Note that this cross-compilation disables cgo by default,
and so the cgo.txt test script isn't run on GOARCH=386.
That seems fine for now, as the test isn't arch-specific.
This testing uncovered one build failure in internal/literals;
the comparison between int and math.MaxUint32 is invalid on 32-bit.
To fix that build failure, use int64 consistently.
One test also incorrectly assumed amd64; it now supports 386 too.
For any other architecture, it's being skipped for now.
I also had to increase the -race test timeout,
as it usually takes 8-9m on GitHub Actions,
and the timeout would sometimes trigger.
Finally, use "go env" rather than "go version" on CI,
which gives us much more useful information,
and also includes Go's own version now via GOVERSION.
Fixes#426.
We've had confused users a handful of times by now.
And it's reasonable to expect flags to be after the command,
as that's how flags work for cmd/go itself.
I don't think we want to mix our flags with Go's,
or start accepting flags in either place.
Both seem like worse solutions long-term, as they can add confusion.
However, we can quickly give a useful hint when a flag is misplaced.
That should get new users unblocked without asking for help.
We use a regular expression for this purpose,
because it doesn't seem like a FlagSet supports what we need;
to detect whether an argument is one of our flags,
without actually applying its value to the flagset.
Our flagset would also error on Go's flags, which we don't want.
We went to great lengths to ensure garble builds are reproducible.
This includes how the tool itself works,
as its behavior should be the same given the same inputs.
However, we made one crucial mistake with the runtime package.
It has go:linkname directives pointing at other packages,
and some of those pointed packages aren't its dependencies.
Imagine two scenarios where garble builds the runtime package:
1) We run "garble build runtime". The way we handle linkname directives
calls listPackage on the target package, to obfuscate the target's
import path and object name. However, since we only obtained build
info of runtime and its deps, calls for some linknames such as
listPackage("sync/atomic") will fail. The linkname directive will
leave its target untouched.
2) We run "garble build std". Unlike the first scenario, all listPackage
calls issued by runtime's linkname directives will succeed, so its
linkname directive targets will be obfuscated.
At best, this can result in inconsistent builds, depending on how the
runtime package was built. At worst, the mismatching object names can
result in errors at link time, if the target packages are actually used.
The modified test reproduces the worst case scenario reliably,
when the fix is reverted:
> env GOCACHE=${WORK}/gocache-empty
> garble build -a runtime
> garble build -o=out_rebuild ./stdimporter
[stderr]
# test/main/stdimporter
JZzQivnl.NtQJu0H3: relocation target JZzQivnl.iioHinYT not defined
JZzQivnl.NtQJu0H3.func9: relocation target JZzQivnl.yz5z0NaH not defined
JZzQivnl.(*ypvqhKiQ).String: relocation target JZzQivnl.eVciBQeI not defined
JZzQivnl.(*ypvqhKiQ).PkgPath: relocation target JZzQivnl.eVciBQeI not defined
[...]
The fix consists of two steps. First, if we're building the runtime and
listPackage fails on a package, that means we ran into scenario 1 above.
To avoid the inconsistency, we fill ListedPackages with "go list [...] std".
This means we'll always build runtime as described in scenario 2 above.
Second, when building packages other than the runtime,
we only allow listPackage to succeed if we're listing a dependency of
the current package.
This ensures we won't run into similar reproducibility bugs in the future.
Finally, re-enable test-gotip on CI since this was the last test flake.
Add a regression test in gogarble.txt,
as that test is already set up with packages to not obfuscate.
This bug manifested in the form of a build failure for GOOS=plan9
with -literals turned on:
[...]/os/file_plan9.go:151:12: invalid operation: cannot call non-function append (variable of type bool)
In this case, the "os" package is not to be obfuscated,
but we would still obfuscate its literals as per the bug.
But, since the package's identifiers were not obfuscated,
names like "append" were not replaced as per ea2e0bdf71,
meaning that the shadowing would still affect us.
Fixes#417.
Piggybacking off of GOPRIVATE is great for a number of reasons:
* People tend to obfuscate private code, whose package paths will
generally be in GOPRIVATE already
* Its meaning and syntax are well understood
* It allows all the flexibility we need without adding our own env var
or config option
However, using GOPRIVATE directly has one main drawback.
It's fairly common to also want to obfuscate public dependencies,
to make the code in private packages even harder to follow.
However, using "GOPRIVATE=*" will result in two main downsides:
* GONOPROXY defaults to GOPRIVATE, so the proxy would be entirely disabled.
Downloading modules, such as when adding or updating dependencies,
or when the local cache is cold, can be less reliable.
* GONOSUMDB defaults to GOPRIVATE, so the sumdb would be entirely disabled.
Adding entries to go.sum, such as when adding or updating dependencies,
can be less secure.
We will continue to consume GOPRIVATE as a fallback,
but we now expect users to set GOGARBLE instead.
The new logic is documented in the README.
While here, rewrite some uses of "private" with "to obfuscate",
to make the code easier to follow and harder to misunderstand.
Fixes#276.
For packages that we alter, we parse and modify the importcfg file.
Parsing is necessary so we can locate obfuscated object files,
which we use to remember what identifiers were obfuscated.
Modifying the files is necessary when we obfuscate import paths,
and those import paths have entries in an importcfg file.
However, we made one crucial mistake when writing the code.
When handling importmap entries such as:
importmap golang.org/x/net/idna=vendor/golang.org/x/net/idna
we would name the two sides beforePath and afterPath, respectively.
They were added to importMap with afterPath as the key,
but when we iterated over the map to write a modified importcfg file,
we would then assume the key is beforepath.
All in all, we would end up writing the opposite direction:
importmap vendor/golang.org/x/net/idna=golang.org/x/net/idna
This would ultimately result in the importmap never being useful,
and some rather confusing error messages such as:
cannot find package golang.org/x/net/idna (using -importcfg)
Add a test case that reproduces this error,
and fix the code so it always uses beforePath as the key.
Note that we were also updating importCfgEntries with such entries.
I could not reproduce any failure when just removing that code,
nor could I explain why it was added there in the first place.
As such, remove that bit of code as well.
Finally, a reasonable question might be why we never noticed the bug.
In practice, such "importmap"s, represented as ImportMap by "go list",
only currently appear for packages vendored into the standard library.
Until very recently, we didn't support obfuscating most of std,
so we would usually not alter the affected importcfg files.
Now that we do parse and modify them, the bug surfaced.
Fixes#408.
In the added test case, "garble -literals build" would fail:
--- FAIL: TestScripts/literals (8.29s)
testscript.go:397:
> env GOPRIVATE=test/main
> garble -literals build
[stderr]
# test/main
Usz1FmFm.go:1: cannot call non-function string (type int), declared at Usz1FmFm.go:1
Usz1FmFm.go:1: string is not a type
Usz1FmFm.go:1: cannot call non-function append (type int), declared at Usz1FmFm.go:1
That is, for input code such as:
var append int
println("foo")
_ = append
We'd end up with obfuscated code like:
var append int
println(func() string {
// obfuscation...
x = append(x, ...)
// obfuscation...
return string(x)
})
_ = append
Which would then break, as the code is shadowing the "append" builtin.
To work around this, always obfuscate variable names, so we end up with:
var mwu1xuNz int
println(func() string {
// obfuscation...
x = append(x, ...)
// obfuscation...
return string(x)
})
_ = mwu1xuNz
This change shouldn't make the quality of our obfuscation stronger,
as local variable names do not currently end up in Go binaries.
However, this does make garble more consistent in treating identifiers,
and it completely avoids any issues related to shadowing builtins.
Moreover, this also paves the way for publishing obfuscated source code,
such as #369.
Fixes#417.
That is, use a very specific build tag and git commit,
and ensure that neither ends up in the binary.
Luckily, we have nothing to do here.
We were already removing _gomod_.go from the build entirely,
and that is still the mechanism that "go build" uses to bundle the data.
Note that the test will still work if git is not installed,
but it will simply not check the VCS side.
Finally, we use "go version -m" to check the existing fields,
which is easier than calling the Go APIs directly.
It seems like "go test" passes on yesterday's Go master, now.
So, enable test-gotip again with that commit hash.
Fixes#385.
Obfuscating newName arguments of linkname directives
with dots in the importpath didn't work before.
We had a test which covers this,
but the corresponding package wasn't actually obfuscated.
Also update linkname directives of public packages,
to allow the package where something is linknamed to to be
obfuscated regardless.
Public packages can now depend on private packages.
We can now use pruned module graphs in go.mod files,
and we no longer need to worry about runtime/internal/sys.
Note that I had to update testdata/mod slightly,
as the new pruned module graphs algorithm downloads an extra go.mod file.
This change also paves the way towards future Go 1.18 support.
Thanks to lu4p for cleaning up two TODOs as well.
Co-Authored-By: lu4p <lu4p@pm.me>
The "reverse" command had many levels of optional arguments:
garble [garble flags] reverse [build flags] [package] [files]
This was pretty confusing,
and could easily lead to people running the command incorrectly:
# note that output.txt isn't a Go package!
garble reverse output.txt
Moreover, it made the handling of Go build flags pretty confusing.
Should the command below work?
garble reverse -tags=mytag
It also made it easy to not notice that one must supply the main package
to properly reverse some text that it produced, like a panic message.
With the package path being implicit,
one could mistakenly provide the wrong package by running garble
in a directory containing a different package.
See #394.
Functions which use reflection on one of their parameters are,
now added to knownReflectAPIs automatically.
This makes most explicit hints for reflection redundant.
Simple protobuf code now works correctly when obfuscated.
Fixes#162Fixes#373
Before, we would just notice direct calls to reflect's TypeOf and
ValueOf. Any other uses of reflection, such as encoding/json or
google.golang.org/protobuf, would require hints as documented by the
README.
Issue #162 outlines some ways we could fix this issue in a general way,
automatically detecting what functions use reflection on their parameters,
even for third party API funcs.
However, that goal is pretty significant in terms of code and effort.
As a temporary improvement, we can expand the list of "known" reflection
APIs via a static table.
Since this table is keyed by "func full name" strings, we could
potentially include third party APIs, such as:
google.golang.org/protobuf/proto.Marshal
However, for now simply include all the std APIs we know about.
If we fail to do the proper fix for automatic detection in the future,
we can then fall back to expanding this global table for third parties.
Update the README's docs, to clarify that the hint is not always
necessary anymore.
Also update the reflect.txt test to stop using the hint for encoding/json,
and to also start testing text/template with a method call.
While at it, I noticed that we weren't testing the println outputs,
as they'd go to stderr - fix that too.
Updates #162.
Otherwise, the added test case would fail, as we don't modify the C code
and so there would be a name mismatch.
In the far future we might start modifying Go names in C code,
similar to what we did for Go assembly,
but right now that seems out of scope and too complex.
An easier fix is to simply record those (hopefully few) names in ignoreObjects.
While at it, recordReflectArgs started to really outgrow its name, as it
also collected expressions used as constants for literal obfuscation.
Give it a better name.
Fixes#366.
With the -literals flag, we try to convert some const declarations to
vars, as long as that doesn't break typechecking. We really only do that
for typed constant strings, really.
There was a quirk: if a numerical constant had a type and used iota, we
would not obfuscate its value, but we would still convert the
declaration from const to var. Since iotas only work within const
declarations, that would break compilation:
> garble -literals build
[stderr]
# test/main
FeWE3zwi.go:19: undefined: iota
exit status 2
To fix the problem, make the logic more conservative: only obfuscate
constant declarations where the values are typed strings, meaning that
any numerical constants are left entirely untouched.
This fixes the build of google.golang.org/protobuf/runtime/protoiface
with -literals turned on.
Back in early April we added initial support for Go 1.17,
working on a commit from master at that time. For that to work, we just
needed to add a couple of packages to runtimeRelated and tweak printFile
a bit to not break the new "//go:build" directives.
A significant amount of changes have landed since, though, and the tests
broke in multiple ways.
Most notably, the new register ABI is enabled by default for GOOS=amd64.
That affected garble indirectly in two ways: there's a new internal
package to add to runtimeRelated, and we must make reverse.txt more
clever in making its output constant across ABIs.
Another noticeable change is that Go 1.17 changes how its own version is
injected into the runtime package. It used to be via a constant in
runtime/internal/sys, such as:
const TheVersion = `devel ...`
Since we couldn't override such constants via the linker's -X flag,
we had to directly alter the declaration while compiling.
Thankfully, Go 1.17 simply uses a "var buildVersion string" in the
runtime package, and its value is injected by the linker.
This means we can now override it with the linker's -X flag.
We make the code to alter TheVersion for Go 1.16 a bit more clever,
to not break the package when building with Go 1.17.
Finally, our hack to work around ambiguous TOOLEXEC_IMPORTPATH values
now only kicks in for non-test packages, since Go 1.17 includes our
upstream fix. Otherwise, some tests would end up with the ".test"
variant suffix added a second time:
test/bar [test/bar.test] [test/bar [test/bar.test].test]
All the code to keep compatibility with Go 1.16.x remains in place.
We're still leaving TODOs to remind ourselves to remove it or simplify
it once we remove support for 1.16.x.
The 1.17 development freeze has already been in place for a month,
and beta1 is due to come this week, so it's unlikely that Go will change
in any considerable way at this point. Hence, we can say that support
for 1.17 is done.
Fixes#347.
Our recent work in fieldToAlias worked well when the embedded field
declaration (using an alias) was in the same package as the use of that
field. We would have the *ast.Ident for the field declaration, so
types.Info.Uses would give us the TypeName for the alias.
Unfortunately, if the declaration was in a dependency package, we did
not have that same *ast.Ident, as we weren't parsing the source code for
dependencies for type-checking. This resulted in us incorrectly
obfuscating the use of such an embedded field:
> garble build
[stderr]
# test/main
JtzmzxWf.go:4: unknown field 'ExternalForeignAlias' in struct literal of type _BdSNiEL.Vcs_smer
To fix this, look through the direct imports of the package defining the
field to find an alias under the exact same name. Not a foolproof
solution, as there's a TODO, but it should work for most cases.
Fixes the obfuscation of google.golang.org/grpc/internal/status, too.
Updates #349.
The added test case, which is obfuscating and linking os/user, would fail
before this fix:
> garble build
[stderr]
# test/main
/usr/lib/go/pkg/tool/linux_amd64/link: running gcc failed: exit status 1
/usr/bin/ld: $WORK/.tmp/go-link-073246656/go.o: in function `Chz0Yfs2._cgo_cmalloc':
go.go:(.text+0x993cc): undefined reference to `Chz0Yfs2.runtime_throw'
/usr/bin/ld: $WORK/.tmp/go-link-073246656/go.o: in function `Chz0Yfs2.tDfhQ8uK':
go.go:(.text+0x99801): undefined reference to `Chz0Yfs2._cgo_runtime_gostring'
/usr/bin/ld: go.go:(.text+0x9982a): undefined reference to `Chz0Yfs2._cgo_runtime_gostring'
/usr/bin/ld: go.go:(.text+0x99853): undefined reference to `Chz0Yfs2._cgo_runtime_gostring'
collect2: error: ld returned 1 exit status
The reason is that we would alter the linkname directives of cgo-generated
code, but we would not obfuscate the code itself at all.
The generated code would end up being transformed into:
//go:linkname zh_oKZIy runtime.throw
func runtime_throw(string)
One can clearly see the error there; handleDirectives obfuscated the
local linkname name, but since transformGo didn't run, the actual Go
declaration was not obfuscated in the same way. Thus, the linker fails
to find a function body for runtime_throw, and fails.
The solution is simple: handleDirectives assumes that it's running on
code being obfuscated, so only run it when transformGo is running.
We can also remove the cgo skip check in handleDirectives, as it never
runs on cgo-generated code now.
Fixes a number of build errors that have been noticed since
907aebd770.
When such an alias name was used to define an embedded field, we handled
that case gracefully via the code using:
tf.info.Uses[node].(*types.TypeName)
Unfortunately, when the same field name was used elsewhere, such as a
composite literal, tf.Info.Uses gave us a *types.Var, not a
*types.TypeName, meaning we could no longer tell if this was an alias,
or what it pointed to.
Thus, we failed to obfuscate the name properly in the added test case:
> garble build
[stderr]
# test/main/sub
xxWZf66u.go:36: unknown field 'foreignAlias' in struct literal of type smhWelwn
It doesn't seem like any of the go/types APIs allows us to obtain the
*types.TypeName directly in this scenario. Thus, use a trick that we
used before: after typechecking, but before obfuscating, record all
embedded struct field *types.Var which are aliases via a map, where the
value holds the *types.TypeName for the alias.
Updates #349.
We already added support for "//go:embed" with string and []byte,
by not obfuscating the "embed" import path.
However, embed.FS was still failing:
> garble build
[stderr]
# test/main
:13: go:embed cannot apply to var of type embed.WtKNvwbN
The compiler detects the type by matching its name to exactly "embed.FS",
so don't obfuscate the name "FS" either.
While at it, ensure that the embed code behaves the same with "go build".
Updates #349.
Before, "garble build -h" would print the same as "garble -h", which is
too much and confusing, as it doesn't tell us much about "build".
Now it's far better, and includes the output of "go build -h":
$ garble build -h
usage: garble [garble flags] build [arguments]
This command wraps "go build". Below is its help:
usage: go build [-o output] [build flags] [packages]
Run 'go help build' for details.
We do the same for "garble reverse -h", since it doesn't wrap a Go tool
command.
I added a second one recently without noticing, since the original was
too far below and I had written commands above it.
Deduplicate them, and put the new env line as early as possible, to
prevent further issues and confusion.
Every now and then, a CI run would fail:
FAIL: testdata/scripts/reflect.txt:7: unexpected match for ["main.go"] in main
These were rare, and very hard to reproduce or debug.
My best guess is that, since "main.go" is a short string and we use
random eight-character obfuscated filenames ending with ".go", it was
possible that the random filename happened to end in "main" in some
cases.
Given the base64 encoding, the chances of a single suffix collision are
about 0.000006%. Note, however, that a single obfuscated build will most
likely obfuscate many filenames, especially for the tests obfuscating
multiple packages. For a single CI run with many tests across three OSs,
the chances of any collision are likely very low, but realistic.
All this has a simple fix: use longer filenames to match with. We choose
"garble_main.go" since it's long enough, but also because it's still
clear it's a "main" Go file, and it's very unlikely to cause conflicts
with filenames in upstream Go given the "garble_" prefix.
Before this change, we would try to never obfuscate alias names. That
was far from ideal, as they can end up in field names via anonymous
fields.
Even then, we would sometimes still fail to build, because we would
inconsistently obfuscate alias names. For example, in the added test
case:
--- FAIL: TestScripts/syntax (0.23s)
testscript.go:397:
> env GOPRIVATE='test/main,private.source'
> garble build
[stderr]
# test/main/sub
Lv_a8gRD.go:15: undefined: KCvSpxmQ
To fix this problem, we set obj to be the TypeName corresponding to the
alias when it is used as an embedded field. We can then make the right
choice when obfuscating the name.
Right now, all aliases will be obfuscated. A TODO exists about not
obfuscating alias names when they're used as embedded fields in a struct
type in the same package, and that package is used for reflection -
since then, the alias name ends up as the field name.
With these changes, the protobuf module now builds.
obfuscatedTypesPackage is used to figure out if a name in a dependency
package was obfuscated or not. For example, if that package used
reflection on a named type, it wasn't obfuscated, so we must have the
same information to not obfuscate the same name downstream.
obfuscatedTypesPackage could return nil if the package was indirectly
imported, though. This can happen if a direct import has a function that
returns an indirect type, or if a direct import exposes a name that's a
type alias to an indirect type.
We sort of dealt with this in two pieces of code by checking for
obfPkg!=nil, but a third one did not have this check and caused a panic
in the added test case:
--- FAIL: TestScripts/reflect (0.81s)
testscript.go:397:
> env GOPRIVATE=test/main
> garble build
[stderr]
# test/main
panic: runtime error: invalid memory address or nil pointer dereference [recovered]
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x20 pc=0x8a5e39]
More importantly though, the nil check only avoids panics. It doesn't
fix the root cause of the problem: that importcfg does not contain
indirectly imported packages. The added test case would still fail, as
we would obfuscate a type in the main package, but not in the indirectly
imported package where the type is defined.
To fix this, resurrect a bit of code from earlier garble versions, which
uses "go list -toolexec=garble" to fetch a package's export file. This
lets us fill the indirect import gaps in importcfg, working around the
problem entirely.
This solution is still not particularly great, so we add a TODO about
possibly rethinking this in the future. It does add some overhead and
complexity, though thankfully indirect imports should be uncommon.
This fixes a few panics while building the protobuf module.
Our allPkgs boolean logic was wrong, because it could still lead to
garble obfuscating a type when used in the main package, but not in its
defining package. The added test case shows such a case.
To fix that, use a package path to only record the named objects from
the target package, which is a narrower operation without this problem,
but still makes all our tests pass.
This makes the google.golang.org/protobuf/internal/filetype package
start building.
I've wanted this to more easily debug build failures.
To not force a build failure in a test script, as that would require
some trickery to remain stable, we use runtime.Caller without printing
the line number. Before this patch, those filenames without line numbers
would not be reversed at all.
If a package A imports package B, and uses reflect.TypeOf on an unnamed
struct type B.T (such as an alias), we don't want to record B.T's fields
as "do not obfuscate". This is for the same reason that we don't if B.T
is a named struct type: the detection only works for the package
defining the type, as otherwise it's inconsistent.
We failed to handle this case well, because we assumed all struct types
would be under some named type. This is not the case for type aliases.
Fortunately, struct fields are named, and as such they are objects.
Check their package too, just like we do for named types.
Fixes another build error when obfuscating the protobuf module.
We add a simplified version of the example above as a test case,
which originated from debugging the protobuf build failure.
The detection of reflection usage is tricky and there are plenty of edge
cases to test for. We definitely want one script for it, rather than
splitting those cases between other scripts like imports.txt and
syntax.txt.
Moreover, those two were rather generic and large, so this helps keep a
balance.
We were using temporary filenames for modified Go and assembly files.
For example, an obfuscated "encoding/json/encode.go" would end up as:
/tmp/garble-shared123/encode.go.456.go
where "123" and "456" are random numbers, usually longer.
This was usually fine for two reasons:
1) We would add "/tmp/garble-shared123/" to -trimpath, so the temporary
directory and its random number would be invisible.
2) We would add "//line" directives to the source files, replacing
the filename with obfuscated versions excluding any random number.
Unfortunately, this broke in multiple ways. Most notably, assembly files
do not have any line directives, and it's not clear that there's any
support for them. So the random number in their basename could end up in
the binary, breaking reproducibility.
Another issue is that the -trimpath addition described above was only
done for cmd/compile, not cmd/asm, so assembly filenames included the
randomized temporary directory.
To fix the issues above, the same "encoding/json/encode.go" would now
end up as:
/tmp/garble-shared123/encoding/json/encode.go
Such a path is still unique even though the "456" random number is gone,
as import paths are unique within a single build.
This fixes issues with the base name of each file, so we no longer rely
on line directives as the only way to remove the second original random
number.
We still rely on -trimpath to get rid of the temporary directory in
filenames. To fix its problem with assembly files, also amend the
-trimpath flag when running the assembler tool.
Finally, add a test that reproducible builds still work when a full
rebuild is done. We choose goprivate.txt for such a test as its
stdimporter package imports a number of std packages, including uses of
assembly and cgo.
For the time being, we don't use such a "full rebuild" reproducibility
test in other test scripts, as this step is expensive, rebuilding many
packages from scratch.
This issue went unnoticed for over a year because such random numbers
"123" and "456" were created when a package was obfuscated, and that
only happened once per package version as long as the build cache was
kept intact.
When clearing the build cache, or forcing a rebuild with -a, one gets
new random numbers, and thus a different binary resulting from the same
build input. That's not something that most users would do regularly,
and our tests did not cover that edge case either, until now.
Fixes#328.
I've tripped over this GODEBUG env var four or five times already.
Since it affects any Go program by making them print tons of runtime
information, it also affects "go env", breaking garble horribly.
To prevent further issues, unset the env var when done.
Packages P1 and P2 can define identical struct types T1 and T2, and one
can convert from type T1 to T2 or vice versa.
The spec defines two identical struct types as:
Two struct types are identical if they have the same sequence of
fields, and if corresponding fields have the same names, and
identical types, and identical tags. Non-exported field names
from different packages are always different.
Unfortunately, garble broke this: since we obfuscated field names
differently depending on the package, cross-package conversions like the
case above would result in typechecking errors.
To fix this, implement Joe Tsai's idea: hash struct field names with the
string representation of the entire struct. This way, identical struct
types will have their field names obfuscated in the same way in all
packages across a build.
Note that we had to refactor "reverse" a bit to start using transformer,
since now it needs to keep track of struct types as well.
This failure was affecting the build of google.golang.org/protobuf,
since it makes regular use of cross-package struct conversions.
Note that the protobuf module still fails to build, but for other
reasons. The package that used to fail now succeeds, so the build gets a
bit further than before. #240 tracks adding relevant third-party Go
modules to CI, so we'll track the other remaining failures there.
Fixes#310.
Obfuscating some std packages for windows/arm triggered a bug; when
encountering a call to runtime·memmove, we'd hash "memmove" with the
current package's action ID.
This is wrong on two levels: First, we aren't obfuscating the runtime
package yet. And second, if we did, we would have to hash the symbol
appropriately, with that package's action ID.
For now, only hashing the local names does the trick. That's all that
the code currently supports, anyway.
Since 1a8e32227f we have been setting line numbers on call sites, and
that line number is always the minimum value for -tiny: 1.
Still not as zero-overhead as the old mechanism which entirely deleted
line numbers, but a useless and small line number is still pretty good.
THey don't show up in stack traces, and if they show up in regular
program output, we should in theory not obfuscate those names via the
detection of reflection.
However, there's one relatively common scenario where obfuscated field
names can appear: in build error messages, when obfuscation fails a
build.
Our previous logic only took care of fairly simple types, such as a
simple struct or a pointer to a struct. If we had a struct embedding
another struct, we'd fail to record the objects for the fields in the
inner struct, and that would lead to miscompilation:
> garble build
[stderr]
# test/main
LZmt64Nm.go:7: outer.InnerField undefined (type *CcUt1wkQ.EmbeddingOuter has no field or method InnerField)
To fix this issue, make the function that records all objects under a
types.Type smarter. Since it now does more than just dealing with
structs, it's also renamed.
Since the function now walks types properly, we get to remove the extra
ast.Inspect in recordReflectArgs, which is nice.
We also make it a method, to avoid the map parameter. A boolean
parameter is also added, since we need this feature to only look at the
current package when looking at reflect calls.
Finally, we add a test case, a simplified version of the original bug
report.
Fixes#315.
The docs for go/importer.ForCompiler say:
The lookup function is called each time the resulting importer
needs to resolve an import path. In this mode the importer can
only be invoked with canonical import paths (not relative or
absolute ones); it is assumed that the translation to canonical
import paths is being done by the client of the importer.
We use a lookup func for two reasons: first, to support modules, and
second, to be able to use our information from "go list -json -export".
However, go/types does not canonicalize import paths before calling
ImportFrom. This is somewhat understandable; it doesn't know whether an
importer was created with a lookup func, and ImportFrom only requires
the input path to be canonicalized in that scenario. When the lookup
func is nil, the importer canonicalizes by itself via go/build.Import.
Before this change, the added crossbuild test would fail:
> garble build net/http
[stderr]
# vendor/golang.org/x/crypto/chacha20
typecheck error: /usr/lib/go/src/vendor/golang.org/x/crypto/chacha20/chacha_generic.go:10:2: could not import crypto/cipher (can't find import: "crypto/cipher")
# vendor/golang.org/x/text/secure/bidirule
typecheck error: /usr/lib/go/src/vendor/golang.org/x/text/secure/bidirule/bidirule.go:12:2: could not import errors (can't find import: "errors")
# vendor/golang.org/x/crypto/cryptobyte
typecheck error: /usr/lib/go/src/vendor/golang.org/x/crypto/cryptobyte/asn1.go:8:16: could not import encoding/asn1 (can't find import: "encoding/asn1")
# vendor/golang.org/x/text/unicode/norm
typecheck error: /usr/lib/go/src/vendor/golang.org/x/text/unicode/norm/composition.go:7:8: could not import unicode/utf8 (can't find import: "unicode/utf8")
This is because we'd fall back to importer.Default, which only knows how
to find packages in $GOROOT/pkg. Those are missing for cross-builds,
unsurprisingly, as those built archives end up in the build cache.
After this change, we properly support importing std-vendored packages,
so we can get rid of the importer.Default workaround. And, by extension,
cross-builds now work as well.
Note that, in the added test script, the full build of the binary fails,
as there seems to be some sort of linker problem:
> garble build
[stderr]
# test/main
d9rqJyxo.uoqIiDs5: relocation target runtime.os9A16A3 not defined
We leave that as a TODO for now, as this change is subtle enough as it
is.
In the worst case scenario, when GOPRIVATE isn't set at all, we would
run these three commands:
* "go env GOPRIVATE", to fetch GOPRIVATE itself
* "go list -m", for GOPRIVATE's fallback
* "go version", to check the version of Go being used
Now that we support Go 1.16 and later, all these three can be obtained
via "go env -json":
$ go env -json GOPRIVATE GOMOD GOVERSION
{
"GOMOD": "/home/mvdan/src/garble/go.mod",
"GOPRIVATE": "",
"GOVERSION": "go1.16.3"
}
Note that we don't get the module path directly, but we can use the
x/mod/modfile Go API to parse it from the GOMOD file cheaply.
Notably, this also simplifies our Go version checking logic, as now we
get just the version string without the "go version" prefix and
"GOOS/GOARCH" suffix we don't care about.
This makes our code a bit more maintainable and robust. When running a
short incremental build, we can also see a small speed-up, as saving two
"go" invocations can save a few milliseconds:
name old time/op new time/op delta
Build/Cache-8 168ms ± 0% 166ms ± 1% -1.26% (p=0.009 n=6+6)
name old bin-B new bin-B delta
Build/Cache-8 6.36M ± 0% 6.36M ± 0% +0.12% (p=0.002 n=6+6)
name old sys-time/op new sys-time/op delta
Build/Cache-8 222ms ± 2% 219ms ± 3% ~ (p=0.589 n=6+6)
name old user-time/op new user-time/op delta
Build/Cache-8 857ms ± 1% 846ms ± 1% -1.31% (p=0.041 n=6+6)
In ce2c45440a, we simplified the code a bit and removed one call to
obfuscatedTypesPackage.
Unfortunately, we introduced a regression; if an exported function is
linknamed to another symbol name, and it's called from an importer
package, we would have a build failure now:
> garble build
[stderr]
# test/main
ZiOACuw7.go:1: undefined: ODC0xN52.BaDqbhkj
This is because the imported package would not hash the original name,
via its ignoreObjects logic. And, since the importer package has no
access to that knowledge, it would hash the same name, and fail to find
it in the final build.
The regression happened because we used to have a types.Scope Lookup
that saved us in this scenario. Add the test, and re-add the Lookup,
this time only for this particular scenario with function names.
Thanks to Andrew LeFevre for reporting and describing the test case.
While at it, replace more uses of "garbled" to "obfuscated".
Since it changes rapidly, especially during merge cycles, and we don't
want CI to surprisingly blow up in our faces from one day to another.
Pin this to a commit from yesterday which works, since some changes
merged today moved where the Go build version is recorded and broke
garble.
While at it, replace "git clone" with a wget of a source archive. This
is much, much faster, mainly because a tarball is significantly smaller.
We now download about 20MiB instead of over 350MiB.
One downside is that, without git, make.bash can't construct a devel
version on its own. For that reason, add a pretty basic manual version
via the VERSION file.
This means that we must not reject custom devel version strings. This is
a good thing anyway, because custom devel strings are already common
when building Go in custom ways. Those people tend to be advanced users,
such as CI, so fall back to assuming they know what they are doing and
don't error.
Plus, starting last week, devel versions in Go master now contain the
major Go version like in build tags, such as "go1.17-commit...", so we
will soon start relying on that instead of parsing dates:
$ go version
go version devel go1.17-a7e16abb22 Thu Apr 8 07:33:58 2021 +0000 linux/amd64
Two bugs were remaining which made the build with -literals of std fail.
First, we were ignoring too many objects in constant expressions,
including type names. This resulted in type names declared in
dependencies which were incorrectly not obfuscated in the current
package:
# go/constant
O1ku7TCe.go:1: undefined: alzLJ5Fd.Word
b0ieEGVQ.go:1: undefined: alzLJ5Fd.Word
LEpgYKdb.go:4: undefined: alzLJ5Fd.Word
FkhHJCfm.go:1: undefined: alzLJ5Fd.Word
This edge case is easy to reproduce, so a test case is added to
literals.txt.
The second issue is trickier; in some packages like os/user, we would
get syntax errors because of comments printed out of place:
../tip/os/user/getgrouplist_unix.go:35:130: syntax error: unexpected newline, expecting comma or )
This is a similar kind of error that we tried to fix with e2f06cce94. In
particular, it's fixed by also setting CallExpr.Rparen in withPos. We
also add many other missing Pos fields for good measure, even though
we're not sure they help just yet.
Unfortunately, all my attempts to minimize this into a reproducible
failure have failed. We can't just copy the failing file from os/user,
as it only builds on some OSs. It seems like it was the perfect mix of
cgo (which adds line directive comments) plus unlucky positioning of
literals.
For that last reason, as well as for ensuring that -literals works well
with a wide variety of software, we add a build of all of std with
-literals when not testing with -short. This is akin to what we do in
goprivate.txt, but with the -literals flag. This does make "go test"
more expensive, but also more thorough.
Fixes#285, hopefully for good this time.
The regular obfuscation process simply modifies some simple nodes, such
as identifiers and strings. In those cases, we modify the nodes
in-place, meaning that their positions remain the same. This hasn't
caused any problems.
Literal obfuscation is trickier. Since we replace one expression with an
entirely different one, we use cursor.Replace. The new expression is
entirely made up on the spot, so it lacks position information.
This was causing problems. For example, in the added test input:
> garble -literals build
[stderr]
# test/main
dgcm4t6w.go:3: misplaced compiler directive
dgcm4t6w.go:4: misplaced compiler directive
dgcm4t6w.go:3: misplaced compiler directive
dgcm4t6w.go:6: misplaced compiler directive
dgcm4t6w.go:7: misplaced compiler directive
dgcm4t6w.go:3: misplaced compiler directive
dgcm4t6w.go:9: misplaced compiler directive
dgcm4t6w.go:3: misplaced compiler directive
dgcm4t6w.go:3: too many errors
The build errors are because we'd move the compiler directives, which
makes the compiler unhappy as they must be directly followed by a
function declaration.
The root cause there seems to be that, since the replacement nodes lack
position information, go/printer would try to estimate its printing
position by adding to the last known position. Since -literals adds
code, this would result in the printer position increasing rapidly, and
potentially printing directive comments earlier than needed.
For now, making the replacement nodes have the same position as the
original node seems to stop go/printer from making this mistake.
It's possible that this workaround won't be bulletproof forever, but it
works well for now, and I don't see a simpler workaround right now.
It would be possible to use fancier mechanisms like go/ast.CommentMap or
dave/dst, but those are a significant amount of added complexity as well.
Fixes#285.
Reduces "go test -short" with a warm build cache from ~9s to ~4s. The
main offender was the use of "-a" in the "garble test" call; I think I
added that flag to rebuild all packages when debugging an error, and
forgot to remove it.
Just the exit status code is used. Printing an error would be confusing,
since we're printing the input too. Plus, the error is binary anyway -
either we changed something, or we did not.
One could make a case for only printing if we changed anything, and
showing an error otherwise. But that would mean buffering the entire
input, which could be problematic or slow when e.g. reversing large log
files.
Fix up a few TODOs, and simplify the way we handle comments.
We now add whitespace around inline /*line*/ directives, to ensure we
don't break programs. A test case is added too.
We now add line directives to call sites, not function declarations,
since those are what actually shows up in stack traces.
It's unclear if we care about any other lines inside functions at all.
This also fixes reversing with -literals, since that feature adds a
significant amount of code which shuffles line numbers around.
Finally, we extend the tests with types, methods, and anonymous
functions, and we make all of them work well.
Updates #5.
In 6898d61637, we switched from using action IDs from "go list
-toolexec=garble" to those from the original "go list". We still wanted
the obfuscation and hashing to change if the version of garble changes,
so we hashed that "original action ID" with garble's own content ID, and
called the new hash "garble action ID".
While working on a different patch, I noticed something weird: with the
new mechanism, adding or removing flags like -literals did not alter
those hashes, unlike the old method. This is because the old method used
ownContentID, which includes such bits of information, but the new
method does not.
Change that, and add a test that locks in the behavior we want. In
seed.txt, we check that a single function name gets hashed in particular
ways in different scenarios.
Note that we use a mix of "cmp" and "! bincmp", since the former has no
negated form.
While at it, the seed.txt test is revamped a bit. Now, we only run with
-literals once, as this test is mainly about -seed. We also declare seed
strings once, as environment variables, which makes it easier to track
what each step is doing.
In particular, the positions within function declarations, including the
positions of call sites to other functions.
Note that this isn't well tested just yet, particularly not with other
features like -literals. We can extend the tests and code over time.
This gets us the core basics.
The issue will be closed once the feature is documented for users, in a
follow-up PR.
Updates #5.
First, stop writing binaries into the current directory, which pollutes
the git clone.
Second, split the benchmark into two. The old benchmark always used the
build cache after the first iteration, meaning that we weren't really
measuring the cost of cold fresh builds.
The new benchmarks show a build with an always-warm cache, and one
without any cache.
Note that NoCache with the main package importing "fmt" took about 4s
wall time, which makes benchmarking too slow. For that reason, the new
bench-nocache program has no std dependencies other than runtime, which
already pulls in half a dozen dependencies we recompile at every
iteration. This reduces the wall time to 2s, which is bearable.
On the other hand, Cache is already fast, so we add a second and
slightly heavier dependency, net/http. The build still takes under 300ms
of wall time. This also helps the Cache benchmark imitate larger rebuilds
with a warm cache.
Longer term, both benchmarks will be useful, because we want both
scenarios to be as efficient as possible.
name time/op
Build/Cache-8 161ms ± 1%
Build/NoCache-8 1.21s ± 1%
name bin-B
Build/Cache-8 6.35M ± 0%
Build/NoCache-8 6.35M ± 0%
name sys-time/op
Build/Cache-8 218ms ± 7%
Build/NoCache-8 522ms ± 4%
name user-time/op
Build/Cache-8 825ms ± 1%
Build/NoCache-8 8.17s ± 1%
Position information was obfuscated with math/rand manually, which meant
that the resulting positions were pretty small like "x.go:34", but they
were also very hard to reverse due to their short length and difficulty
to reproduce.
We now hash them with hashWith and the package's GarbleActionID:
"main.go:203" hashed with 933ad1c700755b7c3a9913c55cade1 to "mwu1xuNz.go"
The input to the hash is the base filename and the byte offset of the
declaration within the file, meaning that it's unique within a package.
The output filename is long enough to allow easy reversal.
The line number is always 1, since the information needed for reversing
is contained entirely within the filename. It doesn't really matter if
we encode data in the filename or line number, but it's easier for us to
use a string.
For #5.
Historically, it was impossible to rename those funcs as the
implementation was in assembly files, and we only transformed Go code.
Now that transformAsm exists, it's only about fifty lines to do some
very basic parsing and rewriting of assembly files.
This fixes the obfuscated builds of multiple std packages, including a
few dependencies of net/http, since they included assembly funcs which
called pure Go functions. Those pure Go functions had their names
obfuscated, breaking the call sites in assembly.
Fixes#258.
Fixes#261.
With the many refactors building up to v0.1.0, we broke "garble test" as
we no longer dealt with test packages well.
Luckily, now that we can depend on TOOLEXEC_IMPORTPATH, we can support
the test command again, as we can always figure out what package we're
currently compiling, without having to track a "main" package.
Note that one major pitfall there is test packages, where
TOOLEXEC_IMPORTPATH does not agree with ImportPath from "go list -json".
However, we can still work around that with a bit of glue code, which is
also copiously documented.
The second change necessary is to consider test packages private
depending on whether their non-test package is private or not. This can
be done via the ForTest field in "go list -json".
The third change is to obfuscate "_testmain.go" files, which are the
code-generated main functions which actually run tests. We used to not
need to obfuscate them, since test function names are never obfuscated
and we used to not obfuscate import paths at compilation time. Now we do
rewrite import paths, so we must do that for "_testmain.go" too.
The fourth change is to re-enable test.txt, and expand it with more
sanity checks and edge cases.
Finally, document "garble test" again.
Fixes#241.
Our use of go-internal's gotooltest.Setup means that "go" is set up as a
top-level command in the scripts, so we can simply "go build" instead of
having to "exec go build". The result is practically the same, but the
scripts are simpler.
While at it, I had left an "exec cat" behind; remove it.
Now that we've dropped support for Go 1.15.x, we can finally rely on
this environment variable for toolexec calls, present in Go 1.16.
Before, we had hacky ways of trying to figure out the current package's
import path, mostly from the -p flag. The biggest rough edge there was
that, for main packages, that was simply the package name, and not its
full import path.
To work around that, we had a restriction on a single main package, so
we could work around that issue. That restriction is now gone.
The new code is simpler, especially because we can set curPkg in a
single place for all toolexec transform funcs.
Since we can always rely on curPkg not being nil now, we can also start
reusing listedPackage.Private and avoid the majority of repeated calls
to isPrivate. The function is cheap, but still not free.
isPrivate itself can also get simpler. We no longer have to worry about
the "main" edge case. Plus, the sanity check for invalid package paths
is now unnecessary; we only got malformed paths from goobj2, and we now
require exact matches with the ImportPath field from "go list -json".
Another effect of clearing up the "main" edge case is that -debugdir now
uses the right directory for main packages. We also start using
consistent debugdir paths in the tests, for the sake of being easier to
read and maintain.
Finally, note that commandReverse did not need the extra call to "go
list -toolexec", as the "shared" call stored in the cache is enough. We
still call toolexecCmd to get said cache, which should probably be
simplified in a future PR.
While at it, replace the use of the "-std" compiler flag with the
Standard field from "go list -json".
This mainly cleans up the few bits of code where we explicitly kept
support for Go 1.15.x. With v0.1.0 released, we can drop support now,
since the next v0.2.0 release will only support Go 1.16.x.
Also updates all modules, including test ones, to 'go 1.16'.
Note that the TOOLEXEC_IMPORTPATH refactor is not done here, despite all
the TODOs about doing so when we drop 1.15 support. This is because that
refactor needs to be done carefully and might have side effects, so it's
best to keep it to a separate commit.
Finally, update the deps.
First, we had some link errors such as:
cannot find package J6OzO8GN (using -importcfg)
This was caused by the code that writes an updated importcfg, which did
not handle import maps well. That code is now fixed, and we also add an
obfuscatedImportPath method for clarity.
Once fixed, we ran into other link errors:
Pw3g97ww.addVW: relocation target Pw3g97ww.addVWlarge not defined
After some digging, the cause of those is assembly code that we do not
yet support obfuscating. #261 tracks that.
Meanwhile, to fix "GOPRIVATE=* garble build" and to be able to have a
test for the original import path bug, we add the packages which use
that form of assembly code to runtimeRelated - math/big and
crypto/sha512. There might be more, but these were the ones found by
trying to link crypto/tls, a fairly common dependency.
Fixes#256.
Sometimes it prints relatively low addresses, like:
(0xbc200,0xdd658)
We required 6 to 8 hex digits, and sometimes it printed 5. Make the
regular expression more conservative.
Fixes#184.
This was pretty much just fixing the README and closing the issue. The
only other noteworthy user-facing change is that, if the Go version is
detected to be too old, we now suggest 1.16.x instead of 1.15.x.
While at it, refactor goversion.txt a bit. I wanted it to print a
clearer "mocking the go build" error if another command was used like
"go build", but I didn't want to learn BAT. So, instead use a simple Go
program and build it, which will work on all platforms. The added
"go build" step barely takes 100ms on my machine, given how simple the
program is.
The [short] line also doesn't seem necessary to me. The entire script
runs in under 200ms for me, so it's well within the realm of "short", at
least compared to many of the other test scripts.
Fixes#124.
Strip a few more traceback printing related functions in the runtime, new for Go 1.16. Also test that GODEBUG=inittrace=1 does not print anything when -tiny is passed.
The asm tool runs twice for a package with assembly. The second time it
does, the path given to the -p flag matters, just like in the compiler,
as we generate an object file.
We don't have a -buildid flag in the asm tool, so obtaining the action
ID to obfuscate the package path with is a bit tricky. We store it from
transformCompile, and read it from transformAsm. See the detailed docs
for more.
This was the last "skip" line in the tests due to Go 1.16. After all PRs
are merged, one last PR documenting that 1.16 is supported will be sent,
closing the issue for good.
It's unclear why this wasn't an issue in Go 1.15. My best guess is that
the ABI changes only happened in Go 1.16, and this causes exported asm
funcs to start showing up in object files with their package paths.
Updates #124.
With a few extra lines, we can keep Go 1.15 support in the table too.
Re-enables the goprivate.txt test for Go 1.16.
While at it, make the script's use of grep a bit simpler with -E, which
also uses the same syntax as Go's regexp. Its skip logic was also buggy,
resulting in the macos results always being empty.
Updates #124.
There are three minor bugs breaking Go 1.16 with the current version of
garble, after the import path obfuscation refactor:
1) Stripping the runtime results in an unused import error. This PR
fixes that.
2) The asm.txt test seems to be broken; something to do with the export
data not being right for the exported assembly func.
3) The obfuscated build of std fails, since our runtimeRelated table was
generated for Go 1.15, not 1.16.
This PR fixes the first issue, adds conditional skip lines for 1.16 for
the other two issues, and enables 1.16 on CI.
Note that 1.16 support is not here just yet, because of the other two
issues. As such, no doc changes.
Updates #124.
The TODO I left there didn't take long to surface as a bug. If the
package path ends with a word containing a hyphen, that's not a valid
identifier, so we end up with invalid Go syntax.
Add that test case, as well as one where an import was already named.
To fix the issue, we just need to use the package name we got from
'go list -json'.
Fixes#243.
We used to rely on a parallel implementation of an object file parser
and writer to be able to obfuscate import paths. After compiling each
package, we would parse the object file, replace the import paths, and
write the updated object file in-place.
That worked well, in most cases. Unfortunately, it had some flaws:
* Complexity. Even when most of the code is maintained in a separate
module, the import_obfuscation.go file was still close to a thousand
lines of code.
* Go compatibility. The object file format changes between Go releases,
so we were supporting Go 1.15, but not 1.16. Fixing the object file
package to work with 1.16 would probably break 1.15 support.
* Bugs. For example, we recently had to add a workaround for #224, since
import paths containing dots after the domain would end up escaped.
Another example is #190, which seems to be caused by the object file
parser or writer corrupting the compiled code and causing segfaults in
some rare edge cases.
Instead, let's drop that method entirely, and force the compiler and
linker to do the work for us. The steps necessary when compiling a
package to obfuscate are:
1) Replace its "package foo" lines with the obfuscated package path. No
need to separate the package path and name, since the obfuscated path
does not contain slashes.
2) Replace the "-p pkg/foo" flag with the obfuscated path.
3) Replace the "import" spec lines with the obfuscated package paths,
for those dependencies which were obfuscated.
4) Replace the "-importcfg [...]" file with a version that uses the
obfuscated paths instead.
The linker also needs that last step, since it also uses an importcfg
file to find object files.
There are three noteworthy drawbacks to this new method:
1) Since we no longer write object files, we can't use them to store
data to be cached. As such, the -debugdir flag goes back to using the
"-a" build flag to always rebuild all packages. On the plus side,
that caching didn't work very well; see #176.
2) The package name "main" remains in all declarations under it, not
just "func main", since we can only rename entire packages. This
seems fine, as it gives little information to the end user.
3) The -tiny mode no longer sets all lines to 0, since it did that by
modifying object files. As a temporary measure, we instead set all
top-level declarations to be on line 1. A TODO is added to hopefully
improve this again in the near future.
The upside is that we get rid of all the issues mentioned before. Plus,
garble now nearly works with Go 1.16, with the exception of two very
minor bugs that look fixable. A follow-up PR will take care of that and
start testing on 1.16.
Fixes#176.
Fixes#190.
First, remove the shuffling of the declarations list within each file.
This is what we used at the very start to shuffle positions. Ever since
we started obfuscating positions via //line comments, that has been
entirely unnecessary.
Second, add a proper test that will fail if we don't obfuscate line
numbers well enough. Filenames were already decently covered by other
tests.
Third, simplify the line obfuscation code. It does not require
astutil.Apply, and ranging over file.Decls is easier.
Finally, also obfuscate the position of top-level vars, since we only
used to do it for top-level funcs. Without that fix, the test would fail
as varLines was unexpectedly sorted.
There was one bug keeping the command below from working:
GOPRIVATE='*' garble build std
The bug is rather obscure; I'm still working on a minimal reproducer
that I can submit upstream, and I'm not yet convinced about where the
bug lives and how it can be fixed.
In short, the command would fail with:
typecheck error: /go/src/crypto/ecdsa/ecdsa.go:122:12: cannot use asn1.SEQUENCE (constant 48 of type asn1.Tag) as asn1.Tag value in argument to b.AddASN1
Note that the error is ambiguous; there are two asn1 packages, but they
are actually mismatching. We can see that by manually adding debug
prints to go/types:
constant: asn1.SEQUENCE (constant 48 of type golang.org/x/crypto/cryptobyte/asn1.Tag)
argument type: vendor/golang.org/x/crypto/cryptobyte/asn1.Tag
It's clear that, for some reason, go/types ends up confused and loading
a vendored and non-vendored version of asn1. There also seems to be no
way to work around this with our lookup function, as it just receives an
import path as a parameter, and returns an object file reader.
For now, work around the issue by *not* using a custom lookup function
in this rare edge case involving vendored dependencies in std packages.
The added code has a lengthy comment explaining the reasoning.
I still intend to investigate this further, but there's no reason to
keep garble failing if we can work around the bug.
Fixes#223.
The added test case used to crash garble:
panic: runtime error: invalid memory address or nil pointer dereference [recovered]
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x8 pc=0x8fe71b]
goroutine 1 [running]:
golang.org/x/tools/go/ast/astutil.Apply.func1(0xc0001e8880, 0xc000221570)
/go/pkg/mod/golang.org/x/tools@v0.0.0-20210115202250-e0d201561e39/go/ast/astutil/rewrite.go:47 +0x97
panic(0x975bc0, 0xd6c610)
/sdk/go1.15.8/src/runtime/panic.go:969 +0x1b9
go/types.(*Named).Obj(...)
/sdk/go1.15.8/src/go/types/type.go:473
mvdan.cc/garble.isTestSignature(0xc0001e7080, 0xa02e84)
/src/garble/main.go:1170 +0x7b
mvdan.cc/garble.(*transformer).transformGo.func2(0xc000122df0, 0xaac301)
/src/garble/main.go:1028 +0xff1
We were assuming that the first parameter was a named type, but that
might not be the case.
This crash was found out in the wild, from which a minimal repro was
written. We add two variants of it to the test data, just in case.
Unexported names are a bit tricky, since they are not listed in the
export data file. Perhaps unsurprisingly, it's only meant to expose
exported objects.
One option would be to go back to adding an extra header to the export
data file, containing the unexported methods in a map[string]T or
[]string. However, we have an easier route: just parse the Go files and
look up the names directly.
This does mean that we parse the Go files every time "reverse" runs,
even if the build cache is warm, but that should not be an issue.
Parsing Go files without any typechecking is very cheap compared to
everything else we do. Plus, we save having to load go/types information
from the build cache, or having to load extra headers from export files.
It should be noted that the obfuscation process does need type
information, mainly to be careful about which names can be obfuscated
and how they should be obfuscated. Neither is a worry here; all names
belong to a single package, and it doesn't matter if some aren't
actually obfuscated, since the string replacements would simply never
trigger in practice.
The test includes an unexported func, to test the new feature. We also
start reversing the obfuscation of import paths. Now, the test's reverse
output is as follows:
goroutine 1 [running]:
runtime/debug.Stack(0x??, 0x??, 0x??)
runtime/debug/stack.go:24 +0x??
test/main/lib.ExportedLibFunc(0x??, 0x??, 0x??, 0x??)
p.go:6 +0x??
main.unexportedMainFunc(...)
C.go:2
main.main()
z.go:3 +0x??
The only major missing feature is positions and filenames. A follow-up
PR will take care of those.
Updates #5.
First, make isPrivate panic on malformed import paths, since that should
never happen. This catches the errors that some users had run into with
packages like gopkg.in/yaml.v2 and github.com/satori/go.uuid:
panic: malformed import path "gopkg.in/garbletest%2ev2": invalid char '%'
This seems to trigger when a module path contains a dot after the first
element, *and* that module is fetched via the proxy. This results in the
toolchain URL-encoding the second dot, and garble ends up seeing that
encoded path.
We reproduce this behavior with a fake gopkg.in module added to the test
module proxy. Using yaml.v2 directly would have been easier, but it's
pretty large. Note that we tried a replace directive, but that does not
trigger the URL-encoding bug.
Also note that we do not obfuscate the gopkg.in package; that's fine, as
the isPrivate path validity check catches the bug either way.
For now, make initImport use url.PathUnescape to work around this issue.
The underlying bug is likely in either the goobj2 fork, or in the
upstream Go toolchain itself.
hashImport also gives a better error if it cannot find a package now,
rather than just an "empty seed" panic.
Finally, the sanity check in isPrivate unearthed the fact that we do not
support garbling test packages at all, since they were invalid paths
which never matched GOPRIVATE. Add an explicit check and TODO about
that.
Fixes#224.
Fixes#228.
In 90fa325da7, the obfuscation logic was changed to use hashes for
exported names, but incremental names starting at just one letter for
unexported names. Presumably, this was done for the sake of binary size.
I argue that this is not a good idea for the default mode for a number
of reasons:
1) It makes reversing of stack traces nearly impossible for unexported
names, since replacing an obfuscated name "c" with "originalName"
would trigger too many false positives by matching single characters.
2) Exported and unexported names aren't different. We need to know how
names were obfuscated at a later time in both cases, thanks to use
cases like -ldflags=-X. Using short names for one but not the other
doesn't make a lot of sense, and makes the logic inconsistent.
3) Shaving off three bytes for unexported names doesn't seem like a huge
deal for the default mode, when we already have -tiny to optimize for
size.
This saves us a bit of work, but most importantly, simplifies the
obfuscation state as we no longer need to carry privateNameMap between
the compile and link stages.
name old time/op new time/op delta
Build-8 153ms ± 2% 150ms ± 2% ~ (p=0.065 n=6+6)
name old bin-B new bin-B delta
Build-8 7.09M ± 0% 7.08M ± 0% -0.24% (p=0.002 n=6+6)
name old sys-time/op new sys-time/op delta
Build-8 296ms ± 5% 277ms ± 6% -6.50% (p=0.026 n=6+6)
name old user-time/op new user-time/op delta
Build-8 562ms ± 1% 558ms ± 3% ~ (p=0.329 n=5+6)
Note that I do not oppose using short names for both exported and
unexported names in the future for -tiny, since reversing of stack
traces will by design not work there. The code can be resurrected from
the git history if we want to improve -tiny that way in the future, as
we'd need to store state in header files again.
Another major cleanup we can do here is to no longer use the
garbledImports map. From a look at obfuscateImports, we hash a package's
import path with its action ID, much like exported names, so we can
simply re-do that hashing for the linker's -X flag.
garbledImports does have some logic to handle duplicate package names,
but it's worth noting that should not affect package paths, as they are
always unique. That area of code could probably do with some
simplification in the future, too.
While at it, make hashWith panic if either parameter is empty.
obfuscateImports was hashing the main package path without a salt due to
a bug, so we want to catch those in the future.
Finally, make some tiny spacing and typo tweaks to the README.
For now, this only implements reversing of exported names which are
hashed with action IDs. Many other kinds of obfuscation, like positions
and private names, are not yet implemented.
Note that we don't document this new command yet on purpose, since it's
not finished.
Some other minor cleanups were done for future changes, such as making
transformLineInfo into a method that also receives the original
filename, and making header names more self-describing.
Updates #5.
testscript already included magic to also account for commands in the
total code coverage. That does not happen with plain tests, since those
only include coverage from the main test process.
The main problem was that, before, indirectly executed commands did not
properly save their coverage profile anywhere for testscript to collect
it at the end. In other words, we only collected coverage from direct
garble executions like "garble -help", but not indirect ones like "go
build -toolexec=garble".
$ go test -coverprofile=cover.out
PASS
coverage: 3.6% of statements
total coverage: 16.6% of statements
ok mvdan.cc/garble 6.453s
After the delicate changes to testscript, any direct or indirect
executions of commands all go through $PATH and properly count towards
the total coverage:
$ go test -coverprofile=cover.out
PASS
coverage: 3.6% of statements
total coverage: 90.5% of statements
ok mvdan.cc/garble 33.258s
Note that we can also get rid of our code to set up $PATH, since
testscript now does it for us.
goversion.txt needed minor tweaks, since we no longer set up $WORK/.bin.
Finally, note that we disable the reuse of $GOCACHE when collecting
coverage information. This is to do "full builds", as otherwise the
cached package builds would result in lower coverage.
Fixes#35.
We tested that init funcs within a single file retain their order, but
not when they are split between many files.
Add one such case, with ten files and the same global-var-append
mechanism.
While at it, add some docs and make each init func take just one line.
It's common for asset bundling code generators to produce huge literals,
for example in strings. Our literal obfuscators are meant for relatively
small string-like literals that a human would write, such as URLs, file
paths, and English text.
I ran some quick experiments, and it seems like "garble build -literals"
appears to hang trying to obfuscate literals starting at 5-20KiB. It's
not really hung; it's just doing a lot of busy work obfuscating those
literals. The code it produces is also far from ideal, so it also takes
some time to finally compile.
The generated code also led to crashes. For example, using "garble build
-literals -tiny" on a package containing literals of over a megabyte,
our use of asthelper to remove comments and shuffle line numbers could
run out of stack memory.
This all points in one direction: we never designed "-literals" to deal
with large sizes. Set a source-code-size limit of 2KiB.
We alter the literals.txt test as well, to include a few 128KiB string
literals. Before this fix, "go test" would seemingly hang on that test
for over a minute (I did not wait any longer). With the fix, those large
literals are not obfuscated, so the test ends in its usual 1-3s.
As said in the const comment, I don't believe any of this is a big
problem. Come Go 1.16, most developers should stop using asset-bundling
code generators and use go:embed instead. If we wanted to somehow
obfuscate those, it would be an entirely separate feature.
And, if someone wants to work on obfuscating truly large literals for
any reason, we need good tests and benchmarks to ensure garble does not
consume CPU for minutes or run out of memory.
I also simplified the generate-literals test command. The only argument
that matters to the script is the filename, since it's used later on.
Fixes#178.
main.go includes a lengthy comment that documents this edge case, why it
happened, and how we are fixing it. To summarize, we should no longer
error with a build error in those cases. Read the comment for details.
A few other minor changes were done to allow writing this patch.
First, the actionID and contentID funcs were renamed, since they started
to collide with variable names.
Second, the logging has been improved a bit, which allowed me to debug
the issue.
Third, the "cache" global shared by all garble sub-processes now
includes the necessary parameters to run "go list -toolexec", including
the path to garble and the build flags being used.
Thanks to lu4p for writing a test case, which also applied gofmt to that
testdata Go file.
Fixes#180.
Closes#181, since it includes its test case.
If code includes a linkname directive pointing at a name in an imported
package, like:
//go:linkname localName importedpackage.RemoteName
func localName()
We should rewrite the comment to replace "RemoteName" with its
obfuscated counterpart, if the package in question was obfuscated and
that name was as well.
We already had some code to handle linkname directives, but only to
ensure that "localName" was never obfuscated. This behavior is kept, to
ensure that the directive applies to the right name. In the future, we
could instead rewrite "localName" in the directive, like we do with
"RemoteName".
Add plenty of tests, too. The linkname directive used to be tested in
imports.txt and syntax.txt, but that was hard to maintain as each file
tested different edge cases.
Now that we have build caching, adding one extra testscript file isn't a
big problem anymoree. Add linkname.txt, which is self-explanatory. The
other two scripts also get a bit less complex.
Fixes#197.
It turns out that the modules we include in testdata/mod via
txtar-addmod don't result in the same h1 hash that one gets when using
proxy.golang.org.
As proof:
$ go clean -modcache && go get -d rsc.io/quote@v1.5.2 && go test -short
[...]
--- FAIL: TestScripts/imports (0.06s)
> garble build -tags buildtag
go list error: exit status 1: verifying rsc.io/quote@v1.5.2: checksum mismatch
downloaded: h1:w5fcysjrx7yqtD/aO+QwRjYZOKnaM9Uh2b40tElTs3Y=
go.sum: h1:3fEykkD9k7lYzXqCYrwGAf7iNhbk4yCjHmKBN9td4L0=
The added comment explains the situation in detail.
For now, simply work around the issue by not sharing GOMODCACHE with the
host.
Previously, we were never obfuscating runtime and its direct
dependencies. Unfortunately, due to linkname, the runtime package is
actually closely related to dozens of other std packages as well.
Until we can obfuscate the runtime and properly support go:linkname
directives, obfuscating fewer std packages is a better outcome than
breaking and not producing any obfuscated code at all.
The added test case is building runtime/pprof, which used to cause
failures:
# runtime/pprof
/go/src/runtime/pprof/label.go:27:21: undefined: context.Context
/go/src/runtime/pprof/label.go:59:21: undefined: context.Context
/go/src/runtime/pprof/label.go:93:16: undefined: context.Context
/go/src/runtime/pprof/label.go:101:20: undefined: context.Context
The net package was also very close to obfuscating properly thanks to
this change, so its test is now run as well. The only other remaining
fix was to not obfuscate fields on cgo types, since those aren't
obfuscated at the moment.
The map is pretty long, but it's only a temporary solution and the
command to obtain the list again is included. Never obfuscating the
entire std library is also an option, but it's a bit unnecessary.
Fixes#134.
Dan Kortschak
Daniel Martí
lu4p
Andrew LeFevre