generalize async fiber abstraction #11114

dicej · 2025-06-23T21:45:01Z

As part of the work implementing the new Component Model async ABI in the wasip3-prototyping repo, I've generalized the FiberFuture abstraction in wasmtime::runtime::store::async_ to support fibers which can either retain exclusive access to the store across suspend points or release it. The latter allows the store to be used by the component-model-async event loop and/or other fibers to run before the original fiber resumes, which is the key to allowing multiple fibers to run concurrently, passing control of the store back and forth.

In the case of Pulley, the above generalization means we also need to give each fiber its own Interpreter so that multiple concurrent fibers don't clobber each other's state.

Concretely, this moves a lot of the code out of async_.rs and into a new fiber.rs submodule which will be shared with the component-model-async implementation.

This also pulls in a new StoreToken<T> utility which has been useful in wasip3-prototyping to safely convert from a &mut dyn VMStore to a StoreContextMut<'a, T> when we previously witnessed a conversion in the other direction.

Note that I've added a 'static bound to the VMStore trait, which simplifies use of &mut dyn VMStore, avoiding thorny lifetime issues.

As part of the work implementing the new Component Model async ABI in the `wasip3-prototyping` repo, I've generalized the `FiberFuture` abstraction in `wasmtime::runtime::store::async_` to support fibers which can either retain exclusive access to the store across suspend points or release it. The latter allows the store to be used by the `component-model-async` event loop and/or other fibers to run before the original fiber resumes, which is the key to allowing multiple fibers to run concurrently, passing control of the store back and forth. In the case of Pulley, the above generalization means we also need to give each fiber its own `Interpreter` so that multiple concurrent fibers don't clobber each other's state. Concretely, this moves a lot of the code out of `async_.rs` and into a new `fiber.rs` submodule which will be shared with the `component-model-async` implementation. This also pulls in a new `StoreToken<T>` utility which has been useful in `wasip3-prototyping` to safely convert from a `&mut dyn VMStore` to a `StoreContextMut<'a, T>` when we previously witnessed a conversion in the other direction. Note that I've added a `'static` bound to the `VMStore` trait, which simplifies use of `&mut dyn VMStore`, avoiding thorny lifetime issues. Signed-off-by: Joel Dice <[email protected]>

This pulls in the latest Component Model async ABI code from the `wasip3-prototyping` repo, including various API refactors and spec updates. This includes all the changes to the `wasmtime` crate from `wasip3-prototyping` _except_ that the `concurrent` submodule and child submodules contain only non-functional stubs. For that reason, and the fact that e.g. `Func::call_async` is now implemented in terms of `Func::call_concurrent`, most of the component model tests are failing. This commit is not meant to be merged as-is; a follow-up commit (to be PR'd separately) will contain the real `concurrent` implementation, at which point the tests will pass again. I'm splitting these into separate PRs to make review easier. Note that this builds on bytecodealliance#11114; only the second commit is new. Signed-off-by: Joel Dice <[email protected]>

alexcrichton

I'll keep taking a look at fiber.rs, but this is at least an initial round of feedback.

crates/wasmtime/src/runtime/store/token.rs

crates/wasmtime/src/runtime/linker.rs

crates/wasmtime/src/runtime/fiber.rs

crates/wasmtime/src/runtime/store.rs

crates/wasmtime/src/runtime/fiber.rs

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. Note that this builds on bytecodealliance#11114 and bytecodealliance#11123; only the third commit is new. Signed-off-by: Joel Dice <[email protected]>

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

As part of my earlier effort to unify the fiber abstractions in the `wasmtime` crate, I changed a `*mut StoreFiber` field to a `&mut StoreFiber`, not realizing that it resulted in a mutable alias at runtime and thus undefined behavior. Miri caught it, fortunately. Signed-off-by: Joel Dice <[email protected]>

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. Note that this builds on bytecodealliance#11114 and bytecodealliance#11123; only the third commit is new. Signed-off-by: Joel Dice <[email protected]>

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

crates/wasmtime/src/runtime/fiber.rs

Main changes: - Make `resume_fiber[_raw]` take a `&mut StoreOpaque` parameter to make its unsafe internals easier to reason about, safety-wise. - Panic if `StoreFiber::drop` is called on an in-progress fiber without having called `StoreFiber::dispose` to gracefully end it first. - (Re)introduce `FiberFuture`, which closes over a `&mut StoreOpaque` and uses it to call `StoreFiber::dispose` on drop. This will require a few more changes to make it usable by `concurrent.rs`, but I'll save those changes for a later PR. Signed-off-by: Joel Dice <[email protected]>

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

May be needed for concurrent bits, but for now not necessary.

It's predicted Miri won't like this, but for now in-repo it's ok with it.

Remove the need for the function entirely and replace it with `resume_fiber`.

Can use stack-based closures/results to transmit the result instead of needing a channel.

The `on_fiber` function is small enough it should be possible to do so.

Leave the fiber non-optional at-rest so it's always available for the destructor.

Small shims, not otherwise public at this time, so remove a layer of indirection.

Helps remove some raw pointers that are held for a long time within `AsyncCx`

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

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. Note that this builds on bytecodealliance#11114 and bytecodealliance#11123; only the third 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#11114, bytecodealliance#11123, and bytecodealliance#11127; only the fourth commit is new. Signed-off-by: Joel Dice <[email protected]>

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

Move management of the async guard range elsewhere to the normal save/restore area.

* Remove the `AsyncCx` type from Wasmtime as it's inherently `unsafe` to use, instead bundle operations directly on a `Store*` reference. * Don't retain pointers-to-pointers within the roughly-equivalent `BlockingContext` created in this PR. Instead when a blocking context is created "take" the metadata from the store to assert exclusive ownership of the pointers. * Refactor how `&mut Context<'_>` is passed around, namely thread it through fiber parameters to model resumption as registering a new context to poll with. * Remove `PollContext` in favor of directly storing a pointer as it's now mostly an empty structure. * Minor refactorings to make things more future-refactorable and/or clear in a few places. * Refactor management of the "current suspend" and "current future context" pointers. These are now null'd out on resumption and asserted null on suspension. * Remove the need for a generic `Reset` structure in the fiber bits as it's a pretty dangerous structure to have in general. The end result of this refactoring is that all usage of `block_on` is now safe and additionally many of the internals of the implementation are safer than they were before

This pulls in the latest Component Model async ABI code from the `wasip3-prototyping` repo, including various API refactors and spec updates. This includes all the changes to the `wasmtime` crate from `wasip3-prototyping` _except_ that the `concurrent` submodule and child submodules contain only non-functional stubs. For that reason, and the fact that e.g. `Func::call_async` is now implemented in terms of `Func::call_concurrent`, most of the component model tests are failing. This commit is not meant to be merged as-is; a follow-up commit (to be PR'd separately) will contain the real `concurrent` implementation, at which point the tests will pass again. I'm splitting these into separate PRs to make review easier. Note that this builds on bytecodealliance#11114; only the second commit is new. Signed-off-by: Joel Dice <[email protected]>

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. Note that this builds on bytecodealliance#11114 and bytecodealliance#11123; only the third 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#11114, bytecodealliance#11123, and bytecodealliance#11127; only the fourth commit is new. Signed-off-by: Joel Dice <[email protected]>

No need for raw pointers with recent refactorings.

alexcrichton

@dicej ok these are all the changes I'd like to make, mind reviewing the changes I made as well before merging? Afterwards feel free to go ahead and queue this for merge

crates/wasmtime/src/runtime/fiber.rs

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

* generalize async fiber abstraction As part of the work implementing the new Component Model async ABI in the `wasip3-prototyping` repo, I've generalized the `FiberFuture` abstraction in `wasmtime::runtime::store::async_` to support fibers which can either retain exclusive access to the store across suspend points or release it. The latter allows the store to be used by the `component-model-async` event loop and/or other fibers to run before the original fiber resumes, which is the key to allowing multiple fibers to run concurrently, passing control of the store back and forth. In the case of Pulley, the above generalization means we also need to give each fiber its own `Interpreter` so that multiple concurrent fibers don't clobber each other's state. Concretely, this moves a lot of the code out of `async_.rs` and into a new `fiber.rs` submodule which will be shared with the `component-model-async` implementation. This also pulls in a new `StoreToken<T>` utility which has been useful in `wasip3-prototyping` to safely convert from a `&mut dyn VMStore` to a `StoreContextMut<'a, T>` when we previously witnessed a conversion in the other direction. Note that I've added a `'static` bound to the `VMStore` trait, which simplifies use of `&mut dyn VMStore`, avoiding thorny lifetime issues. Signed-off-by: Joel Dice <[email protected]> * address review feedback Signed-off-by: Joel Dice <[email protected]> * fix miri-flagged stacked borrow violation As part of my earlier effort to unify the fiber abstractions in the `wasmtime` crate, I changed a `*mut StoreFiber` field to a `&mut StoreFiber`, not realizing that it resulted in a mutable alias at runtime and thus undefined behavior. Miri caught it, fortunately. Signed-off-by: Joel Dice <[email protected]> * remove unneeded `Send` bounds Signed-off-by: Joel Dice <[email protected]> * address more review feedback Main changes: - Make `resume_fiber[_raw]` take a `&mut StoreOpaque` parameter to make its unsafe internals easier to reason about, safety-wise. - Panic if `StoreFiber::drop` is called on an in-progress fiber without having called `StoreFiber::dispose` to gracefully end it first. - (Re)introduce `FiberFuture`, which closes over a `&mut StoreOpaque` and uses it to call `StoreFiber::dispose` on drop. This will require a few more changes to make it usable by `concurrent.rs`, but I'll save those changes for a later PR. Signed-off-by: Joel Dice <[email protected]> * address more review feedback Signed-off-by: Joel Dice <[email protected]> * update `impl Send For StoreFiber` comment Signed-off-by: Joel Dice <[email protected]> * Remove currently-extraneous `Result<()>` from fibers May be needed for concurrent bits, but for now not necessary. * Use safe pointers instead of raw pointers It's predicted Miri won't like this, but for now in-repo it's ok with it. * Fold more responsibility into `resume_fiber_raw` Remove the need for the function entirely and replace it with `resume_fiber`. * Remove channels from async fibers Can use stack-based closures/results to transmit the result instead of needing a channel. * Fold `on_fiber_raw` directly into `on_fiber` The `on_fiber` function is small enough it should be possible to do so. * Don't use `Option` in `FiberFuture` Leave the fiber non-optional at-rest so it's always available for the destructor. * Fold `suspend` functions together Small shims, not otherwise public at this time, so remove a layer of indirection. * Move stack limit management to `FiberResumeState` Helps remove some raw pointers that are held for a long time within `AsyncCx` * add some doc comments to `fiber.rs` Signed-off-by: Joel Dice <[email protected]> * update `fiber.rs` and friends to match CM async requirements This adds a `resolve_or_release` function, which `Instance::resume_fiber` will use when current `concurrent.rs` stub is replaced by a real implementation. Signed-off-by: Joel Dice <[email protected]> * fix non-component-model-async build warnings Signed-off-by: Joel Dice <[email protected]> * make `resume_fiber` private in `fiber.rs` Signed-off-by: Joel Dice <[email protected]> * Shrink `PollContext` state Move management of the async guard range elsewhere to the normal save/restore area. * Refactor `AsyncCx`, reduce `unsafe` * Remove the `AsyncCx` type from Wasmtime as it's inherently `unsafe` to use, instead bundle operations directly on a `Store*` reference. * Don't retain pointers-to-pointers within the roughly-equivalent `BlockingContext` created in this PR. Instead when a blocking context is created "take" the metadata from the store to assert exclusive ownership of the pointers. * Refactor how `&mut Context<'_>` is passed around, namely thread it through fiber parameters to model resumption as registering a new context to poll with. * Remove `PollContext` in favor of directly storing a pointer as it's now mostly an empty structure. * Minor refactorings to make things more future-refactorable and/or clear in a few places. * Refactor management of the "current suspend" and "current future context" pointers. These are now null'd out on resumption and asserted null on suspension. * Remove the need for a generic `Reset` structure in the fiber bits as it's a pretty dangerous structure to have in general. The end result of this refactoring is that all usage of `block_on` is now safe and additionally many of the internals of the implementation are safer than they were before * Adjust some lint attributes * Make manipulation of `AsyncState` safe No need for raw pointers with recent refactorings. * Fix dead code warning * More dead code warnings * Cut down on raw pointers in fiber.rs * Move executor save/restore to normal fiber state save/restore * Bikeshed a method name * update comment in make_fiber Signed-off-by: Joel Dice <[email protected]> * fix machports build Signed-off-by: Joel Dice <[email protected]> --------- Signed-off-by: Joel Dice <[email protected]> Co-authored-by: Alex Crichton <[email protected]>

dicej requested review from a team as code owners June 23, 2025 21:45

dicej requested review from alexcrichton and removed request for a team June 23, 2025 21:45

dicej force-pushed the generalized-fiber-abstraction branch from 5ce9d98 to 07bd52f Compare June 23, 2025 21:52

github-actions bot added the wasmtime:api Related to the API of the `wasmtime` crate itself label Jun 23, 2025

dicej mentioned this pull request Jun 24, 2025

update component-model-async plumbing #11123

Merged

alexcrichton reviewed Jun 24, 2025

View reviewed changes

dicej mentioned this pull request Jun 24, 2025

implement Component Model async ABI #11127

Merged

dicej added 2 commits June 24, 2025 11:38

address review feedback

66ce1cd

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

remove unneeded Send bounds

8b8ecb1

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

alexcrichton reviewed Jun 24, 2025

View reviewed changes

crates/wasmtime/src/runtime/fiber.rs Outdated Show resolved Hide resolved

crates/wasmtime/src/runtime/fiber.rs Outdated Show resolved Hide resolved

dicej and others added 12 commits June 24, 2025 16:24

address more review feedback

45859eb

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

update impl Send For StoreFiber comment

77d0a80

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

Remove currently-extraneous Result<()> from fibers

8a8768e

May be needed for concurrent bits, but for now not necessary.

Use safe pointers instead of raw pointers

b67b8b8

It's predicted Miri won't like this, but for now in-repo it's ok with it.

Fold more responsibility into resume_fiber_raw

fee87ee

Remove the need for the function entirely and replace it with `resume_fiber`.

Remove channels from async fibers

3de8282

Can use stack-based closures/results to transmit the result instead of needing a channel.

Fold on_fiber_raw directly into on_fiber

dbc0b7d

The `on_fiber` function is small enough it should be possible to do so.

Don't use Option in FiberFuture

7562905

Leave the fiber non-optional at-rest so it's always available for the destructor.

Fold suspend functions together

b9926c6

Small shims, not otherwise public at this time, so remove a layer of indirection.

Move stack limit management to FiberResumeState

44c96f8

Helps remove some raw pointers that are held for a long time within `AsyncCx`

add some doc comments to fiber.rs

97e6aef

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

dicej mentioned this pull request Jun 25, 2025

add Component Model async ABI tests #11136

Merged

dicej and others added 5 commits June 25, 2025 16:15

fix non-component-model-async build warnings

8edeb24

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

make resume_fiber private in fiber.rs

3806f68

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

Shrink PollContext state

7afd823

Move management of the async guard range elsewhere to the normal save/restore area.

Adjust some lint attributes

2942248

alexcrichton added 6 commits June 26, 2025 09:57

Make manipulation of AsyncState safe

3e92492

No need for raw pointers with recent refactorings.

Fix dead code warning

f45ea3a

More dead code warnings

37630c0

Cut down on raw pointers in fiber.rs

c28dfaa

Move executor save/restore to normal fiber state save/restore

8adfffb

Bikeshed a method name

141d70e

alexcrichton approved these changes Jun 26, 2025

View reviewed changes

dicej commented Jun 26, 2025

View reviewed changes

crates/wasmtime/src/runtime/fiber.rs Outdated Show resolved Hide resolved

update comment in make_fiber

691e58f

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

dicej enabled auto-merge June 26, 2025 20:59

dicej added this pull request to the merge queue Jun 26, 2025

github-merge-queue bot removed this pull request from the merge queue due to failed status checks Jun 26, 2025

fix machports build

c3cb184

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

dicej enabled auto-merge June 26, 2025 21:38

dicej added this pull request to the merge queue Jun 26, 2025

Merged via the queue into bytecodealliance:main with commit 421136d Jun 26, 2025
41 checks passed

dicej deleted the generalized-fiber-abstraction branch June 26, 2025 22:15

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

generalize async fiber abstraction #11114

generalize async fiber abstraction #11114

Uh oh!

dicej commented Jun 23, 2025

Uh oh!

alexcrichton left a comment

Uh oh!

Uh oh!

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!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

generalize async fiber abstraction #11114

generalize async fiber abstraction #11114

Uh oh!

Conversation

dicej commented Jun 23, 2025

Uh oh!

alexcrichton left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

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!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants