|
|
|
|
// Code generated by scripts/gen_go_std_tables.go; DO NOT EDIT.
|
|
|
|
|
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
// Generated from Go versions [go1.25.0].
|
|
|
|
|
|
|
|
|
|
package main
|
|
|
|
|
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
// runtimeAndDeps contains the runtime package and all of its transitive dependencies
|
|
|
|
|
// as reported by 'go list -deps'.
|
|
|
|
|
var runtimeAndDeps = map[string]bool{
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"internal/abi": true, // go1.25
|
|
|
|
|
"internal/asan": true, // go1.25
|
|
|
|
|
"internal/bytealg": true, // go1.25
|
|
|
|
|
"internal/byteorder": true, // go1.25
|
|
|
|
|
"internal/chacha8rand": true, // go1.25
|
|
|
|
|
"internal/coverage/rtcov": true, // go1.25
|
|
|
|
|
"internal/cpu": true, // go1.25
|
|
|
|
|
"internal/goarch": true, // go1.25
|
|
|
|
|
"internal/godebugs": true, // go1.25
|
|
|
|
|
"internal/goexperiment": true, // go1.25
|
|
|
|
|
"internal/goos": true, // go1.25
|
|
|
|
|
"internal/msan": true, // go1.25
|
|
|
|
|
"internal/profilerecord": true, // go1.25
|
|
|
|
|
"internal/race": true, // go1.25
|
|
|
|
|
"internal/runtime/atomic": true, // go1.25
|
|
|
|
|
"internal/runtime/cgroup": true, // go1.25
|
|
|
|
|
"internal/runtime/exithook": true, // go1.25
|
|
|
|
|
"internal/runtime/gc": true, // go1.25
|
|
|
|
|
"internal/runtime/maps": true, // go1.25
|
|
|
|
|
"internal/runtime/math": true, // go1.25
|
|
|
|
|
"internal/runtime/strconv": true, // go1.25
|
|
|
|
|
"internal/runtime/sys": true, // go1.25
|
|
|
|
|
"internal/runtime/syscall": true, // go1.25
|
|
|
|
|
"internal/stringslite": true, // go1.25
|
|
|
|
|
"internal/trace/tracev2": true, // go1.25
|
|
|
|
|
"runtime": true, // go1.25
|
|
|
|
|
"unsafe": true, // go1.25
|
|
|
|
|
}
|
|
|
|
|
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
// runtimeAndLinknamed contains the runtime package and all the packages
|
|
|
|
|
// which it points to via //go:linkname directives.
|
|
|
|
|
// We need to track these as some are not imported as transitive dependencies,
|
|
|
|
|
// and we need to load these to properly obfuscate the linkname target names.
|
|
|
|
|
//
|
|
|
|
|
// Note that runtimeAndLinknamed may contain duplicates with runtimeAndDeps.
|
|
|
|
|
// This is on purpose; some packages are in runtimeAndDeps via 'go list -deps'
|
|
|
|
|
// but not transitively imported on some platforms, even though they are used
|
|
|
|
|
// from the runtime package via //go:linkname directives on those platforms.
|
|
|
|
|
// To make sure we have coverage on all platforms, we allow duplicates.
|
|
|
|
|
var runtimeAndLinknamed = []string{
|
|
|
|
|
"arena", // go1.25
|
|
|
|
|
"crypto/internal/boring", // go1.25
|
|
|
|
|
"crypto/internal/boring/bcache", // go1.25
|
|
|
|
|
"crypto/internal/fips140", // go1.25
|
|
|
|
|
"crypto/internal/sysrand", // go1.25
|
|
|
|
|
"crypto/rand", // go1.25
|
|
|
|
|
"crypto/x509/internal/macos", // go1.25
|
|
|
|
|
"internal/bytealg", // go1.25
|
|
|
|
|
"internal/coverage/cfile", // go1.25
|
|
|
|
|
"internal/cpu", // go1.25
|
|
|
|
|
"internal/godebug", // go1.25
|
|
|
|
|
"internal/poll", // go1.25
|
|
|
|
|
"internal/race", // go1.25
|
|
|
|
|
"internal/reflectlite", // go1.25
|
|
|
|
|
"internal/runtime/atomic", // go1.25
|
|
|
|
|
"internal/runtime/cgroup", // go1.25
|
|
|
|
|
"internal/runtime/maps", // go1.25
|
|
|
|
|
"internal/sync", // go1.25
|
|
|
|
|
"internal/synctest", // go1.25
|
|
|
|
|
"internal/syscall/unix", // go1.25
|
|
|
|
|
"internal/syscall/windows", // go1.25
|
|
|
|
|
"maps", // go1.25
|
|
|
|
|
"os", // go1.25
|
|
|
|
|
"os/signal", // go1.25
|
|
|
|
|
"plugin", // go1.25
|
|
|
|
|
"reflect", // go1.25
|
|
|
|
|
"runtime", // go1.25
|
|
|
|
|
"runtime/debug", // go1.25
|
|
|
|
|
"runtime/metrics", // go1.25
|
|
|
|
|
"runtime/pprof", // go1.25
|
|
|
|
|
"runtime/trace", // go1.25
|
|
|
|
|
"sync", // go1.25
|
|
|
|
|
"sync/atomic", // go1.25
|
|
|
|
|
"sync_test", // go1.25
|
|
|
|
|
"syscall", // go1.25
|
|
|
|
|
"syscall/js", // go1.25
|
|
|
|
|
"time", // go1.25
|
|
|
|
|
"unique", // go1.25
|
|
|
|
|
"weak", // go1.25
|
|
|
|
|
// The net package linknames to the runtime, not the other way around.
|
|
|
|
|
"net",
|
|
|
|
|
// The testing package uses a //go:linkname directive pointing to testing/synctest,
|
|
|
|
|
// but it doesn't import the package, presumably to avoid an import cycle.
|
|
|
|
|
"testing/synctest",
|
|
|
|
|
}
|
avoid breaking intrinsics when obfuscating names
We obfuscate import paths as well as their declared names.
The compiler treats some packages and APIs in special ways,
and the way it detects those is by looking at import paths and names.
In the past, we have avoided obfuscating some names like embed.FS or
reflect.Value.MethodByName for this reason. Otherwise,
go:embed or the linker's deadcode elimination might be broken.
This matching by path and name also happens with compiler intrinsics.
Intrinsics allow the compiler to rewrite some standard library calls
with small and efficient assembly, depending on the target GOARCH.
For example, math/bits.TrailingZeros32 gets replaced with ssa.OpCtz32,
which on amd64 may result in using the TZCNTL instruction.
We never noticed that we were breaking many of these intrinsics.
The intrinsics for funcs declared in the runtime and its dependencies
still worked properly, as we do not obfuscate those packages yet.
However, for other packages like math/bits and sync/atomic,
the intrinsics were being entirely disabled due to obfuscated names.
Skipping intrinsics is particularly bad for performance,
and it also leads to slightly larger binaries:
│ old │ new │
│ bin-B │ bin-B vs base │
Build-16 5.450Mi ± ∞ ¹ 5.333Mi ± ∞ ¹ -2.15% (p=0.029 n=4)
Finally, the main reason we noticed that intrinsics were broken
is that apparently GOARCH=mips fails to link without them,
as some symbols end up being not defined at all.
This patch fixes builds for the MIPS family of architectures.
Rather than building and linking all of std for every GOARCH,
test that intrinsics work by asking the compiler to print which
intrinsics are being applied, and checking that math/bits gets them.
This fix is relatively unfortunate, as it means we stop obfuscating
about 120 function names and a handful of package paths.
However, fixing builds and intrinsics is much more important.
We can figure out better ways to deal with intrinsics in the future.
Fixes #646.
3 years ago
|
|
|
|
|
|
|
|
var compilerIntrinsics = map[string]map[string]bool{
|
|
|
|
|
"internal/runtime/atomic": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"And": true, // go1.25
|
|
|
|
|
"And32": true, // go1.25
|
|
|
|
|
"And64": true, // go1.25
|
|
|
|
|
"And8": true, // go1.25
|
|
|
|
|
"Anduintptr": true, // go1.25
|
|
|
|
|
"Cas": true, // go1.25
|
|
|
|
|
"Cas64": true, // go1.25
|
|
|
|
|
"CasRel": true, // go1.25
|
|
|
|
|
"Casint32": true, // go1.25
|
|
|
|
|
"Casint64": true, // go1.25
|
|
|
|
|
"Casp1": true, // go1.25
|
|
|
|
|
"Casuintptr": true, // go1.25
|
|
|
|
|
"Load": true, // go1.25
|
|
|
|
|
"Load64": true, // go1.25
|
|
|
|
|
"Load8": true, // go1.25
|
|
|
|
|
"LoadAcq": true, // go1.25
|
|
|
|
|
"LoadAcq64": true, // go1.25
|
|
|
|
|
"LoadAcquintptr": true, // go1.25
|
|
|
|
|
"Loadint32": true, // go1.25
|
|
|
|
|
"Loadint64": true, // go1.25
|
|
|
|
|
"Loadp": true, // go1.25
|
|
|
|
|
"Loaduint": true, // go1.25
|
|
|
|
|
"Loaduintptr": true, // go1.25
|
|
|
|
|
"Or": true, // go1.25
|
|
|
|
|
"Or32": true, // go1.25
|
|
|
|
|
"Or64": true, // go1.25
|
|
|
|
|
"Or8": true, // go1.25
|
|
|
|
|
"Oruintptr": true, // go1.25
|
|
|
|
|
"Store": true, // go1.25
|
|
|
|
|
"Store64": true, // go1.25
|
|
|
|
|
"Store8": true, // go1.25
|
|
|
|
|
"StoreRel": true, // go1.25
|
|
|
|
|
"StoreRel64": true, // go1.25
|
|
|
|
|
"StoreReluintptr": true, // go1.25
|
|
|
|
|
"Storeint32": true, // go1.25
|
|
|
|
|
"Storeint64": true, // go1.25
|
|
|
|
|
"StorepNoWB": true, // go1.25
|
|
|
|
|
"Storeuintptr": true, // go1.25
|
|
|
|
|
"Xadd": true, // go1.25
|
|
|
|
|
"Xadd64": true, // go1.25
|
|
|
|
|
"Xaddint32": true, // go1.25
|
|
|
|
|
"Xaddint64": true, // go1.25
|
|
|
|
|
"Xadduintptr": true, // go1.25
|
|
|
|
|
"Xchg": true, // go1.25
|
|
|
|
|
"Xchg64": true, // go1.25
|
|
|
|
|
"Xchg8": true, // go1.25
|
|
|
|
|
"Xchgint32": true, // go1.25
|
|
|
|
|
"Xchgint64": true, // go1.25
|
|
|
|
|
"Xchguintptr": true, // go1.25
|
|
|
|
|
},
|
|
|
|
|
"internal/runtime/maps": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"bitsetFirst": true, // go1.25
|
|
|
|
|
"bitsetLowestSet": true, // go1.25
|
|
|
|
|
"bitsetRemoveBelow": true, // go1.25
|
|
|
|
|
"bitsetShiftOutLowest": true, // go1.25
|
|
|
|
|
"ctrlGroupMatchEmpty": true, // go1.25
|
|
|
|
|
"ctrlGroupMatchEmptyOrDeleted": true, // go1.25
|
|
|
|
|
"ctrlGroupMatchFull": true, // go1.25
|
|
|
|
|
"ctrlGroupMatchH2": true, // go1.25
|
|
|
|
|
},
|
|
|
|
|
"internal/runtime/math": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"MulUintptr": true, // go1.25
|
|
|
|
|
},
|
|
|
|
|
"internal/runtime/sys": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"Bswap32": true, // go1.25
|
|
|
|
|
"Bswap64": true, // go1.25
|
|
|
|
|
"Len64": true, // go1.25
|
|
|
|
|
"Len8": true, // go1.25
|
|
|
|
|
"OnesCount64": true, // go1.25
|
|
|
|
|
"Prefetch": true, // go1.25
|
|
|
|
|
"PrefetchStreamed": true, // go1.25
|
|
|
|
|
"TrailingZeros32": true, // go1.25
|
|
|
|
|
"TrailingZeros64": true, // go1.25
|
|
|
|
|
"TrailingZeros8": true, // go1.25
|
|
|
|
|
},
|
|
|
|
|
"math": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"Abs": true, // go1.25
|
|
|
|
|
"Ceil": true, // go1.25
|
|
|
|
|
"Copysign": true, // go1.25
|
|
|
|
|
"FMA": true, // go1.25
|
|
|
|
|
"Floor": true, // go1.25
|
|
|
|
|
"Round": true, // go1.25
|
|
|
|
|
"RoundToEven": true, // go1.25
|
|
|
|
|
"Trunc": true, // go1.25
|
|
|
|
|
"sqrt": true, // go1.25
|
|
|
|
|
},
|
|
|
|
|
"math/big": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"mulWW": true, // go1.25
|
|
|
|
|
},
|
|
|
|
|
"math/bits": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"Add": true, // go1.25
|
|
|
|
|
"Add64": true, // go1.25
|
|
|
|
|
"Div": true, // go1.25
|
|
|
|
|
"Div64": true, // go1.25
|
|
|
|
|
"Len": true, // go1.25
|
|
|
|
|
"Len16": true, // go1.25
|
|
|
|
|
"Len32": true, // go1.25
|
|
|
|
|
"Len64": true, // go1.25
|
|
|
|
|
"Len8": true, // go1.25
|
|
|
|
|
"Mul": true, // go1.25
|
|
|
|
|
"Mul64": true, // go1.25
|
|
|
|
|
"OnesCount": true, // go1.25
|
|
|
|
|
"OnesCount16": true, // go1.25
|
|
|
|
|
"OnesCount32": true, // go1.25
|
|
|
|
|
"OnesCount64": true, // go1.25
|
|
|
|
|
"OnesCount8": true, // go1.25
|
|
|
|
|
"Reverse": true, // go1.25
|
|
|
|
|
"Reverse16": true, // go1.25
|
|
|
|
|
"Reverse32": true, // go1.25
|
|
|
|
|
"Reverse64": true, // go1.25
|
|
|
|
|
"Reverse8": true, // go1.25
|
|
|
|
|
"ReverseBytes16": true, // go1.25
|
|
|
|
|
"ReverseBytes32": true, // go1.25
|
|
|
|
|
"ReverseBytes64": true, // go1.25
|
|
|
|
|
"RotateLeft": true, // go1.25
|
|
|
|
|
"RotateLeft16": true, // go1.25
|
|
|
|
|
"RotateLeft32": true, // go1.25
|
|
|
|
|
"RotateLeft64": true, // go1.25
|
|
|
|
|
"RotateLeft8": true, // go1.25
|
|
|
|
|
"Sub": true, // go1.25
|
|
|
|
|
"Sub64": true, // go1.25
|
|
|
|
|
"TrailingZeros16": true, // go1.25
|
|
|
|
|
"TrailingZeros32": true, // go1.25
|
|
|
|
|
"TrailingZeros64": true, // go1.25
|
|
|
|
|
"TrailingZeros8": true, // go1.25
|
|
|
|
|
},
|
|
|
|
|
"runtime": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"publicationBarrier": true, // go1.25
|
|
|
|
|
},
|
|
|
|
|
"sync": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"runtime_LoadAcquintptr": true, // go1.25
|
|
|
|
|
"runtime_StoreReluintptr": true, // go1.25
|
|
|
|
|
},
|
|
|
|
|
"sync/atomic": {
|
add support for Go 1.25 and drop support for 1.24
While strictly speaking it would be okay to leave Go 1.24 support
in place for the time being, we are behind on a few tasks at the moment
so it's best to keep the setup at master simpler for the next release.
Go 1.25 already came out two weeks ago, and it seems to have been
a fairly smooth release, so I don't suspect any end users will have
trouble upgrading to it.
Note that two changes were necessary for garble to work on Go 1.25.0.
First, we stop deduplicating runtimeAndLinknamed with runtimeAndDeps.
Otherwise, for GOOS=windows, internal/runtime/cgroup would be missing
as it is a //go:linkname target from runtime on all platforms,
but it is not transitively imported from runtime on GOOS=windows.
Second, the testing/synctest package is now part of std,
and it is a //go:linkname target from the testing package
but not a transitive import from it. Teach appendListedPackages that,
when loading all packages for a `go test` run, it should load
the new testing/synctest package too.
Fixes #968.
2 months ago
|
|
|
"AddInt32": true, // go1.25
|
|
|
|
|
"AddInt64": true, // go1.25
|
|
|
|
|
"AddUint32": true, // go1.25
|
|
|
|
|
"AddUint64": true, // go1.25
|
|
|
|
|
"AddUintptr": true, // go1.25
|
|
|
|
|
"AndInt32": true, // go1.25
|
|
|
|
|
"AndInt64": true, // go1.25
|
|
|
|
|
"AndUint32": true, // go1.25
|
|
|
|
|
"AndUint64": true, // go1.25
|
|
|
|
|
"AndUintptr": true, // go1.25
|
|
|
|
|
"CompareAndSwapInt32": true, // go1.25
|
|
|
|
|
"CompareAndSwapInt64": true, // go1.25
|
|
|
|
|
"CompareAndSwapUint32": true, // go1.25
|
|
|
|
|
"CompareAndSwapUint64": true, // go1.25
|
|
|
|
|
"CompareAndSwapUintptr": true, // go1.25
|
|
|
|
|
"LoadInt32": true, // go1.25
|
|
|
|
|
"LoadInt64": true, // go1.25
|
|
|
|
|
"LoadPointer": true, // go1.25
|
|
|
|
|
"LoadUint32": true, // go1.25
|
|
|
|
|
"LoadUint64": true, // go1.25
|
|
|
|
|
"LoadUintptr": true, // go1.25
|
|
|
|
|
"OrInt32": true, // go1.25
|
|
|
|
|
"OrInt64": true, // go1.25
|
|
|
|
|
"OrUint32": true, // go1.25
|
|
|
|
|
"OrUint64": true, // go1.25
|
|
|
|
|
"OrUintptr": true, // go1.25
|
|
|
|
|
"StoreInt32": true, // go1.25
|
|
|
|
|
"StoreInt64": true, // go1.25
|
|
|
|
|
"StoreUint32": true, // go1.25
|
|
|
|
|
"StoreUint64": true, // go1.25
|
|
|
|
|
"StoreUintptr": true, // go1.25
|
|
|
|
|
"SwapInt32": true, // go1.25
|
|
|
|
|
"SwapInt64": true, // go1.25
|
|
|
|
|
"SwapUint32": true, // go1.25
|
|
|
|
|
"SwapUint64": true, // go1.25
|
|
|
|
|
"SwapUintptr": true, // go1.25
|
|
|
|
|
},
|
avoid breaking intrinsics when obfuscating names
We obfuscate import paths as well as their declared names.
The compiler treats some packages and APIs in special ways,
and the way it detects those is by looking at import paths and names.
In the past, we have avoided obfuscating some names like embed.FS or
reflect.Value.MethodByName for this reason. Otherwise,
go:embed or the linker's deadcode elimination might be broken.
This matching by path and name also happens with compiler intrinsics.
Intrinsics allow the compiler to rewrite some standard library calls
with small and efficient assembly, depending on the target GOARCH.
For example, math/bits.TrailingZeros32 gets replaced with ssa.OpCtz32,
which on amd64 may result in using the TZCNTL instruction.
We never noticed that we were breaking many of these intrinsics.
The intrinsics for funcs declared in the runtime and its dependencies
still worked properly, as we do not obfuscate those packages yet.
However, for other packages like math/bits and sync/atomic,
the intrinsics were being entirely disabled due to obfuscated names.
Skipping intrinsics is particularly bad for performance,
and it also leads to slightly larger binaries:
│ old │ new │
│ bin-B │ bin-B vs base │
Build-16 5.450Mi ± ∞ ¹ 5.333Mi ± ∞ ¹ -2.15% (p=0.029 n=4)
Finally, the main reason we noticed that intrinsics were broken
is that apparently GOARCH=mips fails to link without them,
as some symbols end up being not defined at all.
This patch fixes builds for the MIPS family of architectures.
Rather than building and linking all of std for every GOARCH,
test that intrinsics work by asking the compiler to print which
intrinsics are being applied, and checking that math/bits gets them.
This fix is relatively unfortunate, as it means we stop obfuscating
about 120 function names and a handful of package paths.
However, fixing builds and intrinsics is much more important.
We can figure out better ways to deal with intrinsics in the future.
Fixes #646.
3 years ago
|
|
|
}
|
|
|
|
|
|
|
|
|
|
var reflectSkipPkg = map[string]bool{
|
|
|
|
|
"fmt": true,
|
|
|
|
|
}
|
