**DO NOT CLOSE** Notification V2#2
Draft
coder0107git wants to merge 133 commits intocoder0107git:notification-hackfrom
Draft
**DO NOT CLOSE** Notification V2#2coder0107git wants to merge 133 commits intocoder0107git:notification-hackfrom
coder0107git wants to merge 133 commits intocoder0107git:notification-hackfrom
Conversation
Clang has added a new warning for virtual specifiers on methods of `final` classes. This cleans them up in preparation for enabling that warning. Bug: chromium:403236787 Change-Id: I1bb144494028f674d0de409a65412af1f084b56c Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6391658 Commit-Queue: Mark Foltz <mfoltz@chromium.org> Reviewed-by: Mark Foltz <mfoltz@chromium.org>
Currently, Chromium openscreen builders are failing for timeout since remote execution is disabled. https://ci.chromium.org/ui/p/openscreen/builders/try/chromium_linux_x64/b8719177567471980433/overview Those builders need to switch from Ninja to Siso. The new properties are necessary for the migration. Bug: 379584977 Change-Id: I52f42dcc247a70cf609e0fcfd6c916ed914ba531 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6436889 Reviewed-by: Mark Foltz <mfoltz@chromium.org>
This patch resolves a longish standing TODO to move the statistics generation code out of the Sender class and into its own, unit tested class. This class is called StatisticsDispatcher. Bug: 298277160 Change-Id: I755beb6c4222f65a94a9b5503a708c431e52bc1b Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6398046 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Muyao Xu <muyaoxu@google.com>
The chromium builders don't use Reclient anymore. Bug: 379584977 Change-Id: I2071ea6c93c3af80e19a82dfd93ddf25fbbf5725 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6439985 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Mark Foltz <mfoltz@chromium.org>
Closed
Clang maps -Wall (or /Wall) to -Weverything on windows, so it causes low-value warnings to trigger during out build. Instead, we should use /W4, which corresponds to gcc's `-Wall -Wextra`. Bug: chromium:410803303 Change-Id: Id512338bf412794e4b7278e7b6332bc44fe74305 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6457698 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Mark Foltz <mfoltz@chromium.org>
This patch improves error messaging around certificate loading, especially private keys. Change-Id: I5c00b44a9bed04ec51826d78805b0cb29f90e855 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6473163 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Muyao Xu <muyaoxu@google.com>
See comment on https://chromium-review.googlesource.com/c/openscreen/+/6457698 This config mostly exists to disable -Wshadow for one cc file. Instead, just build that one file as no_chromium_code. No effective behavior change. Bug: chromium:410803303 Change-Id: I952057f0a31f0a150926c55e65fd6ade89b93807 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6483407 Commit-Queue: Mark Foltz <mfoltz@chromium.org> Reviewed-by: Mark Foltz <mfoltz@chromium.org>
This patch fixes a critical infra failure where the build is currently broken due to a minor speed improvement made to the clang update script. This patch rolls the build and buildtools submodules as well as updates the clang update download script to pull a specific version of the update script, to avoid breaking changes like this in the future. Change-Id: I975ecaf1667cf03c57ac3f7bd15314b1c54281a9 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6544748 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This patch updates the platform implementation to not subscribe to wait events for UDP sockets. It is part of a larger cleanup to improve the amount of time spent in loops in the platform implementation. Change-Id: Ib09474a1ee05629770acd6b2d7893276bb965e79 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6539456 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This patch optimizes a check for Writable handles, ensuring that we don't pass a handle to the write argument for select() if we don't have anything to write. This lowers the amount of cycles spent polling the select loop. This is a not-insignificant change, involving updating the SocketHandleWaiter interface and its subscribers to allow the waiter to poll for whether the socket has pending write events or not. Change-Id: Icbd38a67fbc1fa4856f9605d6306971568dcca5e Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6556928 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This also requires a change in chromium/src. Bug: b:374199735 Change-Id: I76bc61259d939ff8c87aba84650f228de92c5c2e Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6533602 Reviewed-by: Jordan Bayles <jophba@chromium.org> Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Antonio Rivera <antoniori@google.com>
Also updated protobuf revision to match the current version. Bug: 421988683, 421989590, 421989329, 421989630, 421989709, 421988806, 421989809, 421988937 Change-Id: I45e3cbc029bee0f5f04e2fb0c8f76c9415b7eeba Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6768426 Reviewed-by: Takumi Fujimoto <takumif@chromium.org> Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Jordan Brown <rop@google.com>
takumif@ is no longer on the team. Change-Id: Ie42bf4e833253c12346e43434457c60a87e2b9f6 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6882217 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Takumi Fujimoto <takumif@chromium.org>
Python3 is used by default in recipes now and this config is not necessary anymore. Bug: 440235171 Change-Id: I89d420b8481fe5af08704939f74033226fd511ba Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6888255 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Takuto Ikuta <tikuta@chromium.org>
Bug: None Change-Id: I6dc21fff88a8e61806e98defb5178ce4bd8bcc3a Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6896306 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Mark Foltz <mfoltz@chromium.org>
Not needed and deprecated [0]. This CL should be a no-op. [0]: https://chromium.googlesource.com/external/github.com/google/googletest.git/+/a05c0915074bcd1b82f232e081da9bb6c205c28d/googlemock/include/gmock/gmock-actions.h#2046 Bug: chromium:439838457 Test: `openscreen_unittests` in Chromium Change-Id: I73446be5df46235049139edde087dfa17f7ded15 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6897917 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Mark Foltz <mfoltz@chromium.org>
Not needed and deprecated [0]. This CL should be a no-op. [0]: https://chromium.googlesource.com/external/github.com/google/googletest.git/+/a05c0915074bcd1b82f232e081da9bb6c205c28d/googlemock/include/gmock/gmock-actions.h#2046 Bug: chromium:439838457 Test: `openscreen_unittests` in Chromium Change-Id: Ib4ef3f314a6b0348ac276c7145b5f5862b09853c Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6897919 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Mark Foltz <mfoltz@chromium.org>
Not needed and deprecated [0]. This CL should be a no-op. [0]: https://chromium.googlesource.com/external/github.com/google/googletest.git/+/a05c0915074bcd1b82f232e081da9bb6c205c28d/googlemock/include/gmock/gmock-actions.h#2046 Bug: chromium:439838457 Test: `openscreen_unittests` in Chromium Change-Id: If131ba9ea908a1f2918dc9e712821341714a35a1 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6897920 Commit-Queue: Mark Foltz <mfoltz@chromium.org> Reviewed-by: Mark Foltz <mfoltz@chromium.org>
https://chromium.googlesource.com/external/github.com/google/googletest.git/+log/b514bdc898e2..eb2d85edd0bf $ git log b514bdc89..eb2d85edd --date=short --no-merges --format='%ad %ae %s' 2025-08-24 absl-team Remove unused syslog dependency for Fuchsia. 2025-08-19 absl-team Internal header include changes. 2025-08-16 absl-team Deprecate single-argument DoAll and Invoke. 2025-08-11 absl-team Update the document in typed tests to use using-declaration instead of typedef 2025-07-28 absl-team Provide a better error message when ASSERT and SKIP macros are used in methods that return values 2025-07-23 absl-team Skip the predicate on SIGSEGV in death-test on Android builds with API level <= 23. 2025-07-18 absl-team Fix the sample usage of `ConvertGenerator` by removing the type-id name. 2025-07-12 absl-team Automated Code Change 2025-07-04 absl-team Replace internal link in GoogleTest docs 2025-06-27 pizzud gtest: Reword fail_if_no_test_selected_message for sharding and filtering. 2025-06-24 absl-team Add documentation for exception matchers. 2025-06-13 absl-team Although the following paragraph explains there is a better solution, having this technique in the bullet point seems to suggest that this technique is considered as a valid alternative. It would be better to drop it or make it clear that this technique is not recommended. 2025-06-12 absl-team Add UnpackStructImpl for structs with 24, 25 and 26 fields. 2025-06-11 joel.langlois README.md: Fixed broken Markdown link formatting. 2025-06-10 absl-team Automated Code Change 2025-06-09 absl-team In MatcherCast, store the input value as its own type rather than as the Matcher type, to avoid dangling references 2025-06-09 pizzud googletest: Add a flag to fail if no tests were selected to run. 2025-06-05 absl-team Remove "blindly" from the gmock "uninteresting call" message. 2025-05-30 absl-team Fix unified diff headers. 2025-05-30 absl-team Use the provided length in ConditionalPrintAsText 2025-05-25 nextgen.gubka Fix extra ';' after member function definition 2025-05-22 absl-team Updates Google Analytics tracker. 2025-05-22 dcheng Print std::basic_string_view<Char> as a string literal 2025-05-20 dmauro Delete the absl polyfill support for std::any, std::optional and std::variant now that the absl types are aliases of the std types 2025-05-19 absl-team Add UnpackStructImpl specialization for 23 struct members. 2025-05-19 absl-team Use static_cast instead of ImplicitCast_ for character conversions 2025-05-15 dmauro GoogleTest CI Update 2025-05-13 mkruskal Add [[nodiscard]] to GetParam() to prevent accidental misuse. 2025-05-12 absl-team Clarify that the return value of `InvokeArgument` is not the return value of the action. 2025-05-01 absl-team gtest: Remove incorrect documentation of private headers as PIMPL idiom 2025-04-30 dmauro Update documentation for v1.17.0 2025-04-30 dmauro Update GoogleTest dependencies in preparation for release 2025-04-29 lar Use std::memcpy instead of a union to access floating point bits. 2025-04-29 dmauro Fix GCC15 warning that <ciso646> is deprecated in C++17 2025-04-23 absl-team AllOf, AnyOf, Optional: Avoid generating unnecessary match explanations 2025-04-17 dmauro Bump RE2 dependency to 2024-07-02.bcr.1 2025-04-09 jacobsa `gmock-actions`: add a missing conversion operator to `WithArgsAction`. 2025-04-09 jacobsa `gmock-actions`: run clang-format. 2025-03-26 durandal Remove obsolete references to LinuxThreads. 2025-03-21 absl-team Add UnpackStructImpl for structs with 22 fields. 2025-03-21 ototot Fix a typo 2025-03-20 absl-team Add a non-'const' overload for the function invoked by the IgnoreArgs action wrapper. 2025-03-18 wan Improve the failure messages of `ElementsAre()`, `ElementsAreArray()`, etc. 2025-03-11 wan Clarify the polymorphic matcher concept and explain how to define a composite matcher. 2025-03-11 wan Allow `DistanceFrom()` to use user-defined `abs()` by default. 2025-03-07 wan Add a `DistanceFrom()` matcher for general distance comparison. 2025-03-04 absl-team Add a `testing::ConvertGenerator` overload that accepts a converting functor. This allows the use of classes that do not have a converting ctor to the desired type. 2025-02-28 dmauro Try to warn the user when test filters do not match any tests 2025-02-27 pizzud googletest: Add testonly alias of :gtest for testonly cc_libraries to use. 2025-02-27 absl-team Add UnpackStructImpl for structs with 21 fields. 2025-02-26 dmauro Output to the test warning file if no tests are linked. 2025-02-24 wan Adds support for a --gtest_fail_if_no_test_linked flag to fail the test program if no test case is linked in (a common programmer mistake). 2025-02-21 dmauro GoogleTest C++17 cleanups * Use `[[nodiscard]]` instead of GTEST_MUST_USE_RESULT_ * Use `[[maybe_unused]]` instead of GTEST_INTERNAL_ATTRIBUTE_MAYBE_UNUSED * Use `[[deprecated]]` instead of `GTEST_INTERNAL_DEPRECATED` * Remove `GTEST_INTERNAL_NEED_REDUNDANT_CONSTEXPR_DECL` 2025-02-17 jmr Property: Promote note to warning 2025-02-12 dmauro Require C++17 2025-02-10 absl-team Add SaveArgByMove 2025-02-07 dmauro Update docs for v1.16.x branch 2025-02-06 dmauro Update GoogleTest deps in preparation for release 2025-02-06 dmauro Remove empty glob pattern, which is now an error. 2025-01-31 absl-team Pull Regexp syntax out of Death test section in advanced.md 2025-01-31 absl-team Fix extended regular expressions reference URL in advanced.md 2025-01-30 absl-team Export testsuite properties as XML elements, not attributes. 2025-01-16 absl-team Cast mutable lvalue references to const from testing::ResultOf 2025-01-15 absl-team Enable safe matcher casts from `Matcher<const T&>` to `Matcher<T>`. 2025-01-10 hector.dearman Default to color output for ghostty 2025-01-09 absl-team Document unexpected interaction with death test error code and gmock. 2025-01-08 absl-team Add UnpackStructImpl for structs with 20 fields. 2024-12-26 absl-team gtest: Output a canned test case for test suite setup / teardown failures in XML/JSON 2024-12-23 absl-team Put the fake Fuchsia SDK in a module extension 2024-12-16 absl-team the public version already has the const qualifier 2024-12-11 absl-team Update the links to ISTQB glossary. 2024-12-10 absl-team Add support for printing C++20 std::*_ordering types to gtest. Adds feature test macro for C++20 <compare> header, a pretty-printer, and tests. Inexplicably, these types aren't enums, so can't be handled with a switch. 2024-11-15 andy.soffer Fix documentation bug. 2024-11-04 absl-team Fixing a typo in the documentation of IsSubsetOf(). 2024-10-31 absl-team Remove mention of using `ASSERT_...` statements inside custom matcher definitions. 2024-10-27 pbos Use FAIL() in GTEST_SKIP() documentation 2024-10-16 absl-team Avoid creating std::vector<const T> in UnorderedElementsAreArrayMatcher and others. 2024-10-14 absl-team This change adjusts how `ASSERT_NEAR` and `EXPECT_NEAR` treats infinity, such that `ASSERT_NEAR(inf, inf, 0)` passes. This makes the behavior more consistent with `ASSERT_EQ(inf, inf)` which succeeds. 2024-10-04 absl-team Automated Code Change 2024-10-01 absl-team Optional(): Add support for std::optional<>-like types lacking bool conversion. 2024-09-16 absl-team Use matcher's description in AnyOf when matcher has no explanation. 2024-08-29 absl-team Automated Code Change 2024-08-01 dmauro Update main to point to 1.15.2 patch release 2024-07-30 dmauro Remove auto-detection of Python toolchain from MODULE.bazel since it affects downstream users 2024-07-30 absl-team Directly include `gmock.h` and `gtest.h` in gmock-matchers tests. 2024-07-24 absl-team Use matcher's description in AllOf if matcher has no explanation. 2024-07-23 jacobsa gmock-actions: make DoAll convert to OnceAction via custom conversions. 2024-07-23 jacobsa Run clang-format. 2024-07-20 dmauro Change the Bazel build to use the canonical repo names used by Bzlmod and the Bazel Central Registry 2024-07-16 dmauro Update GoogleTest Bazel quickstart for Bzlmod 2024-07-16 dmauro Update main branch to reflect the release of 1.15.0 2024-07-16 absl-team Use matcher's description in AllOf if matcher has no explanation. 2024-07-15 absl-team Use matcher's description in AllOf if matcher has no explanation. Created with: roll-dep third_party/googletest/src Change-Id: I8352afc32791030187946292a2f5aca6fcc8a05a Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6897921 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Mark Foltz <mfoltz@chromium.org>
Bug: 40249958 Change-Id: I44cac4484c5172f71f96699b6dc9e41e632f5fd8 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6909331 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Mark Foltz <mfoltz@chromium.org>
…6e8a8acb91a This picks up a couple *years* worth of BoringSSL changes. The main relevant change here is that generate_build_files.py is no longer needed! Everything it generated is now checked in BoringSSL proper, including a GN source list. $ git log 8d19c850d..26e8a8acb --date=short --no-merges --format='%ad %ae %s' 2025-08-22 davidben bssl-crypto: Fix Ed25519 SPKI parser to check key types 2025-07-09 agl Add a jitter entropy source. 2025-08-15 davidben Set an EVP_PKEY's algorithm and data together 2025-08-15 davidben Make some more half-empty EVP_PKEY states impossible 2025-08-14 davidben Limit EVP_PKEY_set_type to EVP_PKEY_X25519 2025-08-19 davidben Add a test that arbitrary curves can be wrapped in EVP_PKEY 2025-08-15 davidben Stop tracking an ENGINE in EVP_PKEY_CTX 2025-08-13 davidben Rename EVP_PKEY_METHOD to EVP_PKEY_CTX_METHOD 2025-08-15 davidben Pull the EC_GROUP_new_by_curve_name up into EVP_PKEY_CTX_set_ec_paramgen_curve_nid 2025-08-16 davidben Split evp_tests.txt into separate files 2025-08-16 davidben Add a test for SPKI and PKCS8 parsing with unknown algorithms 2025-08-20 chlily Fix markdown syntax in BUILDING.md 2025-08-17 davidben Test async BIO_flush and fix a corner case 2025-08-15 davidben Rename EVP_R_EXPECTING_AN_EC_KEY_KEY to match OpenSSL 2025-08-16 davidben Document the generators for all the MODP groups 2025-08-15 davidben Remove an old, impractical TODO 2025-08-15 davidben Say a bit more in docs about how to use EVP_PKEY_CTX 2025-08-18 davidben Bump version for BCR 2025-08-15 davidben Run through more code in PKCS#8 and SPKI parsers 2025-08-18 davidben Fix typo in comment 2025-08-17 davidben runner: Remove need for an AllCurves value 2025-08-14 davidben Remove redundant copy of EVP_PKEY type 2025-08-15 davidben Remove EVP_PKEY_print_* support for DSA 2025-08-15 davidben Don't support parameterless DSA keys in SPKIs 2025-08-15 davidben Add a couple more no-op compatibility functions 2025-08-14 davidben More consistently reset EVP_PKEYs in free_it 2025-08-14 davidben Unexport the Kyber implementation 2025-08-14 davidben Deprecate EVP_PKEY_set_type 2025-08-14 davidben Fix reference to public key in evp.h docs 2025-08-13 ellyjones pki: add PEMDecode and PEMDecodeSingle 2025-06-17 agl Switch to using a derivation function in CTR-DRBG. 2025-08-13 ellyjones pki: allow span<string_view> for allowed types in PEMTokenizer 2025-08-13 chlily Style guide: mention placement of & for reference types 2025-08-12 davidben Update releasing docs slightly 2025-08-12 chlily Implement MLKEM1024 for TLS 2025-08-06 agl delocate: support 4-bit tweak immediates for AArch64. 2025-08-08 davidben Add a discussion about callbacks to API Conventions 2025-06-17 agl Don't run CTR-DRBG in Kyber tests. 2025-08-07 agl Link to ACVP results for 20250728. 2025-08-07 davidben Bump Bazel deps and version for BCR 2025-08-05 chlily Expand BCM abbreviation in the FIPS.md documentation 2025-08-04 chlily Add IWYU export in nid.h file generated by objects.go 2025-08-04 chlily Fix documentation typo 2025-08-04 chlily Revert Rust config for 32-bit Linux CI/CQ bots 2025-08-04 chlily Run Rust recipe steps on Mac CI/CQ builders and 32-bit Linux 2025-07-31 chlily Run Rust recipe steps on "win64_rel" builder 2025-07-31 chlily Make Windows Rust try builder use a Release build 2025-07-29 chlily Add a default-off win64_rust try builder for testing 2025-07-28 chlily Run Rust recipe steps on "linux" builder 2025-07-28 agl Fix FIPS version number comment. (...) 2023-09-01 alex.gaynor Return the correct value in EVP_CIPHER_CTX_iv_length after EVP_CTRL_AEAD_SET_IVLEN 2023-06-01 yukl Add X25519 bindings for bssl-crypto 2023-06-01 yukl Add ecdh and P256 bindings to bssl-crypto 2023-09-01 davidben Update the warnings on split handshakes and handshake hints 2023-09-01 davidben runner: Check that the shim HRRs echo the session ID 2023-09-01 davidben Reland "Build with C11 on MSVC in the standalone Bazel build" 2023-08-31 yukl Fix cargo clippy and fmt 2023-08-17 scw delocate: accept more directive expressions 2023-08-29 davidben Initialize libunwind cursors from the signal handler's ucontext_t 2023-08-24 davidben Remove another decltype(fclose) 2023-08-28 davidben Save trampoline state in unwind tests more straightforwardly 2023-08-29 davidben Explicitly mark saved registers with .cfi_restore 2023-08-28 bbe Remove undesired OPENSSL_EXPORT's in non-library code 2023-08-28 davidben Fix RBP-based unwind in fiat assembly 2023-08-28 bbe Remove another OPENSSL_EXPORT in an enum 2023-08-24 davidben Silence warn_unused_result warning on write() call 2023-08-25 bbe Remove OPENSSL_EXPORT from enum in fillins/net_errors.h 2023-08-25 bbe Ensure OPENSSL_EXPORT is only on the class not methods. 2023-08-24 bbe Add pki file lists to generated gn build files 2023-06-05 nwadih Add Rust bindings to AES_CTR through EVP_* cipher API's 2023-08-24 davidben Export CBS/CBB unicode functions 2023-08-24 bbe rename fillins/string_util.h because gn is special 2023-08-24 bbe rename fillins/base64.c to handle gn being special 2023-08-24 davidben Also suppress OPENSSL_LINUX for nanolibc 2023-08-24 davidben Add BN_bn2lebinpad and BN_lebin2bn 2023-08-17 davidben Use a callable type for ScopedFILE in settings_writer.cc 2023-06-02 nwadih Add Rust bindings to AES-GCM through the EVP_AEAD_* APIs 2023-08-18 davidben Deduplicate the three copies of OBJ_cmp 2023-08-17 bbe Remove header file accidentally added to sources.cmake 2023-08-17 bbe Sync pki to chromium d740199e083b70d13506973c6f479f0b01165a05 2023-07-25 bbe Move the fuzzers into the fuzz directory and make them build. 2023-08-16 davidben Fix typo in .size directives for aesni_gcm_encrypt 2023-08-16 bbe Hook in ocsp into libpki, and run the ocsp unit tests. 2023-08-16 bbe Update pki to chromium cf9a08ff8be3a3f2d5b13693cc13ef22ab7ee618 2023-08-16 bbe Make a BSSL_CHECK that always aborts 2023-05-19 nwadih Add rust bindings to AES-GCM-SIV through the EVP_AEAD_* API's 2023-08-14 davidben Add EVP_HPKE_KEY_move 2023-08-11 davidben Default to q = (p-1)/2 for DH keygen 2023-08-15 davidben Remove the CRYPTO_is_*_capable_at_runtime indirection 2023-08-14 davidben Select SHA-256 vs SHA-512 explicitly in perlasm 2023-08-14 davidben Hash-to-curve is now RFC 9380 2023-08-14 bbe Add the rest of the new test data files to sources.cmake 2023-08-11 davidben Align DH keygen with NIST's formulation when q is available 2023-08-14 bbe add missing test file to sources.cmake 2023-07-27 davidben Consistently reject large p and large q in DH 2023-08-14 bbe use <errno.h> instead of <sys/errno.h> 2023-08-11 davidben Document and test DH_generate_key's weird key reuse behavior 2023-08-14 bbe Sync pki to chromium 8049b24a3fa617e66c5d3fc0e9322bb07c500f49 2023-07-31 bbe Add a basic sanity test for getentropy() if we are using it. 2023-07-20 yich Add ChromeOS EC related targets Created with: roll-dep third_party/boringssl/src Change-Id: Ibfb51bca122dc6ed2cffa76fe1d2a79e4a1e47fe Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6881127 Reviewed-by: Jordan Bayles <jophba@chromium.org> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Bug: 391445226 Change-Id: I16ee2a97fb106d2e2560991cd1bfbe92913949fa Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/6917871 Reviewed-by: Muyao Xu <muyaoxu@google.com> Reviewed-by: Jordan Bayles <jophba@chromium.org> Commit-Queue: Taesun Yeoum <taesuny@google.com>
My local mac compiler is complaining about incorrect semicolon usage in the standalone sender code. This patch fixes the build by removing the extraneous semicolons. Change-Id: I87d33e48c59906fce203dfeacb37ab8cf57e1262 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7007553 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This patch cleans up the statistics defines file, changing the name to statistics common to more accurately represent its contents, and more tightly grouping related types and classes into the StatisticsEvent struct. This is helpful for the upcoming receiver statistics implementation. Bug: 448199360 Change-Id: I5bc6ee380b52c10fbe839de12844ba1b03eb58c9 Bypass-Check-License: moved a file. Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7008996 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This patch significantly refactors the ClockOffsetEstimator to improve the accuracy and stability of its network latency and clock offset estimations. The previous implementation used a simple weighted moving average, which was susceptible to jitter and slow to adapt to changing network conditions. The core of this change is the introduction of a 1D Kalman filter. This allows the estimator to: - More effectively smooth out network jitter, providing a more stable latency estimate. - Converge more quickly and accurately to the true latency and offset, even in the presence of noise. - Better track clock drift over time. To validate this new design, the unit test suite for the ClockOffsetEstimator has been greatly expanded to cover a wide range of realistic and challenging network scenarios, including: - High network jitter - Continuous clock drift - Bursty packet loss - Sudden, large latency spikes - A mix of audio and video media streams Additionally, a new general-purpose Gmock matcher, `EqualsDuration`, has been added to `testing/util` to provide more readable error messages for std::chrono::duration types in unit tests. Bug: b:448199360 Merge branch 'refactor-event' into clock_offset_estimator_refactor WIP refactor Change-Id: I7a3df074a7e421b051191dcbc9503a2b31dd6fea Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7007555 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Replaces the ClockDriftSmoother's weighted-average logic with a more robust exponential moving average (EMA) filter. This provides more stable and predictable smoothing of the clock offset. This change also includes: - A new, comprehensive unit test suite for the ClockDriftSmoother. - A change to `ClockDriftSmoother::Current()` to return `std::optional` to make the uninitialized state explicit and improve call-site safety. - A new `ExpectDurationNear` test helper to simplify duration comparisons in tests, replacing the more verbose `EXPECT_THAT` with `EqualsDuration`. The `receiver.cc` implementation is updated to safely handle the optional return value from the smoother. Bug: 298085631 Change-Id: I9ae5816c6a6586bed99774a5b3ce21bd01f888be Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7032428 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
I recently installed Open Screen on my new laptop and wanted to update the build instructions. Change-Id: I9a77a15125d815ef7a3a1a9820ae15020876fe26 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7083713 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Chromium has been building openscreen with C++20 for years now, this looks like a leftover. Bug: 388070065 Change-Id: Id03507d446d3dfccb4aa184a4fe4431daf978978 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7124721 Commit-Queue: Victor Vianna <victorvianna@google.com> Reviewed-by: Jordan Bayles <jophba@chromium.org>
Context in the linked bug. See also https://crrev.com/c/7123796 for the recipe-side change. Bug: 458361503 Change-Id: I986d780ffe35ed0f8c975de63de85eb16cdba4b8 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7128379 Reviewed-by: Jordan Bayles <jophba@chromium.org> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Implements the Input event API to support unidirectional flow from Receiver to Sender, using idiomatic C++20 data handling with ByteView. * Create InputProducer (Receiver-side) and InputConsumer (Sender-side) into their respective targets. * Use ByteView for ProcessMessageFromRemote and SendMessageToRemote to avoid raw pointer/size pairs. * Tie InputProducer lifecycle to specific negotiation sessions via ConfiguredReceivers. * Ensure ANSWER is dispatched before OnNegotiated to maintain correct message ordering in tests. Bug: 482442349 Change-Id: Ie65c0eb96939c08f3ca0017a1e31c7cab57ff260 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7553551 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Now that Open Screen has moved to C++20, this change replaces the custom twiddling in YetAnotherBitVector with a modernized BitVector class using the <bit> header (std::popcount, std::countr_zero). Key changes: - Renamed YetAnotherBitVector to BitVector. - Switched to size_t for sizes and positions. - Simplified API to only include methods used by the Sender. - Preserved file history via git mv. Change-Id: I933af7501e565a5fc30438051dba665f0f5f4a5f Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7648931 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This reverts commit c2f2047. Reland "Use abstract Connection in place of TlsConnection" This reverts commit c2f2047. Reason for revert: decided to fix forward. Original change's description: > Revert "Use abstract Connection in place of TlsConnection" > > This reverts commit 92464b5. > > Reason for revert: breaks the autoroller :(. > > https://chromium-review.googlesource.com/c/chromium/src/+/7257264 > > Hate to revert but need to do that to enable the QUICHE roll. > > Happy to prioritize relanding it tomorrow. > > Example failures: > https://ci.chromium.org/ui/p/chromium/builders/try/android-x64-rel/726580/overview > > [28250/51502] CXX obj/components/openscreen_platform/unittests/tls_client_connection_unittest.o > ../../third_party/llvm-build/Release+Asserts/bin/clang++ -MMD -MF obj/components/openscreen_platform/unittests/tls_client_connection_unittest....(too long) > ../../components/openscreen_platform/tls_client_connection_unittest.cc:134:3: error: non-virtual member function marked 'override' hides virtual member function > 134 | MOCK_METHOD(void, OnError, (TlsConnection*, const Error&), (override)); > | ^ > ../../third_party/googletest/src/googlemock/include/gmock/gmock-function-mocker.h:115:3: note: expanded from macro 'MOCK_METHOD' > 115 | GMOCK_PP_VARIADIC_CALL(GMOCK_INTERNAL_MOCK_METHOD_ARG_, __VA_ARGS__) \ > | ^ > ../../third_party/googletest/src/googlemock/include/gmock/internal/gmock-pp.h:49:7: note: expanded from macro 'GMOCK_PP_VARIADIC_CALL' > 49 | GMOCK_PP_CAT(_Macro, GMOCK_PP_NARG(__VA_ARGS__))(__VA_ARGS__)) > | ^ > > > Original change's description: > > Use abstract Connection in place of TlsConnection > > > > Bug: 445928621 > > Change-Id: I485228f0e56ecd736f661194bb0e72d2682344db > > Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7247255 > > Reviewed-by: Muyao Xu <muyaoxu@google.com> > > Commit-Queue: Taesun Yeoum <taesuny@google.com> > > Reviewed-by: Jordan Bayles <jophba@chromium.org> > > Bug: 445928621 > Change-Id: Ia3369a687a06547aea26a7f99ac48d986eddc567 > Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7303817 > Reviewed-by: Nidhi Jaju <nidhijaju@chromium.org> > Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Bug: 445928621 Change-Id: Iafd794e11a644b33ef6b86433eab4f3f2f20d0a1 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7663781 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Muyao Xu <muyaoxu@google.com>
I've been meaning to do this forever and just got around to it. This patch migrates the offer and answer logic to using ErrorOr, instead of one using Error and one using bool. Change-Id: If83060740ac21bf5026b2ac2a78b63951f32ba1f Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7670823 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Muyao Xu <muyaoxu@google.com>
This patch adds basic Input Event API support to the standalone Cast sender and receiver binaries. Input event support may be enabled in these binaries using the new `--enable-input-events` (`-i`) flag. In this initial implementation, the standalone receiver has the ability to get mouse and keyboard events through the SDL API, and then report them to the standalone sender, which current logs all received events to the terminal and supports blitting "mouse down" events as an expanding concentric rig with a 'ping' effect. Bug: 482442349 Change-Id: I9bb86e7a1fd5597594eccc0068246b39e50a4d79 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7551363 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Bug: chromium:492241998 Change-Id: If567f34e7a86eed44503594ab207850e6a6a6964 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7695059 Reviewed-by: Jordan Bayles <jophba@chromium.org> Commit-Queue: Jon Toohill <jtoohill@google.com>
The Answer message previously defined rtp_extensions as a flat std::vector<std::string>. However, RTP extensions are negotiated per-stream (e.g., input_events for video streams). This change updates rtp_extensions in the Answer message to be a std::vector<std::vector<std::string>> that parallels the send_indexes array. It also introduces JSON helpers to parse and serialize array of arrays, updates the schemas and protocol docs, restricts the input_events extension to only apply to video streams, and extracts the input_events literal into a shared constant in cast/streaming/impl/message_constants.h. Bug: 495577323 Change-Id: If148f7783d6b1daeb229cd2eac5fc4ed367c0ba0 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7695772 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Jordan Bayles <jophba@chromium.org>
When build_with_chromium=true, exclude url.cc and the googleurl dep from the util target. Instead, depend on a Chromium-side platform_impl that provides the URL implementation via //url directly. Standalone builds continue using the bundled googleurl. Also guard url_unittest.cc with !build_with_chromium, since the test links against the bundled googleurl implementation which is not available in Chromium builds. The Chromium-side equivalent test lives in components/openscreen_platform:unittests. Changes in main repo: https://chromium-review.googlesource.com/c/chromium/src/+/7696516 Bug: 495648329, 351564777 Change-Id: Ia69951eabc6d0d788d0a97ae6b586da14aa29365 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7696713 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Jason Leo <cgqaq@chromium.org>
- Grouped and alphabetized existing ignore entries by category. - Added ignores for common AI assistants (Gemini, Claude, Cursor, Aider, Copilot). - Added ignore for Perfetto trace files (*.pftrace). Change-Id: Ib29ceef1c378bb5a658b81fc4f85bedad7b1b674 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7709363 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Refactor the Cast Streaming Receiver to eliminate redundant memory copies during frame decryption. Previously, the Receiver assembled encrypted RTP packets into a single contiguous buffer before decryption. Now, decryption is performed piecewise directly from the original network buffers (PayloadChunks) into the consumer's output buffer. Key changes: - Refactored FrameCollector to avoid contiguous buffer assembly and expose raw chunks. - Refactored FrameCrypto to support streaming/chunked decryption via a unified Crypt method. - Updated Receiver::ConsumeNextFrame to use the new zero-copy decryption path. - Benchmarking showed a ~83% reduction in overhead for 2MB frames (saving ~13.5ms per frame on typical development hardware). - Updated related unit tests to match the new chunk-based APIs. Change-Id: Icddb22f33706ce441d664dc24666a6e6cf63b2b8 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7705668 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Mark Foltz <mfoltz@chromium.org>
This patch deletes the Url class, which is currently unused and causing maintenance headaches downstream. Bug: 495648329 Change-Id: Ifae5b5d7e1097d3807542f0c176bc1bbe4ed9a07 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7719179 Commit-Queue: Mark Foltz <mfoltz@chromium.org> Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This patch adds taesuny@ to the OWNERS list as an additional reviewer. Taesun has made some significant CLs for Open Screen: https://chromium-review.googlesource.com/c/openscreen/+/7108400 https://chromium-review.googlesource.com/c/openscreen/+/7247255 https://chromium-review.googlesource.com/c/openscreen/+/7283925 As well as some smaller, single / few line changes: https://chromium-review.googlesource.com/c/openscreen/+/6917871 https://chromium-review.googlesource.com/c/openscreen/+/7128520 https://chromium-review.googlesource.com/c/openscreen/+/7517644 https://chromium-review.googlesource.com/c/openscreen/+/7598315 In addition, Taesun manages the downstream integration of libcast into the Android Cast implementation, so has extensive experience with the library as a consumer. I believe that, in conjuction, this makes him a great candidate for inclusion in our OWNERS list. Change-Id: Ia623d54e61ed0852adf85c13541c313887e2b4b6 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7714244 Reviewed-by: Muyao Xu <muyaoxu@google.com> Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This patch does some significant cleanups of the Sender and Receiver classes, in order to make embedder's lives easier (and ideally our own) around testing and reasoning about these classes. Sender and Receiver are now virtual abstract base classes, making mocking easy and making it possible for tests in libcast and tests in the embedder to not have to worry about implementation details. Previously, it was impossible to for embedders to appropriately test classes they write that depend on the Receiver interface -- since you can't construct a receiver without a packet router, and including the packet router header is banned by GN check. Bypass-Check-License: moved some files around Bug: 255632830 Change-Id: I2347bf370c392e2f663872c49f6fb9f8a94ee20d Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7594201 Reviewed-by: Mark Foltz <mfoltz@chromium.org>
Added ReportFrameDropEvent to Sender to align with the Receiver's ReportPlayoutEvent. This new event captures the details of the frame dropped by the encoder and wires it up to the statistics analyzer for tracking kNumFramesDroppedByEncoder. Bug: b/298205111 Change-Id: I176b3a539c7682ba08912492447243f5ebbde231 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7709359 Reviewed-by: Muyao Xu <muyaoxu@google.com>
This patch enables the no exit time destructors clang check, and fixes violations of this check. A new `NoDestructor` class is added to avoid littering the codebase with `static Foo* kStaticFoo = new Foo()` to avoid declaring exit time destructors. Bug: 40031409 Change-Id: I7b94ee90f810ab8848d85b6548793232c5f65176 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7732861 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Muyao Xu <muyaoxu@google.com>
- Fixes a bug in the E2E tests where fatal logs were ignored due to incorrect tuple unpacking. - Fixes a race condition where cast_receiver could crash during teardown because flat_map::erase_key used OSP_CHECK when a key was not found. - Improves E2E error outputs when a fatal log is encountered. - Fixes Python argument parsing in standalone_e2e.py so the unittest module does not complain. Change-Id: Ia6ba4420990f3edc2af213c0d1b9f0580c367e12 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7746325 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Muyao Xu <muyaoxu@google.com>
MirroringApplication supports two namespaces, kCastWebrtcNamespace and kCastRemotingNamespace. To support custom namespaces, a custom message callback is added to SessionMessenger, where non-standard namespace messages are sent to. Change-Id: I78b7f501aee339047627b96a01fa15265c29cf78 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7731037 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Taesun Yeoum <taesuny@google.com> Reviewed-by: Jordan Bayles <jophba@chromium.org>
This CL ensures that ReceiverSession and SenderSession unregister themselves (and their components) from the Environment upon destruction. Previously, Environment would hold dangling raw pointers to these objects, leading to a Use-After-Free when socket events or statistics collection occurred after a session was destroyed. Bug: b/505815444 Change-Id: I4b3874d0b9f40ea2e7eab940bd3c83f7d08c42ab Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7797696 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Muyao Xu <muyaoxu@google.com>
This patch fixes a macos specific static initalizer issue by reverting a recent refactor in ScopedWakeLockMac. Change-Id: Ib3e4a8c43d37af57acb677d19203843520028165 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7797504 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Muyao Xu <muyaoxu@google.com>
SenderMessage::Parse and other message parsing functions index into const Json::Value& references with string keys without verifying the root is an object. If the parsed JSON resolves to a scalar or an array, jsoncpp aborts the process instead of returning a null value because OpenScreen builds with JSON_USE_EXCEPTION=0. This CL adds explicit .isObject() validation steps to json parsing functions across the cast/streaming and cast/sender namespaces to properly reject malformed arrays/scalars before querying them. Bug: b/505947418, b/505902444 Change-Id: Ife4d19bfb592656a48773e9e7069571b617ce7df Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7801091 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This updates the visibility list for certificate_boringssl to allow all targets within //components/media_router/common/providers/cast/channel to depend on it. This is necessary because a change to the fuzzer_test template in Chromium for android causes intermediate generated targets to inherit the fuzzer's dependencies. This breaks the strict target-name visibility check Bug: 506174854 Change-Id: I83702f3e353d9f76aedf3ede9387de987bca937b Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7793868 Commit-Queue: Edgar Aguilar <eledgar@google.com> Reviewed-by: Mark Foltz <mfoltz@chromium.org> Reviewed-by: Titouan Rigoudy <titouan@chromium.org>
This patch rolls protobuf to 33.6 to patch some vulnerabilities. A new `tools/roll_protobuf.sh` script is also added to make rolling protobuf much easier. This roll required zero manual intervention after running the script. The script also now prunes obviously not needed directories (such as the github directory and not supported languages), reducing the checkout size by 465 KiB / > half a million lines of code. Bug: b/507096835 Change-Id: I9d75f398f96b88fb5f6d7694443c0cc36e375c73 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7808806 Commit-Queue: Jordan Bayles <jophba@chromium.org> SLSA-Policy-Verified: SLSA Policy Verification Service <devtools-gerritcodereview-exitgate@google.com> Reviewed-by: Mark Foltz <mfoltz@chromium.org>
Implement proper write flow control in TlsConnectionPosix to prevent event loop busy-waiting when SSL_write is blocked by a read operation (e.g., during renegotiation, returning SSL_ERROR_WANT_READ). This change: - Adds `is_write_blocked_by_read_` state to TlsConnectionPosix. - Updates `HasPendingWrite()` to return false when the write path is blocked by read, preventing SocketHandleWaiter from polling for writability. - Sets the flag when SSL_write returns SSL_ERROR_WANT_READ. - Resets the flag when TryReceiveMessage() is called (socket is readable), allowing the write path to retry after the read path makes progress. Bug: 492683133 Change-Id: Ie44060036f6446f30f483110496270f0b7813b28 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7814283 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Muyao Xu <muyaoxu@google.com>
…e destructors Refactor static singletons in SenderSession and CastNonce to use openscreen::NoDestructor instead of manually leaking them on the heap using 'new' to avoid exit-time destructors. This change: - Refactors 5 static Error singletons in sender_session.cc to use NoDestructor<Error>. - Refactors the CastNonce singleton in cast_auth_util.cc to use NoDestructor<CastNonce>. - Makes the CastNonce constructor public so NoDestructor can access it (safe as the class is in an anonymous namespace). Bug: 40031409 Bug: 492683133 Change-Id: Ie26adbbc83a4ed0f08cbb3cf7bfe7821d742a08d Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7813205 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Add OSP_CHECK(task_runner_.IsRunningOnTaskRunner()) to all public API entry-points of UdpSocketPosix. This change: - Enforces the documented UdpSocketPosix threading model: "All public methods must be called on the same thread--the one executing the TaskRunner." - Adds checks to Bind(), SetMulticastOutboundInterface(), JoinMulticastGroup(), SendMessage(), and SetDscp(). - Prevents dynamic race conditions and memory corruptions if an embedder invokes these APIs from the wrong thread. Bug: 492683133 Change-Id: I8010ee0e21df3a7a09728fb904167fbaba9b3084 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7814331 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Migrates the bash-based `tools/roll_protobuf.sh` script to Python (`tools/roll_protobuf.py`) to conform with the Google Shell Style Guide, which mandates Python for scripts exceeding 100 lines. The new roll_protobuf script offers: - Standardized cross-platform execution (eliminating bash dependencies like perl, curl, and awk). - Robust error handling and a structured `ProtobufVersion` dataclass. - Rich terminal output utilizing the newly abstracted `tools/cli_utils.py`. - A dedicated unit test suite (`tools/roll_protobuf_test.py`). Additionally, this change performs a sweep of Python tooling across the repository (`tools/*`, `cast/standalone_e2e.py`, and `PRESUBMIT.py`) to align with the modern Google Python Style Guide: - Applied PEP-8 compliant type hinting. - Refactored internal implementation details from `CamelCase` to `_snake_case` to properly enforce private module boundaries. - Re-formatted all scripts using `git cl format` Change-Id: Ib4ec003716e2857f259e9d50db062467de14d554 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7809783 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Jordan Bayles <jophba@chromium.org>
…common Deduplicate FFMPEG and avcodec unique pointer helpers, compatibility macros, and the AvErrorToString implementation between standalone_sender and standalone_receiver. This change: - Creates a new shared 'cast/standalone_common' component. - Moves and consolidates ffmpeg_glue.h/cc and avcodec_glue.h into the new component. - Refactors both standalone_sender and standalone_receiver to depend on this new shared component. - Simplifies OnError in decoder.cc to use AvErrorToString. Bug: 224642520 Bug: 492683133 Change-Id: I93d9dd862fdf8c885929ba23d89803c3e92cc09f Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7814281 Reviewed-by: Mark Foltz <mfoltz@chromium.org> Commit-Queue: Jordan Bayles <jophba@chromium.org>
Introduce Abseil-style thread safety annotations (OSP_GUARDED_BY, etc.) to enable static analysis of mutex-guarded variables during compilation. This resolves b/322734860. This change: - Creates 'util/thread_annotations.h' containing the Clang-specific macro definitions for thread safety analysis. - Audits and annotates all 'std::mutex' usages in Open Screen production code: - TlsDataRouterPosix (accept_socket_mappings_, connections_) - TaskRunnerImpl (tasks_, delayed_tasks_) - SocketHandleWaiter (handles_being_deleted_, handle_mappings_) - UdpSocketReaderPosix (sockets_) - StreamingAv1Encoder (needs_key_frame_, target_bitrate_, encode_queue_) - StreamingVpxEncoder (needs_key_frame_, target_bitrate_, encode_queue_) - Refactors all 'std::unique_lock' usages to 'std::lock_guard' where possible to resolve Clang static analysis limitations. - Fixes a real concurrency bug in video encoders where GetTargetBitrate() was reading 'target_bitrate_' without holding 'mutex_'. - Makes 'mutex_' mutable in encoders and socket reader to allow locking in const methods. - Safe-bypasses Clang's lambda-condvar-loop analysis limitations by using standard 'while' loops or 'OSP_NO_THREAD_SAFETY_ANALYSIS' where manually verified to be safe. Fixed: 322734860 Bug: 492683133 Change-Id: I9a86c0848267c2893ddba9ff319cdce8b02a5812 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7814327 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
…ification Acknowledge and document that SSL_shutdown is already robustly verified by CastSocket E2ETests, resolving b/169966671. This change: - Adds an explicit comment in cast_socket_e2e_test.cc (inside CloseSocketsFromOneEnd) explaining why the kSocketClosedFailure assertion explicitly verifies SSL_shutdown correctness (i.e., if the peer did not call SSL_shutdown, BoringSSL would report a protocol error mapped to kFatalSSLError instead). - Removes the stale TODO comment from tls_connection_posix.cc since E2E coverage is the correct and most robust way to verify this behavior and is already fully implemented. Bug: 169966671 Bug: 492683133 Change-Id: I49ddc19165f2eac3eaafabf44a71479ad1cc035b Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7815047 Reviewed-by: Muyao Xu <muyaoxu@google.com> Commit-Queue: Jordan Bayles <jophba@chromium.org>
This commit replaces the manually maintained copies of Chromium's `licenses.py` and `protoc_wrapper.py` scripts with an automated solution via DEPS hooks, ensuring these scripts remain up-to-date with Chromium's trunk. - Created a generic `tools/download-chrome-file.py` (based on the former `download-clang-update-script.py`). - Updated `DEPS` hooks to use the new downloader script to fetch `licenses.py`, `spdx_writer.py`, and `protoc_wrapper.py` directly from Chromium matching the existing `chrome_version`. - Replaced the deprecated `ScanThirdPartyDirs()` call in `PRESUBMIT.py` with the modern `_DiscoverMetadatas()` API from `licenses.py`. - Removed tracked copies of the scripts from git and added their paths to `.gitignore`. Bypass-Check-License: moved file. Change-Id: I7363d45252be702c75c0aeb5b9d51f42038100b4 Reviewed-on: https://chromium-review.googlesource.com/c/openscreen/+/7815044 Commit-Queue: Jordan Bayles <jophba@chromium.org> Reviewed-by: Mark Foltz <mfoltz@chromium.org>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
17 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Hack to get commit notifications. Do not close!!!