The corresponding command-line option `sapi_isystem` will be ignored for
compatibility.
PiperOrigin-RevId: 439806387
Change-Id: I8ad6d7feed2fba5fca9940281f03cfc757ada5be
This makes `sapi_library()` more compatible with Bazel's native `cc_library()`
rule.
PiperOrigin-RevId: 439512659
Change-Id: If731f600d56db56f78d2897e0c41a200daa93b75
This should fix the build on Debian 10 with LLVM 11.
- Keep order of files to look up in `ls` invocation
- Use `--start-group`/`--end-group` linker options
- Drive-by: use `splitlines()` instead of manual `split("\n")[:-1]`
PiperOrigin-RevId: 439248079
Change-Id: I919bb292ac8a5f514431aa004345f1c6478b1cc9
This change introduces an experimental `generator_version` attribute to the
`sapi_library()` rule. Version `1` will select the current interface
generator, which is based on libclang and Python. Setting the attribute to
version `2`, will select the newer interface generator written in C++ that
uses a full clang compiler frontend for parsing. Both emit equivalent header
output, differences in parsing and/or edge cases notwithstanding.
The default, as of now, is still the old version `1` generator.
Note: CMake allows to select the new interface generator globally by setting
`SAPI_ENABLE_GENERATOR`.
PiperOrigin-RevId: 438765013
Change-Id: I69c49a6bcf1751724edb0bce5c3b2beea2097138
This adds a workspace rule that inspects the current system first and
downloads a suitable version of LLVM/Clang from GitHub if it can't
find one. In the latter case, the necessary parts are build from source,
which can take a while (~10-15m, depending on the build machine).
In order to be found, LLVM/Clang system libraries must be version 11
or higher. On Debian/Ubuntu, install `llvm-13-dev` and `libclang-13-dev`.
The new `llvm_config.bzl` implements this logic. It is loosely based on
upstream's https://github.com/llvm/llvm-project/blob/main/utils/bazel/configure.bzl.
Note that due to the way Bazel separates local repositories, we have to
duplictate some of this code.
PiperOrigin-RevId: 438759950
Change-Id: Ia65f473b4cdef6507e3816bf09794ea10963d87a
This will allow us to experiment with (and subsequently migrate to)
changes to the generated API, possibly incompatible ones.
This change should be a no-op for current builds, as there is only
a single version of Sandboxed API.
PiperOrigin-RevId: 438003314
Change-Id: Ia23ea4360bee0227692d9f5220ab20d85f089ba7
This change merges the internal version of `sapi.bzl` with the external version again:
- Add more docstrings to the various macros
- Skip creation of `.isystem` file, get info from toolchain instead
PiperOrigin-RevId: 437730588
Change-Id: I6f670d32e3d7177a6a160fd24cbee6f8f3ca9503
This fixes the main issue (#118) with stack traces on Fedora, which uses a
`/lib64` and `/usr/lib64`.
PiperOrigin-RevId: 437717858
Change-Id: I6986aa84c2be57ae1d9f8d0cb9b508768d27f1c1
This uses the Google formatting style to format the prototype comments, with an
internal line length of 75, which accomodates the indentation in the generated
API class.
PiperOrigin-RevId: 435303665
Change-Id: I4dcdf0ed773a79ebc55ead3843f07ca8556fd985
This implements a custom compilation database to conditionally add the correct
language flags to the compiler frontend. Otherwise, a C header might receive
`--std=c++17` and fail.
Note: All headers are always processed in C++ mode. We expect that headers of
well-behaved C libraries contain `#ifdef __cplusplus`/`extern "C" {}` guards.
PiperOrigin-RevId: 435302048
Change-Id: Ib84e6e1f301ba434999846a012b3f8c16884648e
The enclosing type is enough to reconstruct the AST when writing the header and this
change avoids emitting the same struct twice.
PiperOrigin-RevId: 435300029
Change-Id: I34bd660db5ba5c68b64cce73ecf2f026727ac57b
This is a follow up to fa9e6e8a5c.
Drive-by:
- Replace deprecated calls to `getNameAsString()`
PiperOrigin-RevId: 435287759
Change-Id: I81d8c2f93b1ab23c781421b114779b7a241e4a7e
We have a SAPI_ASSERT_OK_AND_ASSIGN which corespondents to
SAPI_ASSIGN_OR_RETURN.
We also have SAPI_RETURN_IF_ERROR but we don't have a coresponding
macro for ASSERT.
I think that this completes the API and makes writting tests a little
bit simpler.
The feature is pure optimization, but it requires
additional syscalls.
PiperOrigin-RevId: 432954277
Change-Id: I1f345f8a26c86e09611fd575cb6ee080f24cc717
There are a lot of internal users depending on the old behavior of the
libclang-based generator.
PiperOrigin-RevId: 432281224
Change-Id: If82333fc3001f52de59e57a874f28bf8815d0877
The constexpr functions can be used to ensure that all branches actually compile
(unlike plain preprocessor `#ifdef`s).
PiperOrigin-RevId: 432186834
Change-Id: I1a8d97dac8480fe9d4543b0e9e39540ca1efc8fa
munmap is widely used by sanitizer, but it
probably works for Asan/Msan because it's enabled
by unrelated Allow* call.
Move mprotect to shared part as well. It will be
needed for compress_stack_depot.
PiperOrigin-RevId: 431989551
Change-Id: I7695a2de81d8d0b2112d3308778b2e9a9c7cb596
Use [`direct_headers`](https://bazel.build/rules/lib/CompilationContext#direct_headers)
from the Bazel/Blaze compilation context instead of _all_ transitive headers.
For the clang based generator, this means we don't try to parse
`textual_headers`, which will fail (they are by definition not
stand-alone, after all).
PiperOrigin-RevId: 431899423
Change-Id: I7a9dfa0dd93eba14b506b0e7ca6db3ed59b55dd6
Linking glibc in fully static mode is mostly unsupported. While such binaries
can easily be produced, conflicting symbols will often make them crash at
runtime. This happens because glibc will always (try to) load some dynamically
linked libraries, even when statically linked. This includes things like the
resolver, unicode/locale handling and others.
Internally at Google, this is not a concern due to the way glibc is being built
there. But in order to make all of our tests run in the open-source version of
this code, we need to change strategy a bit.
As a rule of thumb, glibc can safely be linked statically if a program is
resonably simple and does not use any networking of locale dependent
facilities. Calling syscalls directly instead of the corresponding libc
wrappers works as well, of course.
This change adjusts linker flags and sandbox policies to be more compatible
with regular Linux distributions.
Tested:
- `ctest -R '[A-Z].*'` (all SAPI/Sandbox2 tests)
PiperOrigin-RevId: 429025901
Change-Id: I46b677d9eb61080a8fe868002a34a77de287bf2d
- Link `zipe.c` statically (safe)
- Update policy to allow any use of `stat()`
PiperOrigin-RevId: 428971638
Change-Id: Ib0f5f496ea2389582986b41a8830592e6c1d4390
Instead of C++17 structured bindings, use a plain `const auto&` and annotate
arguments with comments instead.
We still support Clang 6.0, as that is the compiler that ships with Ubuntu
18.04 LTS by default.
PiperOrigin-RevId: 428016214
Change-Id: I3a43b2d47c6825ac4425d22018750282cfe23c1b
This fixes a couple of tests in the open source version of the code.
Internally, since we are using a different ELF loader, the page offset
will always be zero. Hence we never notices this was broken.
PiperOrigin-RevId: 427996428
Change-Id: I44c5b5610b074cf69b9f0c5eeb051be50923e351
Note that `//sandboxed_api/sandbox2:stack_trace_test` may still fail for
unrelated reasons, as we are linking libc statically, which is brittle. A
follow-up change will fix this.
PiperOrigin-RevId: 427175045
Change-Id: Ifb5ec2ac3d60f4bcc9708f26c834c83b75e769d7
This partially reverts the zlib change in 41e0ca0. Turns out the
`CMakeLists.txt` that ships with zlib leaves much to be desired.
PiperOrigin-RevId: 424800727
Change-Id: I356e3bb8d18461a52f845baa4913adff6549ef00
Plain `readdir` is preferred and while not (yet) specified in POSIX it is thread-safe for different directory streams in popular implementations.
PiperOrigin-RevId: 423321528
Change-Id: I4e1e842f338ff7d690c36e7f699b2f3637609524
Newer versions of libunwind use `PTRACE_GETREGSET` to obtain register data.
This change should make it easier to upgrade the libunwind dependency.
PiperOrigin-RevId: 420057842
Change-Id: Ib9abbeff574e457009709715f912ba5962033c5d
Raw `SAPI_RAW_PCHECK` may dump core, depending on environment settings
(issue #89).
This is undesirable in the face of invalid command-line arguments.
Signed-off-by: Christian Blichmann <cblichmann@google.com>
- Allow to specify multiple syscalls with `BlockSyscallsWithErrno()`
- Add functions to allow `unlink()` and `rename()` in all their spellings
PiperOrigin-RevId: 414987303
Change-Id: Ic0e680b785e8e3a3498f20e6a7403737e63fe876
__sanitizer_sandbox_on_notify is not tsan specific.
It's empty for other sanitizers now, but we are going to need it soon.
PiperOrigin-RevId: 414873197
Change-Id: I251ac38e5c886980b4baa7f05306643599a25090
Move VecStringToCharPtrArr before fork, so that it cannot deadlock when other thread holds allocation lock.
PiperOrigin-RevId: 414661912
Change-Id: Ie8aa5c36693e6f86c69d67a1da51b7e7ff1ec30b
This change is a first step to make the SAPI variable hierarchy more sensible.
It turns the `Reg<T>` class into a descendant of `Pointable`, but without
making its `PtrXXX()` methods public (hence the `using` statements). Further
changes are needed to restructure this.
There are no functional changes and the class sizes, including vtables, should
not change.
PiperOrigin-RevId: 413333120
Change-Id: I90ceeaeb7aea482016f8f4bee81489d5a9db9ade
Different versions of the `elf.h` header define their own integer types. For
example, even on LP64 systems, a 64-bit ELF integer types may decay into
`unsigned long long` instead of `unsigned long`.
This change replaces the various overloads with a single function template
that is well-defined for all integral types.
PiperOrigin-RevId: 410746713
Change-Id: I4b560f7541802372f01ae3d6f4a56554e51d70c8
Generate syscall jump table without using bpf_helper.
Check that any jump in the user provided policy is within the provided policy.
PiperOrigin-RevId: 409362089
Change-Id: I31493e52cf868e4b184ff79fcb26beeb75f49773
This is mainly so that the structure of the examples follows what we do
internally (not having separate directories).
PiperOrigin-RevId: 402298115
Change-Id: I0f542607b88597572de39532364816f80a076697
This is the first change in a series that will eventually remove Sandboxed
API's use of multiple inheritance.
Drive-by:
- Rename short member names to full words
- Some reformatting
PiperOrigin-RevId: 402270954
Change-Id: I8af46b887921265a371b85603fd158ef3a8fab50
Note: This change allows `MAP_NORESERVE` generally, not just for MSAN. This follows
what we do for `AllowTcMalloc()/AllowSystemMalloc()`
PiperOrigin-RevId: 402231980
Change-Id: Ifa1c6b9f61f636dd6db231dde3765c3b4a40911b
These were previously dependent on an internal-only testing target.
For now, this only works with Bazel, but should enable us to have better test coverage in GitHub actions.
Eventually, all of these shell-based tests should be converted to `cc_test`s.
PiperOrigin-RevId: 400713615
Change-Id: I1cabb5b72977987ef4a1803480f699b58c4d56e9
For OSS, this change should be mostly a no-op. Visible edits are due to
changed order of code and/or includes.
PiperOrigin-RevId: 394177395
Change-Id: I1d32f9fd175579e8f05c051b1307953b249d139d
This mainly a debugging facility.
It makes diagnosing problems where sandboxed process just randomly exits whereas unsandboxed one runs to completion due to differences in the setup/environment much easier.
PiperOrigin-RevId: 391005548
Change-Id: Ia19fe6632748da93c1f4291bb55e895f50a4e2b0
Otherwise starting forkserver multiple times will result in zombie processes lingering around.
PiperOrigin-RevId: 388926497
Change-Id: Ia9947cce3d9e909edd709b0d3525e1ae8b8bbc51
Also really own `exec_fd_` as previously if the executor is destructed without calling `StartSubProcess` the file descriptor would leak.
PiperOrigin-RevId: 388901766
Change-Id: I6bbb15ced37a0a832ec5a5228452a3d54ef46ee9
This was missing a friend declaration in order to actually compile.
It's now being used in the "stringop" example, so we test it as well.
Drive-by:
- Do not copy the proto's bytes the constructor, but use `std::move`
PiperOrigin-RevId: 387774353
Change-Id: Ic8824af911ac744e2e68130e1f4673c4dddd4939
Calling `Terminate()` issues additional syscalls that may clobber the `errno`
value. Reordering the log statements ensures we actually log the initial error
in `read()`/`write()`.
PiperOrigin-RevId: 387576942
Change-Id: I0f9c8c6001e6dc4ca098abe02cd251029f92a737
1. In many cases, sandboxes need to allow /proc/stat and /proc/cpuinfo so that
get_nprocs(3) will work; otherwise, per-CPU logic can't determine how many CPUs
there are. Unfortunately, some of those sandboxes also disable namespaces. The
solution is to provide two functions: AllowRestartableSequencesWithProcFiles(),
which allows syscalls and files; and AllowRestartableSequences(), which allows
syscalls only. Sandboxes should usually call the former; sandboxes that disable
namespaces should instead call the latter and are responsible for allowing the
files via the deprecated Fs mechanism.
2. Make the mmap(2) policy evaluate prot AND flags, not prot OR flags.
3. Order the code and the comments identically for better readability.
PiperOrigin-RevId: 386414028
Change-Id: I016b1854ed1da9c9bcff7b351c5e0041093b8193
Ideally, we'd seal the embedded SAPI binary using fcntl(). However, in rare
cases, adding the file seals `F_SEAL_SEAL | F_SEAL_SHRINK | F_SEAL_GROW |
F_SEAL_WRITE` results in `EBUSY` errors.
This is likely because of an interaction of `SEAL_WRITE` with pending writes
to the mapped memory region (see `memfd_wait_for_pins()` in Linux'
`mm/memfd.c`). Since `fsync()` is a no-op on memfds, it doesn't help to
ameliorate the problem.
On systems where it is enabled, ksmd might also be a source of pending writes.
PiperOrigin-RevId: 385741435
Change-Id: I21bd6a9039be4b6298774e837ce3628180ed91a8
The existing function signature took a `unique_ptr<>` owning a vector, and
took `nullptr` to mean an empty set of capabilities. This is more naturally
modeled by taking the vector directly and `std::move`-ing it.
PiperOrigin-RevId: 384214849
Change-Id: I177f04a06803ae00429b19a1f3f12e7be04d2908
- Assign to `*mutable_XXX()` instead of looping
- Use a const ref for capabilities
PiperOrigin-RevId: 384192675
Change-Id: I4db3d0c8ce0d7f6acc9fd486a2409962516b5fe7
This bug only manifests if a lot of fds are open when global forkserver is started.
If the allocated exec_fd number was equal Comms::kSandbox2ClientCommsFD then it would be replaced by the comms fd and result in EACCESS at execveat.
PiperOrigin-RevId: 380805414
Change-Id: I31427fa929abfc60890477b55790cc14c749f7f5
Recenly, Debian based distribution kernels started activating the Tomoyo Linux
Security Module by default. Even if it is not used, this changes the behavior
of `/dev/fd` (pointing to `/proc/self/fd` by default), which Sandbox2 needs during
`execveat()`.
As a result, Sandbox2 and Sandboxed API always fail without one of the following
conditions
- `/proc` mounted within the sandboxee
- `/dev` mounted
- `/dev/fd` symlinked to `/proc/self/fd` in the sandboxee's mount namespace
Some code pointers to upstream Linux 5.12.2:
- https://elixir.bootlin.com/linux/v5.12.2/source/fs/exec.c#L1775
- https://elixir.bootlin.com/linux/v5.12.2/source/security/tomoyo/tomoyo.c#L107
- https://elixir.bootlin.com/linux/v5.12.2/source/security/tomoyo/domain.c#L729
To find out whether your system has Tomoyo enabled, use this command, similar to
what this change does in code:
```
$ cat /sys/kernel/security/lsm | grep tomoyo && echo "Tomoyo active"
capability,yama,apparmor,tomoyo
Tomoyo active
```
The config setting `CONFIG_DEFAULT_SECURITY` controls which LSMs are built into
the kernel by default.
PiperOrigin-RevId: 372919524
Change-Id: I2181819c04f15f57d96c44ea9977d0def4a1b623
Depending on architecture and optimization level, the compiler may choose to
not generate full stack frames, even with no-inline and no tail-call
attributes.
PiperOrigin-RevId: 372339987
Change-Id: I42043131bbb6092ff234e80ae9047f7a2bf31161
This fixes tests for PPC, where the tail-call optimization would consistently
remove 'violate()' from the stack trace.
PiperOrigin-RevId: 371103794
Change-Id: Ifb1a7d588a455041a6b0f3c763276ed44de47e60
On x86 `long double` has 10 bytes of meaningful data, but `sizeof(long double)` is 16 - the remaining bytes are random garbage.
Roll forward after fixing a bug in the original commit.
PiperOrigin-RevId: 368170639
Change-Id: I4a1d2d95b92eed6b71c37145726f7320cfc00ba0
On x86 `long double` has 10 bytes of meaningful data, but `sizeof(long double)` is 16 - the remaining bytes are random garbage.
PiperOrigin-RevId: 367423349
Change-Id: I769b3444ce4fa60f941ccd2115b0b09ccc809f13
This is needed for some compiler versions where `absl::string_view` == `std::string_view`.
PiperOrigin-RevId: 367392064
Change-Id: Id91d23510501df4745f386475ef9049d94062e1b
This changes the workflow definition so that we always try to install
compiler toolchains that we need.
See https://github.com/actions/virtual-environments/issues/2950 for more
context.
Drive-by:
- Mini fix to enable compilation under Clang 6.0
Signed-off-by: Christian Blichmann <cblichmann@google.com>
Now unwinding will properly handle binaries inside bind-mounted directories.
Drive-by:
- Get rid of n^2 path handling
- Get rid of namespace alias
PiperOrigin-RevId: 358353666
Change-Id: Ieec7690ec6a1ae6d358de375220566b69e8cb094
Bazel readily enforces header visiblity for each target, CMake is more lenient.
PiperOrigin-RevId: 355407845
Change-Id: Ic59fa2162db8456d4c5cf4205c0fe42cc79874a9
The former is to fix compilation on GCC 7, the latter to satisfy MSAN.
PiperOrigin-RevId: 355114355
Change-Id: I5c89a65df16fe9338bcfa24b2e48c246d240ce62
Using C++17 means we can get rid of many `#ifdef`s by using `if constexpr`.
This way, we ensure that both branches compile and still retain zero runtime
overhead.
Note that open source builds of Sandboxed API do not ship with sanitizer
configurations yet. This will be added in follow-up changes.
PiperOrigin-RevId: 354932160
Change-Id: I3678dffc47ea873919f0a8c01f3a7d999fc29a5b
Also accept `absl::string_view` and `absl::Span<const std::string>` arguments.
Drive-by:
- Move using declaration into namespace
PiperOrigin-RevId: 354271016
Change-Id: Iadd873377e51cac7fa3800aab1f9e85ff94bd4e9
