Conversation
…ancelling (bytecodealliance#1638) `cancel_inter_task_stream_read` issued a synchronous `stream.cancel-read` on the inter-task wakeup stream while that stream was still a member of the task's waitable set, only calling `remove_waitable` afterwards. Per the Component Model (component-model#647), a synchronous `{stream,future}.cancel-{read,write}` traps if the waitable is still in a waitable set, for the same reason synchronous reads/writes do. Runtimes that enforce this trap (e.g. recent Wasmtime) therefore fault here during ordinary async operation. Reorder so the stream leaves the waitable set before the synchronous cancel, matching the unregister-then-cancel ordering already used by the general `WaitableOperation::cancel` path. The cancel result is discarded as before, so behavior is otherwise unchanged.
vigoo
force-pushed
the
golem-outline-lift-v0.58.0
branch
from
June 25, 2026 15:01
03966d4 to
29a94e3
Compare
* Evolve the wasip3 async C ABI for tasks This commit is an evolution of the C ABI used to managed task-related infrastructure in WASIp3 with the goal of solving bytecodealliance#1618. The basic problem of bytecodealliance#1618 is that waitables in Rust aren't guaranteed to be polled within the context of the original task. For example by mixing an `async` Rust export and `block_on` it's possible to "cross the wires" and poll in one context while dropping/completing in another context. This can lead to buggy situations where a waitable is left in a set, not added to an appropriate set, or generally mis-managed. The solution here is to enhance the current C ABI of task management with clone/drop operations. Notably this enables waitables to retain a strong reference to the task state as opposed to always consulting what the current task in. This fixes a few situations such as: * When dropping a half-finished waitable it no longer needs to be dropped in the context of the original task. Dropping will unregister the waitable from a task that it was originally registered with. * When a waitable is moved from one task to another it needs to implicitly de-register with the previous task, and this was not previously done. Now with a retained strong reference it's able to clear out previous state upon re-registering with a new task. This change requires some finesse as this needs to be ABI-stable to work with previous versions of the `wit-bindgen` crate. The runtime support additionally can't assume that the new ABI bits are available and instead needs to handle the previous ABI as well. Not too too bad, in the end, though. This additionally did some refactoring of the state associated with async tasks to juggle things around and better represent the raw pointers/`Arc`/etc from before. Closes bytecodealliance#1618 * Document the C ABI * Fix disabled compile * Just one version marker
* feat(cpp): add map type support * fix(cpp): add operator< to wit::string for std::map key support and fix rustfmt wit::string lacked comparison operators, causing compilation failures when used as a std::map key. Also fixes rustfmt formatting issues. * fix(cpp): const_cast map keys during lowering for ownership transfer std::map keys are const, but the ABI lowering needs to call leak() on string keys to transfer ownership to the flat buffer. Use const_cast since the map is consumed during the lowering operation. * fix(cpp): copy map keys by value instead of const_cast during lowering const_cast fails when the map key type differs between contexts (e.g. std::string_view in imports vs wit::string in exports). Copying by value works universally and is safe since the map is consumed. * fix(cpp): remove map runtime test until wasi-sdk supports map types wasm-component-ld bundled with wasi-sdk 30 doesn't support the map type encoding (0x63) in the component model binary format. The C++ map codegen is still validated by the codegen test. The runtime test can be re-added when wasi-sdk ships a compatible wasm-component-ld. * refactor(cpp): use wit::map instead of std::map for map types std::map requires per-entry rb-tree allocation during lift; wit::map mirrors the canonical ABI layout (flat pair array) so lift is a single allocation, matching the existing wit::vector / wit::string pattern. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * refactor(cpp): drop redundant size_t cast in MapLower wit::map::size() and std::span::size() already return size_t, so the C-style cast was a no-op. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * refactor(cpp): skip guest-dealloc loop when body is empty Avoids an unused `base` local and the `(void) base;` suppression when the element has no nested heap-owned fields (e.g. list<u32>, map<char, u32>). Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * refactor(cpp): use static_cast for MapLower realloc result The cast converts void* (from cabi_realloc) to uint8_t*, which is a well-defined static_cast and doesn't need a C-style cast. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * fix(cpp): zero length when moving wit::map Leaves the moved-from map in a fully consistent state so any later inspection of size()/empty() reflects the empty invariant, not just the destructor's short-circuit on a null data_ pointer. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * style(cpp): break long single-line methods in wit::map Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * refactor(cpp): drop wit::map operator[] Index-based subscript on a map is a footgun: integral keys would silently compile while doing positional access. Iteration via range-for and pointer access via data() are sufficient for codegen and don't carry the same expectation mismatch. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * refactor(cpp): drop std::span accessors from wit::map A span of pairs is a vector-shaped view that doesn't fit a map abstraction; codegen sites that need a flat pair span build it explicitly from data() and size() at the call site. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * feat(cpp): introduce wit::map_view for borrowed map arguments Borrowed map arguments were surfacing as std::span<std::pair<K,V> const>, which carries vector-shaped affordances (positional indexing, span conversions) on what is conceptually a map. wit::map_view is a borrow- only counterpart to wit::map that mimics map semantics rather than vector semantics. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * refactor(cpp): rename wit::map to wit::unordered_map Component-model map<K,V> carries no ordering or hashing guarantee, and the type's API surface deliberately mirrors std::unordered_map (size, empty, begin/end) plus bindings-construction primitives, so the name should reflect the unordered semantics rather than std::map's ordered ones. Companion borrow type renamed to wit::unordered_map_view. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * fix(cpp): align map iteration variable with renamed _base Adjacent block bodies (string/list/option dealloc) now reference _base after main's rename, so MapLower / MapLift / GuestDeallocateMap must declare the per-entry pointer under the same name to compile. Adds the matching (void) _base; suppression so loops still compile when the inner body doesn't reference it. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> * refactor(cpp): remove unordered_map_view and update map handling The `unordered_map_view` class has been removed to streamline the API, as it was a borrow-only handle that mimicked map semantics. The code now directly utilizes `std::span` for borrowed map arguments, ensuring a clearer distinction between map and vector semantics. Additionally, the handling of dependencies has been updated to include `<span>` where necessary. Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> --------- Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com>
…e#1649) * Fix exception safety violation in AbiBuffer::advance * Add more explanation
Signed-off-by: Andrew Steurer <94206073+asteurer@users.noreply.github.com>
* Move all async tests to general folder WASIp3 and component-model-async have shipped and are stable, so these are no longer expected to fail. * Try to fix CI failure * Try to fix Go in CI * Fix codegen test expectation * Another round of attempting ignores * Disable broken tests I'm tired of trying to work around them, just disable them. * Tweak CI configuration
* Update wasm-tools dependencies Additionally update C intrinsics for threading as they've been renamed slightly. * Remove unused import
[automatically-tag-and-release-this-commit] Co-authored-by: Auto Release Process <auto-release-process@users.noreply.github.com>
* Add support for named implements in Go Signed-off-by: Jeremy Fleitz <jeremy@cosmonic.com> * clippy: use package.clear() Signed-off-by: Jeremy Fleitz <jeremy@cosmonic.com> * support implements for cpp Signed-off-by: Jeremy Fleitz <jeremy@cosmonic.com> * linter fix Signed-off-by: Jeremy Fleitz <jeremy@cosmonic.com> * update to go_package_name Signed-off-by: Jeremy Fleitz <jeremy@cosmonic.com> * linter update Signed-off-by: Jeremy Fleitz <jeremy@cosmonic.com> --------- Signed-off-by: Jeremy Fleitz <jeremy@cosmonic.com>
vigoo
force-pushed
the
golem-outline-lift-v0.58.0
branch
from
July 21, 2026 17:29
78a7464 to
4407232
Compare
Amp-Thread-ID: https://ampcode.com/threads/T-019ff03e-1c09-74b3-ac44-bc3b987a4cda Co-authored-by: Amp <amp@ampcode.com>
Amp-Thread-ID: https://ampcode.com/threads/T-01a056e3-cbd8-72ad-b4dc-9861765f11b2 Co-authored-by: Amp <amp@ampcode.com>
Amp-Thread-ID: https://ampcode.com/threads/T-01a0615e-d01b-74a6-8eb0-cc4a683fc166 Co-authored-by: Amp <amp@ampcode.com>
Close materialized local stream sinks without masking producer errors, add runtime coverage for normal-return EOF and producer traps, and teach the runtime harness to validate expected failures by message. Amp-Thread-ID: https://ampcode.com/threads/T-01a061e4-35fe-778e-863e-5f99777b0290 Co-authored-by: Daniel Vigovszky <me@vigoo.dev>
Amp-Thread-ID: https://ampcode.com/threads/T-01a061e4-35fe-778e-863e-5f99777b0290 Co-authored-by: Daniel Vigovszky <me@vigoo.dev>
Apply the generated canonical byte budget before retaining a caller view. Keep nested endpoints serialized and bound other payloads by both element count and canonical size. Cover the WASI filesystem import direction and full/partial component transfers. Amp-Thread-ID: https://ampcode.com/threads/T-01a0c98e-3797-728e-b811-76f4ada72d7b Co-authored-by: Daniel Vigovszky <me@vigoo.dev>
Share canonical memory lowering of plain record and variant results, including nested metadata graphs. Keep async task.return borrowing, resource aggregates, import arguments and nested payload vtables inline. Test memory layout, post-return, aliases and mixed sync/async exports. Amp-Thread-ID: https://ampcode.com/threads/T-01a0c98e-3797-728e-b811-76f4ada72d7b Co-authored-by: Daniel Vigovszky <me@vigoo.dev>
Amp-Thread-ID: https://ampcode.com/threads/T-01a0c87f-eeaa-77b9-a78f-21483cb9b54d Co-authored-by: Amp <amp@ampcode.com>
Outline async export lifts and borrow-aware result lowering while retaining scratch buffers through task.return. Share list-only post-return cleanup without changing resource transfer or cancellation ownership. Amp-Thread-ID: https://ampcode.com/threads/T-01a0d24c-3562-702a-afc6-ab09e639a67a Co-authored-by: Daniel Vigovszky <me@vigoo.dev>
Register memory-lowered children of flat task returns as well as indirect results. Keep nested and outer canonical buffers alive through task.return, then release each buffer exactly once. Amp-Thread-ID: https://ampcode.com/threads/T-01a0d24c-3562-702a-afc6-ab09e639a67a Co-authored-by: Daniel Vigovszky <me@vigoo.dev>
vigoo
added a commit
that referenced
this pull request
Sep 24, 2026
Revert 55421c4 and its dependent owning-method-chaining follow-up 389bd60. Restore exactly the pre-merge d1d1637 tree, retaining all changes from fork PRs #1 and #2, including the ABI and resource cleanup optimizations. Defer the upstream update and MoonBit stable-toolchain/cancellation migration to GOL-662. Amp-Thread-ID: https://ampcode.com/threads/T-01a0d330-9bd5-7748-b23a-cc862e50bf25 Co-authored-by: Amp <amp@ampcode.com>
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
9 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.
No description provided.