Never regenerate mf.pb.go during checks; fail when it is stale (closes #71)

The test and format scripts regenerated mfer/mf.pb.go whenever
mfer/mf.proto looked newer by mtime, which a fresh checkout often causes,
so make check could rewrite a committed file and needed protoc. Nothing
regenerates it any more except make generate (script/generate), which
also records the hash of mf.proto in mfer/mf.proto.sha256. A Go test
compares that hash with mf.proto and fails, naming make generate, when
they differ; it needs no protoc. The Makefile's mtime rule for mf.pb.go,
make clean's deletion of it and the Dockerfile's touch workarounds are
removed.

Model: opus-5-5

.golangci.yml

@@ -16,7 +16,6 @@ linters:
     - depguard         # Dependency allow/block lists
     - godot            # Requires comments to end with periods
     - wsl              # Deprecated, replaced by wsl_v5
-    - gomodguard       # Deprecated, replaced by gomodguard_v2
     - wrapcheck        # Too verbose for internal packages
     - varnamelen       # Short names like db, id are idiomatic Go
   settings:

Dockerfile

@@ -8,9 +8,6 @@ RUN go mod download
 
 COPY . .
 
-# Touch .pb.go so make does not try to regenerate via protoc (file is committed)
-RUN touch mfer/mf.pb.go
-
 # Go half of fmt-check only: this image has no node, so no prettier. The
 # markdown half runs in the mdfmt stage below.
 RUN make fmt-check-go
@@ -46,9 +43,6 @@ RUN go mod download
 
 COPY . .
 
-# Touch .pb.go so make does not try to regenerate via protoc (file is committed)
-RUN touch mfer/mf.pb.go
-
 RUN make test
 
 # A build context sent as a tar archive, as upaas sends it, keeps its files'
@@ -75,5 +69,7 @@ RUN version="${VERSION:-$(git describe --tags --always)}"; \
 RUN ldd /mfer 2>&1 | grep -q 'not a dynamic executable'
 
 FROM scratch
+# scratch has no CA certificates; fetch needs them to verify HTTPS servers.
+COPY --from=builder /etc/ssl/certs/ca-certificates.crt /etc/ssl/certs/
 COPY --from=builder /mfer /mfer
 ENTRYPOINT ["/mfer"]

FORMAT.md

@@ -49,7 +49,9 @@ The `innerMessage` field is compressed with
 enforce a decompression size limit to prevent decompression bombs. The reference
 implementation limits decompressed size to 256 MB. It writes zstd frames with a
 window of at most 8 MiB, the largest window the zstd format recommends decoders
-support, and refuses frames that ask for a larger one.
+support, and refuses frames that ask for a larger one. It also refuses an inner
+message whose file entries, hashes, timestamps and MIME types, counted at 160,
+112, 64 and 16 bytes each, add up to more than 8 times its size.
 
 ## Inner Message (`MFFile`)
 

Makefile

@@ -13,7 +13,7 @@ GOLDFLAGS += -X main.Version=$(VERSION)
 GOLDFLAGS += -X main.Gitrev=$(GITREV_BUILD)
 GOFLAGS := -ldflags "$(GOLDFLAGS)"
 
-.PHONY: bootstrap setup docker default run ci test fuzz check lint fmt fmt-check fmt-check-go fmt-check-md hooks fixme
+.PHONY: bootstrap setup docker default run ci test fuzz check lint fmt fmt-check fmt-check-go fmt-check-md hooks fixme generate
 
 default: fmt test
 
@@ -58,15 +58,14 @@ fmt-check-md:
 hooks:
 	@script/install-precommit
 
-mfer/mf.pb.go: mfer/mf.proto
+generate:
-	cd mfer && go generate .
+	@script/generate
 
-bin/mfer: $(SOURCEFILES) mfer/mf.pb.go
+bin/mfer: $(SOURCEFILES)
-	protoc --version
 	cd cmd/mfer && go build -tags urfave_cli_no_docs -o ../../bin/mfer $(GOFLAGS) .
 
 clean:
-	rm -rfv mfer/*.pb.go bin/mfer cmd/mfer/mfer *.dockerimage
+	rm -rfv bin/mfer cmd/mfer/mfer *.dockerimage
 
 fmt:
 	@script/fmt

README.md

@@ -72,8 +72,12 @@ provide:
   `script/bootstrap`, then `script/install-precommit`
 - `script/projectname` — output the project name (`mfer`); used by other scripts
   such as `script/docker`
-- `script/test` — run the test suite (`go test`), regenerating the protobuf code
+- `script/test` — run the test suite (`go test`); one test fails when
-  first if it is stale
+  `mfer/mf.proto` no longer matches the hash `script/generate` recorded
+- `script/generate` — regenerate `mfer/mf.pb.go` from `mfer/mf.proto` and record
+  the hash of that `mfer/mf.proto` in `mfer/mf.proto.sha256`; needs `protoc` and
+  `protoc-gen-go`, and is the only thing that regenerates the committed
+  `mfer/mf.pb.go`
 - `script/fuzz` — fuzz the manifest parser for one minute; run by hand
   (`make fuzz`), never by CI, while `script/test` runs its committed seed corpus
   as ordinary tests

go.mod

@@ -12,13 +12,13 @@ require (
 	github.com/pterm/pterm v0.12.35
 	github.com/spf13/afero v1.8.0
 	github.com/stretchr/testify v1.8.1
-	github.com/urfave/cli/v2 v2.23.6
+	github.com/urfave/cli/v2 v2.27.7
 	google.golang.org/protobuf v1.28.1
 )
 
 require (
 	github.com/atomicgo/cursor v0.0.1 // indirect
-	github.com/cpuguy83/go-md2man/v2 v2.0.2 // indirect
+	github.com/cpuguy83/go-md2man/v2 v2.0.7 // indirect
 	github.com/fatih/color v1.7.0 // indirect
 	github.com/gookit/color v1.4.2 // indirect
 	github.com/klauspost/cpuid/v2 v2.0.9 // indirect
@@ -34,7 +34,7 @@ require (
 	github.com/russross/blackfriday/v2 v2.1.0 // indirect
 	github.com/spaolacci/murmur3 v1.1.0 // indirect
 	github.com/xo/terminfo v0.0.0-20210125001918-ca9a967f8778 // indirect
-	github.com/xrash/smetrics v0.0.0-20201216005158-039620a65673 // indirect
+	github.com/xrash/smetrics v0.0.0-20240521201337-686a1a2994c1 // indirect
 	golang.org/x/crypto v0.0.0-20220525230936-793ad666bf5e // indirect
 	golang.org/x/sys v0.1.0 // indirect
 	golang.org/x/term v0.0.0-20210927222741-03fcf44c2211 // indirect

go.sum

@@ -61,8 +61,8 @@ github.com/client9/misspell v0.3.4/go.mod h1:qj6jICC3Q7zFZvVWo7KLAzC3yx5G7kyvSDk
 github.com/cncf/udpa/go v0.0.0-20191209042840-269d4d468f6f/go.mod h1:M8M6+tZqaGXZJjfX53e64911xZQV5JYwmTeXPW+k8Sc=
 github.com/cncf/udpa/go v0.0.0-20200629203442-efcf912fb354/go.mod h1:WmhPx2Nbnhtbo57+VJT5O0JRkEi1Wbu0z5j0R8u5Hbk=
 github.com/cncf/udpa/go v0.0.0-20201120205902-5459f2c99403/go.mod h1:WmhPx2Nbnhtbo57+VJT5O0JRkEi1Wbu0z5j0R8u5Hbk=
-github.com/cpuguy83/go-md2man/v2 v2.0.2 h1:p1EgwI/C7NhT0JmVkwCD2ZBK8j4aeHQX2pMHHBfMQ6w=
+github.com/cpuguy83/go-md2man/v2 v2.0.7 h1:zbFlGlXEAKlwXpmvle3d8Oe3YnkKIK4xSRTd3sHPnBo=
-github.com/cpuguy83/go-md2man/v2 v2.0.2/go.mod h1:tgQtvFlXSQOSOSIRvRPT7W67SCa46tRHOmNcaadrF8o=
+github.com/cpuguy83/go-md2man/v2 v2.0.7/go.mod h1:oOW0eioCTA6cOiMLiUPZOpcVxMig6NIQQ7OS05n1F4g=
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
@@ -231,12 +231,12 @@ github.com/tj/go-buffer v1.1.0/go.mod h1:iyiJpfFcR2B9sXu7KvjbT9fpM4mOelRSDTbntVj
 github.com/tj/go-elastic v0.0.0-20171221160941-36157cbbebc2/go.mod h1:WjeM0Oo1eNAjXGDx2yma7uG2XoyRZTq1uv3M/o7imD0=
 github.com/tj/go-kinesis v0.0.0-20171128231115-08b17f58cb1b/go.mod h1:/yhzCV0xPfx6jb1bBgRFjl5lytqVqZXEaeqWP8lTEao=
 github.com/tj/go-spin v1.1.0/go.mod h1:Mg1mzmePZm4dva8Qz60H2lHwmJ2loum4VIrLgVnKwh4=
-github.com/urfave/cli/v2 v2.23.6 h1:iWmtKD+prGo1nKUtLO0Wg4z9esfBM4rAV4QRLQiEmJ4=
+github.com/urfave/cli/v2 v2.27.7 h1:bH59vdhbjLv3LAvIu6gd0usJHgoTTPhCFib8qqOwXYU=
-github.com/urfave/cli/v2 v2.23.6/go.mod h1:GHupkWPMM0M/sj1a2b4wUrWBPzazNrIjouW6fmdJLxc=
+github.com/urfave/cli/v2 v2.27.7/go.mod h1:CyNAG/xg+iAOg0N4MPGZqVmv2rCoP267496AOXUZjA4=
 github.com/xo/terminfo v0.0.0-20210125001918-ca9a967f8778 h1:QldyIu/L63oPpyvQmHgvgickp1Yw510KJOqX7H24mg8=
 github.com/xo/terminfo v0.0.0-20210125001918-ca9a967f8778/go.mod h1:2MuV+tbUrU1zIOPMxZ5EncGwgmMJsa+9ucAQZXxsObs=
-github.com/xrash/smetrics v0.0.0-20201216005158-039620a65673 h1:bAn7/zixMGCfxrRTfdpNzjtPYqr8smhKouy9mxVdGPU=
+github.com/xrash/smetrics v0.0.0-20240521201337-686a1a2994c1 h1:gEOO8jv9F4OT7lGCjxCBTO/36wtF6j2nSip77qHd4x4=
-github.com/xrash/smetrics v0.0.0-20201216005158-039620a65673/go.mod h1:N3UwUGtsrSj3ccvlPHLoLsHnpR27oXr4ZE984MbSER8=
+github.com/xrash/smetrics v0.0.0-20240521201337-686a1a2994c1/go.mod h1:Ohn+xnUBiLI6FVj/9LpzZWtj1/D6lUovWYBkxHVV3aM=
 github.com/yuin/goldmark v1.1.25/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.1.27/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.1.32/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=

internal/cli/entry_test.go

@@ -8,6 +8,8 @@ import (
 	"io"
 	"math/rand"
 	"os"
+	"slices"
+	"strings"
 	"sync"
 	"testing"
 
@@ -30,6 +32,7 @@ const (
 	testFlagBase    = "--base"
 	testFlagNoExtra = "--no-extra-files"
 	testFlagVersion = "--version"
+	testFlagVerbose = "--verbose"
 )
 
 var errSimulatedWrite = errors.New("simulated write failure")
@@ -42,20 +45,32 @@ var errSimulatedWrite = errors.New("simulated write failure")
 var runMu sync.Mutex
 
 // runCLI invokes RunWithOptions while holding runMu so parallel tests
-// capture their own output. Before releasing the lock it points the
+// capture their own output, and returns its exit code.
-// process-global logger at io.Discard: other tests log outside the lock
-// (manifest loads, scans), and those lines must not land in this run's
-// buffers once it has returned and its test is reading them.
 func runCLI(opts *RunOptions) int {
+	exitCode, _ := runCLIWithLevel(opts)
+
+	return exitCode
+}
+
+// runCLIWithLevel is runCLI that also returns the log level the run left
+// set, read while runMu still keeps other runs from changing it. Each run
+// starts at the default level, as a new process does. Before releasing the
+// lock it points the process-global logger at io.Discard: other tests log
+// outside the lock (manifest loads, scans), and those lines must not land in
+// this run's buffers once it has returned and its test is reading them.
+func runCLIWithLevel(opts *RunOptions) (int, log.Level) {
 	runMu.Lock()
 	defer runMu.Unlock()
 
+	log.SetLevel(log.InfoLevel)
+
 	exitCode := RunWithOptions(opts)
+	level := log.GetLevel()
 
 	log.SetOutput(io.Discard, io.Discard)
 	log.Init()
 
-	return exitCode
+	return exitCode, level
 }
 
 func TestMain(m *testing.M) {
@@ -235,6 +250,90 @@ func TestRootVerbosityFlags(t *testing.T) {
 	})
 }
 
+// commandsTakingVerbose returns the command lines -v can follow: the root and
+// the generate, check, freshen and fetch subcommands.
+func commandsTakingVerbose() [][]string {
+	return [][]string{
+		{testApp},
+		{testApp, cmdGenerate},
+		{testApp, cmdCheck},
+		{testApp, cmdFreshen},
+		{testApp, cmdFetch},
+	}
+}
+
+// TestVerboseCount asserts that one -v or --verbose gives verbose output and
+// two -v give debug output (issue #125). urfave/cli before v2.25.5 counted a
+// flag given by its alias twice, so one -v gave debug output.
+func TestVerboseCount(t *testing.T) {
+	t.Parallel()
+
+	cases := []struct {
+		flags []string
+		want  log.Level
+	}{
+		{[]string{"-v"}, log.VerboseLevel},
+		{[]string{testFlagVerbose}, log.VerboseLevel},
+		{[]string{"-v", "-v"}, log.DebugLevel},
+	}
+
+	for _, command := range commandsTakingVerbose() {
+		for _, tc := range cases {
+			args := slices.Concat(command, tc.flags)
+
+			t.Run(strings.Join(args, " "), func(t *testing.T) {
+				t.Parallel()
+
+				_, level := runCLIWithLevel(testOpts(args, afero.NewMemMapFs()))
+				assert.Equal(t, tc.want, level)
+			})
+		}
+	}
+}
+
+// TestCombinedShortVerboseRefused asserts that -vv is refused (issue #125):
+// single-letter flags do not combine, so the -v help text says -v -v.
+func TestCombinedShortVerboseRefused(t *testing.T) {
+	t.Parallel()
+
+	for _, command := range commandsTakingVerbose() {
+		args := slices.Concat(command, []string{"-vv"})
+
+		t.Run(strings.Join(args, " "), func(t *testing.T) {
+			t.Parallel()
+
+			opts := testOpts(args, afero.NewMemMapFs())
+			exitCode, level := runCLIWithLevel(opts)
+
+			assert.Equal(t, 1, exitCode)
+			assert.Contains(t, testStderr(t, opts), "flag provided but not defined: -vv")
+			assert.Equal(t, log.InfoLevel, level)
+		})
+	}
+}
+
+// TestShortAndLongVerboseRefused asserts that -v and --verbose given together
+// are refused (issue #125): urfave/cli v2 refuses a flag given under two of its
+// names, and one flag with an alias keeps help and parsing simple.
+func TestShortAndLongVerboseRefused(t *testing.T) {
+	t.Parallel()
+
+	for _, command := range commandsTakingVerbose() {
+		args := slices.Concat(command, []string{"-v", testFlagVerbose})
+
+		t.Run(strings.Join(args, " "), func(t *testing.T) {
+			t.Parallel()
+
+			opts := testOpts(args, afero.NewMemMapFs())
+			exitCode, level := runCLIWithLevel(opts)
+
+			assert.Equal(t, 1, exitCode)
+			assert.Contains(t, testStderr(t, opts), "Cannot use two forms of the same flag")
+			assert.Equal(t, log.InfoLevel, level)
+		})
+	}
+}
+
 func TestHelpCommand(t *testing.T) {
 	t.Parallel()
 

internal/cli/mfer.go

@@ -17,6 +17,7 @@ import (
 const (
 	cmdGenerate = "generate"
 	cmdCheck    = "check"
+	cmdFreshen  = "freshen"
 	cmdExport   = "export"
 	cmdFetch    = "fetch"
 	cmdVersion  = "version"
@@ -118,7 +119,7 @@ func commonFlags() []cli.Flag {
 		&cli.BoolFlag{
 			Name:    "verbose",
 			Aliases: []string{"v"},
-			Usage:   "Increase verbosity (-v for verbose, -vv for debug)",
+			Usage:   "Increase verbosity (-v for verbose, -v -v for debug)",
 			Count:   new(int),
 		},
 		&cli.BoolFlag{
@@ -131,9 +132,10 @@ func commonFlags() []cli.Flag {
 
 func (mfa *CLIApp) generateCommand() *cli.Command {
 	return &cli.Command{
-		Name:    cmdGenerate,
+		Name:      cmdGenerate,
-		Aliases: []string{"gen"},
+		Aliases:   []string{"gen"},
-		Usage:   "Generate manifest file",
+		Usage:     "Generate manifest file",
+		ArgsUsage: "[path ...]",
 		Action: func(c *cli.Context) error {
 			mfa.setVerbosity(c)
 			mfa.printBanner()
@@ -227,7 +229,7 @@ func (mfa *CLIApp) checkCommand() *cli.Command {
 
 func (mfa *CLIApp) freshenCommand() *cli.Command {
 	return &cli.Command{
-		Name:      "freshen",
+		Name:      cmdFreshen,
 		Usage:     "Update manifest with changed, new, and removed files",
 		ArgsUsage: manifestArgsUsage,
 		Action: func(c *cli.Context) error {
@@ -324,8 +326,9 @@ func (mfa *CLIApp) listCommand() *cli.Command {
 
 func (mfa *CLIApp) fetchCommand() *cli.Command {
 	return &cli.Command{
-		Name:  cmdFetch,
+		Name:      cmdFetch,
-		Usage: "fetch manifest and referenced files",
+		Usage:     "fetch manifest and referenced files",
+		ArgsUsage: "URL",
 		Action: func(c *cli.Context) error {
 			mfa.setVerbosity(c)
 			mfa.printBanner()

mfer/constants.go

@@ -17,4 +17,27 @@ const (
 
 	// uuidLength is the length in bytes of a binary UUID.
 	uuidLength = 16
+
+	// Numbers in mf.proto of MFFile.files and of the MFFilePath fields
+	// that decoding sets aside a fixed amount of memory for.
+	filesFieldNumber    = 101
+	hashesFieldNumber   = 3
+	mimeTypeFieldNumber = 301
+	mtimeFieldNumber    = 302
+	ctimeFieldNumber    = 303
+
+	// Bytes decoding sets aside for each file entry, hash, timestamp and
+	// MIME type, however short its encoding. checkDecodedSize refuses an
+	// inner message for which these add up to more than maxDecodedGrowth
+	// times its size.
+	decodedFileEntrySize = 160
+	decodedHashSize      = 112
+	decodedTimestampSize = 64
+	decodedMIMETypeSize  = 16
+
+	// Each file entry mfer writes holds a path of at least one byte, a
+	// 34-byte SHA-256 multihash and a modification time: at least 47 bytes,
+	// counted at 336. So its manifests add up to at most about 7.15 times
+	// their size, and this limit is about 12% above that.
+	maxDecodedGrowth = 8
 )

mfer/deserialize.go

@@ -11,6 +11,7 @@ import (
 	"github.com/google/uuid"
 	"github.com/klauspost/compress/zstd"
 	"github.com/spf13/afero"
+	"google.golang.org/protobuf/encoding/protowire"
 	"google.golang.org/protobuf/proto"
 	"sneak.berlin/go/mfer/internal/bork"
 	"sneak.berlin/go/mfer/internal/log"
@@ -27,6 +28,8 @@ var (
 	errUUIDMismatch         = errors.New("outer and inner UUID mismatch")
 	errInvalidFileFormat    = errors.New("invalid file format")
 	errInvalidManifestPath  = errors.New("manifest contains invalid path")
+	errDecodedTooLarge      = errors.New(
+		"manifest would take too much memory to decode")
 )
 
 // validateUUID checks that the byte slice is a valid UUID (16 bytes, parseable).
@@ -154,6 +157,85 @@ func (m *manifest) decompressInner() ([]byte, error) {
 	return dat, nil
 }
 
+// checkDecodedSize refuses an encoded inner message whose file entries,
+// hashes, timestamps and MIME types would take more than maxDecodedGrowth
+// times its size to decode. Decoding sets aside a fixed amount for each,
+// however short its encoding, so a message of empty ones would take about
+// 50 times its size.
+func checkDecodedSize(inner []byte) error {
+	limit := maxDecodedGrowth * int64(len(inner))
+
+	var decoded int64
+
+	add := func(size int64) error {
+		decoded += size
+		if decoded > limit {
+			return errDecodedTooLarge
+		}
+
+		return nil
+	}
+
+	return forEachBytesField(inner, func(num protowire.Number, entry []byte) error {
+		if num != filesFieldNumber {
+			return nil
+		}
+
+		err := add(decodedFileEntrySize)
+		if err != nil {
+			return err
+		}
+
+		return forEachBytesField(entry, func(num protowire.Number, _ []byte) error {
+			if num == hashesFieldNumber {
+				return add(decodedHashSize)
+			}
+
+			if num == mtimeFieldNumber || num == ctimeFieldNumber {
+				return add(decodedTimestampSize)
+			}
+
+			if num == mimeTypeFieldNumber {
+				return add(decodedMIMETypeSize)
+			}
+
+			return nil
+		})
+	})
+}
+
+// forEachBytesField calls fn with the number and value of each
+// length-delimited field in the encoded message msg, and fails if msg is
+// malformed.
+func forEachBytesField(
+	msg []byte, fn func(num protowire.Number, value []byte) error,
+) error {
+	for len(msg) > 0 {
+		num, wireType, tagLen := protowire.ConsumeTag(msg)
+		if tagLen < 0 {
+			return protowire.ParseError(tagLen)
+		}
+
+		valueLen := protowire.ConsumeFieldValue(num, wireType, msg[tagLen:])
+		if valueLen < 0 {
+			return protowire.ParseError(valueLen)
+		}
+
+		if wireType == protowire.BytesType {
+			value, _ := protowire.ConsumeBytes(msg[tagLen:])
+
+			err := fn(num, value)
+			if err != nil {
+				return err
+			}
+		}
+
+		msg = msg[tagLen+valueLen:]
+	}
+
+	return nil
+}
+
 func (m *manifest) deserializeInner() error {
 	err := m.validateOuterHeader()
 	if err != nil {
@@ -177,10 +259,16 @@ func (m *manifest) deserializeInner() error {
 		return bork.ErrFileTruncated
 	}
 
+	err = checkDecodedSize(dat)
+	if err != nil {
+		return fmt.Errorf("deserialize: unmarshal inner: %w", err)
+	}
+
 	// Deserialize inner message
 	m.pbInner = new(MFFile)
 
-	err = proto.Unmarshal(dat, m.pbInner)
+	// Unknown fields would cost memory; mfer never writes a loaded manifest out.
+	err = proto.UnmarshalOptions{DiscardUnknown: true}.Unmarshal(dat, m.pbInner)
 	if err != nil {
 		return fmt.Errorf("deserialize: unmarshal inner: %w", err)
 	}
@@ -249,7 +337,8 @@ func NewManifestFromReader(input io.Reader) (*manifest, error) {
 	// deserialize outer:
 	m.pbOuter = new(MFFileOuter)
 
-	err = proto.Unmarshal(dat, m.pbOuter)
+	// Unknown fields would cost memory; mfer never writes a loaded manifest out.
+	err = proto.UnmarshalOptions{DiscardUnknown: true}.Unmarshal(dat, m.pbOuter)
 	if err != nil {
 		return nil, err
 	}

mfer/deserialize_fuzz_test.go

@@ -54,9 +54,15 @@ func FuzzNewManifestFromReader(f *testing.F) {
 		}
 
 		// It also keeps a few copies of its input. Buffers grow by
-		// copying, so reaching those sizes allocates a few times them in
+		// copying, so reaching those sizes allocates up to about six times
-		// total: sixteen times the input and the decompressed data leaves
+		// them in total. Decoding the decompressed data takes up to
-		// room for that.
+		// maxDecodedGrowth times its size for file entries, hashes,
+		// timestamps and MIME types, and drops fields it does not know.
+		// The strings and bytes it copies out of it, such as many one-byte
+		// values in one hash, take up to about five times more under the
+		// race detector, which pads every small copy to 16 bytes, and about
+		// half that without it. Twenty times the input and the
+		// decompressed data leaves room for all of that.
 		//
 		// The decoder also sets aside a new buffer of one to two times the
 		// window for each frame that asks for a larger window than the
@@ -71,8 +77,9 @@ func FuzzNewManifestFromReader(f *testing.F) {
 		// fails if the decoder accepts windows of twice zstdWindowSize; the
 		// seed whose two frames together exceed MaxDecompressedSize fails
 		// if the decoder decodes them in full instead of stopping at the
-		// declared size.
+		// declared size; the seeds of empty file entries and of a file
-		limit := 16*(uint64(len(data))+decompressed) + 24*zstdWindowSize
+		// entry of empty hashes fail if the parser decodes them.
+		limit := 20*(uint64(len(data))+decompressed) + 24*zstdWindowSize
 
 		allocated := after.TotalAlloc - before.TotalAlloc
 		if allocated > limit {

mfer/deserialize_path_test.go

@@ -6,7 +6,10 @@ import (
 	"context"
 	"crypto/sha256"
 	"fmt"
+	"strconv"
+	"strings"
 	"testing"
+	"time"
 
 	"github.com/google/uuid"
 	"github.com/klauspost/compress/zstd"
@@ -114,6 +117,113 @@ func TestDeserializeRejectsInvalidEntryPaths(t *testing.T) {
 	}
 }
 
+// Entries of a path, an empty hash, an empty MIME type and empty modification
+// and change times are counted at 416 bytes each (160 + 112 + 16 + 64 + 64)
+// and take 16 bytes plus the path to encode. A 35-character path makes that
+// 51 bytes, about 8.2 times: refused, and leaving any one of the five
+// uncounted, even the MIME type, brings it under 8. A 37-character path makes
+// it 53 bytes, about 7.8 times: loaded.
+func TestDeserializeRefusesEntriesThatDecodeTooLarge(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		pathLen int
+		refused bool
+	}{
+		{35, true},
+		{37, false},
+	}
+
+	for _, tt := range tests {
+		t.Run(strconv.Itoa(tt.pathLen), func(t *testing.T) {
+			t.Parallel()
+
+			entry := protowire.AppendTag(nil, 1, protowire.BytesType) // MFFilePath.path
+			entry = protowire.AppendString(entry, strings.Repeat("a", tt.pathLen))
+			entry = protowire.AppendTag(entry, 3, protowire.BytesType) // MFFilePath.hashes
+			entry = protowire.AppendBytes(entry, nil)
+			entry = protowire.AppendTag(entry, 301, protowire.BytesType) // MFFilePath.mimeType
+			entry = protowire.AppendBytes(entry, nil)
+			entry = protowire.AppendTag(entry, 302, protowire.BytesType) // MFFilePath.mtime
+			entry = protowire.AppendBytes(entry, nil)
+			entry = protowire.AppendTag(entry, 303, protowire.BytesType) // MFFilePath.ctime
+			entry = protowire.AppendBytes(entry, nil)
+
+			id := uuid.New()
+			inner := protowire.AppendTag(nil, 102, protowire.BytesType) // MFFile.uuid
+			inner = protowire.AppendBytes(inner, id[:])
+
+			for range 1000 {
+				inner = protowire.AppendTag(inner, 101, protowire.BytesType) // MFFile.files
+				inner = protowire.AppendBytes(inner, entry)
+			}
+
+			_, err := NewManifestFromReader(bytes.NewReader(wrapInner(t, id, inner)))
+			if tt.refused {
+				require.ErrorIs(t, err, errDecodedTooLarge)
+			} else {
+				require.NoError(t, err)
+			}
+		})
+	}
+}
+
+// Fields the decoder does not know are dropped, in the outer message, the inner
+// message and a file entry, so that they take no memory once loaded.
+func TestDeserializeDropsUnknownFields(t *testing.T) {
+	t.Parallel()
+
+	unknown := protowire.AppendTag(nil, 99, protowire.BytesType) // in no message
+	unknown = protowire.AppendBytes(unknown, []byte("not known"))
+
+	entry := protowire.AppendTag(nil, 1, protowire.BytesType) // MFFilePath.path
+	entry = protowire.AppendString(entry, "a")
+	entry = append(entry, unknown...)
+
+	id := uuid.New()
+	inner := protowire.AppendTag(nil, 101, protowire.BytesType) // MFFile.files
+	inner = protowire.AppendBytes(inner, entry)
+	inner = protowire.AppendTag(inner, 102, protowire.BytesType) // MFFile.uuid
+	inner = protowire.AppendBytes(inner, id[:])
+	inner = append(inner, unknown...)
+
+	data := wrapInner(t, id, inner)
+	data = append(data, unknown...) // the outer message ends the file
+
+	m, err := NewManifestFromReader(bytes.NewReader(data))
+	require.NoError(t, err)
+	require.Len(t, m.Files(), 1)
+	assert.Empty(t, m.pbOuter.ProtoReflect().GetUnknown())
+	assert.Empty(t, m.pbInner.ProtoReflect().GetUnknown())
+	assert.Empty(t, m.Files()[0].ProtoReflect().GetUnknown())
+}
+
+// Many empty files with names of at most three characters and modification
+// times at the epoch make about the densest manifest mfer writes: it takes
+// about 7 times its size to decode, and still loads. A signature would not
+// change the inner message, so none is added.
+func TestDeserializeLoadsDensestManifest(t *testing.T) {
+	t.Parallel()
+
+	hash := make([]byte, 34) // multihash: 2-byte prefix + 32-byte SHA-256
+
+	b := NewBuilder()
+	b.SetIncludeTimestamps(true)
+
+	const files = 10000
+	for i := range files {
+		name := RelFilePath(strconv.FormatInt(int64(i), 36))
+		require.NoError(t, b.AddFileWithHash(name, 0, ModTime(time.Unix(0, 0)), hash))
+	}
+
+	var buf bytes.Buffer
+	require.NoError(t, b.Build(context.Background(), &buf))
+
+	m, err := NewManifestFromReader(&buf)
+	require.NoError(t, err)
+	assert.Len(t, m.Files(), files)
+}
+
 func TestDeserializeValidManifestRoundTrips(t *testing.T) {
 	t.Parallel()
 

mfer/mf.proto.sha256

@@ -0,0 +1 @@
+103901c42b94396aa7ae128fd503ef693a4b7a03b2169481f25fda3d2c254e00  mf.proto

mfer/mf_test.go

@@ -0,0 +1,29 @@
+package mfer_test
+
+import (
+	"crypto/sha256"
+	"encoding/hex"
+	"os"
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+// mf.pb.go is generated from mf.proto and committed. `make generate`
+// records the hash of the mf.proto it generated from in mf.proto.sha256.
+func TestGeneratedCodeMatchesProto(t *testing.T) {
+	t.Parallel()
+
+	proto, err := os.ReadFile("mf.proto")
+	require.NoError(t, err)
+
+	recorded, err := os.ReadFile("mf.proto.sha256")
+	require.NoError(t, err)
+
+	recordedHash, _, _ := strings.Cut(string(recorded), " ")
+	sum := sha256.Sum256(proto)
+	require.Equal(t, recordedHash, hex.EncodeToString(sum[:]),
+		"mfer/mf.proto has changed since mfer/mf.pb.go was generated "+
+			"from it: run `make generate` and commit the result")
+}

mfer/testdata/fuzz/FuzzNewManifestFromReader/empty-file-entries

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("ZNAVSRFG\xa8\x06\x01\xb0\x06\x01\xb8\x06\xff\xff\xff\x03\xc2\x06 {\x16\xbdu\xa0\xa2\x11\xfcH\xef*\x1b7\r\x99\xefb\x04\x02g\n\xa9\xf3B5\xe5p\x96\x8c\x8c\xac\x0e\xca\x06\x10\x03Q\xb2\xd0\x19`F\xc1\xb1\xc0Z\xf4x\xf4g^\xba\f\xa1\x06(\xb5/\xfd\x04h\x04\x01\x00d\x01\xb2\x06\x10\x03Q\xb2\xd0\x19`F\xc1\xb1\xc0Z\xf4x\xf4g^\xaa\x06\x00\x01T\x13\x024\xce\xff\rL\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15M\x00\x00\x00\x01T\x00\x044\xfc\xff\x153\xea\a\xb4")

mfer/testdata/fuzz/FuzzNewManifestFromReader/file-entry-of-empty-hashes

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("ZNAVSRFG\xa8\x06\x01\xb0\x06\x01\xb8\x06\xff\xff\xff\x03\xc2\x06 .\xcd\x11|0\xfcP\xe5\x1b\xe3\xc6Ӡ\xcdڤx\xcd\x169t\x1a9~ǽB\xc9\xe8G`\x05\xca\x06\x10\x11*!\x0e\x95EF\xb8\xbd\x9f\xde\x12MF\r\x99\xba\f\xa6\x06(\xb5/\xfd\x04h,\x01\x00\xb4\x01\xb2\x06\x10\x11*!\x0e\x95EF\xb8\xbd\x9f\xde\x12MF\r\x99\xaa\x06\xe6\xff\xff\x03\x1a\x00\x01T\x14\x024\x8b\xff\x17L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15L\x00\x00\x00\x01T\x00\x044\xfd\xff\x15M\x00\x00\x00\x01T\x00\x044\xfc\xff\x15\x02\xd1.\xe3")

script/fmt

@@ -5,19 +5,8 @@ set -eu
 SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
 ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
 
-# Regenerate mfer/mf.pb.go from mfer/mf.proto if it is missing or stale
-# (mirrors the old Makefile prerequisite; the generated file is
-# committed, so this is normally a no-op).
-ensure_pb() {
-    if [ ! -f mfer/mf.pb.go ] ||
-        [ -n "$(find mfer/mf.proto -newer mfer/mf.pb.go 2>/dev/null)" ]; then
-        (cd mfer && go generate .)
-    fi
-}
-
 main() {
     cd "$ROOT"
-    ensure_pb
     gofumpt -l -w mfer internal cmd
     # Markdown and JSON, over the same file set script/fmt-check verifies.
     "$SCRIPT_DIR/prettier" --write

script/fmt-check-go

@@ -6,19 +6,8 @@ set -eu
 
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 
-# Regenerate mfer/mf.pb.go from mfer/mf.proto if it is missing or stale
-# (mirrors the old Makefile prerequisite; the generated file is
-# committed, so this is normally a no-op).
-ensure_pb() {
-    if [ ! -f mfer/mf.pb.go ] ||
-        [ -n "$(find mfer/mf.proto -newer mfer/mf.pb.go 2>/dev/null)" ]; then
-        (cd mfer && go generate .)
-    fi
-}
-
 main() {
     cd "$ROOT"
-    ensure_pb
     if [ -n "$(gofmt -l .)" ]; then
         echo "gofmt: files need formatting:" >&2
         gofmt -l . >&2

script/generate

@@ -0,0 +1,17 @@
+#!/bin/sh
+# script/generate: regenerate mfer/mf.pb.go from mfer/mf.proto, and record
+# the hash of that mf.proto in mfer/mf.proto.sha256. Nothing else
+# regenerates mf.pb.go: it is committed, so building and checking need no
+# protoc. Needs protoc and protoc-gen-go on PATH. A test fails while
+# mf.proto no longer matches the recorded hash.
+set -eu
+
+ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
+
+main() {
+    cd "$ROOT/mfer"
+    go generate .
+    shasum -a 256 mf.proto >mf.proto.sha256
+}
+
+main "$@"

script/test

@@ -4,19 +4,8 @@ set -eu
 
 ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
 
-# Regenerate mfer/mf.pb.go from mfer/mf.proto if it is missing or stale
-# (mirrors the old Makefile prerequisite; the generated file is
-# committed, so this is normally a no-op).
-ensure_pb() {
-    if [ ! -f mfer/mf.pb.go ] ||
-        [ -n "$(find mfer/mf.proto -newer mfer/mf.pb.go 2>/dev/null)" ]; then
-        (cd mfer && go generate .)
-    fi
-}
-
 main() {
     cd "$ROOT"
-    ensure_pb
     go test -timeout 30s -race -cover ./... ||
         {
             echo "--- Rerunning with -v for details ---"