Open
Description
Go version
go1.24.2 linux/amd64
Output of go env in your module/workspace:
AR='ar'
CC='gcc'
CGO_CFLAGS='-O2 -g'
CGO_CPPFLAGS=''
CGO_CXXFLAGS='-O2 -g'
CGO_ENABLED='1'
CGO_FFLAGS='-O2 -g'
CGO_LDFLAGS='-O2 -g'
CXX='g++'
GCCGO='gccgo'
GO111MODULE=''
GOAMD64='v1'
GOARCH='amd64'
GOAUTH='netrc'
GOBIN=''
GOCACHE='/home/user/.cache/go-build'
GOCACHEPROG=''
GODEBUG=''
GOENV='/home/user/.config/go/env'
GOEXE=''
GOEXPERIMENT=''
GOFIPS140='off'
GOFLAGS=''
GOGCCFLAGS='-fPIC -m64 -pthread -Wl,--no-gc-sections -fmessage-length=0 -ffile-prefix-map=/tmp/go-build114624775=/tmp/go-build -gno-record-gcc-switches'
GOHOSTARCH='amd64'
GOHOSTOS='linux'
GOINSECURE=''
GOMOD='/home/user/go/src/github.com/holiman/goevmlab/go.mod'
GOMODCACHE='/home/user/go/pkg/mod'
GONOPROXY=''
GONOSUMDB=''
GOOS='linux'
GOPATH='/home/user/go'
GOPRIVATE=''
GOPROXY='https://proxy.golang.org,direct'
GOROOT='/usr/local/go'
GOSUMDB='sum.golang.org'
GOTELEMETRY='local'
GOTELEMETRYDIR='/home/user/.config/go/telemetry'
GOTMPDIR=''
GOTOOLCHAIN='auto'
GOTOOLDIR='/usr/local/go/pkg/tool/linux_amd64'
GOVCS=''
GOVERSION='go1.24.2'
GOWORK=''
PKG_CONFIG='pkg-config'What did you do?
I want to build a fuzztest as a standalone binary, for usage in a docker container. However,
the binary complains that "the test binary was not built with coverage instrumentation, so fuzzing will run without coverage guidance and may be inefficient".
I want to enable the coverage instrumentation. Repro steps, below.
Example test:
package main
import (
"bytes"
"crypto/sha256"
"testing"
)
func FuzzFoobar(f *testing.F) {
f.Fuzz(func(t *testing.T, data []byte) {
// Invoke some arbitrary methods
data = sha256.New().Sum(bytes.ToLower(data))
if len(data) == 0 {
panic(1)
}
})
}Running it regularly works fine:
$ go test . -fuzz Foo
fuzz: elapsed: 0s, gathering baseline coverage: 0/66 completed
fuzz: elapsed: 0s, gathering baseline coverage: 66/66 completed, now fuzzing with 8 workers
^Cfuzz: elapsed: 3s, execs: 107197 (39642/sec), new interesting: 8 (total: 74)
Compiling it yields complaint:
$ go test -c .
$ ./testfuzz.test -test.fuzz=Foo -test.fuzzcachedir=/tmp
warning: the test binary was not built with coverage instrumentation, so fuzzing will run without coverage guidance and may be inefficient
warning: starting with empty corpus
fuzz: elapsed: 0s, execs: 0 (0/sec)
If I try to use the -cover when compiling, it does not help:
$ go test -cover -c .
$ ./testfuzz.test -test.fuzz=Foo -test.fuzzcachedir=/tmp
warning: the test binary was not built with coverage instrumentation, so fuzzing will run without coverage guidance and may be inefficient
warning: starting with empty corpus
fuzz: elapsed: 0s, execs: 0 (0/sec)
However, if I add -coverpkg, then something happens. (This is a bit odd, because without specifying coverpkg, it should target all packages):
$ go test -cover -coverpkg=main,crypto/sha256,bytes,testing -c . && ./testfuzz.test -test.fuzz=Foo -test.fuzzcachedir=/tmp
warning: no packages being tested depend on matches for pattern main
warning: the test binary was not built with coverage instrumentation, so fuzzing will run without coverage guidance and may be inefficient
warning: starting with empty corpus
fuzz: elapsed: 0s, execs: 0 (0/sec)
^Cfuzz: elapsed: 3s, execs: 461777 (173901/sec)
PASS
coverage: 20.1% of statements in main, crypto/sha256, bytes, testing
It still complains about lack of coverage instrumentation, but it also emits coverage data afterwards...
Also, if I specify a too large -coverpkg expression, it leads to some internal crash:
$ go test -cover -coverpkg=... -c . && ./testfuzz.test -test.fuzz=Foo -test.fuzzcachedir=/tmp
warning: the test binary was not built with coverage instrumentation, so fuzzing will run without coverage guidance and may be inefficient
warning: starting with empty corpus
fuzz: elapsed: 0s, execs: 0 (0/sec)
^Cfuzz: elapsed: 1s, execs: 107855 (103023/sec)
PASS
internal error in coverage meta-data tracking:
encountered bad pkgID: 0 at slot: 488 fnID: 6 numCtrs: 1
list of hard-coded runtime package IDs needs revising.
[see the comment on the 'rtPkgs' var in
<goroot>/src/internal/coverage/pkid.go]
registered list:
slot: 0 path='internal/asan'
slot: 1 path='internal/byteorder'
slot: 2 path='internal/coverage/rtcov'
slot: 3 path='internal/cpu' hard-coded id: -2
slot: 4 path='internal/bytealg' hard-coded id: -11
slot: 5 path='internal/goarch' hard-coded id: -3
slot: 6 path='internal/abi' hard-coded id: -8
slot: 7 path='internal/chacha8rand' hard-coded id: -6
slot: 8 path='internal/godebugs'
slot: 9 path='internal/goexperiment' hard-coded id: -12
slot: 10 path='internal/goos' hard-coded id: -5
slot: 11 path='internal/msan'
slot: 12 path='internal/profilerecord'
slot: 13 path='internal/race'
slot: 14 path='internal/runtime/atomic' hard-coded id: -4
slot: 15 path='internal/runtime/exithook'
slot: 16 path='internal/runtime/math' hard-coded id: -10
slot: 17 path='internal/runtime/sys' hard-coded id: -7
slot: 18 path='internal/runtime/maps' hard-coded id: -9
slot: 19 path='internal/runtime/syscall' hard-coded id: -13
slot: 20 path='internal/stringslite' hard-coded id: -14
slot: 21 path='runtime' hard-coded id: -15
slot: 22 path='cmp'
slot: 23 path='crypto/internal/boring/sig'
slot: 24 path='crypto/internal/fips140/alias'
slot: 25 path='crypto/internal/fips140/subtle'
slot: 26 path='encoding'
slot: 27 path='crypto/internal/fips140deps/byteorder'
slot: 28 path='internal/coverage'
slot: 29 path='internal/coverage/calloc'
slot: 30 path='internal/coverage/uleb128'
slot: 31 path='crypto/internal/fips140deps/cpu'
slot: 32 path='internal/itoa'
slot: 33 path='internal/unsafeheader'
slot: 34 path='math/bits'
slot: 35 path='math'
slot: 36 path='crypto/subtle'
slot: 37 path='internal/reflectlite'
slot: 38 path='errors'
slot: 39 path='internal/oserror'
slot: 40 path='iter'
slot: 41 path='maps'
slot: 42 path='math/rand/v2'
slot: 43 path='slices'
slot: 44 path='sort'
slot: 45 path='sync/atomic'
slot: 46 path='internal/sync'
slot: 47 path='sync'
slot: 48 path='internal/bisect'
slot: 49 path='internal/godebug'
slot: 50 path='crypto/internal/fips140deps/godebug'
slot: 51 path='internal/testlog'
slot: 52 path='io'
slot: 53 path='crypto/internal/randutil'
slot: 54 path='hash'
slot: 55 path='hash/fnv'
slot: 56 path='math/rand'
slot: 57 path='syscall'
slot: 58 path='internal/syscall/execenv'
slot: 59 path='internal/syscall/unix'
slot: 60 path='time'
slot: 61 path='context'
slot: 62 path='internal/poll'
slot: 63 path='unicode'
slot: 64 path='unicode/utf16'
slot: 65 path='unicode/utf8'
slot: 66 path='bytes'
slot: 67 path='path'
slot: 68 path='io/fs'
slot: 69 path='internal/filepathlite'
slot: 70 path='os'
slot: 71 path='crypto/internal/sysrand'
slot: 72 path='crypto/internal/entropy'
slot: 73 path='strconv'
slot: 74 path='crypto'
slot: 75 path='reflect'
slot: 76 path='encoding/binary'
slot: 77 path='encoding/base64'
slot: 78 path='internal/fmtsort'
slot: 79 path='fmt'
slot: 80 path='internal/coverage/cmerge'
slot: 81 path='internal/coverage/slicereader'
slot: 82 path='internal/coverage/slicewriter'
slot: 83 path='internal/coverage/stringtab'
slot: 84 path='internal/coverage/decodecounter'
slot: 85 path='runtime/trace'
slot: 86 path='strings'
slot: 87 path='bufio'
slot: 88 path='crypto/internal/fips140'
slot: 89 path='crypto/internal/fips140/sha3'
slot: 90 path='crypto/internal/impl'
slot: 91 path='crypto/internal/fips140/sha256'
slot: 92 path='crypto/internal/fips140/sha512'
slot: 93 path='crypto/internal/fips140/hmac'
slot: 94 path='crypto/internal/fips140/check'
slot: 95 path='crypto/internal/fips140/aes'
slot: 96 path='crypto/internal/fips140/drbg'
slot: 97 path='crypto/internal/fips140/aes/gcm'
slot: 98 path='crypto/internal/fips140only'
slot: 99 path='crypto/cipher'
slot: 100 path='crypto/internal/boring'
slot: 101 path='crypto/sha256'
slot: 102 path='encoding/json'
slot: 103 path='flag'
slot: 104 path='internal/coverage/decodemeta'
slot: 105 path='internal/coverage/encodecounter'
slot: 106 path='internal/coverage/encodemeta'
slot: 107 path='internal/sysinfo'
slot: 108 path='path/filepath'
slot: 109 path='regexp/syntax'
slot: 110 path='regexp'
slot: 111 path='internal/coverage/pods'
slot: 112 path='runtime/debug'
slot: 113 path='testing'
slot: 114 path='text/tabwriter'
slot: 115 path='internal/coverage/cformat'
slot: 116 path='internal/coverage/cfile'
slot: 117 path='runtime/coverage'
remap table:
from -4 to 14
from -7 to 17
from -9 to 18
from -15 to 21
from -11 to 4
from -3 to 5
from -6 to 7
from -12 to 9
from -5 to 10
from -10 to 16
from -13 to 19
from -14 to 20
from -2 to 3
from -8 to 6
coverage: 24.0% of statements in ...
What did you see happen?
Standalone fuzzing-binaries are less efficient than fuzz-tests executed via go test.
What did you expect to see?
I'd like to be able to compile fuzz-tests as standalone binaries, to execute in docker-containers which do not have the go runtime.