 - MOVD (R15), R8 + MOVD 32(RSP), R1 + MOVD (R1)(R8<<3), R8 sequenceDecs_decodeSync_amd64_skip_update: // Adjust offset - MOVD s+0(FP), R1 - MOVD 8(RSP), R12 - CMP $0x01, R0 - BLS sequenceDecs_decodeSync_amd64_adjust_offsetB_1_or_0 - ADD $144, R1, R15 - VLD1 (R15), [V0.B16] - MOVD R12, 144(R1) - ADD $152, R1, R15 - VST1 [V0.B16], (R15) - JMP sequenceDecs_decodeSync_amd64_after_adjust + MOVD s+0(FP), R1 + MOVD 48(RSP), R12 + CMP $0x01, R0 + BLS sequenceDecs_decodeSync_amd64_adjust_offsetB_1_or_0 + FMOVQ 144(R1), F0 + MOVD R12, 144(R1) + FMOVQ F0, 152(R1) + JMP sequenceDecs_decodeSync_amd64_after_adjust sequenceDecs_decodeSync_amd64_adjust_offsetB_1_or_0: - MOVD 24(RSP), R16 + MOVD 64(RSP), R16 CMP $0x00000000, R16 BNE sequenceDecs_decodeSync_amd64_adjust_offset_maybezero ADD $1, R12, R12 @@ -1643,16 +1556,11 @@ sequenceDecs_decodeSync_amd64_adjust_skip: MOVD R13, R12 sequenceDecs_decodeSync_amd64_after_adjust: - MOVD R12, 8(RSP) + MOVD R12, 48(RSP) // Check values - MOVD 16(RSP), R0 - MOVD 24(RSP), R1 - ADD R1, R0, R13 - MOVD s+0(FP), R14 - MOVD 256(R14), R16 - ADD R13, R16, R16 - MOVD R16, 256(R14) + MOVD 56(RSP), R0 + MOVD 64(RSP), R1 MOVD ctx+16(FP), R13 MOVD 104(R13), R16 SUBS R1, R16, R16 @@ -1666,231 +1574,240 @@ sequenceDecs_decodeSync_amd64_after_adjust: BNE sequenceDecs_decodeSync_amd64_error_match_len_ofs_mismatch sequenceDecs_decodeSync_amd64_match_len_ofs_ok: - MOVD 24(RSP), R0 - MOVD 8(RSP), R1 - MOVD 16(RSP), R12 - // Check if we have enough space in s.out - ADD R12, R0, R13 + ADD R0, R1, R13 + ADD $16, R13, R13 ADD R9, R13, R13 - MOVD 32(RSP), R16 + MOVD 72(RSP), R16 CMP R16, R13 BHI error_not_enough_space // Copy literals - TST R0, R0 - BEQ check_offset - MOVD $0, R13 + FMOVQ (R10), F0 + FMOVQ F0, (R9) + CMP $0x10, R1 + BLS copy_1_end + MOVD $0x00000010, R13 copy_1: - ADD R13, R10, R15 - VLD1 (R15), [V0.B16] - ADD R13, R9, R15 - VST1 [V0.B16], (R15) - ADD $0x10, R13, R13 - CMP R0, R13 - BLO copy_1 - ADD R0, R10, R10 - ADD R0, R9, R9 - ADD R0, R11, R11 + ADD R13, R10, R15 + FMOVQ (R15), F0 + ADD R13, R9, R15 + FMOVQ F0, (R15) + ADD $0x10, R13, R13 + CMP R1, R13 + BLO copy_1 + +copy_1_end: + ADD R1, R10, R10 + ADD R1, R9, R9 + ADD R1, R11, R11 // Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize) -check_offset: - MOVD R11, R0 - MOVD 40(RSP), R16 - ADD R16, R0, R0 - CMP R0, R1 + MOVD R11, R1 + MOVD 80(RSP), R16 + ADD R16, R1, R1 + CMP R1, R12 BGT error_match_off_too_big - MOVD 56(RSP), R16 - CMP R16, R1 + MOVD 96(RSP), R16 + CMP R16, R12 BGT error_match_off_too_big // Copy match from history - MOVD R1, R0 - SUBS R11, R0, R0 + MOVD R12, R1 + SUBS R11, R1, R1 BLS copy_match - MOVD 48(RSP), R13 - SUB R0, R13, R13 - CMP R0, R12 + MOVD 88(RSP), R13 + SUB R1, R13, R13 + CMP R1, R0 BGT copy_all_from_history - MOVD R12, R0 - SUBS $0x10, R0, R0 + MOVD R0, R1 + SUBS $0x10, R1, R1 BLO copy_4_small copy_4_loop: - VLD1 (R13), [V0.B16] - VST1 [V0.B16], (R9) - ADD $0x10, R13, R13 - ADD $0x10, R9, R9 - SUBS $0x10, R0, R0 - BHS copy_4_loop - ADD R0, R13, R13 - ADD $16, R13, R13 - ADD R0, R9, R9 - ADD $16, R9, R9 - ADD $-16, R13, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R9, R15 - VST1 [V0.B16], (R15) - JMP copy_4_end + FMOVQ (R13), F0 + FMOVQ F0, (R9) + ADD $0x10, R13, R13 + ADD $0x10, R9, R9 + SUBS $0x10, R1, R1 + BHS copy_4_loop + ADD R1, R13, R13 + ADD $16, R13, R13 + ADD R1, R9, R9 + ADD $16, R9, R9 + FMOVQ -16(R13), F0 + FMOVQ F0, -16(R9) + JMP copy_4_end copy_4_small: - CMP $0x03, R12 + CMP $0x03, R0 BEQ copy_4_move_3 - CMP $0x08, R12 + CMP $0x08, R0 BLO copy_4_move_4through7 JMP copy_4_move_8through16 copy_4_move_3: - MOVH (R13), R0 - MOVB 2(R13), R1 - MOVH R0, (R9) - MOVB R1, 2(R9) - ADD R12, R13, R13 - ADD R12, R9, R9 - JMP copy_4_end + MOVHU (R13), R1 + MOVBU 2(R13), R16 + BFI $0, R16, $8, R12 + MOVH R1, (R9) + MOVB R12, 2(R9) + ADD R0, R13, R13 + ADD R0, R9, R9 + JMP copy_4_end copy_4_move_4through7: - MOVWU (R13), R0 - ADD R12, R13, R15 - MOVWU -4(R15), R1 - MOVW R0, (R9) - ADD R12, R9, R15 - MOVW R1, -4(R15) - ADD R12, R13, R13 - ADD R12, R9, R9 + MOVWU (R13), R1 + ADD R0, R13, R15 + MOVWU -4(R15), R12 + MOVW R1, (R9) + ADD R0, R9, R15 + MOVW R12, -4(R15) + ADD R0, R13, R13 + ADD R0, R9, R9 JMP copy_4_end copy_4_move_8through16: - MOVD (R13), R0 - ADD R12, R13, R15 - MOVD -8(R15), R1 - MOVD R0, (R9) - ADD R12, R9, R15 - MOVD R1, -8(R15) - ADD R12, R13, R13 - ADD R12, R9, R9 + MOVD (R13), R1 + ADD R0, R13, R15 + MOVD -8(R15), R12 + MOVD R1, (R9) + ADD R0, R9, R15 + MOVD R12, -8(R15) + ADD R0, R13, R13 + ADD R0, R9, R9 copy_4_end: - ADD R12, R11, R11 + ADD R0, R11, R11 JMP handle_loop JMP loop_finished copy_all_from_history: - MOVD R0, R14 + MOVD R1, R14 SUBS $0x10, R14, R14 BLO copy_5_small copy_5_loop: - VLD1 (R13), [V0.B16] - VST1 [V0.B16], (R9) - ADD $0x10, R13, R13 - ADD $0x10, R9, R9 - SUBS $0x10, R14, R14 - BHS copy_5_loop - ADD R14, R13, R13 - ADD $16, R13, R13 - ADD R14, R9, R9 - ADD $16, R9, R9 - ADD $-16, R13, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R9, R15 - VST1 [V0.B16], (R15) - JMP copy_5_end + FMOVQ (R13), F0 + FMOVQ F0, (R9) + ADD $0x10, R13, R13 + ADD $0x10, R9, R9 + SUBS $0x10, R14, R14 + BHS copy_5_loop + ADD R14, R13, R13 + ADD $16, R13, R13 + ADD R14, R9, R9 + ADD $16, R9, R9 + FMOVQ -16(R13), F0 + FMOVQ F0, -16(R9) + JMP copy_5_end copy_5_small: - CMP $0x03, R0 + CMP $0x03, R1 BEQ copy_5_move_3 BLO copy_5_move_1or2 - CMP $0x08, R0 + CMP $0x08, R1 BLO copy_5_move_4through7 JMP copy_5_move_8through16 copy_5_move_1or2: - MOVB (R13), R14 - ADD R0, R13, R15 - MOVB -1(R15), R4 - MOVB R14, (R9) - ADD R0, R9, R15 - MOVB R4, -1(R15) - ADD R0, R13, R13 - ADD R0, R9, R9 - JMP copy_5_end + MOVBU (R13), R16 + BFI $0, R16, $8, R14 + ADD R1, R13, R15 + MOVBU -1(R15), R16 + BFI $0, R16, $8, R4 + MOVB R14, (R9) + ADD R1, R9, R15 + MOVB R4, -1(R15) + ADD R1, R13, R13 + ADD R1, R9, R9 + JMP copy_5_end copy_5_move_3: - MOVH (R13), R14 - MOVB 2(R13), R4 - MOVH R14, (R9) - MOVB R4, 2(R9) - ADD R0, R13, R13 - ADD R0, R9, R9 - JMP copy_5_end + MOVHU (R13), R14 + MOVBU 2(R13), R16 + BFI $0, R16, $8, R4 + MOVH R14, (R9) + MOVB R4, 2(R9) + ADD R1, R13, R13 + ADD R1, R9, R9 + JMP copy_5_end copy_5_move_4through7: MOVWU (R13), R14 - ADD R0, R13, R15 + ADD R1, R13, R15 MOVWU -4(R15), R4 MOVW R14, (R9) - ADD R0, R9, R15 + ADD R1, R9, R15 MOVW R4, -4(R15) - ADD R0, R13, R13 - ADD R0, R9, R9 + ADD R1, R13, R13 + ADD R1, R9, R9 JMP copy_5_end copy_5_move_8through16: MOVD (R13), R14 - ADD R0, R13, R15 + ADD R1, R13, R15 MOVD -8(R15), R4 MOVD R14, (R9) - ADD R0, R9, R15 + ADD R1, R9, R15 MOVD R4, -8(R15) - ADD R0, R13, R13 - ADD R0, R9, R9 + ADD R1, R13, R13 + ADD R1, R9, R9 copy_5_end: - ADD R0, R11, R11 - SUB R0, R12, R12 + ADD R1, R11, R11 + SUB R1, R0, R0 // Copy match from the current buffer copy_match: - MOVD R9, R0 - SUB R1, R0, R0 + MOVD R9, R1 + SUB R12, R1, R1 // ml <= mo - CMP R1, R12 + CMP R12, R0 BHI copy_overlapping_match // Copy non-overlapping match - ADD R12, R11, R11 - MOVD R9, R1 - ADD R12, R9, R9 + ADD R0, R11, R11 + FMOVQ (R1), F0 + FMOVQ F0, (R9) + CMP $0x10, R0 + BHI copy_2_long + ADD R0, R9, R9 + JMP handle_loop + +copy_2_long: + ADD $16, R9, R12 + ADD R0, R9, R9 + ADD $0x10, R1, R1 + SUB $0x10, R0, R0 copy_2: - VLD1 (R0), [V0.B16] - VST1 [V0.B16], (R1) - ADD $0x10, R0, R0 - ADD $0x10, R1, R1 - SUBS $0x10, R12, R12 - BHI copy_2 - JMP handle_loop + FMOVQ (R1), F0 + FMOVQ F0, (R12) + ADD $0x10, R1, R1 + ADD $0x10, R12, R12 + SUBS $0x10, R0, R0 + BHI copy_2 + JMP handle_loop // Copy overlapping match copy_overlapping_match: - ADD R12, R11, R11 + ADD R0, R11, R11 copy_slow_3: - MOVB (R0), R1 - MOVB R1, (R9) - ADD $1, R0, R0 - ADD $1, R9, R9 - SUBS $1, R12, R12 - BNE copy_slow_3 + MOVBU (R1), R12 + MOVB R12, (R9) + ADD $1, R1, R1 + ADD $1, R9, R9 + SUBS $1, R0, R0 + BNE copy_slow_3 handle_loop: - MOVD ctx+16(FP), R0 - MOVD 96(R0), R16 + MOVD 40(RSP), R16 SUBS $1, R16, R16 - MOVD R16, 96(R0) + MOVD R16, 40(RSP) BPL sequenceDecs_decodeSync_amd64_main_loop loop_finished: @@ -1899,6 +1816,16 @@ loop_finished: MOVB R3, 40(R0) MOVD R5, 32(R0) + // s.seqSize += outPosition - ctx.outPosition + MOVD ctx+16(FP), R0 + MOVD s+0(FP), R1 + MOVD R11, R2 + MOVD 136(R0), R16 + SUB R16, R2, R2 + MOVD 256(R1), R16 + ADD R2, R16, R16 + MOVD R16, 256(R1) + // Update the context MOVD ctx+16(FP), R0 MOVD R11, 136(R0) @@ -1913,7 +1840,7 @@ loop_finished: // Return with match length error sequenceDecs_decodeSync_amd64_error_match_len_ofs_mismatch: - MOVD 16(RSP), R0 + MOVD 56(RSP), R0 MOVD ctx+16(FP), R1 MOVD R0, 216(R1) MOVD $0x00000001, R16 @@ -1923,7 +1850,7 @@ sequenceDecs_decodeSync_amd64_error_match_len_ofs_mismatch: // Return with match too long error sequenceDecs_decodeSync_amd64_error_match_len_too_big: MOVD ctx+16(FP), R0 - MOVD 16(RSP), R1 + MOVD 56(RSP), R1 MOVD R1, 216(R0) MOVD $0x00000002, R16 MOVD R16, ret+24(FP) @@ -1932,7 +1859,7 @@ sequenceDecs_decodeSync_amd64_error_match_len_too_big: // Return with match offset too long error error_match_off_too_big: MOVD ctx+16(FP), R0 - MOVD 8(RSP), R1 + MOVD 48(RSP), R1 MOVD R1, 224(R0) MOVD R11, 136(R0) MOVD $0x00000003, R16 @@ -1942,7 +1869,7 @@ error_match_off_too_big: // Return with not enough literals error error_not_enough_literals: MOVD ctx+16(FP), R0 - MOVD 24(RSP), R1 + MOVD 64(RSP), R1 MOVD R1, 208(R0) MOVD $0x00000004, R16 MOVD R16, ret+24(FP) @@ -1957,9 +1884,9 @@ error_overread: // Return with not enough output space error error_not_enough_space: MOVD ctx+16(FP), R0 - MOVD 24(RSP), R1 + MOVD 64(RSP), R1 MOVD R1, 208(R0) - MOVD 16(RSP), R1 + MOVD 56(RSP), R1 MOVD R1, 216(R0) MOVD R11, 136(R0) MOVD $0x00000005, R16 @@ -1970,54 +1897,61 @@ error_not_enough_space: // func sequenceDecs_decodeSync_safe_amd64(s *sequenceDecs, br *bitReader, ctx *decodeSyncAsmContext) int // Requires: CMOV, SSE -TEXT ·sequenceDecs_decodeSync_safe_arm64(SB), $64-32 +TEXT ·sequenceDecs_decodeSync_safe_arm64(SB), $96-32 MOVD br+8(FP), R1 MOVD 24(R1), R2 MOVBU 40(R1), R3 MOVD (R1), R0 MOVD 32(R1), R5 ADD R5, R0, R0 - MOVD R0, (RSP) + MOVD R0, 8(RSP) MOVD ctx+16(FP), R0 MOVD 72(R0), R6 MOVD 80(R0), R7 MOVD 88(R0), R8 - MOVD $0, R1 - MOVD R1, 8(RSP) + MOVD (R0), R1 MOVD R1, 16(RSP) + MOVD 24(R0), R1 MOVD R1, 24(RSP) + MOVD 48(R0), R1 + MOVD R1, 32(RSP) + MOVD 96(R0), R1 + MOVD R1, 40(RSP) + MOVD $0, R1 + MOVD R1, 48(RSP) + MOVD R1, 56(RSP) + MOVD R1, 64(RSP) MOVD 112(R0), R9 MOVD 128(R0), R1 - MOVD R1, 32(RSP) + MOVD R1, 72(RSP) MOVD 144(R0), R10 MOVD 136(R0), R11 MOVD 200(R0), R1 - MOVD R1, 56(RSP) + MOVD R1, 96(RSP) MOVD 176(R0), R1 - MOVD R1, 48(RSP) + MOVD R1, 88(RSP) MOVD 184(R0), R0 - MOVD R0, 40(RSP) - MOVD 40(RSP), R0 - MOVD 48(RSP), R16 + MOVD R0, 80(RSP) + MOVD 80(RSP), R0 + MOVD 88(RSP), R16 ADD R0, R16, R16 - MOVD R16, 48(RSP) + MOVD R16, 88(RSP) // Calculate pointer to s.out[cap(s.out)] (a past-end pointer) - MOVD 32(RSP), R16 + MOVD 72(RSP), R16 ADD R9, R16, R16 - MOVD R16, 32(RSP) + MOVD R16, 72(RSP) // outBase += outPosition ADD R11, R9, R9 sequenceDecs_decodeSync_safe_amd64_main_loop: - MOVD (RSP), R12 + MOVD 8(RSP), R12 // Fill bitreader to have enough for the offset and match length. CMP $0x08, R5 BLT sequenceDecs_decodeSync_safe_amd64_fill_byte_by_byte - MOVD R3, R0 - LSR $0x03, R0, R0 + LSR $0x03, R3, R0 SUB R0, R12, R12 MOVD (R12), R2 SUB R0, R5, R5 @@ -2044,51 +1978,32 @@ sequenceDecs_decodeSync_safe_amd64_fill_check_overread: sequenceDecs_decodeSync_safe_amd64_fill_end: // Update offset MOVD R8, R0 - MOVD R3, R1 - MOVD R2, R13 - LSL R1, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decodeSync_safe_amd64_of_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decodeSync_safe_amd64_of_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decodeSync_safe_amd64_of_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R13, R13 ADD R13, R0, R0 - -sequenceDecs_decodeSync_safe_amd64_of_update_zero: - MOVD R0, 8(RSP) + MOVD R0, 48(RSP) // Update match length MOVD R7, R0 - MOVD R3, R1 - MOVD R2, R13 - LSL R1, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decodeSync_safe_amd64_ml_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decodeSync_safe_amd64_ml_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decodeSync_safe_amd64_ml_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R13, R13 ADD R13, R0, R0 - -sequenceDecs_decodeSync_safe_amd64_ml_update_zero: - MOVD R0, 16(RSP) + MOVD R0, 56(RSP) // Fill bitreader to have enough for the remaining CMP $0x08, R5 BLT sequenceDecs_decodeSync_safe_amd64_fill_2_byte_by_byte - MOVD R3, R0 - LSR $0x03, R0, R0 + LSR $0x03, R3, R0 SUB R0, R12, R12 MOVD (R12), R2 SUB R0, R5, R5 @@ -2115,113 +2030,78 @@ sequenceDecs_decodeSync_safe_amd64_fill_2_check_overread: sequenceDecs_decodeSync_safe_amd64_fill_2_end: // Update literal length MOVD R6, R0 - MOVD R3, R1 - MOVD R2, R13 - LSL R1, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 UBFX $8, R0, $8, R1 LSR $0x20, R0, R0 - TST R1, R1 - BEQ sequenceDecs_decodeSync_safe_amd64_ll_update_zero ADD R1, R3, R3 - CMP $0x40, R3 - BHI sequenceDecs_decodeSync_safe_amd64_ll_update_zero - CMP $0x40, R1 - BHS sequenceDecs_decodeSync_safe_amd64_ll_update_zero - NEG R1, R1 + EOR $0x3f, R1, R1 LSR R1, R13, R13 ADD R13, R0, R0 - -sequenceDecs_decodeSync_safe_amd64_ll_update_zero: - MOVD R0, 24(RSP) + MOVD R0, 64(RSP) // Fill bitreader for state updates - MOVD R12, (RSP) - MOVD R8, R0 - LSR $0x08, R0, R0 - MOVBU R0, R0 - MOVD ctx+16(FP), R1 - MOVD 96(R1), R16 - CMP $0x00, R16 - BEQ sequenceDecs_decodeSync_safe_amd64_skip_update + MOVD R12, 8(RSP) + UBFX $8, R8, $8, R0 + MOVD 40(RSP), R16 + CMP $0x00, R16 + BEQ sequenceDecs_decodeSync_safe_amd64_skip_update // Update Literal Length State MOVBU R6, R12 LSRW $0x10, R6, R6 - ADD R12, R3, R1 - MOVD R2, R13 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R13, R13 - MOVD $0x00000001, R14 - MOVB R12, R1 - LSLW R1, R14, R14 - SUBW $1, R14, R14 - AND R14, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 + ADD R12, R3, R3 + EOR $0x3f, R12, R12 + LSR R12, R13, R13 ADD R13, R6, R6 // Load ctx.llTable - MOVD ctx+16(FP), R1 - MOVD (R1), R1 - ADD R6<<3, R1, R15 - MOVD (R15), R6 + MOVD 16(RSP), R1 + MOVD (R1)(R6<<3), R6 // Update Match Length State MOVBU R7, R12 LSRW $0x10, R7, R7 - ADD R12, R3, R1 - MOVD R2, R13 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R13, R13 - MOVD $0x00000001, R14 - MOVB R12, R1 - LSLW R1, R14, R14 - SUBW $1, R14, R14 - AND R14, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 + ADD R12, R3, R3 + EOR $0x3f, R12, R12 + LSR R12, R13, R13 ADD R13, R7, R7 // Load ctx.mlTable - MOVD ctx+16(FP), R1 - MOVD 24(R1), R1 - ADD R7<<3, R1, R15 - MOVD (R15), R7 + MOVD 24(RSP), R1 + MOVD (R1)(R7<<3), R7 // Update Offset State MOVBU R8, R12 LSRW $0x10, R8, R8 - ADD R12, R3, R1 - MOVD R2, R13 - MOVD R1, R3 - NEG R1, R16 - ROR R16, R13, R13 - MOVD $0x00000001, R14 - MOVB R12, R1 - LSLW R1, R14, R14 - SUBW $1, R14, R14 - AND R14, R13, R13 + LSL R3, R2, R13 + LSR $0x01, R13, R13 + ADD R12, R3, R3 + EOR $0x3f, R12, R12 + LSR R12, R13, R13 ADD R13, R8, R8 // Load ctx.ofTable - MOVD ctx+16(FP), R1 - MOVD 48(R1), R1 - ADD R8<<3, R1, R15 - MOVD (R15), R8 + MOVD 32(RSP), R1 + MOVD (R1)(R8<<3), R8 sequenceDecs_decodeSync_safe_amd64_skip_update: // Adjust offset - MOVD s+0(FP), R1 - MOVD 8(RSP), R12 - CMP $0x01, R0 - BLS sequenceDecs_decodeSync_safe_amd64_adjust_offsetB_1_or_0 - ADD $144, R1, R15 - VLD1 (R15), [V0.B16] - MOVD R12, 144(R1) - ADD $152, R1, R15 - VST1 [V0.B16], (R15) - JMP sequenceDecs_decodeSync_safe_amd64_after_adjust + MOVD s+0(FP), R1 + MOVD 48(RSP), R12 + CMP $0x01, R0 + BLS sequenceDecs_decodeSync_safe_amd64_adjust_offsetB_1_or_0 + FMOVQ 144(R1), F0 + MOVD R12, 144(R1) + FMOVQ F0, 152(R1) + JMP sequenceDecs_decodeSync_safe_amd64_after_adjust sequenceDecs_decodeSync_safe_amd64_adjust_offsetB_1_or_0: - MOVD 24(RSP), R16 + MOVD 64(RSP), R16 CMP $0x00000000, R16 BNE sequenceDecs_decodeSync_safe_amd64_adjust_offset_maybezero ADD $1, R12, R12 @@ -2259,16 +2139,11 @@ sequenceDecs_decodeSync_safe_amd64_adjust_skip: MOVD R13, R12 sequenceDecs_decodeSync_safe_amd64_after_adjust: - MOVD R12, 8(RSP) + MOVD R12, 48(RSP) // Check values - MOVD 16(RSP), R0 - MOVD 24(RSP), R1 - ADD R1, R0, R13 - MOVD s+0(FP), R14 - MOVD 256(R14), R16 - ADD R13, R16, R16 - MOVD R16, 256(R14) + MOVD 56(RSP), R0 + MOVD 64(RSP), R1 MOVD ctx+16(FP), R13 MOVD 104(R13), R16 SUBS R1, R16, R16 @@ -2282,351 +2157,348 @@ sequenceDecs_decodeSync_safe_amd64_after_adjust: BNE sequenceDecs_decodeSync_safe_amd64_error_match_len_ofs_mismatch sequenceDecs_decodeSync_safe_amd64_match_len_ofs_ok: - MOVD 24(RSP), R0 - MOVD 8(RSP), R1 - MOVD 16(RSP), R12 - // Check if we have enough space in s.out - ADD R12, R0, R13 + ADD R0, R1, R13 ADD R9, R13, R13 - MOVD 32(RSP), R16 + MOVD 72(RSP), R16 CMP R16, R13 BHI error_not_enough_space // Copy literals - TST R0, R0 + TST R1, R1 BEQ check_offset - MOVD R0, R13 + MOVD R1, R13 SUBS $0x10, R13, R13 BLO copy_1_small copy_1_loop: - VLD1 (R10), [V0.B16] - VST1 [V0.B16], (R9) - ADD $0x10, R10, R10 - ADD $0x10, R9, R9 - SUBS $0x10, R13, R13 - BHS copy_1_loop - ADD R13, R10, R10 - ADD $16, R10, R10 - ADD R13, R9, R9 - ADD $16, R9, R9 - ADD $-16, R10, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R9, R15 - VST1 [V0.B16], (R15) - JMP copy_1_end + FMOVQ (R10), F0 + FMOVQ F0, (R9) + ADD $0x10, R10, R10 + ADD $0x10, R9, R9 + SUBS $0x10, R13, R13 + BHS copy_1_loop + ADD R13, R10, R10 + ADD $16, R10, R10 + ADD R13, R9, R9 + ADD $16, R9, R9 + FMOVQ -16(R10), F0 + FMOVQ F0, -16(R9) + JMP copy_1_end copy_1_small: - CMP $0x03, R0 + CMP $0x03, R1 BEQ copy_1_move_3 BLO copy_1_move_1or2 - CMP $0x08, R0 + CMP $0x08, R1 BLO copy_1_move_4through7 JMP copy_1_move_8through16 copy_1_move_1or2: - MOVB (R10), R13 - ADD R0, R10, R15 - MOVB -1(R15), R14 - MOVB R13, (R9) - ADD R0, R9, R15 - MOVB R14, -1(R15) - ADD R0, R10, R10 - ADD R0, R9, R9 - JMP copy_1_end + MOVBU (R10), R16 + BFI $0, R16, $8, R13 + ADD R1, R10, R15 + MOVBU -1(R15), R16 + BFI $0, R16, $8, R14 + MOVB R13, (R9) + ADD R1, R9, R15 + MOVB R14, -1(R15) + ADD R1, R10, R10 + ADD R1, R9, R9 + JMP copy_1_end copy_1_move_3: - MOVH (R10), R13 - MOVB 2(R10), R14 - MOVH R13, (R9) - MOVB R14, 2(R9) - ADD R0, R10, R10 - ADD R0, R9, R9 - JMP copy_1_end + MOVHU (R10), R13 + MOVBU 2(R10), R16 + BFI $0, R16, $8, R14 + MOVH R13, (R9) + MOVB R14, 2(R9) + ADD R1, R10, R10 + ADD R1, R9, R9 + JMP copy_1_end copy_1_move_4through7: MOVWU (R10), R13 - ADD R0, R10, R15 + ADD R1, R10, R15 MOVWU -4(R15), R14 MOVW R13, (R9) - ADD R0, R9, R15 + ADD R1, R9, R15 MOVW R14, -4(R15) - ADD R0, R10, R10 - ADD R0, R9, R9 + ADD R1, R10, R10 + ADD R1, R9, R9 JMP copy_1_end copy_1_move_8through16: MOVD (R10), R13 - ADD R0, R10, R15 + ADD R1, R10, R15 MOVD -8(R15), R14 MOVD R13, (R9) - ADD R0, R9, R15 + ADD R1, R9, R15 MOVD R14, -8(R15) - ADD R0, R10, R10 - ADD R0, R9, R9 + ADD R1, R10, R10 + ADD R1, R9, R9 copy_1_end: - ADD R0, R11, R11 + ADD R1, R11, R11 // Malformed input if seq.mo > t+len(hist) || seq.mo > s.windowSize) check_offset: - MOVD R11, R0 - MOVD 40(RSP), R16 - ADD R16, R0, R0 - CMP R0, R1 + MOVD R11, R1 + MOVD 80(RSP), R16 + ADD R16, R1, R1 + CMP R1, R12 BGT error_match_off_too_big - MOVD 56(RSP), R16 - CMP R16, R1 + MOVD 96(RSP), R16 + CMP R16, R12 BGT error_match_off_too_big // Copy match from history - MOVD R1, R0 - SUBS R11, R0, R0 + MOVD R12, R1 + SUBS R11, R1, R1 BLS copy_match - MOVD 48(RSP), R13 - SUB R0, R13, R13 - CMP R0, R12 + MOVD 88(RSP), R13 + SUB R1, R13, R13 + CMP R1, R0 BGT copy_all_from_history - MOVD R12, R0 - SUBS $0x10, R0, R0 + MOVD R0, R1 + SUBS $0x10, R1, R1 BLO copy_4_small copy_4_loop: - VLD1 (R13), [V0.B16] - VST1 [V0.B16], (R9) - ADD $0x10, R13, R13 - ADD $0x10, R9, R9 - SUBS $0x10, R0, R0 - BHS copy_4_loop - ADD R0, R13, R13 - ADD $16, R13, R13 - ADD R0, R9, R9 - ADD $16, R9, R9 - ADD $-16, R13, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R9, R15 - VST1 [V0.B16], (R15) - JMP copy_4_end + FMOVQ (R13), F0 + FMOVQ F0, (R9) + ADD $0x10, R13, R13 + ADD $0x10, R9, R9 + SUBS $0x10, R1, R1 + BHS copy_4_loop + ADD R1, R13, R13 + ADD $16, R13, R13 + ADD R1, R9, R9 + ADD $16, R9, R9 + FMOVQ -16(R13), F0 + FMOVQ F0, -16(R9) + JMP copy_4_end copy_4_small: - CMP $0x03, R12 + CMP $0x03, R0 BEQ copy_4_move_3 - CMP $0x08, R12 + CMP $0x08, R0 BLO copy_4_move_4through7 JMP copy_4_move_8through16 copy_4_move_3: - MOVH (R13), R0 - MOVB 2(R13), R1 - MOVH R0, (R9) - MOVB R1, 2(R9) - ADD R12, R13, R13 - ADD R12, R9, R9 - JMP copy_4_end + MOVHU (R13), R1 + MOVBU 2(R13), R16 + BFI $0, R16, $8, R12 + MOVH R1, (R9) + MOVB R12, 2(R9) + ADD R0, R13, R13 + ADD R0, R9, R9 + JMP copy_4_end copy_4_move_4through7: - MOVWU (R13), R0 - ADD R12, R13, R15 - MOVWU -4(R15), R1 - MOVW R0, (R9) - ADD R12, R9, R15 - MOVW R1, -4(R15) - ADD R12, R13, R13 - ADD R12, R9, R9 + MOVWU (R13), R1 + ADD R0, R13, R15 + MOVWU -4(R15), R12 + MOVW R1, (R9) + ADD R0, R9, R15 + MOVW R12, -4(R15) + ADD R0, R13, R13 + ADD R0, R9, R9 JMP copy_4_end copy_4_move_8through16: - MOVD (R13), R0 - ADD R12, R13, R15 - MOVD -8(R15), R1 - MOVD R0, (R9) - ADD R12, R9, R15 - MOVD R1, -8(R15) - ADD R12, R13, R13 - ADD R12, R9, R9 + MOVD (R13), R1 + ADD R0, R13, R15 + MOVD -8(R15), R12 + MOVD R1, (R9) + ADD R0, R9, R15 + MOVD R12, -8(R15) + ADD R0, R13, R13 + ADD R0, R9, R9 copy_4_end: - ADD R12, R11, R11 + ADD R0, R11, R11 JMP handle_loop JMP loop_finished copy_all_from_history: - MOVD R0, R14 + MOVD R1, R14 SUBS $0x10, R14, R14 BLO copy_5_small copy_5_loop: - VLD1 (R13), [V0.B16] - VST1 [V0.B16], (R9) - ADD $0x10, R13, R13 - ADD $0x10, R9, R9 - SUBS $0x10, R14, R14 - BHS copy_5_loop - ADD R14, R13, R13 - ADD $16, R13, R13 - ADD R14, R9, R9 - ADD $16, R9, R9 - ADD $-16, R13, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R9, R15 - VST1 [V0.B16], (R15) - JMP copy_5_end + FMOVQ (R13), F0 + FMOVQ F0, (R9) + ADD $0x10, R13, R13 + ADD $0x10, R9, R9 + SUBS $0x10, R14, R14 + BHS copy_5_loop + ADD R14, R13, R13 + ADD $16, R13, R13 + ADD R14, R9, R9 + ADD $16, R9, R9 + FMOVQ -16(R13), F0 + FMOVQ F0, -16(R9) + JMP copy_5_end copy_5_small: - CMP $0x03, R0 + CMP $0x03, R1 BEQ copy_5_move_3 BLO copy_5_move_1or2 - CMP $0x08, R0 + CMP $0x08, R1 BLO copy_5_move_4through7 JMP copy_5_move_8through16 copy_5_move_1or2: - MOVB (R13), R14 - ADD R0, R13, R15 - MOVB -1(R15), R4 - MOVB R14, (R9) - ADD R0, R9, R15 - MOVB R4, -1(R15) - ADD R0, R13, R13 - ADD R0, R9, R9 - JMP copy_5_end + MOVBU (R13), R16 + BFI $0, R16, $8, R14 + ADD R1, R13, R15 + MOVBU -1(R15), R16 + BFI $0, R16, $8, R4 + MOVB R14, (R9) + ADD R1, R9, R15 + MOVB R4, -1(R15) + ADD R1, R13, R13 + ADD R1, R9, R9 + JMP copy_5_end copy_5_move_3: - MOVH (R13), R14 - MOVB 2(R13), R4 - MOVH R14, (R9) - MOVB R4, 2(R9) - ADD R0, R13, R13 - ADD R0, R9, R9 - JMP copy_5_end + MOVHU (R13), R14 + MOVBU 2(R13), R16 + BFI $0, R16, $8, R4 + MOVH R14, (R9) + MOVB R4, 2(R9) + ADD R1, R13, R13 + ADD R1, R9, R9 + JMP copy_5_end copy_5_move_4through7: MOVWU (R13), R14 - ADD R0, R13, R15 + ADD R1, R13, R15 MOVWU -4(R15), R4 MOVW R14, (R9) - ADD R0, R9, R15 + ADD R1, R9, R15 MOVW R4, -4(R15) - ADD R0, R13, R13 - ADD R0, R9, R9 + ADD R1, R13, R13 + ADD R1, R9, R9 JMP copy_5_end copy_5_move_8through16: MOVD (R13), R14 - ADD R0, R13, R15 + ADD R1, R13, R15 MOVD -8(R15), R4 MOVD R14, (R9) - ADD R0, R9, R15 + ADD R1, R9, R15 MOVD R4, -8(R15) - ADD R0, R13, R13 - ADD R0, R9, R9 + ADD R1, R13, R13 + ADD R1, R9, R9 copy_5_end: - ADD R0, R11, R11 - SUB R0, R12, R12 + ADD R1, R11, R11 + SUB R1, R0, R0 // Copy match from the current buffer copy_match: - MOVD R9, R0 - SUB R1, R0, R0 + MOVD R9, R1 + SUB R12, R1, R1 // ml <= mo - CMP R1, R12 + CMP R12, R0 BHI copy_overlapping_match // Copy non-overlapping match - ADD R12, R11, R11 - MOVD R12, R1 - SUBS $0x10, R1, R1 + ADD R0, R11, R11 + MOVD R0, R12 + SUBS $0x10, R12, R12 BLO copy_2_small copy_2_loop: - VLD1 (R0), [V0.B16] - VST1 [V0.B16], (R9) - ADD $0x10, R0, R0 - ADD $0x10, R9, R9 - SUBS $0x10, R1, R1 - BHS copy_2_loop - ADD R1, R0, R0 - ADD $16, R0, R0 - ADD R1, R9, R9 - ADD $16, R9, R9 - ADD $-16, R0, R15 - VLD1 (R15), [V0.B16] - ADD $-16, R9, R15 - VST1 [V0.B16], (R15) - JMP copy_2_end + FMOVQ (R1), F0 + FMOVQ F0, (R9) + ADD $0x10, R1, R1 + ADD $0x10, R9, R9 + SUBS $0x10, R12, R12 + BHS copy_2_loop + ADD R12, R1, R1 + ADD $16, R1, R1 + ADD R12, R9, R9 + ADD $16, R9, R9 + FMOVQ -16(R1), F0 + FMOVQ F0, -16(R9) + JMP copy_2_end copy_2_small: - CMP $0x03, R12 + CMP $0x03, R0 BEQ copy_2_move_3 BLO copy_2_move_1or2 - CMP $0x08, R12 + CMP $0x08, R0 BLO copy_2_move_4through7 JMP copy_2_move_8through16 copy_2_move_1or2: - MOVB (R0), R1 - ADD R12, R0, R15 - MOVB -1(R15), R13 - MOVB R1, (R9) - ADD R12, R9, R15 - MOVB R13, -1(R15) - ADD R12, R0, R0 - ADD R12, R9, R9 - JMP copy_2_end + MOVBU (R1), R16 + BFI $0, R16, $8, R12 + ADD R0, R1, R15 + MOVBU -1(R15), R16 + BFI $0, R16, $8, R13 + MOVB R12, (R9) + ADD R0, R9, R15 + MOVB R13, -1(R15) + ADD R0, R1, R1 + ADD R0, R9, R9 + JMP copy_2_end copy_2_move_3: - MOVH (R0), R1 - MOVB 2(R0), R13 - MOVH R1, (R9) - MOVB R13, 2(R9) - ADD R12, R0, R0 - ADD R12, R9, R9 - JMP copy_2_end + MOVHU (R1), R12 + MOVBU 2(R1), R16 + BFI $0, R16, $8, R13 + MOVH R12, (R9) + MOVB R13, 2(R9) + ADD R0, R1, R1 + ADD R0, R9, R9 + JMP copy_2_end copy_2_move_4through7: - MOVWU (R0), R1 - ADD R12, R0, R15 + MOVWU (R1), R12 + ADD R0, R1, R15 MOVWU -4(R15), R13 - MOVW R1, (R9) - ADD R12, R9, R15 + MOVW R12, (R9) + ADD R0, R9, R15 MOVW R13, -4(R15) - ADD R12, R0, R0 - ADD R12, R9, R9 + ADD R0, R1, R1 + ADD R0, R9, R9 JMP copy_2_end copy_2_move_8through16: - MOVD (R0), R1 - ADD R12, R0, R15 + MOVD (R1), R12 + ADD R0, R1, R15 MOVD -8(R15), R13 - MOVD R1, (R9) - ADD R12, R9, R15 + MOVD R12, (R9) + ADD R0, R9, R15 MOVD R13, -8(R15) - ADD R12, R0, R0 - ADD R12, R9, R9 + ADD R0, R1, R1 + ADD R0, R9, R9 copy_2_end: JMP handle_loop // Copy overlapping match copy_overlapping_match: - ADD R12, R11, R11 + ADD R0, R11, R11 copy_slow_3: - MOVB (R0), R1 - MOVB R1, (R9) - ADD $1, R0, R0 - ADD $1, R9, R9 - SUBS $1, R12, R12 - BNE copy_slow_3 + MOVBU (R1), R12 + MOVB R12, (R9) + ADD $1, R1, R1 + ADD $1, R9, R9 + SUBS $1, R0, R0 + BNE copy_slow_3 handle_loop: - MOVD ctx+16(FP), R0 - MOVD 96(R0), R16 + MOVD 40(RSP), R16 SUBS $1, R16, R16 - MOVD R16, 96(R0) + MOVD R16, 40(RSP) BPL sequenceDecs_decodeSync_safe_amd64_main_loop loop_finished: @@ -2635,6 +2507,16 @@ loop_finished: MOVB R3, 40(R0) MOVD R5, 32(R0) + // s.seqSize += outPosition - ctx.outPosition + MOVD ctx+16(FP), R0 + MOVD s+0(FP), R1 + MOVD R11, R2 + MOVD 136(R0), R16 + SUB R16, R2, R2 + MOVD 256(R1), R16 + ADD R2, R16, R16 + MOVD R16, 256(R1) + // Update the context MOVD ctx+16(FP), R0 MOVD R11, 136(R0) @@ -2649,7 +2531,7 @@ loop_finished: // Return with match length error sequenceDecs_decodeSync_safe_amd64_error_match_len_ofs_mismatch: - MOVD 16(RSP), R0 + MOVD 56(RSP), R0 MOVD ctx+16(FP), R1 MOVD R0, 216(R1) MOVD $0x00000001, R16 @@ -2659,7 +2541,7 @@ sequenceDecs_decodeSync_safe_amd64_error_match_len_ofs_mismatch: // Return with match too long error sequenceDecs_decodeSync_safe_amd64_error_match_len_too_big: MOVD ctx+16(FP), R0 - MOVD 16(RSP), R1 + MOVD 56(RSP), R1 MOVD R1, 216(R0) MOVD $0x00000002, R16 MOVD R16, ret+24(FP) @@ -2668,7 +2550,7 @@ sequenceDecs_decodeSync_safe_amd64_error_match_len_too_big: // Return with match offset too long error error_match_off_too_big: MOVD ctx+16(FP), R0 - MOVD 8(RSP), R1 + MOVD 48(RSP), R1 MOVD R1, 224(R0) MOVD R11, 136(R0) MOVD $0x00000003, R16 @@ -2678,7 +2560,7 @@ error_match_off_too_big: // Return with not enough literals error error_not_enough_literals: MOVD ctx+16(FP), R0 - MOVD 24(RSP), R1 + MOVD 64(RSP), R1 MOVD R1, 208(R0) MOVD $0x00000004, R16 MOVD R16, ret+24(FP) @@ -2693,9 +2575,9 @@ error_overread: // Return with not enough output space error error_not_enough_space: MOVD ctx+16(FP), R0 - MOVD 24(RSP), R1 + MOVD 64(RSP), R1 MOVD R1, 208(R0) - MOVD 16(RSP), R1 + MOVD 56(RSP), R1 MOVD R1, 216(R0) MOVD R11, 136(R0) MOVD $0x00000005, R16 diff --git a/vendor/github.com/klauspost/compress/zstd/seqdec_asm.go b/vendor/github.com/klauspost/compress/zstd/seqdec_asm.go index 55405f3914..94559c459a 100644 --- a/vendor/github.com/klauspost/compress/zstd/seqdec_asm.go +++ b/vendor/github.com/klauspost/compress/zstd/seqdec_asm.go @@ -77,6 +77,36 @@ const errorNotEnoughSpace = 5 // error reported when bits are overread. const errorOverread = 6 +// useSafeDecodeSync reports whether decodeSyncSimple must use the bounds-exact +// ("safe") copy variants for the current buffer geometry. +// +// When the output and literal buffers have compressedBlockOverAlloc (16) +// bytes of slack past their logical use, the assembly may use extended +// memory copies that read and write in 16-byte blocks, overrunning the end +// of a literal run or match by up to 15 bytes; an empty literal run still +// copies one block, reading and writing the full 16 bytes of slack past +// its position. Otherwise it must use the bounds-exact ("safe") copies. +// This mirrors the analogous, always-dynamic selection in executeSimple below. +// +// The unsafe copies were disabled in #644 (2022) as a mitigation for a +// crash, but that crash's root cause — an unguarded bitReader overread that +// produced out-of-range match offsets/lengths — was fixed three days later +// in #645, which also added the fuzz corpus that has guarded this path since. +// See #1168. An asan-instrumented fuzz job (see .github/workflows/go.yml) +// covers the extended-copy path, which -race and plain fuzzing cannot. +func (s *sequenceDecs) useSafeDecodeSync() bool { + if s.maxSyncLen == 0 && cap(s.out)-len(s.out) < maxCompressedBlockSizeAlloc { + return true + } + if s.maxSyncLen > 0 && cap(s.out)-len(s.out)-compressedBlockOverAlloc < int(s.maxSyncLen) { + return true + } + if cap(s.literals) < len(s.literals)+compressedBlockOverAlloc { + return true + } + return false +} + // decode sequences from the stream with the provided history but without a dictionary. func (s *sequenceDecs) decodeSyncSimple(hist []byte) (bool, error) { if len(s.dict) > 0 { @@ -86,9 +116,7 @@ func (s *sequenceDecs) decodeSyncSimple(hist []byte) (bool, error) { return false, nil } - // FIXME: Using unsafe memory copies leads to rare, random crashes - // with fuzz testing. It is therefore disabled for now. - const useSafe = true + useSafe := s.useSafeDecodeSync() br := s.br diff --git a/vendor/github.com/prometheus/client_golang/internal/github.com/golang/gddo/httputil/header/header.go b/vendor/github.com/prometheus/client_golang/internal/github.com/golang/gddo/httputil/header/header.go index 820bf436ab..2cd16f13a4 100644 --- a/vendor/github.com/prometheus/client_golang/internal/github.com/golang/gddo/httputil/header/header.go +++ b/vendor/github.com/prometheus/client_golang/internal/github.com/golang/gddo/httputil/header/header.go @@ -39,7 +39,7 @@ func init() { // token = 1* // qdtext = > - for c := 0; c < 256; c++ { + for c := range 256 { var t octetType isCtl := c <= 31 || c == 127 isChar := 0 <= c && c <= 127 diff --git a/vendor/github.com/prometheus/client_golang/prometheus/build_info_collector.go b/vendor/github.com/prometheus/client_golang/prometheus/build_info_collector.go index 450189f35e..b80b964c9c 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/build_info_collector.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/build_info_collector.go @@ -32,7 +32,8 @@ func NewBuildInfoCollector() Collector { "Build information about the main Go module.", nil, Labels{"path": path, "version": version, "checksum": sum}, ), - GaugeValue, 1)} + GaugeValue, 1, + )} c.init(c.self) return c } diff --git a/vendor/github.com/prometheus/client_golang/prometheus/collectors/go_collector_latest.go b/vendor/github.com/prometheus/client_golang/prometheus/collectors/go_collector_latest.go index d29ade654f..c5a5d50aa0 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/collectors/go_collector_latest.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/collectors/go_collector_latest.go @@ -11,9 +11,6 @@ // See the License for the specific language governing permissions and // limitations under the License. -//go:build go1.17 -// +build go1.17 - package collectors import ( diff --git a/vendor/github.com/prometheus/client_golang/prometheus/desc.go b/vendor/github.com/prometheus/client_golang/prometheus/desc.go index a3c92e7a4c..881d91dee0 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/desc.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/desc.go @@ -21,7 +21,6 @@ import ( "github.com/cespare/xxhash/v2" dto "github.com/prometheus/client_model/go" "github.com/prometheus/common/model" - "google.golang.org/protobuf/proto" "github.com/prometheus/client_golang/prometheus/internal" ) @@ -181,8 +180,8 @@ func (v2) NewDesc(fqName, help string, variableLabels ConstrainableLabels, const d.constLabelPairs = make([]*dto.LabelPair, 0, len(constLabels)) for n, v := range constLabels { d.constLabelPairs = append(d.constLabelPairs, &dto.LabelPair{ - Name: proto.String(n), - Value: proto.String(v), + Name: new(n), + Value: new(v), }) } sort.Sort(internal.LabelPairSorter(d.constLabelPairs)) diff --git a/vendor/github.com/prometheus/client_golang/prometheus/get_pid.go b/vendor/github.com/prometheus/client_golang/prometheus/get_pid.go index 614fd61be9..c7be2a0ca7 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/get_pid.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/get_pid.go @@ -12,7 +12,6 @@ // limitations under the License. //go:build !js || wasm -// +build !js wasm package prometheus diff --git a/vendor/github.com/prometheus/client_golang/prometheus/get_pid_gopherjs.go b/vendor/github.com/prometheus/client_golang/prometheus/get_pid_gopherjs.go index eaf8059ee1..0ddc8c6eed 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/get_pid_gopherjs.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/get_pid_gopherjs.go @@ -12,7 +12,6 @@ // limitations under the License. //go:build js && !wasm -// +build js,!wasm package prometheus diff --git a/vendor/github.com/prometheus/client_golang/prometheus/go_collector.go b/vendor/github.com/prometheus/client_golang/prometheus/go_collector.go index 520cbd7d41..a4725849f4 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/go_collector.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/go_collector.go @@ -211,23 +211,28 @@ func newBaseGoCollector() baseGoCollector { goroutinesDesc: NewDesc( "go_goroutines", "Number of goroutines that currently exist.", - nil, nil), + nil, nil, + ), threadsDesc: NewDesc( "go_threads", "Number of OS threads created.", - nil, nil), + nil, nil, + ), gcDesc: NewDesc( "go_gc_duration_seconds", "A summary of the wall-time pause (stop-the-world) duration in garbage collection cycles.", - nil, nil), + nil, nil, + ), gcLastTimeDesc: NewDesc( "go_memstats_last_gc_time_seconds", "Number of seconds since 1970 of last garbage collection.", - nil, nil), + nil, nil, + ), goInfoDesc: NewDesc( "go_info", "Information about the Go environment.", - nil, Labels{"version": runtime.Version()}), + nil, Labels{"version": runtime.Version()}, + ), } } diff --git a/vendor/github.com/prometheus/client_golang/prometheus/go_collector_latest.go b/vendor/github.com/prometheus/client_golang/prometheus/go_collector_latest.go index 1db1c4be09..23091e1774 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/go_collector_latest.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/go_collector_latest.go @@ -11,9 +11,6 @@ // See the License for the specific language governing permissions and // limitations under the License. -//go:build go1.17 -// +build go1.17 - package prometheus import ( @@ -27,7 +24,6 @@ import ( "github.com/prometheus/client_golang/prometheus/internal" dto "github.com/prometheus/client_model/go" - "google.golang.org/protobuf/proto" ) const ( @@ -573,14 +569,14 @@ func (h *batchHistogram) Write(out *dto.Metric) error { // by obtaining the next float64 value down, in order. upperBound := math.Nextafter(h.buckets[i+1], h.buckets[i]) dtoBuckets = append(dtoBuckets, &dto.Bucket{ - CumulativeCount: proto.Uint64(totalCount), - UpperBound: proto.Float64(upperBound), + CumulativeCount: new(totalCount), + UpperBound: new(upperBound), }) } out.Histogram = &dto.Histogram{ Bucket: dtoBuckets, - SampleCount: proto.Uint64(totalCount), - SampleSum: proto.Float64(sum), + SampleCount: new(totalCount), + SampleSum: new(sum), } return nil } diff --git a/vendor/github.com/prometheus/client_golang/prometheus/histogram.go b/vendor/github.com/prometheus/client_golang/prometheus/histogram.go index 88bae3b32c..1d6ca1614f 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/histogram.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/histogram.go @@ -803,8 +803,8 @@ func (h *histogram) Write(out *dto.Metric) error { his := &dto.Histogram{ Bucket: make([]*dto.Bucket, len(h.upperBounds)), - SampleCount: proto.Uint64(count), - SampleSum: proto.Float64(math.Float64frombits(atomic.LoadUint64(&coldCounts.sumBits))), + SampleCount: new(count), + SampleSum: new(math.Float64frombits(atomic.LoadUint64(&coldCounts.sumBits))), CreatedTimestamp: timestamppb.New(h.lastResetTime), } out.Histogram = his @@ -814,8 +814,8 @@ func (h *histogram) Write(out *dto.Metric) error { for i, upperBound := range h.upperBounds { cumCount += atomic.LoadUint64(&coldCounts.buckets[i]) his.Bucket[i] = &dto.Bucket{ - CumulativeCount: proto.Uint64(cumCount), - UpperBound: proto.Float64(upperBound), + CumulativeCount: new(cumCount), + UpperBound: new(upperBound), } if e := h.exemplars[i].Load(); e != nil { his.Bucket[i].Exemplar = e.(*dto.Exemplar) @@ -824,15 +824,15 @@ func (h *histogram) Write(out *dto.Metric) error { // If there is an exemplar for the +Inf bucket, we have to add that bucket explicitly. if e := h.exemplars[len(h.upperBounds)].Load(); e != nil { b := &dto.Bucket{ - CumulativeCount: proto.Uint64(count), - UpperBound: proto.Float64(math.Inf(1)), + CumulativeCount: new(count), + UpperBound: new(math.Inf(1)), Exemplar: e.(*dto.Exemplar), } his.Bucket = append(his.Bucket, b) } if h.nativeHistogramSchema > math.MinInt32 { - his.ZeroThreshold = proto.Float64(math.Float64frombits(atomic.LoadUint64(&coldCounts.nativeHistogramZeroThresholdBits))) - his.Schema = proto.Int32(atomic.LoadInt32(&coldCounts.nativeHistogramSchema)) + his.ZeroThreshold = new(math.Float64frombits(atomic.LoadUint64(&coldCounts.nativeHistogramZeroThresholdBits))) + his.Schema = new(atomic.LoadInt32(&coldCounts.nativeHistogramSchema)) zeroBucket := atomic.LoadUint64(&coldCounts.nativeHistogramZeroBucket) defer func() { @@ -840,7 +840,7 @@ func (h *histogram) Write(out *dto.Metric) error { coldCounts.nativeHistogramBucketsNegative.Range(addAndReset(&hotCounts.nativeHistogramBucketsNegative, &hotCounts.nativeHistogramBucketsNumber)) }() - his.ZeroCount = proto.Uint64(zeroBucket) + his.ZeroCount = new(zeroBucket) his.NegativeSpan, his.NegativeDelta = makeBuckets(&coldCounts.nativeHistogramBucketsNegative) his.PositiveSpan, his.PositiveDelta = makeBuckets(&coldCounts.nativeHistogramBucketsPositive) @@ -1331,12 +1331,12 @@ func (h *constHistogram) Write(out *dto.Metric) error { buckets := make([]*dto.Bucket, 0, len(h.buckets)) - his.SampleCount = proto.Uint64(h.count) - his.SampleSum = proto.Float64(h.sum) + his.SampleCount = new(h.count) + his.SampleSum = new(h.sum) for upperBound, count := range h.buckets { buckets = append(buckets, &dto.Bucket{ - CumulativeCount: proto.Uint64(count), - UpperBound: proto.Float64(upperBound), + CumulativeCount: new(count), + UpperBound: new(upperBound), }) } @@ -1481,7 +1481,7 @@ func pickSchema(bucketFactor float64) int32 { func makeBuckets(buckets *sync.Map) ([]*dto.BucketSpan, []int64) { var ii []int - buckets.Range(func(k, v any) bool { + buckets.Range(func(k, _ any) bool { ii = append(ii, k.(int)) return true }) @@ -1518,13 +1518,13 @@ func makeBuckets(buckets *sync.Map) ([]*dto.BucketSpan, []int64) { // at the very beginning, or because we have found a gap // of more than two buckets. spans = append(spans, &dto.BucketSpan{ - Offset: proto.Int32(iDelta), + Offset: new(iDelta), Length: proto.Uint32(0), }) } else { // We have found a small gap (or no gap at all). // Insert empty buckets as needed. - for j := int32(0); j < iDelta; j++ { + for range iDelta { appendDelta(0) } } @@ -1570,7 +1570,7 @@ func addAndReset(hotBuckets *sync.Map, bucketNumber *uint32) func(k, v any) bool } func deleteSyncMap(m *sync.Map) { - m.Range(func(k, v any) bool { + m.Range(func(k, _ any) bool { m.Delete(k) return true }) @@ -1578,7 +1578,7 @@ func deleteSyncMap(m *sync.Map) { func findSmallestKey(m *sync.Map) int { result := math.MaxInt32 - m.Range(func(k, v any) bool { + m.Range(func(k, _ any) bool { key := k.(int) if key < result { result = key @@ -1956,7 +1956,7 @@ func NewConstNativeHistogram( PositiveSpan: PositiveSpan, PositiveDelta: PositiveDelta, - ZeroCount: proto.Uint64(zeroBucket), + ZeroCount: new(zeroBucket), }, labelPairs: MakeLabelPairs(desc, labelValues), } @@ -2044,13 +2044,13 @@ func makeBucketsFromMap(buckets map[int]int64) ([]*dto.BucketSpan, []int64) { // at the very beginning, or because we have found a gap // of more than two buckets. spans = append(spans, &dto.BucketSpan{ - Offset: proto.Int32(iDelta), + Offset: new(iDelta), Length: proto.Uint32(0), }) } else { // We have found a small gap (or no gap at all). // Insert empty buckets as needed. - for j := int32(0); j < iDelta; j++ { + for range iDelta { appendDelta(0) } } diff --git a/vendor/github.com/prometheus/client_golang/prometheus/internal/go_runtime_metrics.go b/vendor/github.com/prometheus/client_golang/prometheus/internal/go_runtime_metrics.go index d273b6640e..e38dfe0adc 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/internal/go_runtime_metrics.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/internal/go_runtime_metrics.go @@ -11,9 +11,6 @@ // See the License for the specific language governing permissions and // limitations under the License. -//go:build go1.17 -// +build go1.17 - package internal import ( diff --git a/vendor/github.com/prometheus/client_golang/prometheus/metric.go b/vendor/github.com/prometheus/client_golang/prometheus/metric.go index c5cb90adf8..e4bbf3b8ec 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/metric.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/metric.go @@ -22,7 +22,6 @@ import ( dto "github.com/prometheus/client_model/go" "github.com/prometheus/common/model" - "google.golang.org/protobuf/proto" ) var separatorByteSlice = []byte{model.SeparatorByte} // For convenient use with xxhash. @@ -153,7 +152,7 @@ type timestampedMetric struct { func (m timestampedMetric) Write(pb *dto.Metric) error { e := m.Metric.Write(pb) - pb.TimestampMs = proto.Int64(m.t.Unix()*1000 + int64(m.t.Nanosecond()/1000000)) + pb.TimestampMs = new(m.t.Unix()*1000 + int64(m.t.Nanosecond()/1000000)) return e } @@ -209,8 +208,8 @@ func (m *withExemplarsMetric) Write(pb *dto.Metric) error { } else { // The +Inf bucket should be explicitly added if there is an exemplar for it, similar to non-const histogram logic in https://github.com/prometheus/client_golang/blob/main/prometheus/histogram.go#L357-L365. b := &dto.Bucket{ - CumulativeCount: proto.Uint64(h.GetSampleCount()), - UpperBound: proto.Float64(math.Inf(1)), + CumulativeCount: new(h.GetSampleCount()), + UpperBound: new(math.Inf(1)), Exemplar: e, } h.Bucket = append(h.Bucket, b) diff --git a/vendor/github.com/prometheus/client_golang/prometheus/num_threads.go b/vendor/github.com/prometheus/client_golang/prometheus/num_threads.go index 7c12b21087..7ffac21340 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/num_threads.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/num_threads.go @@ -12,7 +12,6 @@ // limitations under the License. //go:build !js || wasm -// +build !js wasm package prometheus diff --git a/vendor/github.com/prometheus/client_golang/prometheus/num_threads_gopherjs.go b/vendor/github.com/prometheus/client_golang/prometheus/num_threads_gopherjs.go index 7348df01df..65f4263f9f 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/num_threads_gopherjs.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/num_threads_gopherjs.go @@ -12,7 +12,6 @@ // limitations under the License. //go:build js && !wasm -// +build js,!wasm package prometheus diff --git a/vendor/github.com/prometheus/client_golang/prometheus/process_collector_darwin.go b/vendor/github.com/prometheus/client_golang/prometheus/process_collector_darwin.go index 2b16298f40..46e7e45aab 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/process_collector_darwin.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/process_collector_darwin.go @@ -54,21 +54,22 @@ func getOpenFileCount() (float64, error) { // benefits, however, include fewer system calls and not failing when at the // open file soft limit. - if dir, err := os.Open("/dev/fd"); err != nil { + dir, err := os.Open("/dev/fd") + if err != nil { return 0.0, err - } else { - defer dir.Close() + } + defer dir.Close() - // Avoid ReadDir(), as it calls stat(2) on each descriptor. Not only is - // that info not used, but KQUEUE descriptors fail stat(2), which causes - // the whole method to fail. - if names, err := dir.Readdirnames(0); err != nil { - return 0.0, err - } else { - // Subtract 1 to ignore the open /dev/fd descriptor above. - return float64(len(names) - 1), nil - } + // Avoid ReadDir(), as it calls stat(2) on each descriptor. Not only is + // that info not used, but KQUEUE descriptors fail stat(2), which causes + // the whole method to fail. + names, err := dir.Readdirnames(0) + if err != nil { + return 0.0, err } + + // Subtract 1 to ignore the open /dev/fd descriptor above. + return float64(len(names) - 1), nil } func (c *processCollector) processCollect(ch chan<- Metric) { diff --git a/vendor/github.com/prometheus/client_golang/prometheus/process_collector_not_supported.go b/vendor/github.com/prometheus/client_golang/prometheus/process_collector_not_supported.go index 7732b7f376..8650b6f4a7 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/process_collector_not_supported.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/process_collector_not_supported.go @@ -12,7 +12,6 @@ // limitations under the License. //go:build wasip1 || js || ios -// +build wasip1 js ios package prometheus diff --git a/vendor/github.com/prometheus/client_golang/prometheus/process_collector_procfsenabled.go b/vendor/github.com/prometheus/client_golang/prometheus/process_collector_procfsenabled.go index 8074f70f5d..56c97d62ef 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/process_collector_procfsenabled.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/process_collector_procfsenabled.go @@ -12,7 +12,6 @@ // limitations under the License. //go:build !windows && !js && !wasip1 && !darwin -// +build !windows,!js,!wasip1,!darwin package prometheus diff --git a/vendor/github.com/prometheus/client_golang/prometheus/promhttp/http.go b/vendor/github.com/prometheus/client_golang/prometheus/promhttp/http.go index c28af5ce2b..0af3deebfe 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/promhttp/http.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/promhttp/http.go @@ -51,10 +51,12 @@ import ( ) const ( + acceptHeader = "Accept" contentTypeHeader = "Content-Type" contentEncodingHeader = "Content-Encoding" acceptEncodingHeader = "Accept-Encoding" processStartTimeHeader = "Process-Start-Time-Unix" + varyHeader = "Vary" ) // Compression represents the content encodings handlers support for the HTTP @@ -70,9 +72,8 @@ const ( func defaultCompressionFormats() []Compression { if internal.NewZstdWriter != nil { return []Compression{Identity, Gzip, Zstd} - } else { - return []Compression{Identity, Gzip} } + return []Compression{Identity, Gzip} } var gzipPool = sync.Pool{ @@ -315,11 +316,20 @@ func HandlerForTransactional(reg prometheus.TransactionalGatherer, opts HandlerO } } - var contentType expfmt.Format - if opts.EnableOpenMetrics { - contentType = expfmt.NegotiateIncludingOpenMetrics(req.Header) - } else { - contentType = expfmt.Negotiate(req.Header) + var ( + contentType expfmt.Format + acceptedFormats = opts.AcceptedFormats + ) + if len(acceptedFormats) == 0 { + acceptedFormats = expfmt.DefaultAcceptedFormats() + if opts.EnableOpenMetrics { + acceptedFormats = expfmt.DefaultOpenMetricsAcceptedFormats() + } + } + contentType = expfmt.NegotiateAccept(req.Header, acceptedFormats...) + rsp.Header().Add(varyHeader, acceptHeader) + if len(compressions) > 0 { + rsp.Header().Add(varyHeader, acceptEncodingHeader) } rsp.Header().Set(contentTypeHeader, string(contentType)) @@ -598,6 +608,8 @@ type HandlerOpts struct { // a trailing ".0" if they would otherwise look like integer numbers // (which changes the identity of the resulting series on the Prometheus // server). + // + // Note that setting AcceptedFormats overrides this option. EnableOpenMetrics bool // EnableOpenMetricsTextCreatedSamples specifies if this handler should add, extra, synthetic // Created Timestamps for counters, histograms and summaries, which for the current @@ -622,6 +634,18 @@ type HandlerOpts struct { // NOTE: This feature is experimental and not covered by OpenMetrics or Prometheus // exposition format. ProcessStartTime time.Time + // AcceptedFormats allows customizing the formats the handler will negotiate + // with the client based on the request's Accept header. + // The client's Accept header and q-values take precedence, but the order of + // formats in AcceptedFormats specifies the server's preference order used as a + // tie-breaker (e.g. for wildcard "*/*" requests or equal q-values) and fallback. + // If empty, the default set is used (expfmt.DefaultAcceptedFormats()). + // If empty and EnableOpenMetrics is true, stable OpenMetrics formats are negotiated + // (expfmt.DefaultOpenMetricsAcceptedFormats()). + // This is useful for testing experimental formats (e.g. OpenMetrics 2.0) + // or restricting the formats you want to allow. + // Using EnableOpenMetrics with an empty AcceptedFormats is still the recommended option for the majority of cases. + AcceptedFormats []expfmt.Format } // httpError removes any content-encoding header and then calls http.Error with diff --git a/vendor/github.com/prometheus/client_golang/prometheus/promhttp/option.go b/vendor/github.com/prometheus/client_golang/prometheus/promhttp/option.go index d4c0954f36..258a2a5b55 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/promhttp/option.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/promhttp/option.go @@ -41,7 +41,7 @@ type options struct { func defaultOptions() *options { return &options{ - getExemplarFn: func(req *http.Request) prometheus.Labels { return nil }, + getExemplarFn: func(_ *http.Request) prometheus.Labels { return nil }, extraLabelsFromRequest: map[string]LabelValueFromRequest{}, } } diff --git a/vendor/github.com/prometheus/client_golang/prometheus/registry.go b/vendor/github.com/prometheus/client_golang/prometheus/registry.go index ed0681c8b4..4151b59c48 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/registry.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/registry.go @@ -17,6 +17,7 @@ import ( "bytes" "errors" "fmt" + "maps" "os" "path/filepath" "runtime" @@ -31,7 +32,6 @@ import ( "github.com/cespare/xxhash/v2" dto "github.com/prometheus/client_model/go" "github.com/prometheus/common/expfmt" - "google.golang.org/protobuf/proto" ) const ( @@ -296,6 +296,7 @@ func (r *Registry) Register(c Collector) error { defer func() { // Drain channel in case of premature return to not leak a goroutine. for range descChan { + continue } r.mtx.Unlock() }() @@ -369,9 +370,7 @@ func (r *Registry) Register(c Collector) error { for hash := range newDescIDs { r.descIDs[hash] = struct{}{} } - for name, dimHash := range newDimHashesByName { - r.dimHashesByName[name] = dimHash - } + maps.Copy(r.dimHashesByName, newDimHashesByName) return nil } @@ -476,7 +475,7 @@ func (r *Registry) Gather() ([]*dto.MetricFamily, error) { for { select { case collector := <-checkedCollectors: - safeErrs.Append((safeCollect(collector, checkedMetricChan))) + safeErrs.Append(safeCollect(collector, checkedMetricChan)) case collector := <-uncheckedCollectors: safeErrs.Append(safeCollect(collector, uncheckedMetricChan)) default: @@ -502,10 +501,12 @@ func (r *Registry) Gather() ([]*dto.MetricFamily, error) { defer func() { if checkedMetricChan != nil { for range checkedMetricChan { + continue } } if uncheckedMetricChan != nil { for range uncheckedMetricChan { + continue } } }() @@ -727,10 +728,10 @@ func processMetric( } } else { // New name. metricFamily = &dto.MetricFamily{} - metricFamily.Name = proto.String(desc.fqName) - metricFamily.Help = proto.String(desc.help) + metricFamily.Name = new(desc.fqName) + metricFamily.Help = new(desc.help) if desc.unit != "" { - metricFamily.Unit = proto.String(desc.unit) + metricFamily.Unit = new(desc.unit) } // TODO(beorn7): Simplify switch once Desc has type. switch { @@ -956,7 +957,8 @@ func checkMetricConsistency( if !utf8.ValidString(labelPair.GetValue()) { return fmt.Errorf( "collected metric %q { %s} has a label named %q whose value is not utf8: %#v", - name, dtoMetric, labelName, labelPair.GetValue()) + name, dtoMetric, labelName, labelPair.GetValue(), + ) } previousLabelName = labelName } @@ -981,7 +983,7 @@ func checkMetricConsistency( h.Write(separatorByteSlice) } if dtoMetric.TimestampMs != nil { - h.WriteString(strconv.FormatInt(*(dtoMetric.TimestampMs), 10)) + h.WriteString(strconv.FormatInt(*dtoMetric.TimestampMs, 10)) h.Write(separatorByteSlice) } hSum := h.Sum64() @@ -1013,7 +1015,7 @@ func checkDescConsistency( copy(lpsFromDesc, desc.constLabelPairs) for _, l := range desc.variableLabels.names { lpsFromDesc = append(lpsFromDesc, &dto.LabelPair{ - Name: proto.String(l), + Name: new(l), }) } if len(lpsFromDesc) != len(dtoMetric.Label) { diff --git a/vendor/github.com/prometheus/client_golang/prometheus/summary.go b/vendor/github.com/prometheus/client_golang/prometheus/summary.go index c12b8d13d4..d92850ca01 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/summary.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/summary.go @@ -25,7 +25,6 @@ import ( dto "github.com/prometheus/client_model/go" "github.com/beorn7/perks/quantile" - "google.golang.org/protobuf/proto" "google.golang.org/protobuf/types/known/timestamppb" ) @@ -337,8 +336,8 @@ func (s *summary) Write(out *dto.Metric) error { s.bufMtx.Unlock() s.flushColdBuf() - sum.SampleCount = proto.Uint64(s.cnt) - sum.SampleSum = proto.Float64(s.sum) + sum.SampleCount = new(s.cnt) + sum.SampleSum = new(s.sum) for _, rank := range s.sortedObjectives { var q float64 @@ -348,8 +347,8 @@ func (s *summary) Write(out *dto.Metric) error { q = s.headStream.Query(rank) } qs = append(qs, &dto.Quantile{ - Quantile: proto.Float64(rank), - Value: proto.Float64(q), + Quantile: new(rank), + Value: new(q), }) } @@ -512,8 +511,8 @@ func (s *noObjectivesSummary) Write(out *dto.Metric) error { } sum := &dto.Summary{ - SampleCount: proto.Uint64(count), - SampleSum: proto.Float64(math.Float64frombits(atomic.LoadUint64(&coldCounts.sumBits))), + SampleCount: new(count), + SampleSum: new(math.Float64frombits(atomic.LoadUint64(&coldCounts.sumBits))), CreatedTimestamp: s.createdTs, } @@ -718,13 +717,13 @@ func (s *constSummary) Write(out *dto.Metric) error { } qs := make([]*dto.Quantile, 0, len(s.quantiles)) - sum.SampleCount = proto.Uint64(s.count) - sum.SampleSum = proto.Float64(s.sum) + sum.SampleCount = new(s.count) + sum.SampleSum = new(s.sum) for rank, q := range s.quantiles { qs = append(qs, &dto.Quantile{ - Quantile: proto.Float64(rank), - Value: proto.Float64(q), + Quantile: new(rank), + Value: new(q), }) } diff --git a/vendor/github.com/prometheus/client_golang/prometheus/value.go b/vendor/github.com/prometheus/client_golang/prometheus/value.go index cc23011fad..c856e4f336 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/value.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/value.go @@ -23,7 +23,6 @@ import ( "github.com/prometheus/client_golang/prometheus/internal" dto "github.com/prometheus/client_model/go" - "google.golang.org/protobuf/proto" "google.golang.org/protobuf/types/known/timestamppb" ) @@ -196,11 +195,11 @@ func populateMetric( m.Label = labelPairs switch t { case CounterValue: - m.Counter = &dto.Counter{Value: proto.Float64(v), Exemplar: e, CreatedTimestamp: ct} + m.Counter = &dto.Counter{Value: new(v), Exemplar: e, CreatedTimestamp: ct} case GaugeValue: - m.Gauge = &dto.Gauge{Value: proto.Float64(v)} + m.Gauge = &dto.Gauge{Value: new(v)} case UntypedValue: - m.Untyped = &dto.Untyped{Value: proto.Float64(v)} + m.Untyped = &dto.Untyped{Value: new(v)} default: return fmt.Errorf("encountered unknown type %v", t) } @@ -227,8 +226,8 @@ func MakeLabelPairs(desc *Desc, labelValues []string) []*dto.LabelPair { labelPairs := make([]*dto.LabelPair, 0, totalLen) for i, l := range desc.variableLabels.names { labelPairs = append(labelPairs, &dto.LabelPair{ - Name: proto.String(l), - Value: proto.String(labelValues[i]), + Name: new(l), + Value: new(labelValues[i]), }) } labelPairs = append(labelPairs, desc.constLabelPairs...) @@ -244,7 +243,7 @@ const ExemplarMaxRunes = 128 // number of runes in the label names and values exceeds ExemplarMaxRunes. func newExemplar(value float64, ts time.Time, l Labels) (*dto.Exemplar, error) { e := &dto.Exemplar{} - e.Value = proto.Float64(value) + e.Value = new(value) tsProto := timestamppb.New(ts) if err := tsProto.CheckValid(); err != nil { return nil, err @@ -262,8 +261,8 @@ func newExemplar(value float64, ts time.Time, l Labels) (*dto.Exemplar, error) { } runes += utf8.RuneCountInString(value) labelPairs = append(labelPairs, &dto.LabelPair{ - Name: proto.String(name), - Value: proto.String(value), + Name: new(name), + Value: new(value), }) } if runes > ExemplarMaxRunes { diff --git a/vendor/github.com/prometheus/client_golang/prometheus/wrap.go b/vendor/github.com/prometheus/client_golang/prometheus/wrap.go index 697f55558b..cc9ae8f20b 100644 --- a/vendor/github.com/prometheus/client_golang/prometheus/wrap.go +++ b/vendor/github.com/prometheus/client_golang/prometheus/wrap.go @@ -20,7 +20,6 @@ import ( "github.com/prometheus/client_golang/prometheus/internal" dto "github.com/prometheus/client_model/go" - "google.golang.org/protobuf/proto" ) // WrapRegistererWith returns a Registerer wrapping the provided @@ -212,8 +211,8 @@ func (m *wrappingMetric) Write(out *dto.Metric) error { } for ln, lv := range m.labels { out.Label = append(out.Label, &dto.LabelPair{ - Name: proto.String(ln), - Value: proto.String(lv), + Name: new(ln), + Value: new(lv), }) } sort.Sort(internal.LabelPairSorter(out.Label)) diff --git a/vendor/github.com/prometheus/common/expfmt/decode.go b/vendor/github.com/prometheus/common/expfmt/decode.go index 8f8dc65d38..0339d8c1d0 100644 --- a/vendor/github.com/prometheus/common/expfmt/decode.go +++ b/vendor/github.com/prometheus/common/expfmt/decode.go @@ -72,13 +72,15 @@ func ResponseFormat(h http.Header) Format { // NewDecoder returns a new decoder based on the given input format. Metric // names are validated based on the provided Format -- if the format requires -// escaping, raditional Prometheues validity checking is used. Otherwise, names +// escaping, traditional Prometheus validity checking is used. Otherwise, names // are checked for UTF-8 validity. Supported formats include delimited protobuf -// and Prometheus text format. For historical reasons, this decoder fallbacks -// to classic text decoding for any other format. This decoder does not fully -// support OpenMetrics although it may often succeed due to the similarities -// between the formats. This decoder may not support the latest features of -// Prometheus text format and is not intended for high-performance applications. +// and Prometheus text format. For historical reasons, this decoder falls back +// to classic text decoding for other legacy formats, but returns an error for +// unsupported formats such as OpenMetrics 2.0. This decoder does not fully +// support OpenMetrics although it may often succeed for OpenMetrics 1.0 due to +// the similarities between the formats. This decoder may not support the latest +// features of Prometheus text format and is not intended for high-performance +// applications. // See: https://github.com/prometheus/common/issues/812 func NewDecoder(r io.Reader, format Format) Decoder { scheme := model.LegacyValidation @@ -90,6 +92,11 @@ func NewDecoder(r io.Reader, format Format) Decoder { return &protoDecoder{r: bufio.NewReader(r), s: scheme} case TypeProtoText, TypeProtoCompact: return &errDecoder{err: fmt.Errorf("format %s not supported for decoding", format)} + case TypeOpenMetrics: + _, params, err := mime.ParseMediaType(string(format)) + if err == nil && params["version"] == OpenMetricsVersion_2_0_0 { + return &errDecoder{err: fmt.Errorf("format %s not supported for decoding", format)} + } } return &textDecoder{r: r, s: scheme} } diff --git a/vendor/github.com/prometheus/common/expfmt/encode.go b/vendor/github.com/prometheus/common/expfmt/encode.go index 73c24dfbc9..651e69a2c8 100644 --- a/vendor/github.com/prometheus/common/expfmt/encode.go +++ b/vendor/github.com/prometheus/common/expfmt/encode.go @@ -16,7 +16,9 @@ package expfmt import ( "fmt" "io" + "mime" "net/http" + "slices" "github.com/munnerz/goautoneg" dto "github.com/prometheus/client_model/go" @@ -26,6 +28,55 @@ import ( "github.com/prometheus/common/model" ) +var formatToAccept = map[Format]goautoneg.Accept{} + +func init() { + for _, f := range []Format{ + FmtText, + FmtProtoDelim, + FmtProtoText, + FmtProtoCompact, + FmtOpenMetrics_1_0_0, + FmtOpenMetrics_2_0_0, + FmtOpenMetrics_0_0_1, + } { + if parsed := goautoneg.ParseAccept(string(f)); len(parsed) > 0 { + formatToAccept[f] = parsed[0] + } + } +} + +var ( + defaultAcceptedFormats = []Format{ + FmtProtoDelim, + FmtProtoText, + FmtProtoCompact, + FmtText, + } + + defaultOpenMetricsAcceptedFormats = []Format{ + FmtOpenMetrics_1_0_0, + FmtOpenMetrics_0_0_1, + FmtProtoDelim, + FmtProtoText, + FmtProtoCompact, + FmtText, + } +) + +// DefaultAcceptedFormats returns a copy of standard accepted formats for Negotiate, +// ordered by preference (delimited protobuf, protobuf text, compact protobuf text, +// and Prometheus text format). +func DefaultAcceptedFormats() []Format { + return slices.Clone(defaultAcceptedFormats) +} + +// DefaultOpenMetricsAcceptedFormats returns a copy of standard accepted formats including +// stable OpenMetrics formats, ordered by preference. +func DefaultOpenMetricsAcceptedFormats() []Format { + return slices.Clone(defaultOpenMetricsAcceptedFormats) +} + // Encoder types encode metric families into an underlying wire protocol. type Encoder interface { Encode(*dto.MetricFamily) error @@ -57,11 +108,30 @@ func (ec encoderCloser) Close() error { // Negotiate returns the Content-Type based on the given Accept header. If no // appropriate accepted type is found, FmtText is returned (which is the -// Prometheus text format). This function will never negotiate FmtOpenMetrics, -// as the support is still experimental. To include the option to negotiate -// FmtOpenMetrics, use NegotiateIncludingOpenMetrics. +// Prometheus text format). +// +// Deprecated: Use NegotiateAccept(h, DefaultAcceptedFormats()...) +// or specify only the formats supported by your server. func Negotiate(h http.Header) Format { - escapingScheme := Format(fmt.Sprintf("; escaping=%s", Format(model.NameEscapingScheme.String()))) + return NegotiateAccept(h, defaultAcceptedFormats...) +} + +// NegotiateIncludingOpenMetrics works like Negotiate but includes all stable +// FmtOpenMetrics formats as an option for the result. +// +// Deprecated: Use NegotiateAccept(h, DefaultOpenMetricsAcceptedFormats()...) +// or specify only the formats supported by your server. +func NegotiateIncludingOpenMetrics(h http.Header) Format { + return NegotiateAccept(h, defaultOpenMetricsAcceptedFormats...) +} + +// NegotiateAccept returns the Content-Type based on the given Accept header +// and the list of accepted Formats provided by the caller, in order of preference. +// If no accepted format matches the Accept header, it falls back to the text +// format if present in the accepted list, or the first accepted format (or FmtText +// if accepted is empty). +func NegotiateAccept(h http.Header, accepted ...Format) Format { + escapingScheme := Format("; escaping=" + model.NameEscapingScheme.String()) for _, ac := range goautoneg.ParseAccept(h.Get(hdrAccept)) { if escapeParam := ac.Params[model.EscapingKey]; escapeParam != "" { switch Format(escapeParam) { @@ -71,63 +141,66 @@ func Negotiate(h http.Header) Format { // If the escaping parameter is unknown, ignore it. } } - ver := ac.Params["version"] - if ac.Type+"/"+ac.SubType == ProtoType && ac.Params["proto"] == ProtoProtocol { - switch ac.Params["encoding"] { - case "delimited": - return FmtProtoDelim + escapingScheme - case "text": - return FmtProtoText + escapingScheme - case "compact-text": - return FmtProtoCompact + escapingScheme + + for _, f := range accepted { + if matchFormat(ac, f) { + return f + escapingScheme } } - if ac.Type == "text" && ac.SubType == "plain" && (ver == TextVersion || ver == "") { - return FmtText + escapingScheme + } + for _, f := range accepted { + if f.FormatType() == TypeTextPlain { + return f + escapingScheme } } + if len(accepted) > 0 { + return accepted[0] + escapingScheme + } return FmtText + escapingScheme } -// NegotiateIncludingOpenMetrics works like Negotiate but includes -// FmtOpenMetrics as an option for the result. Note that this function is -// temporary and will disappear once FmtOpenMetrics is fully supported and as -// such may be negotiated by the normal Negotiate function. -func NegotiateIncludingOpenMetrics(h http.Header) Format { - escapingScheme := Format(fmt.Sprintf("; escaping=%s", Format(model.NameEscapingScheme.String()))) - for _, ac := range goautoneg.ParseAccept(h.Get(hdrAccept)) { - if escapeParam := ac.Params[model.EscapingKey]; escapeParam != "" { - switch Format(escapeParam) { - case model.AllowUTF8, model.EscapeUnderscores, model.EscapeDots, model.EscapeValues: - escapingScheme = Format("; escaping=" + escapeParam) - default: - // If the escaping parameter is unknown, ignore it. - } +// matchFormat checks if a parsed accept clause matches a given Format. +func matchFormat(ac goautoneg.Accept, f Format) bool { + target, ok := formatToAccept[f] + if !ok { + parsed := goautoneg.ParseAccept(string(f)) + if len(parsed) == 0 { + return false } - ver := ac.Params["version"] - if ac.Type+"/"+ac.SubType == ProtoType && ac.Params["proto"] == ProtoProtocol { - switch ac.Params["encoding"] { - case "delimited": - return FmtProtoDelim + escapingScheme - case "text": - return FmtProtoText + escapingScheme - case "compact-text": - return FmtProtoCompact + escapingScheme - } + target = parsed[0] + } + + if ac.Type != "*" && ac.Type != target.Type { + return false + } + if ac.SubType != "*" && ac.SubType != target.SubType { + return false + } + + // Default OpenMetrics version to OpenMetricsVersion_0_0_1. + acVersion := ac.Params["version"] + if acVersion == "" && ac.Type+"/"+ac.SubType == OpenMetricsType { + acVersion = OpenMetricsVersion_0_0_1 + } + if acVersion == "" && ac.Type == "text" && ac.SubType == "plain" { + acVersion = TextVersion + } + + // General param matching. + for k, v := range target.Params { + if k == "charset" { + continue } - if ac.Type == "text" && ac.SubType == "plain" && (ver == TextVersion || ver == "") { - return FmtText + escapingScheme + acVal := ac.Params[k] + if k == "version" { + acVal = acVersion } - if ac.Type+"/"+ac.SubType == OpenMetricsType && (ver == OpenMetricsVersion_0_0_1 || ver == OpenMetricsVersion_1_0_0 || ver == "") { - switch ver { - case OpenMetricsVersion_1_0_0: - return FmtOpenMetrics_1_0_0 + escapingScheme - default: - return FmtOpenMetrics_0_0_1 + escapingScheme - } + if acVal != v { + return false } } - return FmtText + escapingScheme + + return true } // NewEncoder returns a new encoder based on content type negotiation. All @@ -181,6 +254,19 @@ func NewEncoder(w io.Writer, format Format, options ...EncoderOption) Encoder { close: func() error { return nil }, } case TypeOpenMetrics: + _, params, err := mime.ParseMediaType(string(format)) + if err == nil && params["version"] == OpenMetricsVersion_2_0_0 { + return encoderCloser{ + encode: func(v *dto.MetricFamily) error { + _, err := MetricFamilyToOpenMetrics20(w, model.EscapeMetricFamily(v, escapingScheme), options...) + return err + }, + close: func() error { + _, err := FinalizeOpenMetrics(w) + return err + }, + } + } return encoderCloser{ encode: func(v *dto.MetricFamily) error { _, err := MetricFamilyToOpenMetrics(w, model.EscapeMetricFamily(v, escapingScheme), options...) diff --git a/vendor/github.com/prometheus/common/expfmt/expfmt.go b/vendor/github.com/prometheus/common/expfmt/expfmt.go index 10bf35708c..ea9a0a2a04 100644 --- a/vendor/github.com/prometheus/common/expfmt/expfmt.go +++ b/vendor/github.com/prometheus/common/expfmt/expfmt.go @@ -16,50 +16,48 @@ package expfmt import ( "errors" + "mime" "strings" "github.com/prometheus/common/model" ) -// Format specifies the HTTP content type of the different wire protocols. -type Format string - // Constants to assemble the Content-Type values for the different wire -// protocols. The Content-Type strings here are all for the legacy exposition -// formats, where valid characters for metric names and label names are limited. -// Support for arbitrary UTF-8 characters in those names is already partially -// implemented in this module (see model.ValidationScheme), but to actually use -// it on the wire, new content-type strings will have to be agreed upon and -// added here. +// protocols. const ( TextVersion = "0.0.4" ProtoType = `application/vnd.google.protobuf` ProtoProtocol = `io.prometheus.client.MetricFamily` - // Deprecated: Use expfmt.NewFormat(expfmt.TypeProtoCompact) instead. - ProtoFmt = ProtoType + "; proto=" + ProtoProtocol + ";" + ProtoFmt = ProtoType + "; proto=" + ProtoProtocol + ";" + OpenMetricsType = `application/openmetrics-text` //nolint:revive // Allow for underscores. OpenMetricsVersion_0_0_1 = "0.0.1" //nolint:revive // Allow for underscores. OpenMetricsVersion_1_0_0 = "1.0.0" + //nolint:revive // Allow for underscores. + OpenMetricsVersion_2_0_0 = "2.0.0" +) - // The Content-Type values for the different wire protocols. Do not do direct - // comparisons to these constants, instead use the comparison functions. - // - // Deprecated: Use expfmt.NewFormat(expfmt.TypeUnknown) instead. - FmtUnknown Format = `` - // Deprecated: Use expfmt.NewFormat(expfmt.TypeTextPlain) instead. - FmtText Format = `text/plain; version=` + TextVersion + `; charset=utf-8` - // Deprecated: Use expfmt.NewFormat(expfmt.TypeProtoDelim) instead. - FmtProtoDelim Format = ProtoFmt + ` encoding=delimited` - // Deprecated: Use expfmt.NewFormat(expfmt.TypeProtoText) instead. - FmtProtoText Format = ProtoFmt + ` encoding=text` - // Deprecated: Use expfmt.NewFormat(expfmt.TypeProtoCompact) instead. +// Format specifies the HTTP content type of the different wire protocols. +// The Content-Type values for the different wire protocols represent +// baseline Content-Types used for HTTP headers and content negotiation. +// +// Because Content-Types on the wire may include dynamic parameters +// (such as "; escaping=...") or whitespace variations, avoid comparing Format +// values using direct equality (==). +type Format string + +const ( + FmtUnknown Format = `` + FmtText Format = `text/plain; version=` + TextVersion + `; charset=utf-8` + FmtProtoDelim Format = ProtoFmt + ` encoding=delimited` + FmtProtoText Format = ProtoFmt + ` encoding=text` FmtProtoCompact Format = ProtoFmt + ` encoding=compact-text` - // Deprecated: Use expfmt.NewFormat(expfmt.TypeOpenMetrics) instead. + //nolint:revive // Allow for underscores. + FmtOpenMetrics_2_0_0 Format = OpenMetricsType + `; version=` + OpenMetricsVersion_2_0_0 + `; charset=utf-8` //nolint:revive // Allow for underscores. FmtOpenMetrics_1_0_0 Format = OpenMetricsType + `; version=` + OpenMetricsVersion_1_0_0 + `; charset=utf-8` - // Deprecated: Use expfmt.NewFormat(expfmt.TypeOpenMetrics) instead. //nolint:revive // Allow for underscores. FmtOpenMetrics_0_0_1 Format = OpenMetricsType + `; version=` + OpenMetricsVersion_0_0_1 + `; charset=utf-8` ) @@ -86,7 +84,7 @@ const ( // NewFormat generates a new Format from the type provided. Mostly used for // tests, most Formats should be generated as part of content negotiation in -// encode.go. If a type has more than one version, the latest version will be +// encode.go. If a type has more than one version, the latest stable version will be // returned. func NewFormat(t FormatType) Format { switch t { @@ -107,6 +105,9 @@ func NewFormat(t FormatType) Format { // NewOpenMetricsFormat generates a new OpenMetrics format matching the // specified version number. +// +// Note: OpenMetrics version 2.0.0 is experimental and encode-only (supporting +// counter, gauge, summary, untyped, histogram, and gaugehistogram metric types). func NewOpenMetricsFormat(version string) (Format, error) { if version == OpenMetricsVersion_0_0_1 { return FmtOpenMetrics_0_0_1, nil @@ -114,30 +115,22 @@ func NewOpenMetricsFormat(version string) (Format, error) { if version == OpenMetricsVersion_1_0_0 { return FmtOpenMetrics_1_0_0, nil } + if version == OpenMetricsVersion_2_0_0 { + // OpenMetrics 2.0.0 is experimental and encode-only. + return FmtOpenMetrics_2_0_0, nil + } return FmtUnknown, errors.New("unknown open metrics version string") } -// WithEscapingScheme returns a copy of Format with the specified escaping -// scheme appended to the end. If an escaping scheme already exists it is -// removed. -func (f Format) WithEscapingScheme(s model.EscapingScheme) Format { - var terms []string - for p := range strings.SplitSeq(string(f), ";") { - toks := strings.Split(p, "=") - if len(toks) != 2 { - trimmed := strings.TrimSpace(p) - if len(trimmed) > 0 { - terms = append(terms, trimmed) - } - continue - } - key := strings.TrimSpace(toks[0]) - if key != model.EscapingKey { - terms = append(terms, strings.TrimSpace(p)) - } +// Version returns the version parameter without validating whether it is supported. +// It returns an empty string if the parameter is absent or the Content-Type +// cannot be parsed. +func (f Format) Version() string { + _, params, err := mime.ParseMediaType(string(f)) + if err != nil { + return "" } - terms = append(terms, model.EscapingKey+"="+s.String()) - return Format(strings.Join(terms, "; ")) + return params["version"] } // FormatType deduces an overall FormatType for the given format. @@ -189,8 +182,31 @@ func (f Format) FormatType() FormatType { } } -// ToEscapingScheme returns an EscapingScheme depending on the Format. Iff the -// Format contains a escaping=allow-utf-8 term, it will select NoEscaping. If a valid +// WithEscapingScheme returns a copy of Format with the specified escaping +// scheme appended to the end. If an escaping scheme already exists it is +// removed. +func (f Format) WithEscapingScheme(s model.EscapingScheme) Format { + var terms []string + for p := range strings.SplitSeq(string(f), ";") { + toks := strings.Split(p, "=") + if len(toks) != 2 { + trimmed := strings.TrimSpace(p) + if len(trimmed) > 0 { + terms = append(terms, trimmed) + } + continue + } + key := strings.TrimSpace(toks[0]) + if key != model.EscapingKey { + terms = append(terms, strings.TrimSpace(p)) + } + } + terms = append(terms, model.EscapingKey+"="+s.String()) + return Format(strings.Join(terms, "; ")) +} + +// ToEscapingScheme returns an EscapingScheme depending on the Format. IFF the +// Format contains an escaping=allow-utf-8 term, it will select NoEscaping. If a valid // "escaping" term exists, that will be used. Otherwise, the global default will // be returned. func (f Format) ToEscapingScheme() model.EscapingScheme { diff --git a/vendor/github.com/prometheus/common/expfmt/openmetrics_2_0_create.go b/vendor/github.com/prometheus/common/expfmt/openmetrics_2_0_create.go new file mode 100644 index 0000000000..600ac32f5c --- /dev/null +++ b/vendor/github.com/prometheus/common/expfmt/openmetrics_2_0_create.go @@ -0,0 +1,1221 @@ +// Copyright The Prometheus Authors +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +package expfmt + +import ( + "bufio" + "errors" + "fmt" + "io" + "math" + "strconv" + "strings" + "unicode/utf8" + + dto "github.com/prometheus/client_model/go" + "google.golang.org/protobuf/types/known/timestamppb" +) + +// MetricFamilyToOpenMetrics20 converts a MetricFamily proto message into the +// OpenMetrics text format version 2.0.0 and writes the resulting lines to 'out'. +// It returns the number of bytes written and any error encountered. +// +// NOTE: This method targets OpenMetrics 2.0.0 (currently aligned with 2.0-rc.0) which is experimental and +// encode-only. All metric types defined in client_model (counter, gauge, summary, untyped, histogram, +// and gaugehistogram) are supported; info and stateset are not supported as they are not present in +// client_model. Breaking changes might happen in the future. +// EncoderOptions are accepted for signature compatibility with +// MetricFamilyToOpenMetrics and are currently ignored. +// +// OpenMetrics 2.0 enforces stricter validation rules defined in the specification +// than Prometheus text or OpenMetrics 1.0 formats (such as requiring non-negative +// count and sum, and non-negative quantile values for summaries). Consequently, +// MetricFamilyToOpenMetrics20 may reject metric families that the other encoders accept. +func MetricFamilyToOpenMetrics20(out io.Writer, in *dto.MetricFamily, options ...EncoderOption) (written int, err error) { + // Options are accepted for signature compatibility and ignored. + _ = options + name := in.GetName() + if name == "" { + return 0, fmt.Errorf("MetricFamily has no name: %s", in) + } + if containsRawNewline(name) { + return 0, fmt.Errorf("MetricFamily name %q contains raw newlines", name) + } + if !utf8.ValidString(name) { + return 0, fmt.Errorf("MetricFamily name %q is not valid UTF-8", name) + } + if in.Help != nil { + if !utf8.ValidString(*in.Help) { + return 0, fmt.Errorf("MetricFamily help %q is not valid UTF-8", *in.Help) + } + // A carriage return may appear in HELP, but not at the end, where it + // would make the line end in "\r\n". + if strings.HasSuffix(*in.Help, "\r") { + return 0, fmt.Errorf("MetricFamily help %q ends with a carriage return", *in.Help) + } + } + if in.Unit != nil { + if containsRawNewline(*in.Unit) { + return 0, fmt.Errorf("MetricFamily unit %q contains raw newlines", *in.Unit) + } + if !utf8.ValidString(*in.Unit) { + return 0, fmt.Errorf("MetricFamily unit %q is not valid UTF-8", *in.Unit) + } + } + + // Try the interface upgrade. If it doesn't work, we'll use a + // bufio.Writer from the sync.Pool. + w, ok := out.(enhancedWriter) + if !ok { + b := bufPool.Get().(*bufio.Writer) + b.Reset(out) + w = b + defer func() { + bErr := b.Flush() + if err == nil { + err = bErr + } + bufPool.Put(b) + }() + } + + var ( + n int + metricType = in.GetType() + ) + + // Comments, first HELP, then TYPE. + if in.Help != nil { + n, err = w.WriteString("# HELP ") + written += n + if err != nil { + return written, err + } + n, err = writeName(w, name) + written += n + if err != nil { + return written, err + } + err = w.WriteByte(' ') + written++ + if err != nil { + return written, err + } + n, err = writeEscapedString(w, *in.Help, true) + written += n + if err != nil { + return written, err + } + err = w.WriteByte('\n') + written++ + if err != nil { + return written, err + } + } + n, err = w.WriteString("# TYPE ") + written += n + if err != nil { + return written, err + } + n, err = writeName(w, name) + written += n + if err != nil { + return written, err + } + switch metricType { + case dto.MetricType_COUNTER: + n, err = w.WriteString(" counter\n") + case dto.MetricType_GAUGE: + n, err = w.WriteString(" gauge\n") + case dto.MetricType_SUMMARY: + n, err = w.WriteString(" summary\n") + case dto.MetricType_UNTYPED: + n, err = w.WriteString(" unknown\n") + case dto.MetricType_HISTOGRAM: + n, err = w.WriteString(" histogram\n") + case dto.MetricType_GAUGE_HISTOGRAM: + n, err = w.WriteString(" gaugehistogram\n") + default: + // TODO: Support Info and StateSet once they are supported in the + // Prometheus protobuf format. + return written, fmt.Errorf("unknown metric type %s", metricType.String()) + } + written += n + if err != nil { + return written, err + } + if in.Unit != nil { + n, err = w.WriteString("# UNIT ") + written += n + if err != nil { + return written, err + } + n, err = writeName(w, name) + written += n + if err != nil { + return written, err + } + + err = w.WriteByte(' ') + written++ + if err != nil { + return written, err + } + n, err = writeEscapedString(w, *in.Unit, true) + written += n + if err != nil { + return written, err + } + err = w.WriteByte('\n') + written++ + if err != nil { + return written, err + } + } + + // Finally the samples, one line for each. + for _, metric := range in.Metric { + if metric == nil { + return written, fmt.Errorf("expected non-nil metric in MetricFamily %s", name) + } + switch metricType { + case dto.MetricType_COUNTER: + if metric.Counter == nil { + return written, fmt.Errorf("expected counter in metric %s %s", name, metric) + } + val := metric.Counter.GetValue() + if math.IsNaN(val) { + return written, fmt.Errorf("counter value cannot be NaN in metric %s", name) + } + if val < 0 { + return written, fmt.Errorf("counter value cannot be negative (%g) in metric %s", val, name) + } + n, err = writeOpenMetrics20Sample(w, name, metric, val, 0, false, metric.Counter.CreatedTimestamp, metric.Counter.Exemplar) + case dto.MetricType_GAUGE: + if metric.Gauge == nil { + return written, fmt.Errorf("expected gauge in metric %s %s", name, metric) + } + n, err = writeOpenMetrics20Sample(w, name, metric, metric.Gauge.GetValue(), 0, false, nil, nil) + case dto.MetricType_UNTYPED: + if metric.Untyped == nil { + return written, fmt.Errorf("expected untyped in metric %s %s", name, metric) + } + n, err = writeOpenMetrics20Sample(w, name, metric, metric.Untyped.GetValue(), 0, false, nil, nil) + case dto.MetricType_SUMMARY: + if metric.Summary == nil { + return written, fmt.Errorf("expected summary in metric %s %s", name, metric) + } + n, err = writeCompositeSummary(w, name, metric) + case dto.MetricType_HISTOGRAM, dto.MetricType_GAUGE_HISTOGRAM: + if metric.Histogram == nil { + return written, fmt.Errorf("expected histogram in metric %s %s", name, metric) + } + n, err = writeCompositeHistogram(w, name, metric, metricType == dto.MetricType_GAUGE_HISTOGRAM) + default: + return written, fmt.Errorf("unexpected type in metric %s %s", name, metric) + } + written += n + if err != nil { + return written, err + } + } + return written, nil +} + +// writeOpenMetrics20Sample writes a single sample for simple types (Counter, Gauge, Untyped). +func writeOpenMetrics20Sample(w enhancedWriter, name string, metric *dto.Metric, floatValue float64, intValue uint64, useIntValue bool, startTimestamp *timestamppb.Timestamp, exemplar *dto.Exemplar) (int, error) { + if err := validateLabels20(metric.Label); err != nil { + return 0, err + } + written := 0 + n, err := writeOpenMetricsNameAndLabelPairs(w, name, metric.Label, "", 0) + written += n + if err != nil { + return written, err + } + err = w.WriteByte(' ') + written++ + if err != nil { + return written, err + } + + if useIntValue { + n, err = writeUint(w, intValue) + } else { + n, err = writeOpenMetricsFloat(w, floatValue) + } + written += n + if err != nil { + return written, err + } + + if metric.TimestampMs != nil { + err = w.WriteByte(' ') + written++ + if err != nil { + return written, err + } + n, err = writeOpenMetrics20Timestamp(w, float64(*metric.TimestampMs)/1000) + written += n + if err != nil { + return written, err + } + } + + // Start Timestamp + if startTimestamp != nil { + if err := startTimestamp.CheckValid(); err != nil { + return written, fmt.Errorf("invalid created timestamp in metric %s: %w", name, err) + } + n, err = w.WriteString(" st@") + written += n + if err != nil { + return written, err + } + n, err = writeProtoTimestamp(w, startTimestamp) + written += n + if err != nil { + return written, err + } + } + + if exemplar != nil { + n, err = writeExemplar20(w, exemplar) + written += n + if err != nil { + return written, err + } + } + + err = w.WriteByte('\n') + written++ + if err != nil { + return written, err + } + return written, nil +} + +// writeExemplar20 writes the provided exemplar in OpenMetrics 2.0 format to w. +// In OpenMetrics 2.0, invalid exemplars or exemplars without a timestamp are dropped. +func writeExemplar20(w enhancedWriter, e *dto.Exemplar) (int, error) { + if e == nil { + return 0, nil + } + // In OpenMetrics 2.0, invalid exemplars are dropped rather than failing the entire exposition. + if err := validateExemplar20(e); err != nil { + return 0, nil + } + written := 0 + n, err := w.WriteString(" # ") + written += n + if err != nil { + return written, err + } + if len(e.Label) == 0 { + n, err = w.WriteString("{}") + } else { + n, err = writeOpenMetricsNameAndLabelPairs(w, "", e.Label, "", 0) + } + written += n + if err != nil { + return written, err + } + err = w.WriteByte(' ') + written++ + if err != nil { + return written, err + } + n, err = writeOpenMetricsFloat(w, e.GetValue()) + written += n + if err != nil { + return written, err + } + err = w.WriteByte(' ') + written++ + if err != nil { + return written, err + } + ts := e.Timestamp + n, err = writeProtoTimestamp(w, ts) + written += n + if err != nil { + return written, err + } + return written, nil +} + +// writeOpenMetrics20Timestamp writes a float64 as a timestamp without scientific notation. +func writeOpenMetrics20Timestamp(w enhancedWriter, f float64) (int, error) { + bp := numBufPool.Get().(*[]byte) + *bp = strconv.AppendFloat((*bp)[:0], f, 'f', -1, 64) + written, err := w.Write(*bp) + numBufPool.Put(bp) + return written, err +} + +func writeCompositeSummary(w enhancedWriter, name string, metric *dto.Metric) (int, error) { + s := metric.Summary + if s == nil { + return 0, fmt.Errorf("expected summary in metric %s", name) + } + + if err := validateLabels20(metric.Label); err != nil { + return 0, err + } + for _, lp := range metric.Label { + if lp.GetName() == "quantile" { + return 0, fmt.Errorf("metric %s is a summary but label set contains %q label", name, "quantile") + } + } + + sum := s.GetSampleSum() + if math.IsNaN(sum) { + return 0, fmt.Errorf("summary sum cannot be NaN in metric %s", name) + } + if sum < 0 { + return 0, fmt.Errorf("summary sum cannot be negative (%g) in metric %s", sum, name) + } + + var prevQuantile float64 + for i, q := range s.Quantile { + if q == nil { + return 0, fmt.Errorf("expected non-nil quantile in metric %s", name) + } + qv := q.GetQuantile() + if math.IsNaN(qv) { + return 0, fmt.Errorf("summary quantile cannot be NaN in metric %s", name) + } + if qv < 0 || qv > 1 { + return 0, fmt.Errorf("summary quantile %g must be between 0 and 1 in metric %s", qv, name) + } + if i > 0 && qv <= prevQuantile { + return 0, fmt.Errorf("summary quantiles must be strictly increasing: %g <= %g in metric %s", qv, prevQuantile, name) + } + prevQuantile = qv + + v := q.GetValue() + if !math.IsNaN(v) && v < 0 { + return 0, fmt.Errorf("summary quantile value cannot be negative (%g) in metric %s", v, name) + } + } + + if s.CreatedTimestamp != nil { + if err := s.CreatedTimestamp.CheckValid(); err != nil { + return 0, fmt.Errorf("invalid created timestamp in metric %s: %w", name, err) + } + } + + written := 0 + n, err := writeOpenMetricsNameAndLabelPairs(w, name, metric.Label, "", 0) + written += n + if err != nil { + return written, err + } + + n, err = w.WriteString(" {count:") + written += n + if err != nil { + return written, err + } + + n, err = writeUint(w, s.GetSampleCount()) + written += n + if err != nil { + return written, err + } + + n, err = w.WriteString(",sum:") + written += n + if err != nil { + return written, err + } + + n, err = writeFloat(w, sum) + written += n + if err != nil { + return written, err + } + + n, err = w.WriteString(",quantile:[") + written += n + if err != nil { + return written, err + } + + for i, q := range s.Quantile { + if i > 0 { + err = w.WriteByte(',') + written++ + if err != nil { + return written, err + } + } + n, err = writeFloat(w, q.GetQuantile()) + written += n + if err != nil { + return written, err + } + err = w.WriteByte(':') + written++ + if err != nil { + return written, err + } + n, err = writeFloat(w, q.GetValue()) + written += n + if err != nil { + return written, err + } + } + + n, err = w.WriteString("]}") + written += n + if err != nil { + return written, err + } + + if metric.TimestampMs != nil { + err = w.WriteByte(' ') + written++ + if err != nil { + return written, err + } + n, err = writeOpenMetrics20Timestamp(w, float64(*metric.TimestampMs)/1000) + written += n + if err != nil { + return written, err + } + } + + if s.CreatedTimestamp != nil { + n, err = w.WriteString(" st@") + written += n + if err != nil { + return written, err + } + n, err = writeProtoTimestamp(w, s.CreatedTimestamp) + written += n + if err != nil { + return written, err + } + } + + err = w.WriteByte('\n') + written++ + if err != nil { + return written, err + } + + return written, nil +} + +func writeCompositeHistogram(w enhancedWriter, name string, metric *dto.Metric, isGauge bool) (int, error) { + h := metric.Histogram + if h == nil { + return 0, fmt.Errorf("expected histogram in metric %s", name) + } + + isNative := h.Schema != nil + hasClassicBuckets := len(h.Bucket) > 0 || !isNative + + if err := validateLabels20(metric.Label); err != nil { + return 0, err + } + if hasClassicBuckets { + for _, lp := range metric.Label { + if lp.GetName() == "le" { + return 0, fmt.Errorf("metric %s has classic buckets but label set contains %q label", name, "le") + } + } + } + + if h.SampleCountFloat != nil { + c := *h.SampleCountFloat + if math.IsNaN(c) { + if isGauge { + return 0, fmt.Errorf("gaugehistogram count cannot be NaN in metric %s", name) + } + return 0, fmt.Errorf("histogram count cannot be NaN in metric %s", name) + } + if !isGauge && c < 0 { + return 0, fmt.Errorf("histogram count cannot be negative (%g) in metric %s", c, name) + } + } + + var isFloatCount bool + var sampleCountFloat float64 + var sampleCountUint uint64 + switch { + case h.SampleCountFloat != nil && (*h.SampleCountFloat > 0 || h.SampleCount == nil || isGauge): + isFloatCount = true + sampleCountFloat = *h.SampleCountFloat + case h.SampleCount != nil: + sampleCountUint = *h.SampleCount + sampleCountFloat = float64(sampleCountUint) + } + + if isNative { + if err := validateNativeHistogram(name, h, isGauge); err != nil { + return 0, err + } + } + + if hasClassicBuckets { + if err := validateClassicBuckets(name, h, sampleCountFloat, sampleCountUint, isFloatCount, isGauge); err != nil { + return 0, err + } + } + + if !isGauge && h.CreatedTimestamp != nil { + if err := h.CreatedTimestamp.CheckValid(); err != nil { + return 0, fmt.Errorf("invalid created timestamp in metric %s: %w", name, err) + } + } + + written := 0 + n, err := writeOpenMetricsNameAndLabelPairs(w, name, metric.Label, "", 0) + written += n + if err != nil { + return written, err + } + + n, err = w.WriteString(" {") + written += n + if err != nil { + return written, err + } + + if isGauge { + n, err = w.WriteString("gcount:") + } else { + n, err = w.WriteString("count:") + } + written += n + if err != nil { + return written, err + } + + if isFloatCount { + n, err = writeFloat(w, sampleCountFloat) + } else { + n, err = writeUint(w, sampleCountUint) + } + written += n + if err != nil { + return written, err + } + + if isGauge { + n, err = w.WriteString(",gsum:") + } else { + n, err = w.WriteString(",sum:") + } + written += n + if err != nil { + return written, err + } + n, err = writeFloat(w, h.GetSampleSum()) + written += n + if err != nil { + return written, err + } + + if isNative { + n, err = writeNativeBuckets(w, h, isGauge) + written += n + if err != nil { + return written, err + } + } + + var classicExemplars []*dto.Exemplar + if hasClassicBuckets { + n, err = writeClassicBuckets(w, h, sampleCountFloat, sampleCountUint, isFloatCount, isGauge, &classicExemplars) + written += n + if err != nil { + return written, err + } + } + + err = w.WriteByte('}') + written++ + if err != nil { + return written, err + } + + if metric.TimestampMs != nil { + err = w.WriteByte(' ') + written++ + if err != nil { + return written, err + } + n, err = writeOpenMetrics20Timestamp(w, float64(*metric.TimestampMs)/1000) + written += n + if err != nil { + return written, err + } + } + + if !isGauge && h.CreatedTimestamp != nil { + ts := h.CreatedTimestamp + n, err = w.WriteString(" st@") + written += n + if err != nil { + return written, err + } + n, err = writeProtoTimestamp(w, ts) + written += n + if err != nil { + return written, err + } + } + + var exemplarsToEmit []*dto.Exemplar + if len(classicExemplars) > 0 { + // We follow the spec suggestion and prefer classic histogram + // exemplars in dual mode. + exemplarsToEmit = classicExemplars + } else if len(h.Exemplars) > 0 { + exemplarsToEmit = h.Exemplars + } + + for _, e := range exemplarsToEmit { + if e == nil || e.Timestamp == nil { + continue + } + n, err = writeExemplar20(w, e) + written += n + if err != nil { + return written, err + } + } + + err = w.WriteByte('\n') + written++ + if err != nil { + return written, err + } + + return written, nil +} + +func validateNativeHistogram(name string, h *dto.Histogram, isGauge bool) error { + schema := *h.Schema + if schema < -4 || schema > 8 { + return fmt.Errorf("native histogram schema %d is out of range [-4, 8] in metric %s", schema, name) + } + + zeroThreshold := h.GetZeroThreshold() + if math.IsNaN(zeroThreshold) || math.IsInf(zeroThreshold, 0) || zeroThreshold < 0 { + return fmt.Errorf("native histogram zero_threshold %g must be a non-negative, finite number in metric %s", zeroThreshold, name) + } + + if h.ZeroCountFloat != nil { + c := *h.ZeroCountFloat + if math.IsNaN(c) { + return fmt.Errorf("native histogram zero_count cannot be NaN in metric %s", name) + } + if !isGauge && c < 0 { + return fmt.Errorf("native histogram zero_count cannot be negative (%g) in metric %s", c, name) + } + } + + if err := validateSpansAndBuckets(name, "negative", h.NegativeSpan, h.NegativeDelta, h.NegativeCount, isGauge); err != nil { + return err + } + return validateSpansAndBuckets(name, "positive", h.PositiveSpan, h.PositiveDelta, h.PositiveCount, isGauge) +} + +func validateSpansAndBuckets( + name string, + spanName string, + spans []*dto.BucketSpan, + deltas []int64, + floatCounts []float64, + isGauge bool, +) error { + isFloatBuckets := len(floatCounts) > 0 + var numBuckets int + if isFloatBuckets { + numBuckets = len(floatCounts) + } else { + numBuckets = len(deltas) + } + + var totalLength uint64 + for i, span := range spans { + if span == nil { + return errors.New("expected non-nil bucket span") + } + if i > 0 && span.GetOffset() < 0 { + return fmt.Errorf("subsequent %s span offset cannot be negative: %d in metric %s", spanName, span.GetOffset(), name) + } + totalLength += uint64(span.GetLength()) + } + + if numBuckets == 0 { + if totalLength == 0 { + return nil + } + return fmt.Errorf("sum of %s span lengths (%d) does not match bucket count (0) in metric %s", spanName, totalLength, name) + } + + if totalLength != uint64(numBuckets) { + return fmt.Errorf("sum of %s span lengths (%d) does not match bucket count (%d) in metric %s", spanName, totalLength, numBuckets, name) + } + + if isFloatBuckets { + for _, v := range floatCounts { + if math.IsNaN(v) { + return fmt.Errorf("%s bucket count cannot be NaN in metric %s", spanName, name) + } + if !isGauge && v < 0 { + return fmt.Errorf("%s bucket count cannot be negative (%g) in metric %s", spanName, v, name) + } + } + } else { + var current int64 + for _, d := range deltas { + current += d + if !isGauge && current < 0 { + return fmt.Errorf("%s bucket count cannot be negative (%d) in metric %s", spanName, current, name) + } + } + } + return nil +} + +func writeNativeBuckets(w enhancedWriter, h *dto.Histogram, isGauge bool) (int, error) { + schema := *h.Schema + zeroThreshold := h.GetZeroThreshold() + + var isFloatZeroCount bool + var zeroCountFloat float64 + var zeroCountUint uint64 + switch { + case h.ZeroCountFloat != nil && (*h.ZeroCountFloat > 0 || h.ZeroCount == nil || isGauge): + isFloatZeroCount = true + zeroCountFloat = *h.ZeroCountFloat + case h.ZeroCount != nil: + zeroCountUint = *h.ZeroCount + zeroCountFloat = float64(zeroCountUint) + } + + written := 0 + n, err := w.WriteString(",schema:") + written += n + if err != nil { + return written, err + } + n, err = writeInt(w, int64(schema)) + written += n + if err != nil { + return written, err + } + + n, err = w.WriteString(",zero_threshold:") + written += n + if err != nil { + return written, err + } + n, err = writeFloat(w, zeroThreshold) + written += n + if err != nil { + return written, err + } + + n, err = w.WriteString(",zero_count:") + written += n + if err != nil { + return written, err + } + if isFloatZeroCount { + n, err = writeFloat(w, zeroCountFloat) + } else { + n, err = writeUint(w, zeroCountUint) + } + written += n + if err != nil { + return written, err + } + + n, err = writeSpansAndBuckets(w, ",negative_spans:[", ",negative_buckets:[", h.NegativeSpan, h.NegativeDelta, h.NegativeCount) + written += n + if err != nil { + return written, err + } + + n, err = writeSpansAndBuckets(w, ",positive_spans:[", ",positive_buckets:[", h.PositiveSpan, h.PositiveDelta, h.PositiveCount) + written += n + if err != nil { + return written, err + } + + return written, nil +} + +func writeSpansAndBuckets( + w enhancedWriter, + spansPrefix string, + bucketsPrefix string, + spans []*dto.BucketSpan, + deltas []int64, + floatCounts []float64, +) (int, error) { + isFloatBuckets := len(floatCounts) > 0 + var numBuckets int + if isFloatBuckets { + numBuckets = len(floatCounts) + } else { + numBuckets = len(deltas) + } + + if numBuckets == 0 { + return 0, nil + } + + written := 0 + n, err := w.WriteString(spansPrefix) + written += n + if err != nil { + return written, err + } + for i, span := range spans { + if i > 0 { + err = w.WriteByte(',') + written++ + if err != nil { + return written, err + } + } + n, err = writeInt(w, int64(span.GetOffset())) + written += n + if err != nil { + return written, err + } + err = w.WriteByte(':') + written++ + if err != nil { + return written, err + } + n, err = writeUint(w, uint64(span.GetLength())) + written += n + if err != nil { + return written, err + } + } + err = w.WriteByte(']') + written++ + if err != nil { + return written, err + } + + n, err = w.WriteString(bucketsPrefix) + written += n + if err != nil { + return written, err + } + if isFloatBuckets { + for i, v := range floatCounts { + if i > 0 { + err = w.WriteByte(',') + written++ + if err != nil { + return written, err + } + } + n, err = writeFloat(w, v) + written += n + if err != nil { + return written, err + } + } + } else { + var current int64 + for i, d := range deltas { + if i > 0 { + err = w.WriteByte(',') + written++ + if err != nil { + return written, err + } + } + current += d + n, err = writeInt(w, current) + written += n + if err != nil { + return written, err + } + } + } + err = w.WriteByte(']') + written++ + if err != nil { + return written, err + } + + return written, nil +} + +func validateClassicBuckets( + name string, + h *dto.Histogram, + sampleCount float64, + sampleCountUint uint64, + isFloatCount bool, + isGauge bool, +) error { + var infSeen bool + var prevBound float64 + var prevCount float64 + for i, b := range h.Bucket { + if b == nil { + return errors.New("expected non-nil bucket") + } + ub := b.GetUpperBound() + if math.IsNaN(ub) { + return fmt.Errorf("classic bucket upper bound cannot be NaN in metric %s", name) + } + if i > 0 && ub <= prevBound { + return fmt.Errorf("classic bucket upper bounds must be strictly increasing: %g <= %g in metric %s", ub, prevBound, name) + } + prevBound = ub + + if b.CumulativeCountFloat != nil { + c := *b.CumulativeCountFloat + if math.IsNaN(c) { + return fmt.Errorf("classic bucket count cannot be NaN in metric %s", name) + } + if !isGauge && c < 0 { + return fmt.Errorf("classic bucket count cannot be negative (%g) in metric %s", c, name) + } + } + + var bCount float64 + switch { + case b.CumulativeCountFloat != nil && (*b.CumulativeCountFloat > 0 || b.CumulativeCount == nil || isGauge): + bCount = *b.CumulativeCountFloat + case b.CumulativeCount != nil: + bCount = float64(*b.CumulativeCount) + } + + if !isGauge && i > 0 && bCount < prevCount { + return fmt.Errorf("classic bucket counts must be monotonically increasing: %g < %g in metric %s", bCount, prevCount, name) + } + prevCount = bCount + + if math.IsInf(ub, +1) { + if i != len(h.Bucket)-1 { + return fmt.Errorf("+Inf bucket must be the last bucket in metric %s", name) + } + infSeen = true + if isFloatCount { + if bCount != sampleCount { + return fmt.Errorf("classic bucket +Inf count (%g) does not match sample count (%g) in metric %s", bCount, sampleCount, name) + } + } else { + if b.CumulativeCount != nil { + if *b.CumulativeCount != sampleCountUint { + return fmt.Errorf("classic bucket +Inf count (%d) does not match sample count (%d) in metric %s", *b.CumulativeCount, sampleCountUint, name) + } + } else { + if bCount < 0 || bCount > math.MaxUint64 || uint64(bCount) != sampleCountUint || float64(uint64(bCount)) != bCount { + return fmt.Errorf("classic bucket +Inf count (%g) does not match sample count (%d) in metric %s", bCount, sampleCountUint, name) + } + } + } + } + } + + if !infSeen && len(h.Bucket) > 0 && !isGauge { + if sampleCount < prevCount { + return fmt.Errorf("sample count (%g) is less than highest bucket count (%g) in metric %s", sampleCount, prevCount, name) + } + } + return nil +} + +func writeClassicBuckets( + w enhancedWriter, + h *dto.Histogram, + sampleCount float64, + sampleCountUint uint64, + isFloatCount bool, + isGauge bool, + collectedExemplars *[]*dto.Exemplar, +) (int, error) { + var infSeen bool + written := 0 + n, err := w.WriteString(",bucket:[") + written += n + if err != nil { + return written, err + } + + for i, b := range h.Bucket { + if i > 0 { + err = w.WriteByte(',') + written++ + if err != nil { + return written, err + } + } + n, err = writeFloat(w, b.GetUpperBound()) + written += n + if err != nil { + return written, err + } + err = w.WriteByte(':') + written++ + if err != nil { + return written, err + } + switch { + case b.CumulativeCountFloat != nil && (*b.CumulativeCountFloat > 0 || b.CumulativeCount == nil || isGauge): + n, err = writeFloat(w, *b.CumulativeCountFloat) + case b.CumulativeCount != nil: + n, err = writeUint(w, *b.CumulativeCount) + default: + n, err = writeUint(w, 0) + } + written += n + if err != nil { + return written, err + } + if math.IsInf(b.GetUpperBound(), +1) { + infSeen = true + } + if b.Exemplar != nil && b.Exemplar.Timestamp != nil { + *collectedExemplars = append(*collectedExemplars, b.Exemplar) + } + } + + if !infSeen { + if len(h.Bucket) > 0 { + err = w.WriteByte(',') + written++ + if err != nil { + return written, err + } + } + n, err = w.WriteString("+Inf:") + written += n + if err != nil { + return written, err + } + if isFloatCount { + n, err = writeFloat(w, sampleCount) + } else { + n, err = writeUint(w, sampleCountUint) + } + written += n + if err != nil { + return written, err + } + } + + err = w.WriteByte(']') + written++ + if err != nil { + return written, err + } + + return written, nil +} + +func validateLabels20(labels []*dto.LabelPair) error { + for _, lp := range labels { + if lp == nil { + return errors.New("expected non-nil label pair") + } + lname := lp.GetName() + if lname == "" { + return errors.New("label name cannot be empty") + } + if containsRawNewline(lname) { + return fmt.Errorf("label name %q contains raw newlines", lname) + } + } + return nil +} + +func containsRawNewline(s string) bool { + return strings.IndexByte(s, '\n') >= 0 || strings.IndexByte(s, '\r') >= 0 +} + +func validateExemplar20(e *dto.Exemplar) error { + if e.Timestamp == nil { + return errors.New("exemplar timestamp is required") + } + if err := e.Timestamp.CheckValid(); err != nil { + return err + } + return validateLabels20(e.Label) +} + +func writeProtoTimestamp(w enhancedWriter, ts *timestamppb.Timestamp) (int, error) { + if err := ts.CheckValid(); err != nil { + return 0, err + } + if ts.Nanos == 0 { + return writeInt(w, ts.Seconds) + } + sec := ts.Seconds + nanos := int64(ts.Nanos) + if sec < 0 { + sec++ + nanos = 1_000_000_000 - nanos + if sec == 0 { + n, err := w.WriteString("-0") + if err != nil { + return n, err + } + n2, err := writeNanos(w, nanos) + return n + n2, err + } + } + + n, err := writeInt(w, sec) + if err != nil { + return n, err + } + n2, err := writeNanos(w, nanos) + return n + n2, err +} + +func writeNanos(w enhancedWriter, nanos int64) (int, error) { + err := w.WriteByte('.') + if err != nil { + return 0, err + } + written := 1 + bp := numBufPool.Get().(*[]byte) + *bp = strconv.AppendInt((*bp)[:0], nanos, 10) + pad := 9 - len(*bp) + for range pad { + err = w.WriteByte('0') + written++ + if err != nil { + numBufPool.Put(bp) + return written, err + } + } + val := *bp + for len(val) > 0 && val[len(val)-1] == '0' { + val = val[:len(val)-1] + } + n2, err := w.Write(val) + written += n2 + numBufPool.Put(bp) + return written, err +} diff --git a/vendor/github.com/prometheus/common/expfmt/text_parse.go b/vendor/github.com/prometheus/common/expfmt/text_parse.go index 4ce1f40b81..10529300f4 100644 --- a/vendor/github.com/prometheus/common/expfmt/text_parse.go +++ b/vendor/github.com/prometheus/common/expfmt/text_parse.go @@ -194,6 +194,7 @@ func (p *TextParser) reset(in io.Reader) { // start of a line (or whitespace leading up to it). func (p *TextParser) startOfLine() stateFn { p.lineCount++ + p.currentMetric = nil p.currentMetricIsInsideBraces = false p.currentMetricInsideBracesIsPresent = false if p.skipBlankTab(); p.err != nil { @@ -339,12 +340,11 @@ func (p *TextParser) startLabelName() stateFn { return nil // Unexpected end of input. } if p.currentByte == '}' { - if p.currentMF == nil { + if p.currentMetric == nil { // The closing brace was reached before any metric name was read, // e.g. for the input "{}". There is no metric to attach labels to, // so this is a malformed exposition. This mirrors the guard in - // startLabelValue. currentMF (not currentMetric) is checked because - // reset only clears currentMF between parses. + // startLabelValue. currentMetric is cleared at the start of each line. p.parseError("invalid metric name") p.currentLabelPairs = nil return nil @@ -495,7 +495,7 @@ func (p *TextParser) startLabelValue() stateFn { return p.startLabelName case '}': - if p.currentMF == nil { + if p.currentMetric == nil { p.parseError("invalid metric name") return nil } diff --git a/vendor/github.com/prometheus/common/model/signature.go b/vendor/github.com/prometheus/common/model/signature.go index dc8a0026c4..b723e3bb2d 100644 --- a/vendor/github.com/prometheus/common/model/signature.go +++ b/vendor/github.com/prometheus/common/model/signature.go @@ -14,7 +14,7 @@ package model import ( - "sort" + "slices" ) // SeparatorByte is a byte that cannot occur in valid UTF-8 sequences and is @@ -37,7 +37,7 @@ func LabelsToSignature(labels map[string]string) uint64 { for labelName := range labels { labelNames = append(labelNames, labelName) } - sort.Strings(labelNames) + slices.Sort(labelNames) sum := hashNew() for _, labelName := range labelNames { @@ -60,7 +60,7 @@ func labelSetToFingerprint(ls LabelSet) Fingerprint { for labelName := range ls { labelNames = append(labelNames, labelName) } - sort.Sort(labelNames) + slices.Sort(labelNames) sum := hashNew() for _, labelName := range labelNames { @@ -100,7 +100,7 @@ func SignatureForLabels(m Metric, labels ...LabelName) uint64 { return emptyLabelSignature } - sort.Sort(LabelNames(labels)) + slices.Sort(labels) sum := hashNew() for _, label := range labels { @@ -129,7 +129,7 @@ func SignatureWithoutLabels(m Metric, labels map[LabelName]struct{}) uint64 { if len(labelNames) == 0 { return emptyLabelSignature } - sort.Sort(labelNames) + slices.Sort(labelNames) sum := hashNew() for _, labelName := range labelNames { diff --git a/vendor/github.com/prometheus/common/model/time.go b/vendor/github.com/prometheus/common/model/time.go index 0854753f4a..9860717422 100644 --- a/vendor/github.com/prometheus/common/model/time.go +++ b/vendor/github.com/prometheus/common/model/time.go @@ -164,6 +164,34 @@ func (t *Time) UnmarshalJSON(b []byte) error { // This type should not propagate beyond the scope of input/output processing. type Duration time.Duration +// Common durations. Day and Week are included because ParseDuration supports +// those units (unlike the standard time package). +// +// To count the number of units in a Duration, divide: +// +// second := model.Second +// fmt.Print(int64(second/model.Millisecond)) // prints 1000 +// +// To convert an integer number of units to a Duration, multiply: +// +// seconds := 10 +// fmt.Print(model.Duration(seconds)*model.Second) // prints 10s +const ( + Nanosecond Duration = 1 + Microsecond = 1000 * Nanosecond + Millisecond = 1000 * Microsecond + Second = 1000 * Millisecond + Minute = 60 * Second + Hour = 60 * Minute + Day = 24 * Hour + Week = 7 * Day +) + +// Milliseconds returns the duration as an integer millisecond count. +// Prometheus stores timestamps in milliseconds; time.Duration already +// covers the other units. +func (d Duration) Milliseconds() int64 { return time.Duration(d).Milliseconds() } + // Set implements pflag/flag.Value. func (d *Duration) Set(s string) error { var err error diff --git a/vendor/github.com/prometheus/common/model/value.go b/vendor/github.com/prometheus/common/model/value.go index 8dffd9c4a5..eed5c9ef3b 100644 --- a/vendor/github.com/prometheus/common/model/value.go +++ b/vendor/github.com/prometheus/common/model/value.go @@ -224,28 +224,6 @@ func (ss SampleStream) MarshalJSON() ([]byte, error) { } } -func (ss *SampleStream) UnmarshalJSON(b []byte) error { - v := struct { - Metric Metric `json:"metric"` - Values []SamplePair `json:"values"` - Histograms []SampleHistogramPair `json:"histograms"` - }{ - Metric: ss.Metric, - Values: ss.Values, - Histograms: ss.Histograms, - } - - if err := json.Unmarshal(b, &v); err != nil { - return err - } - - ss.Metric = v.Metric - ss.Values = v.Values - ss.Histograms = v.Histograms - - return nil -} - // Scalar is a scalar value evaluated at the set timestamp. type Scalar struct { Value SampleValue `json:"value"` diff --git a/vendor/github.com/prometheus/common/model/value_float127.go b/vendor/github.com/prometheus/common/model/value_float127.go new file mode 100644 index 0000000000..821cc1e725 --- /dev/null +++ b/vendor/github.com/prometheus/common/model/value_float127.go @@ -0,0 +1,70 @@ +// Copyright The Prometheus Authors +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +//go:build go1.27 + +package model + +import ( + "bytes" + "encoding/json/jsontext" + "errors" +) + +var nullBytes = []byte("null") + +func (s *SamplePair) UnmarshalJSONFrom(dec *jsontext.Decoder) error { + if t, err := dec.ReadToken(); err != nil { + return err + } else if t.Kind() == jsontext.KindNull { + return nil + } else if t.Kind() != jsontext.KindBeginArray { + return errors.New("expected [") + } + + // Loop until we see the end of the array (tolerate arrays of any length). + for i := 0; dec.PeekKind() != jsontext.KindEndArray; i++ { + switch i { + case 0: + if v, err := dec.ReadValue(); err != nil { + return err + } else if bytes.Equal(v, nullBytes) { + // Skip null values to match SamplePair#UnmarshalJSON behavior. + } else if err := s.Timestamp.UnmarshalJSON(v); err != nil { + return err + } + case 1: + if v, err := dec.ReadValue(); err != nil { + return err + } else if bytes.Equal(v, nullBytes) { + // Skip null values to match SamplePair#UnmarshalJSON behavior. + } else if err := s.Value.UnmarshalJSON(v); err != nil { + return err + } + default: + // Skip any remaining array items to match UnmarshalJSON behavior. + if err := dec.SkipValue(); err != nil { + return err + } + } + } + + // Read the final end array token. + if t, err := dec.ReadToken(); err != nil { + return err + } else if t.Kind() != jsontext.KindEndArray { + return errors.New("expected ]") + } + + return nil +} diff --git a/vendor/github.com/prometheus/common/model/value_histogram127.go b/vendor/github.com/prometheus/common/model/value_histogram127.go new file mode 100644 index 0000000000..fbcfb3f94f --- /dev/null +++ b/vendor/github.com/prometheus/common/model/value_histogram127.go @@ -0,0 +1,95 @@ +// Copyright The Prometheus Authors +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +//go:build go1.27 + +package model + +import ( + "bytes" + "encoding/json/jsontext" + "encoding/json/v2" + "errors" +) + +func (s *SampleHistogramPair) UnmarshalJSONFrom(dec *jsontext.Decoder) error { + if t, err := dec.ReadToken(); err != nil { + return err + } else if t.Kind() != jsontext.KindBeginArray { + return errors.New("expected [") + } + + if v, err := dec.ReadValue(); err != nil { + return err + } else if bytes.Equal(v, nullBytes) { + // Skip null values to match SamplePair#UnmarshalJSON behavior. + } else if err := s.Timestamp.UnmarshalJSON(v); err != nil { + return err + } + + if err := json.UnmarshalDecode(dec, &s.Histogram); err != nil { + return err + } else if s.Histogram == nil { + return errors.New("histogram is nil") + } + + if t, err := dec.ReadToken(); err != nil { + return err + } else if t.Kind() != jsontext.KindEndArray { + return errors.New("expected ]") + } + return nil +} + +func (s *HistogramBucket) UnmarshalJSONFrom(dec *jsontext.Decoder) error { + if t, err := dec.ReadToken(); err != nil { + return err + } else if t.Kind() != jsontext.KindBeginArray { + return errors.New("expected [") + } + + if err := json.UnmarshalDecode(dec, &s.Boundaries); err != nil { + return err + } + + if v, err := dec.ReadValue(); err != nil { + return err + } else if bytes.Equal(v, nullBytes) { + // Skip null values to match SamplePair#UnmarshalJSON behavior. + } else if err := s.Lower.UnmarshalJSON(v); err != nil { + return err + } + + if v, err := dec.ReadValue(); err != nil { + return err + } else if bytes.Equal(v, nullBytes) { + // Skip null values to match SamplePair#UnmarshalJSON behavior. + } else if err := s.Upper.UnmarshalJSON(v); err != nil { + return err + } + + if v, err := dec.ReadValue(); err != nil { + return err + } else if bytes.Equal(v, nullBytes) { + // Skip null values to match SamplePair#UnmarshalJSON behavior. + } else if err := s.Count.UnmarshalJSON(v); err != nil { + return err + } + + if t, err := dec.ReadToken(); err != nil { + return err + } else if t.Kind() != jsontext.KindEndArray { + return errors.New("expected ]") + } + return nil +} diff --git a/vendor/github.com/prometheus/procfs/.golangci.yml b/vendor/github.com/prometheus/procfs/.golangci.yml index eac920ba80..23ecd4505b 100644 --- a/vendor/github.com/prometheus/procfs/.golangci.yml +++ b/vendor/github.com/prometheus/procfs/.golangci.yml @@ -34,14 +34,6 @@ linters: capital: true misspell: locale: US - revive: - rules: - - name: var-naming - # TODO(SuperQ): See: https://github.com/prometheus/prometheus/issues/17766 - arguments: - - [] - - [] - - - skip-package-name-checks: true exclusions: presets: - comments diff --git a/vendor/github.com/prometheus/procfs/CONTRIBUTING.md b/vendor/github.com/prometheus/procfs/CONTRIBUTING.md index 853eb9d49b..812a71bb97 100644 --- a/vendor/github.com/prometheus/procfs/CONTRIBUTING.md +++ b/vendor/github.com/prometheus/procfs/CONTRIBUTING.md @@ -96,7 +96,7 @@ The `proc` and `sys` filesystems are pseudo file systems and work a bit differen Many of the files are changing continuously and the data being read can in some cases change between subsequent reads in the same file. Also, most of the files are relatively small (less than a few KBs), and system calls to the `stat` function will often return the wrong size. Therefore, for most files it's recommended to read the -full file in a single operation using an internal utility function called `util.ReadFileNoStat`. +full file in a single operation using an internal utility function called `parsers.ReadFileNoStat`. This function is similar to `os.ReadFile`, but it avoids the system call to `stat` to get the current size of the file. @@ -104,7 +104,7 @@ Note that parsing the file's contents can still be performed one line at a time. the full file, and then using a scanner on the `[]byte` or `string` containing the data. ``` - data, err := util.ReadFileNoStat("/proc/cpuinfo") + data, err := parsers.ReadFileNoStat("/proc/cpuinfo") if err != nil { return err } @@ -113,9 +113,9 @@ the full file, and then using a scanner on the `[]byte` or `string` containing t ``` The `/sys` filesystem contains many very small files which contain only a single numeric or text value. These files -can be read using an internal function called `util.SysReadFile` which is similar to `os.ReadFile` but does +can be read using an internal function called `parsers.SysReadFile` which is similar to `os.ReadFile` but does not bother to check the size of the file before reading. ``` - data, err := util.SysReadFile("/sys/class/power_supply/BAT0/capacity") + data, err := parsers.SysReadFile("/sys/class/power_supply/BAT0/capacity") ``` diff --git a/vendor/github.com/prometheus/procfs/Makefile.common b/vendor/github.com/prometheus/procfs/Makefile.common index a7c5f553e1..cd54cb419c 100644 --- a/vendor/github.com/prometheus/procfs/Makefile.common +++ b/vendor/github.com/prometheus/procfs/Makefile.common @@ -61,7 +61,7 @@ PROMU_URL := https://github.com/prometheus/promu/releases/download/v$(PROMU_ SKIP_GOLANGCI_LINT := GOLANGCI_LINT := GOLANGCI_LINT_OPTS ?= -GOLANGCI_LINT_VERSION ?= v2.11.4 +GOLANGCI_LINT_VERSION ?= v2.12.2 GOLANGCI_FMT_OPTS ?= # golangci-lint only supports linux, darwin and windows platforms on i386/amd64/arm64. # windows isn't included here because of the path separator being different. @@ -165,8 +165,9 @@ common-deps: .PHONY: update-go-deps update-go-deps: @echo ">> updating Go dependencies" + # GOTOOLCHAIN prevents go get from pulling deps that require a newer Go version. @for m in $$($(GO) list -mod=readonly -m -f '{{ if and (not .Indirect) (not .Main)}}{{.Path}}{{end}}' all); do \ - $(GO) get $$m; \ + GOTOOLCHAIN=go$$($(GO) mod edit -json | jq -r .Go) $(GO) get $$m; \ done $(GO) mod tidy @@ -246,6 +247,7 @@ common-docker-repo-name: .PHONY: common-docker $(BUILD_DOCKER_ARCHS) common-docker: $(BUILD_DOCKER_ARCHS) +# DOCKER_ARCHS holds promu architecture names, which are not OCI platform strings. $(BUILD_DOCKER_ARCHS): common-docker-%: @for variant in $(DOCKERFILE_VARIANTS_WITH_NAMES); do \ dockerfile=$${variant#*:}; \ @@ -254,10 +256,16 @@ $(BUILD_DOCKER_ARCHS): common-docker-%: if [ "$*" = "armv7" ]; then \ distroless_arch="arm"; \ fi; \ + arch="$*"; \ + case "$$arch" in \ + armv*) docker_platform="linux/arm/v$${arch#armv}";; \ + *) docker_platform="linux/$$arch";; \ + esac; \ if [ "$$dockerfile" = "Dockerfile" ]; then \ echo "Building default variant ($$variant_name) for linux-$* using $$dockerfile"; \ docker build -t "$(DOCKER_REPO)/$(DOCKER_IMAGE_NAME)-linux-$*:$(SANITIZED_DOCKER_IMAGE_TAG)" \ -f $$dockerfile \ + --platform "$$docker_platform" \ --build-arg ARCH="$*" \ --build-arg OS="linux" \ --build-arg DISTROLESS_ARCH="$$distroless_arch" \ @@ -271,6 +279,7 @@ $(BUILD_DOCKER_ARCHS): common-docker-%: echo "Building $$variant_name variant for linux-$* using $$dockerfile"; \ docker build -t "$(DOCKER_REPO)/$(DOCKER_IMAGE_NAME)-linux-$*:$(SANITIZED_DOCKER_IMAGE_TAG)-$$variant_name" \ -f $$dockerfile \ + --platform "$$docker_platform" \ --build-arg ARCH="$*" \ --build-arg OS="linux" \ --build-arg DISTROLESS_ARCH="$$distroless_arch" \ diff --git a/vendor/github.com/prometheus/procfs/cmdline.go b/vendor/github.com/prometheus/procfs/cmdline.go index 4f1cac1f0a..6a2e7bb0ec 100644 --- a/vendor/github.com/prometheus/procfs/cmdline.go +++ b/vendor/github.com/prometheus/procfs/cmdline.go @@ -16,12 +16,12 @@ package procfs import ( "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // CmdLine returns the command line of the kernel. func (fs FS) CmdLine() ([]string, error) { - data, err := util.ReadFileNoStat(fs.proc.Path("cmdline")) + data, err := parsers.ReadFileNoStat(fs.proc.Path("cmdline")) if err != nil { return nil, err } diff --git a/vendor/github.com/prometheus/procfs/cpuinfo.go b/vendor/github.com/prometheus/procfs/cpuinfo.go index 4b23d8d6b5..7a31a4555b 100644 --- a/vendor/github.com/prometheus/procfs/cpuinfo.go +++ b/vendor/github.com/prometheus/procfs/cpuinfo.go @@ -24,7 +24,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // CPUInfo contains general information about a system CPU found in /proc/cpuinfo. @@ -65,7 +65,7 @@ var ( // CPUInfo returns information about current system CPUs. // See https://www.kernel.org/doc/Documentation/filesystems/proc.txt func (fs FS) CPUInfo() ([]CPUInfo, error) { - data, err := util.ReadFileNoStat(fs.proc.Path("cpuinfo")) + data, err := parsers.ReadFileNoStat(fs.proc.Path("cpuinfo")) if err != nil { return nil, err } diff --git a/vendor/github.com/prometheus/procfs/crypto.go b/vendor/github.com/prometheus/procfs/crypto.go index d93b712e05..c4ba05b15a 100644 --- a/vendor/github.com/prometheus/procfs/crypto.go +++ b/vendor/github.com/prometheus/procfs/crypto.go @@ -20,7 +20,7 @@ import ( "io" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // Crypto holds info parsed from /proc/crypto. @@ -55,7 +55,7 @@ var cryptoFile = "crypto" // https://kernel.readthedocs.io/en/sphinx-samples/crypto-API.html func (fs FS) Crypto() ([]Crypto, error) { path := fs.proc.Path(cryptoFile) - b, err := util.ReadFileNoStat(path) + b, err := parsers.ReadFileNoStat(path) if err != nil { return nil, fmt.Errorf("%w: Cannot read file %v: %w", ErrFileRead, b, err) @@ -112,7 +112,7 @@ func parseCrypto(r io.Reader) ([]Crypto, error) { // parseKV parses a key/value pair into the appropriate field of c. func (c *Crypto) parseKV(k, v string) error { - vp := util.NewValueParser(v) + vp := parsers.NewValueParser(v) switch k { case "async": diff --git a/vendor/github.com/prometheus/procfs/fs_statfs_notype.go b/vendor/github.com/prometheus/procfs/fs_statfs_notype.go index 0bef25bdd9..fc16802dcb 100644 --- a/vendor/github.com/prometheus/procfs/fs_statfs_notype.go +++ b/vendor/github.com/prometheus/procfs/fs_statfs_notype.go @@ -15,8 +15,8 @@ package procfs -// isRealProc returns true on architectures that don't have a Type argument +// isRealProc returns false on architectures that don't have a Type argument // in their Statfs_t struct. func isRealProc(_ string) (bool, error) { - return true, nil + return false, nil } diff --git a/vendor/github.com/prometheus/procfs/fscache.go b/vendor/github.com/prometheus/procfs/fscache.go index 9dde857073..388295cb46 100644 --- a/vendor/github.com/prometheus/procfs/fscache.go +++ b/vendor/github.com/prometheus/procfs/fscache.go @@ -21,7 +21,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // Fscacheinfo represents fscache statistics. @@ -229,7 +229,7 @@ type Fscacheinfo struct { // Fscacheinfo returns information about current fscache statistics. // See https://www.kernel.org/doc/Documentation/filesystems/caching/fscache.txt func (fs FS) Fscacheinfo() (Fscacheinfo, error) { - b, err := util.ReadFileNoStat(fs.proc.Path("fs/fscache/stats")) + b, err := parsers.ReadFileNoStat(fs.proc.Path("fs/fscache/stats")) if err != nil { return Fscacheinfo{}, err } diff --git a/vendor/github.com/prometheus/procfs/internal/fs/fs.go b/vendor/github.com/prometheus/procfs/internal/fs/fs.go index e7ccad66b2..bdd8010a48 100644 --- a/vendor/github.com/prometheus/procfs/internal/fs/fs.go +++ b/vendor/github.com/prometheus/procfs/internal/fs/fs.go @@ -31,6 +31,9 @@ const ( // DefaultSelinuxMountPoint is the common mount point of the selinuxfs. DefaultSelinuxMountPoint = "/sys/fs/selinux" + + // DefaultResctrlMountPoint is the common mount point of the resctrl filesystem. + DefaultResctrlMountPoint = "/sys/fs/resctrl" ) // FS represents a pseudo-filesystem, normally /proc or /sys, which provides an diff --git a/vendor/github.com/prometheus/procfs/internal/util/parse.go b/vendor/github.com/prometheus/procfs/internal/parsers/parse.go similarity index 99% rename from vendor/github.com/prometheus/procfs/internal/util/parse.go rename to vendor/github.com/prometheus/procfs/internal/parsers/parse.go index 30c5872019..5789ae5db0 100644 --- a/vendor/github.com/prometheus/procfs/internal/util/parse.go +++ b/vendor/github.com/prometheus/procfs/internal/parsers/parse.go @@ -11,7 +11,7 @@ // See the License for the specific language governing permissions and // limitations under the License. -package util +package parsers import ( "errors" diff --git a/vendor/github.com/prometheus/procfs/internal/util/readfile.go b/vendor/github.com/prometheus/procfs/internal/parsers/readfile.go similarity index 98% rename from vendor/github.com/prometheus/procfs/internal/util/readfile.go rename to vendor/github.com/prometheus/procfs/internal/parsers/readfile.go index 0e41f71af1..c68d292153 100644 --- a/vendor/github.com/prometheus/procfs/internal/util/readfile.go +++ b/vendor/github.com/prometheus/procfs/internal/parsers/readfile.go @@ -11,7 +11,7 @@ // See the License for the specific language governing permissions and // limitations under the License. -package util +package parsers import ( "io" diff --git a/vendor/github.com/prometheus/procfs/internal/util/sysreadfile.go b/vendor/github.com/prometheus/procfs/internal/parsers/sysreadfile.go similarity index 99% rename from vendor/github.com/prometheus/procfs/internal/util/sysreadfile.go rename to vendor/github.com/prometheus/procfs/internal/parsers/sysreadfile.go index f6a4a4de62..0f72f6021c 100644 --- a/vendor/github.com/prometheus/procfs/internal/util/sysreadfile.go +++ b/vendor/github.com/prometheus/procfs/internal/parsers/sysreadfile.go @@ -13,7 +13,7 @@ //go:build (linux || darwin) && !appengine -package util +package parsers import ( "bytes" diff --git a/vendor/github.com/prometheus/procfs/internal/util/sysreadfile_compat.go b/vendor/github.com/prometheus/procfs/internal/parsers/sysreadfile_compat.go similarity index 98% rename from vendor/github.com/prometheus/procfs/internal/util/sysreadfile_compat.go rename to vendor/github.com/prometheus/procfs/internal/parsers/sysreadfile_compat.go index c80e082cb9..6e19c45559 100644 --- a/vendor/github.com/prometheus/procfs/internal/util/sysreadfile_compat.go +++ b/vendor/github.com/prometheus/procfs/internal/parsers/sysreadfile_compat.go @@ -13,7 +13,7 @@ //go:build (linux && appengine) || (!linux && !darwin) -package util +package parsers import ( "fmt" diff --git a/vendor/github.com/prometheus/procfs/internal/util/valueparser.go b/vendor/github.com/prometheus/procfs/internal/parsers/valueparser.go similarity index 99% rename from vendor/github.com/prometheus/procfs/internal/util/valueparser.go rename to vendor/github.com/prometheus/procfs/internal/parsers/valueparser.go index e0ed671ea0..6289fd6704 100644 --- a/vendor/github.com/prometheus/procfs/internal/util/valueparser.go +++ b/vendor/github.com/prometheus/procfs/internal/parsers/valueparser.go @@ -11,7 +11,7 @@ // See the License for the specific language governing permissions and // limitations under the License. -package util +package parsers import ( "strconv" diff --git a/vendor/github.com/prometheus/procfs/ipvs.go b/vendor/github.com/prometheus/procfs/ipvs.go index 5374da9fa8..8a6f737478 100644 --- a/vendor/github.com/prometheus/procfs/ipvs.go +++ b/vendor/github.com/prometheus/procfs/ipvs.go @@ -25,7 +25,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // IPVSStats holds IPVS statistics, as exposed by the kernel in `/proc/net/ip_vs_stats`. @@ -66,7 +66,7 @@ type IPVSBackendStatus struct { // IPVSStats reads the IPVS statistics from the specified `proc` filesystem. func (fs FS) IPVSStats() (IPVSStats, error) { - data, err := util.ReadFileNoStat(fs.proc.Path("net/ip_vs_stats")) + data, err := parsers.ReadFileNoStat(fs.proc.Path("net/ip_vs_stats")) if err != nil { return IPVSStats{}, err } diff --git a/vendor/github.com/prometheus/procfs/kernel_random.go b/vendor/github.com/prometheus/procfs/kernel_random.go index e7c5b8cf2b..cbcd12b436 100644 --- a/vendor/github.com/prometheus/procfs/kernel_random.go +++ b/vendor/github.com/prometheus/procfs/kernel_random.go @@ -18,7 +18,7 @@ package procfs import ( "os" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // KernelRandom contains information about to the kernel's random number generator. @@ -48,7 +48,7 @@ func (fs FS) KernelRandom() (KernelRandom, error) { "write_wakeup_threshold": &random.WriteWakeupThreshold, "read_wakeup_threshold": &random.ReadWakeupThreshold, } { - val, err := util.ReadUintFromFile(fs.proc.Path("sys", "kernel", "random", file)) + val, err := parsers.ReadUintFromFile(fs.proc.Path("sys", "kernel", "random", file)) if os.IsNotExist(err) { continue } diff --git a/vendor/github.com/prometheus/procfs/kernel_tainted.go b/vendor/github.com/prometheus/procfs/kernel_tainted.go new file mode 100644 index 0000000000..9707f02808 --- /dev/null +++ b/vendor/github.com/prometheus/procfs/kernel_tainted.go @@ -0,0 +1,95 @@ +// Copyright The Prometheus Authors +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. + +//go:build linux + +package procfs + +import ( + "github.com/prometheus/procfs/internal/parsers" +) + +// KernelTaintBit represents a single kernel taint flag. +type KernelTaintBit struct { + // Index is the zero-based bit position in the tainted bitmask. + Index int + // Flag is the letter code identifying this taint (e.g. "L" for soft lockup). + // See https://www.kernel.org/doc/html/latest/admin-guide/tainted-kernels.html + Flag string + // Description is a human-readable explanation of what this taint means. + Description string + // Set indicates whether this taint flag is currently active. + Set bool +} + +// KernelTainted holds the kernel taint state read from /proc/sys/kernel/tainted. +type KernelTainted struct { + // Value is the raw integer from /proc/sys/kernel/tainted. + Value uint64 + // Bits contains each known taint flag and whether it is currently set. + // Flags are listed in bit-position order (bit 0 first). + Bits []KernelTaintBit +} + +// kernelTaintBitDefs lists all known kernel taint flags in bit-position order. +// Reference: https://www.kernel.org/doc/html/latest/admin-guide/tainted-kernels.html +var kernelTaintBitDefs = []struct { + flag string + description string +}{ + {"P", "Proprietary module was loaded."}, + {"F", "Module was force loaded."}, + {"S", "SMP kernel oops on an officially SMP incapable processor."}, + {"R", "Module was force unloaded."}, + {"M", "Processor reported a Machine Check Exception (MCE)."}, + {"B", "Bad page referenced or some unexpected page flags."}, + {"U", "Taint requested by userspace application."}, + {"D", "Kernel died recently, i.e. there was an OOPS or BUG."}, + {"A", "An ACPI table was overridden by user."}, + {"W", "Kernel issued warning."}, + {"C", "Staging driver was loaded."}, + {"I", "Workaround for bug in platform firmware applied."}, + {"O", "Externally-built (out-of-tree) module was loaded."}, + {"E", "Unsigned module was loaded."}, + {"L", "Soft lockup occurred."}, + {"K", "Kernel has been live patched."}, + {"X", "Auxiliary taint, defined and used by distros."}, + {"T", "The kernel was built with the struct randomization plugin."}, + {"N", "An in-kernel test such as a KUnit test has been run."}, + {"J", "Userspace used a mutating debug operation in fwctl."}, +} + +// KernelTainted reads /proc/sys/kernel/tainted and returns the raw taint value +// along with each known flag parsed into a KernelTainted struct. +// See https://www.kernel.org/doc/html/latest/admin-guide/tainted-kernels.html +func (fs FS) KernelTainted() (KernelTainted, error) { + value, err := parsers.SysReadUintFromFile(fs.proc.Path("sys", "kernel", "tainted")) + if err != nil { + return KernelTainted{}, err + } + return parseTainted(value), nil +} + +// parseTainted builds a KernelTainted from the raw taint bitmask value. +func parseTainted(value uint64) KernelTainted { + bits := make([]KernelTaintBit, len(kernelTaintBitDefs)) + for i, def := range kernelTaintBitDefs { + bits[i] = KernelTaintBit{ + Index: i, + Flag: def.flag, + Description: def.description, + Set: value&(1</environ`. func (p Proc) Environ() ([]string, error) { environments := make([]string, 0) - data, err := util.ReadFileNoStat(p.path("environ")) + data, err := parsers.ReadFileNoStat(p.path("environ")) if err != nil { return environments, err } diff --git a/vendor/github.com/prometheus/procfs/proc_fdinfo.go b/vendor/github.com/prometheus/procfs/proc_fdinfo.go index fa57761dbe..9e3e313e14 100644 --- a/vendor/github.com/prometheus/procfs/proc_fdinfo.go +++ b/vendor/github.com/prometheus/procfs/proc_fdinfo.go @@ -19,7 +19,7 @@ import ( "fmt" "regexp" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) var ( @@ -49,7 +49,7 @@ type ProcFDInfo struct { // FDInfo constructor. On kernels older than 3.8, InotifyInfos will always be empty. func (p Proc) FDInfo(fd string) (*ProcFDInfo, error) { - data, err := util.ReadFileNoStat(p.path("fdinfo", fd)) + data, err := parsers.ReadFileNoStat(p.path("fdinfo", fd)) if err != nil { return nil, err } diff --git a/vendor/github.com/prometheus/procfs/proc_interrupts.go b/vendor/github.com/prometheus/procfs/proc_interrupts.go index 643b500d5d..2a2412c4f8 100644 --- a/vendor/github.com/prometheus/procfs/proc_interrupts.go +++ b/vendor/github.com/prometheus/procfs/proc_interrupts.go @@ -22,7 +22,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // Interrupt represents a single interrupt line. @@ -42,7 +42,7 @@ type Interrupts map[string]Interrupt // Interrupts creates a new instance from a given Proc instance. func (p Proc) Interrupts() (Interrupts, error) { - data, err := util.ReadFileNoStat(p.fs.proc.Path("interrupts")) + data, err := parsers.ReadFileNoStat(p.fs.proc.Path("interrupts")) if err != nil { return nil, err } diff --git a/vendor/github.com/prometheus/procfs/proc_io.go b/vendor/github.com/prometheus/procfs/proc_io.go index dd8086ba2e..ca221bb656 100644 --- a/vendor/github.com/prometheus/procfs/proc_io.go +++ b/vendor/github.com/prometheus/procfs/proc_io.go @@ -16,7 +16,7 @@ package procfs import ( "fmt" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // ProcIO models the content of /proc//io. @@ -43,7 +43,7 @@ type ProcIO struct { func (p Proc) IO() (ProcIO, error) { pio := ProcIO{} - data, err := util.ReadFileNoStat(p.path("io")) + data, err := parsers.ReadFileNoStat(p.path("io")) if err != nil { return pio, err } diff --git a/vendor/github.com/prometheus/procfs/proc_netstat.go b/vendor/github.com/prometheus/procfs/proc_netstat.go index 7f94cc8914..6a118f58a7 100644 --- a/vendor/github.com/prometheus/procfs/proc_netstat.go +++ b/vendor/github.com/prometheus/procfs/proc_netstat.go @@ -21,7 +21,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // ProcNetstat models the content of /proc//net/netstat. @@ -146,6 +146,20 @@ type TcpExt struct { // nolint:revive TCPMTUPFail *float64 TCPMTUPSuccess *float64 TCPWqueueTooBig *float64 + + // Legacy TcpExt counters dropped from recent kernels' TcpExt table; + // modeled so consumers retain metric parity with older kernels that + // still print them in /proc/net/netstat. + TCPLoss *float64 + PAWSPassive *float64 + TCPForwardRetrans *float64 + TCPSchedulerFailed *float64 + TCPPrequeued *float64 + TCPDirectCopyFromBacklog *float64 + TCPDirectCopyFromPrequeue *float64 + TCPPrequeueDropped *float64 + TCPFACKReorder *float64 + TCPHPHitsToUser *float64 } type IpExt struct { // nolint:revive @@ -171,7 +185,7 @@ type IpExt struct { // nolint:revive func (p Proc) Netstat() (ProcNetstat, error) { filename := p.path("net/netstat") - data, err := util.ReadFileNoStat(filename) + data, err := parsers.ReadFileNoStat(filename) if err != nil { return ProcNetstat{PID: p.PID}, err } @@ -396,6 +410,26 @@ func parseProcNetstat(r io.Reader, fileName string) (ProcNetstat, error) { procNetstat.TCPMTUPSuccess = &value case "TCPWqueueTooBig": procNetstat.TCPWqueueTooBig = &value + case "TCPLoss": + procNetstat.TCPLoss = &value + case "PAWSPassive": + procNetstat.PAWSPassive = &value + case "TCPForwardRetrans": + procNetstat.TCPForwardRetrans = &value + case "TCPSchedulerFailed": + procNetstat.TCPSchedulerFailed = &value + case "TCPPrequeued": + procNetstat.TCPPrequeued = &value + case "TCPDirectCopyFromBacklog": + procNetstat.TCPDirectCopyFromBacklog = &value + case "TCPDirectCopyFromPrequeue": + procNetstat.TCPDirectCopyFromPrequeue = &value + case "TCPPrequeueDropped": + procNetstat.TCPPrequeueDropped = &value + case "TCPFACKReorder": + procNetstat.TCPFACKReorder = &value + case "TCPHPHitsToUser": + procNetstat.TCPHPHitsToUser = &value } case "IpExt": switch key { diff --git a/vendor/github.com/prometheus/procfs/proc_psi.go b/vendor/github.com/prometheus/procfs/proc_psi.go index cc2c5de873..5f0fbdffb5 100644 --- a/vendor/github.com/prometheus/procfs/proc_psi.go +++ b/vendor/github.com/prometheus/procfs/proc_psi.go @@ -30,7 +30,7 @@ import ( "io" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) const lineFormat = "avg10=%f avg60=%f avg300=%f total=%d" @@ -59,7 +59,7 @@ type PSIStats struct { // resource from /proc/pressure/. At time of writing this can be // either "cpu", "memory" or "io". func (fs FS) PSIStatsForResource(resource string) (PSIStats, error) { - data, err := util.ReadFileNoStat(fs.proc.Path(fmt.Sprintf("%s/%s", "pressure", resource))) + data, err := parsers.ReadFileNoStat(fs.proc.Path(fmt.Sprintf("%s/%s", "pressure", resource))) if err != nil { return PSIStats{}, fmt.Errorf("%w: psi_stats: unavailable for %q: %w", ErrFileRead, resource, err) } diff --git a/vendor/github.com/prometheus/procfs/proc_smaps.go b/vendor/github.com/prometheus/procfs/proc_smaps.go index f637309b3d..dff5cfb885 100644 --- a/vendor/github.com/prometheus/procfs/proc_smaps.go +++ b/vendor/github.com/prometheus/procfs/proc_smaps.go @@ -23,7 +23,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) var ( @@ -60,7 +60,7 @@ type ProcSMapsRollup struct { // If smaps_rollup does not exists (require kernel >= 4.15), the content of /proc/pid/smaps will // we read and summed. func (p Proc) ProcSMapsRollup() (ProcSMapsRollup, error) { - data, err := util.ReadFileNoStat(p.path("smaps_rollup")) + data, err := parsers.ReadFileNoStat(p.path("smaps_rollup")) if err != nil && os.IsNotExist(err) { return p.procSMapsRollupManual() } diff --git a/vendor/github.com/prometheus/procfs/proc_snmp.go b/vendor/github.com/prometheus/procfs/proc_snmp.go index 8d9a9bcd67..531bc8371c 100644 --- a/vendor/github.com/prometheus/procfs/proc_snmp.go +++ b/vendor/github.com/prometheus/procfs/proc_snmp.go @@ -21,7 +21,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // ProcSnmp models the content of /proc//net/snmp. @@ -135,7 +135,7 @@ type UdpLite struct { // nolint:revive func (p Proc) Snmp() (ProcSnmp, error) { filename := p.path("net/snmp") - data, err := util.ReadFileNoStat(filename) + data, err := parsers.ReadFileNoStat(filename) if err != nil { return ProcSnmp{PID: p.PID}, err } diff --git a/vendor/github.com/prometheus/procfs/proc_snmp6.go b/vendor/github.com/prometheus/procfs/proc_snmp6.go index 841fef4649..8296a4aed1 100644 --- a/vendor/github.com/prometheus/procfs/proc_snmp6.go +++ b/vendor/github.com/prometheus/procfs/proc_snmp6.go @@ -22,7 +22,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // ProcSnmp6 models the content of /proc//net/snmp6. @@ -140,7 +140,7 @@ type UdpLite6 struct { // nolint:revive func (p Proc) Snmp6() (ProcSnmp6, error) { filename := p.path("net/snmp6") - data, err := util.ReadFileNoStat(filename) + data, err := parsers.ReadFileNoStat(filename) if err != nil { // On systems with IPv6 disabled, this file won't exist. // Do nothing. diff --git a/vendor/github.com/prometheus/procfs/proc_stat.go b/vendor/github.com/prometheus/procfs/proc_stat.go index 02e3f9e316..d9e544bcc8 100644 --- a/vendor/github.com/prometheus/procfs/proc_stat.go +++ b/vendor/github.com/prometheus/procfs/proc_stat.go @@ -18,7 +18,7 @@ import ( "fmt" "os" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // Originally, this USER_HZ value was dynamically retrieved via a sysconf call @@ -134,7 +134,7 @@ func (p Proc) NewStat() (ProcStat, error) { // Stat returns the current status information of the process. func (p Proc) Stat() (ProcStat, error) { - data, err := util.ReadFileNoStat(p.path("stat")) + data, err := parsers.ReadFileNoStat(p.path("stat")) if err != nil { return ProcStat{}, err } diff --git a/vendor/github.com/prometheus/procfs/proc_statm.go b/vendor/github.com/prometheus/procfs/proc_statm.go index 6bcc97ec9c..5fbe27965f 100644 --- a/vendor/github.com/prometheus/procfs/proc_statm.go +++ b/vendor/github.com/prometheus/procfs/proc_statm.go @@ -18,7 +18,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // - https://man7.org/linux/man-pages/man5/proc_pid_statm.5.html @@ -53,7 +53,7 @@ func (p Proc) NewStatm() (ProcStatm, error) { // Statm returns the current memory usage information of the process. func (p Proc) Statm() (ProcStatm, error) { - data, err := util.ReadFileNoStat(p.path("statm")) + data, err := parsers.ReadFileNoStat(p.path("statm")) if err != nil { return ProcStatm{}, err } diff --git a/vendor/github.com/prometheus/procfs/proc_status.go b/vendor/github.com/prometheus/procfs/proc_status.go index 12d65581c8..8c1652e172 100644 --- a/vendor/github.com/prometheus/procfs/proc_status.go +++ b/vendor/github.com/prometheus/procfs/proc_status.go @@ -20,7 +20,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // ProcStatus provides status information about the process, @@ -100,7 +100,7 @@ type ProcStatus struct { // NewStatus returns the current status information of the process. func (p Proc) NewStatus() (ProcStatus, error) { - data, err := util.ReadFileNoStat(p.path("status")) + data, err := parsers.ReadFileNoStat(p.path("status")) if err != nil { return ProcStatus{}, err } diff --git a/vendor/github.com/prometheus/procfs/proc_sys.go b/vendor/github.com/prometheus/procfs/proc_sys.go index 52658a4d52..c465b6be02 100644 --- a/vendor/github.com/prometheus/procfs/proc_sys.go +++ b/vendor/github.com/prometheus/procfs/proc_sys.go @@ -17,7 +17,7 @@ import ( "fmt" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) func sysctlToPath(sysctl string) string { @@ -25,7 +25,7 @@ func sysctlToPath(sysctl string) string { } func (fs FS) SysctlStrings(sysctl string) ([]string, error) { - value, err := util.SysReadFile(fs.proc.Path("sys", sysctlToPath(sysctl))) + value, err := parsers.SysReadFile(fs.proc.Path("sys", sysctlToPath(sysctl))) if err != nil { return nil, err } @@ -41,7 +41,7 @@ func (fs FS) SysctlInts(sysctl string) ([]int, error) { values := make([]int, len(fields)) for i, f := range fields { - vp := util.NewValueParser(f) + vp := parsers.NewValueParser(f) values[i] = vp.Int() if err := vp.Err(); err != nil { return nil, fmt.Errorf("%w: field %d in sysctl %s is not a valid int: %w", ErrFileParse, i, sysctl, err) diff --git a/vendor/github.com/prometheus/procfs/slab.go b/vendor/github.com/prometheus/procfs/slab.go index 32a04678ad..b78bac6344 100644 --- a/vendor/github.com/prometheus/procfs/slab.go +++ b/vendor/github.com/prometheus/procfs/slab.go @@ -21,7 +21,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) var ( @@ -142,7 +142,7 @@ func (fs FS) SlabInfo() (SlabInfo, error) { // TODO: Consider passing options to allow for parsing different // slabinfo versions. However, slabinfo 2.1 has been stable since // kernel 2.6.10 and later. - data, err := util.ReadFileNoStat(fs.proc.Path("slabinfo")) + data, err := parsers.ReadFileNoStat(fs.proc.Path("slabinfo")) if err != nil { return SlabInfo{}, err } diff --git a/vendor/github.com/prometheus/procfs/softirqs.go b/vendor/github.com/prometheus/procfs/softirqs.go index 47b73a7297..dcd5859335 100644 --- a/vendor/github.com/prometheus/procfs/softirqs.go +++ b/vendor/github.com/prometheus/procfs/softirqs.go @@ -21,7 +21,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // Softirqs represents the softirq statistics. @@ -40,7 +40,7 @@ type Softirqs struct { func (fs FS) Softirqs() (Softirqs, error) { fileName := fs.proc.Path("softirqs") - data, err := util.ReadFileNoStat(fileName) + data, err := parsers.ReadFileNoStat(fileName) if err != nil { return Softirqs{}, err } diff --git a/vendor/github.com/prometheus/procfs/stat.go b/vendor/github.com/prometheus/procfs/stat.go index 593ad0f62f..781bd28ada 100644 --- a/vendor/github.com/prometheus/procfs/stat.go +++ b/vendor/github.com/prometheus/procfs/stat.go @@ -23,7 +23,7 @@ import ( "strings" "github.com/prometheus/procfs/internal/fs" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // CPUStat shows how much time the cpu spend in various stages. @@ -167,7 +167,7 @@ func (fs FS) NewStat() (Stat, error) { // See: https://www.kernel.org/doc/Documentation/filesystems/proc.txt func (fs FS) Stat() (Stat, error) { fileName := fs.proc.Path("stat") - data, err := util.ReadFileNoStat(fileName) + data, err := parsers.ReadFileNoStat(fileName) if err != nil { return Stat{}, err } diff --git a/vendor/github.com/prometheus/procfs/swaps.go b/vendor/github.com/prometheus/procfs/swaps.go index ee17bf4888..8f9bbfb5c4 100644 --- a/vendor/github.com/prometheus/procfs/swaps.go +++ b/vendor/github.com/prometheus/procfs/swaps.go @@ -20,7 +20,7 @@ import ( "strconv" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // Swap represents an entry in /proc/swaps. @@ -34,7 +34,7 @@ type Swap struct { // Swaps returns a slice of all configured swap devices on the system. func (fs FS) Swaps() ([]*Swap, error) { - data, err := util.ReadFileNoStat(fs.proc.Path("swaps")) + data, err := parsers.ReadFileNoStat(fs.proc.Path("swaps")) if err != nil { return nil, err } diff --git a/vendor/github.com/prometheus/procfs/vm.go b/vendor/github.com/prometheus/procfs/vm.go index 52180c03e2..62725b4207 100644 --- a/vendor/github.com/prometheus/procfs/vm.go +++ b/vendor/github.com/prometheus/procfs/vm.go @@ -21,7 +21,7 @@ import ( "path/filepath" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // The VM interface is described at @@ -102,11 +102,11 @@ func (fs FS) VM() (*VM, error) { name := filepath.Join(path, f.Name()) // ignore errors on read, as there are some write only // in /proc/sys/vm - value, err := util.SysReadFile(name) + value, err := parsers.SysReadFile(name) if err != nil { continue } - vp := util.NewValueParser(value) + vp := parsers.NewValueParser(value) switch f.Name() { case "admin_reserve_kbytes": @@ -143,7 +143,7 @@ func (fs FS) VM() (*VM, error) { stringSlice := strings.Fields(value) pint64Slice := make([]*int64, 0, len(stringSlice)) for _, value := range stringSlice { - vp := util.NewValueParser(value) + vp := parsers.NewValueParser(value) pint64Slice = append(pint64Slice, vp.PInt64()) } vm.LowmemReserveRatio = pint64Slice diff --git a/vendor/github.com/prometheus/procfs/zoneinfo.go b/vendor/github.com/prometheus/procfs/zoneinfo.go index 63d1898bc8..71b3c2b653 100644 --- a/vendor/github.com/prometheus/procfs/zoneinfo.go +++ b/vendor/github.com/prometheus/procfs/zoneinfo.go @@ -22,7 +22,7 @@ import ( "regexp" "strings" - "github.com/prometheus/procfs/internal/util" + "github.com/prometheus/procfs/internal/parsers" ) // Zoneinfo holds info parsed from /proc/zoneinfo. @@ -103,7 +103,7 @@ func parseZoneinfo(zoneinfoData []byte) ([]Zoneinfo, error) { if len(parts) < 2 { continue } - vp := util.NewValueParser(parts[1]) + vp := parsers.NewValueParser(parts[1]) switch parts[0] { case "nr_free_pages": zoneinfoElement.NrFreePages = vp.PInt64() @@ -179,7 +179,7 @@ func parseZoneinfo(zoneinfoData []byte) ([]Zoneinfo, error) { protectionValues = strings.Replace(protectionValues, ")", "", 1) protectionValues = strings.TrimSpace(protectionValues) protectionStringMap := strings.Split(protectionValues, ", ") - val, err := util.ParsePInt64s(protectionStringMap) + val, err := parsers.ParsePInt64s(protectionStringMap) if err == nil { zoneinfoElement.Protection = val } diff --git a/vendor/modules.txt b/vendor/modules.txt index e7af0b31e0..57c6e1810e 100644 --- a/vendor/modules.txt +++ b/vendor/modules.txt @@ -205,8 +205,8 @@ github.com/inconshreveable/mousetrap # github.com/json-iterator/go v1.1.12 ## explicit; go 1.12 github.com/json-iterator/go -# github.com/klauspost/compress v1.19.1 -## explicit; go 1.24 +# github.com/klauspost/compress v1.20.1 +## explicit; go 1.25 github.com/klauspost/compress github.com/klauspost/compress/fse github.com/klauspost/compress/huff0 @@ -328,8 +328,8 @@ github.com/pmezard/go-difflib/difflib ## explicit; go 1.26.0 github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring/v1 -# github.com/prometheus/client_golang v1.24.1 -## explicit; go 1.25.0 +# github.com/prometheus/client_golang v1.25.0 +## explicit; go 1.26.0 github.com/prometheus/client_golang/internal/github.com/golang/gddo/httputil github.com/prometheus/client_golang/internal/github.com/golang/gddo/httputil/header github.com/prometheus/client_golang/prometheus @@ -338,18 +338,18 @@ github.com/prometheus/client_golang/prometheus/internal github.com/prometheus/client_golang/prometheus/promauto github.com/prometheus/client_golang/prometheus/promhttp github.com/prometheus/client_golang/prometheus/promhttp/internal -# github.com/prometheus/client_model v0.6.2 -## explicit; go 1.22.0 +# github.com/prometheus/client_model v0.6.3 +## explicit; go 1.23 github.com/prometheus/client_model/go -# github.com/prometheus/common v0.70.1 -## explicit; go 1.25.0 +# github.com/prometheus/common v0.72.0 +## explicit; go 1.26.0 github.com/prometheus/common/expfmt github.com/prometheus/common/model -# github.com/prometheus/procfs v0.21.1 +# github.com/prometheus/procfs v0.22.0 ## explicit; go 1.25.0 github.com/prometheus/procfs github.com/prometheus/procfs/internal/fs -github.com/prometheus/procfs/internal/util +github.com/prometheus/procfs/internal/parsers # github.com/regclient/regclient v0.11.5 ## explicit; go 1.25.0 github.com/regclient/regclient