implement Component Model async ABI #11127

dicej · 2025-06-24T17:19:00Z

This commit replaces the stub functions and types in wasmtime::runtime::component::concurrent and its submodules with the working implementation developed in the wasip3-prototyping repo. For ease of review, it does not include any new tests; I'll add those in a follow-up commit.

This pulls in the tests from the `wasip3-prototyping` repo, minus the ones requiring WASIp3 support in `wasmtime-wasi[-http]`, which will be PR'd separately. Note that this builds on bytecodealliance#11114, bytecodealliance#11123, and bytecodealliance#11127; only the fourth commit is new. Signed-off-by: Joel Dice <[email protected]>

This pulls in the tests from the `wasip3-prototyping` repo, minus the ones requiring WASIp3 support in `wasmtime-wasi[-http]`, which will be PR'd separately. Note that this builds on bytecodealliance#11123 and bytecodealliance#11127; only the most recent commit is new. Signed-off-by: Joel Dice <[email protected]>

github-actions · 2025-06-27T18:49:28Z

Subscribe to Label Action

cc @fitzgen

This issue or pull request has been labeled: "fuzzing", "wasmtime:api", "wasmtime:c-api"

Thus the following users have been cc'd because of the following labels:

fitzgen: fuzzing

To subscribe or unsubscribe from this label, edit the .github/subscribe-to-label.json configuration file.

Learn more.

This commit is extracted from from review of bytecodealliance#11123 and bytecodealliance#11127. While not literally present in those PRs it's my own personal conclusion that it's best to just go ahead and add these bounds at the "base" of the component trait hierarchy. The current implementation in bytecodealliance#11123 adds bounds in many locations and this would remove the need to add bounds everywhere and instead have everything inherited through the `Lift` and `Lower` traits. This raises the question of: why? The main conclusion that I've reached leading to this change is that Wasmtime currently will store `R`, a return value, on the stack during the lowering process back into linear memory. This might involve allocation, however, meaning that wasm can be invoked and a context switch could happen. For Wasmtime's `unsafe impl` of `Send` and `Sync` on fibers to be sound it requires that this stack-local variable is also `Send` and `Sync` as it's an entirely user-provided type. Thus I've concluded that for results it's always required for these to be both `Send` and `Sync` (or at the very least, `Send`). Given that I've gone ahead and updated to require both `Send` and `Sync` for both params and results. This is not expected to actually have any impact in practice since all primitives are already `Send`/`Sync` (minus `Rc` impls all removed here) and all `bindgen!`-generated types are compositions of `Send`/`Sync` primitives meaning that they're also `Send` and `Sync`.

alexcrichton · 2025-06-30T20:00:37Z

@dicej on this one thing I want to write down before I forget (I've forgotten this many times so far...)

Index allocation for futures/streams/error-context I believe are going to need to be updated here. I believe the current state of this PR is such that there's one indexing namespace for all resources within a component, and then there's another index namespace for futures/streams/error-context. Spec-wise I believe it's required we unify these index namespaces. I realize that this is an artifact of the history of the implementation here, but it's something that I want to be sure to write down and not forget.

Effectively my naive expectation would be that this type grows to include Stream/Future/ErrorContext variants and a table or similar in ConcurrentState will end up being removed.

* Add `Send`/`Sync` bounds to `ComponentType` This commit is extracted from from review of #11123 and #11127. While not literally present in those PRs it's my own personal conclusion that it's best to just go ahead and add these bounds at the "base" of the component trait hierarchy. The current implementation in #11123 adds bounds in many locations and this would remove the need to add bounds everywhere and instead have everything inherited through the `Lift` and `Lower` traits. This raises the question of: why? The main conclusion that I've reached leading to this change is that Wasmtime currently will store `R`, a return value, on the stack during the lowering process back into linear memory. This might involve allocation, however, meaning that wasm can be invoked and a context switch could happen. For Wasmtime's `unsafe impl` of `Send` and `Sync` on fibers to be sound it requires that this stack-local variable is also `Send` and `Sync` as it's an entirely user-provided type. Thus I've concluded that for results it's always required for these to be both `Send` and `Sync` (or at the very least, `Send`). Given that I've gone ahead and updated to require both `Send` and `Sync` for both params and results. This is not expected to actually have any impact in practice since all primitives are already `Send`/`Sync` (minus `Rc` impls all removed here) and all `bindgen!`-generated types are compositions of `Send`/`Sync` primitives meaning that they're also `Send` and `Sync`. * Remove some now-unnecessary bounds * Promote to doc comments

This pulls in the tests from the `wasip3-prototyping` repo, minus the ones requiring WASIp3 support in `wasmtime-wasi[-http]`, which will be PR'd separately. Note that this builds on bytecodealliance#11123 and bytecodealliance#11127; only the most recent commit is new. Signed-off-by: Joel Dice <[email protected]>

Signed-off-by: Joel Dice <[email protected]>

This restores the original behavior of requiring explicit post-return calls for `call[_async]` invocations. Signed-off-by: Joel Dice <[email protected]>

This pulls in the tests from the `wasip3-prototyping` repo, minus the ones requiring WASIp3 support in `wasmtime-wasi[-http]`, which will be PR'd separately. Note that this builds on bytecodealliance#11123 and bytecodealliance#11127; only the most recent commit is new. Signed-off-by: Joel Dice <[email protected]>

Mostly move unnecessary state out of the `Arc`.

We weren't always disposing of guest or host task handles once they became unreachable. This adds a couple of hidden methods which integration tests may use to guard against use-after-delete, double-delete, and leak bugs regarding waitable handles. It also tightens up handle management in `concurrent.rs` to ensure those tests pass. Signed-off-by: Joel Dice <[email protected]>

Don't rely on all buffers to handle `TypeId` and assertions and such, instead have a helper type which is the one location of the assertions and everything else can stay typed.

Mostly move away from raw pointers and instead use utilities like `&[MaybeUninit<T>]`. Also make `WriteBuffer` an `unsafe` trait after absorbing the `TakeBuffer` trait. Update all safety-related comments here and there too.

crates/wasmtime/src/runtime/component/concurrent/futures_and_streams.rs

crates/wasmtime/src/runtime/component/concurrent/futures_and_streams/buffers.rs

crates/wasmtime/Cargo.toml

crates/wasmtime/src/runtime/component/concurrent/futures_and_streams/buffers.rs

This pulls in the tests from the `wasip3-prototyping` repo, minus the ones requiring WASIp3 support in `wasmtime-wasi[-http]`, which will be PR'd separately. Note that this builds on bytecodealliance#11123 and bytecodealliance#11127; only the most recent commit is new. Signed-off-by: Joel Dice <[email protected]>

This avoids the need for an `Arc`. Signed-off-by: Joel Dice <[email protected]>

Signed-off-by: Joel Dice <[email protected]>

This gates interop with the `bytes` crate, making it optional and non-default. Signed-off-by: Joel Dice <[email protected]>

This pulls in the tests from the `wasip3-prototyping` repo, minus the ones requiring WASIp3 support in `wasmtime-wasi[-http]`, which will be PR'd separately. Note that this builds on bytecodealliance#11123 and bytecodealliance#11127; only the most recent commit is new. Signed-off-by: Joel Dice <[email protected]>

alexcrichton

Ok I believe we've filed issues for follow-up tasks but otherwise this is an off-by-default feature and in a pretty good spot otherwise, so I'm going to flag for merge. Thanks again @dicej!

dicej requested review from a team as code owners June 24, 2025 17:19

dicej requested review from fitzgen and removed request for a team June 24, 2025 17:19

dicej force-pushed the cm-async-implementation branch 2 times, most recently from 4fd3b8e to c4020df Compare June 24, 2025 18:36

dicej requested review from alexcrichton and removed request for fitzgen June 24, 2025 18:37

github-actions bot added the wasmtime:c-api Issues pertaining to the C API. label Jun 24, 2025

dicej mentioned this pull request Jun 25, 2025

add Component Model async ABI tests #11136

Merged

dicej force-pushed the cm-async-implementation branch from c4020df to 8d7e0af Compare June 26, 2025 22:58

alexcrichton force-pushed the cm-async-implementation branch from 8d7e0af to bee8ccf Compare June 27, 2025 15:48

alexcrichton mentioned this pull request Jun 27, 2025

update component-model-async plumbing #11123

Merged

github-actions bot added fuzzing Issues related to our fuzzing infrastructure wasmtime:api Related to the API of the `wasmtime` crate itself labels Jun 27, 2025

alexcrichton force-pushed the cm-async-implementation branch from bee8ccf to 26c0fbf Compare June 30, 2025 15:03

alexcrichton mentioned this pull request Jun 30, 2025

Add Send/Sync bounds to ComponentType #11160

Merged

alexcrichton mentioned this pull request Jun 30, 2025

Fill out implementations for {Future,Stream,ErrorContext}Any #11161

Open

alexcrichton force-pushed the cm-async-implementation branch from db7f20a to a4e9243 Compare July 1, 2025 22:01

dicej mentioned this pull request Jul 7, 2025

Return whether task was cancelled from yield intrinsic #11191

Open

dicej added 4 commits July 7, 2025 11:52

reference issue 11191 in yield TODO comments

6da1914

Signed-off-by: Joel Dice <[email protected]>

replace dummy_waker with Waker::noop

25edf02

Signed-off-by: Joel Dice <[email protected]>

remove obsolete AsyncState::spawned_tasks field

c2fe750

Signed-off-by: Joel Dice <[email protected]>

only call post-return automatically for call_concurrent

1bc010d

This restores the original behavior of requiring explicit post-return calls for `call[_async]` invocations. Signed-off-by: Joel Dice <[email protected]>

alexcrichton and others added 7 commits July 8, 2025 07:35

Merge remote-tracking branch 'origin/main' into cm-async-implementation

87bd321

Favor function arguments before closures

757df9e

Simplify watch a bit

2e60e92

Mostly move unnecessary state out of the `Arc`.

Encapsulate type erasure in stream buffers

7eeee3a

Don't rely on all buffers to handle `TypeId` and assertions and such, instead have a helper type which is the one location of the assertions and everything else can stay typed.

Remove some methods ending in underscores

6c4f74f

Refactor unsafety in buffers.rs

b5864e5

Mostly move away from raw pointers and instead use utilities like `&[MaybeUninit<T>]`. Also make `WriteBuffer` an `unsafe` trait after absorbing the `TakeBuffer` trait. Update all safety-related comments here and there too.

alexcrichton reviewed Jul 8, 2025

View reviewed changes

remove task on drop in TypedFunc::call_async

27b4158

This avoids the need for an `Arc`. Signed-off-by: Joel Dice <[email protected]>

dicej mentioned this pull request Jul 9, 2025

call_async performance regression with component-model-async feature enabled #11209

Open

dicej added 3 commits July 9, 2025 15:43

remove obsolete clause from FutureReader::read docs

30272e3

Signed-off-by: Joel Dice <[email protected]>

unhide and expand docs for WriteBuffer and ReadBuffer

cde24cb

Signed-off-by: Joel Dice <[email protected]>

add optional component-model-async-bytes feature

471d3c9

This gates interop with the `bytes` crate, making it optional and non-default. Signed-off-by: Joel Dice <[email protected]>

This was referenced Jul 11, 2025

Change Func::call_concurrent signature to accept param and result slices #11218

Open

Refactor {Stream,Future}{Reader,Writer} method signatures and implementations #11219

Open

alexcrichton mentioned this pull request Jul 11, 2025

Bikeshed naming of {Read,Write}Buffer in component-model-async implementation #11223

Open

alexcrichton approved these changes Jul 11, 2025

View reviewed changes

alexcrichton added this pull request to the merge queue Jul 11, 2025

alexcrichton mentioned this pull request Jul 11, 2025

Non-component-model-async execution paths are nontrivial and no longer tested #11227

Open

rvolosatovs mentioned this pull request Jul 11, 2025

feat: begin wasip3 implementation #11221

Draft

Merged via the queue into bytecodealliance:main with commit fa70f02 Jul 11, 2025
42 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

implement Component Model async ABI #11127

implement Component Model async ABI #11127

Uh oh!

dicej commented Jun 24, 2025 •

edited

Loading

Uh oh!

github-actions bot commented Jun 27, 2025

Uh oh!

alexcrichton commented Jun 30, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

alexcrichton left a comment

Uh oh!

Uh oh!

Uh oh!

implement Component Model async ABI #11127

implement Component Model async ABI #11127

Uh oh!

Conversation

dicej commented Jun 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Jun 27, 2025

Subscribe to Label Action

Uh oh!

alexcrichton commented Jun 30, 2025

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

alexcrichton left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

dicej commented Jun 24, 2025 •

edited

Loading