chore: refresh imported plugin sources from upstream 0.17.0-131 (#45)#50
Merged
Conversation
Implement Ecdsa::PublicKey::verify() using EVP_PKEY_public_check. Part of #2669. Signed-off-by: Parth Dagia <parth.24bcs10414@sst.scaler.com>
GCC's -Wshadow=local errors because the z_stream unique_ptr locals in the init/copy host functions share the HostZStream/DestZStream names with the SyncRun lambda parameters. Rename the locals to NewZStream; the lambda parameter naming stays consistent across the file. Assisted-by: GitHub Copilot CLI (Claude Fable 5) Signed-off-by: hydai <z54981220@gmail.com>
The histogram-printing loops reuse i inside the polling loop that already declares it, which errors under GCC's -Wshadow=local and Clang's -Wshadow. Rename the inner loop variables to j. Assisted-by: GitHub Copilot CLI (Claude Fable 5) Signed-off-by: hydai <z54981220@gmail.com>
… target The opencvmini test target pulls the OpenCV include path in through the plugin's INCLUDE_DIRECTORIES property as a regular include, so Clang's -Wshadow-field errors on parameters shadowing members inherited from cv::Mat inside OpenCV's own mat.inl.hpp. Reuse the wasmedge_suppress_shadow_warnings helper on the test target, matching how the other third-party dependencies are handled. Assisted-by: GitHub Copilot CLI (Claude Fable 5) Signed-off-by: hydai <z54981220@gmail.com>
Signed-off-by: hydai <z54981220@gmail.com>
…ion (#4940) Additionally, remove unused functions from tts/tts_core.h exports. Signed-off-by: Pranjal Kole <pranjal.kole7@gmail.com>
…tion Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
…rtualization Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
…fix devirtualization Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
…lization Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
…5008) * fix(plugin/wasi-logging): validate full string extent in Log::body getPointer<char *> only checks sizeof(char) == 1 byte, but CxtLen and MsgLen bytes are then consumed via string_view comparisons and logger calls. Passing CxtPtr + CxtLen or MsgPtr + MsgLen past the linear memory boundary crashes the host process through the guard page. Replace both with getStringView(ptr, len) which calls checkAccessBound(offset, len) over the full extent before returning. Signed-off-by: Yashika <ssyashika1311@gmail.com> * test(plugin/wasi-logging): add OOB regression tests for Log::body Verify that CxtPtr + CxtLen and MsgPtr + MsgLen extending past the linear memory boundary return HostFuncError instead of crashing the host process through the guard page. Signed-off-by: Yashika <ssyashika1311@gmail.com> * test(plugin/wasi-logging): add WAT reproducer for OOB crash in Log::body Signed-off-by: Yashika <ssyashika1311@gmail.com> * test(plugin/wasi-logging): remove unused WAT reproducer file The poc_oob_log.wat file was not referenced by the test suite and was flagged as unrelated in review. The OOB behavior is already covered by the C++ regression tests in wasi_logging.cpp. Signed-off-by: Yashika <ssyashika1311@gmail.com> --------- Signed-off-by: Yashika <ssyashika1311@gmail.com>
…tualization Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
The previous EXPECT_* checks will allow a dereference on calling getHostFunc(), which may lead segfaults instead of a failing cleanly. Call sites with no guard at all are hardened the same way, inserting the ASSERT_NE/ASSERT_TRUE pair before getHostFunc() across the FFmpeg test files the earlier pass missed. Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
…perate Two comments labelled the lookup as wasm_bpf_map_fd_by_name while the export actually fetched is wasm_bpf_map_operate, a copy-paste leftover that misleads when navigating the test. Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
…dings Replacing the concrete dynamic_cast host-function lookups with getHostFunc() dropped the only check that each export was wired to the correct, same-signature host function: a call that copies bytes into memory or returns a scalar still reports ErrNo::Success when mis-bound. Read the written output back or pin a distinguishing value so every export binding stays observable. wasmedge_ffmpeg name and scalar readbacks: - avPixfmt: av_color_range_name (pc), av_color_transfer_name (smpte170m), av_color_space_name / av_color_primaries_name (bt709), av_pix_format_name (yuv420p). - avCodec: avcodec_get_name (h264) and avcodec_get_long_name. - avSampleFmt: av_get_sample_fmt_name (u8). - avInputFormat: name (mov,mp4,...) and long_name (QuickTime / MOV). - avfilter: avfilter_name (abuffer), description, pad_get_name. - avDictionary: av_dict_get_key_value copies KEY/VALUE; read both back. - avError: av_strerror ran with no arguments; pass the real buffer and assert. - av_log_get_flags ran with a stray argument; call the setter with an empty result span and the getter with none, then assert the flag. - utils: initEmptyFrame asserts the av_frame_alloc run() result. wasmedge_ffmpeg version, configuration/license, and codec-kind sweep: - version exports (avutil, avcodec, avformat, avfilter, swresample, swscale): assert the major-version byte (>> 16) is nonzero, so a swap with a configuration/license length getter no longer passes. - configuration/license copiers (all six modules): read back the buffer and require the configure-flag marker "--" to be present for configuration and absent for license, distinguishing the pair on both sides. - avformatContext duration: replace the tautological check with the sample asset's exact duration. - avcodec find_decoder_by_name / find_encoder_by_name: confirm the returned codec kind via av_codec_is_decoder / av_codec_is_encoder. wasmedge_process: - GetStdErr compares the StdErr buffer instead of the empty StdOut. - GetExitCode seeds a nonzero Env.ExitCode and asserts it. Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
… duration initFormatCtx opens the sample with avformat_open_input only, so AVFormatContext::duration stays AV_NOPTS_VALUE until avformat_find_stream_info runs. The prior EXPECT_EQ(..., 2000000) assumed a probed container duration and failed on every ffmpeg CI job; assert the value the getter actually returns. Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
…dings observable After dropping the concrete dynamic_cast lookups, several exports were only checked for success or a non-negative result, so a same-signature binding swap would pass undetected. Pin distinguishing values verified against ffmpeg 7.1: - avPixfmt log2_chromaw/h: use YUV422P (1/0) so a width<->height swap fails. - avSampleFmt bytes_per_sample: guest sample format 1 (U8) is exactly 1 byte. - avutil log_set_level: call with an empty result span, assert run() succeeds, and use a non-default level so the getter proves the setter executed. - avutil channel-layout helpers: assert nb_channels and the default mono layout, which differ between the two exports. - avutil av_gettime vs av_gettime_relative: separate the wall-clock epoch from the smaller monotonic clock. - swscale isSupportedInput/Output: PAL8 is input-only, distinguishing the pair. - avStream duration/start_time/nb_frames/disposition: assert the sample's exact per-stream values. Assisted-by: Claude (Anthropic) Signed-off-by: hydai <z54981220@gmail.com>
Super-linter summary
All files and directories linted successfully For more information, see the GitHub Actions workflow run Powered by Super-linter |
5 tasks
0yi0
added a commit
that referenced
this pull request
Jun 26, 2026
Upstream deprecated the wasmedge_llmc plugin (WasmEdge/WasmEdge#4964). Rather than migrate it into cpp-plugins, remove it: drop the residual plugins/wasmedge_llmc and test/plugins/wasmedge_llmc directories and the add_subdirectory stanzas in plugins/CMakeLists.txt and test/plugins/CMakeLists.txt. libpiper (wasi_nn) is unrelated and kept. Rebased onto main after the refresh re-import (#50), whose upstream deprecation commit had already emptied the plugin to stub CMakeLists. Closes #36 Signed-off-by: Yi Huang <yi@secondstate.io> Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
6 tasks
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
5 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.
PR summary
Refresh re-import of the C++ plugin sources/tests/CI-utils from WasmEdge core, pinned to
0.17.0-131-g3e752187(3e752187). Brings the 17-commit delta that landed in the imported paths since the original import (PR #44, based67087e). Follow-up to #45 (the migration-window re-import procedure); same recipe as #44, replayed onto the currentmain.Implementation design
3e752187→git filter-repoover the feat: import plugin sources, tests, and CI utils from WasmEdge core #44 path list (plugins,test/plugins,thirdparty/wasi_crypto,utils/docker,utils/ffmpeg,utils/opencvmini,utils/wasi-crypto,utils/wasi-nn,utils/wasi-test,utils/build_libpiper.sh) → 640 filtered commits, 0 merges.git rebase --onto main --root --empty=drop --committer-date-is-author-date: the 623 commits already onmain(from feat: import plugin sources, tests, and CI utils from WasmEdge core #44) drop as already-applied/empty; the 17 new commits replay ontomain. Original authors, dates, messages, and DCO sign-offs preserved; committer rewritten (rebase semantics).Design decisions
0.17.0-131-g3e752187), not a floatingmaster/HEAD — reproducible.--empty=dropyields the delta: becausemainalready contains feat: import plugin sources, tests, and CI utils from WasmEdge core #44's import, re-running the recipe naturally produces only the new commits (verified: 17, not 640).wasmedge_llmcdeprecation is included as-is (chore(plugin/llmc): deprecate wasmedge-llmc plugin). cpp-plugins will remove llmc directly rather than carry it — tracked in feat: migrate wasmedge_llmc #36 / tracking: migrate C++ plugins from WasmEdge core (WasmEdge/WasmEdge#3840) #22; intentionally NOT done in this faithful re-import.Commit slicing
17 upstream commits, no authored commits — this PR is exactly the filtered delta (
git log main..HEAD).Test plan
plugins/,test/plugins/,thirdparty/wasi_crypto, selectedutils/) — nothing outsidegit rev-list --count main..HEAD= 17; 0 merge commits (linear)plugins/) — exit 0fix(test/plugins/wasi_nn): remove concrete types…andchore(plugin/llmc): deprecate…carry the upstream squash-merge artifact (author rewritten to a GitHub noreply / different identity while the sign-off keeps the real email). Same as feat: import plugin sources, tests, and CI utils from WasmEdge core #44; not introduced here. Resolve via the "Set DCO to pass" button on the Checks tab (per docs: import guidelines for future re-imports + PR #44 leftovers #45).Non-goals / afterwards
wasmedge_llmc(deprecated upstream) → tracked in feat: migrate wasmedge_llmc #36 / tracking: migrate C++ plugins from WasmEdge core (WasmEdge/WasmEdge#3840) #22, separate from this mirror🤖 Generated by Claude Opus 4.8 (1M context) with Claude Code