Provide cabi_realloc on wasm32-wasip2 by default #122411

alexcrichton · 2024-03-12T22:28:39Z

This commit provides a component model intrinsic in the standard library
by default on the wasm32-wasip2 target. This intrinsic is not
required by the component model itself but is quite common to use, for
example it's needed if a wasm module receives a string or a list.

The intention of this commit is to provide an overridable definition in
the standard library through a weak definition of this function. That
means that downstream crates can provide their own customized and more
specific versions if they'd like, but the standard library's version
should suffice for general-purpose use.

rustbot · 2024-03-12T22:28:47Z

r? @m-ou-se

rustbot has assigned @m-ou-se.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

alexcrichton · 2024-03-12T22:29:25Z

cc @sunfishcode and @rylev

I've tested this locally where the wit-bindgen test suite passes with the wasm32-wasip2 target from this PR

This commit adds some initial support for `wasm32-wasip2` in `wit-bindgen`. Everything works as-is but one of the benefits of a new target is that we can move runtime bits such as `cabi_realloc` into the standard library rather than in this crate. This PR sets up infrastructure to test the `wasm32-wasip2` target and additionally assumes rust-lang/rust#122411 for providing the implementation of `cabi_realloc`.

sunfishcode · 2024-03-12T23:05:04Z

library/std/src/sys/pal/mod.rs

    } else if #[cfg(all(target_os = "wasi", target_env = "p2"))] {
        mod wasip2;
        pub use self::wasip2::*;
+    } else if #[cfg(target_os = "wasi")] {
+        mod wasi;


Would it make sense to rename this module to wasip1?

Good point! I'll do that as a follow-up

library/std/src/sys/pal/wasip2/cabi_realloc.rs

bors · 2024-03-13T10:15:16Z

☔ The latest upstream changes (presumably #122423) made this pull request unmergeable. Please resolve the merge conflicts.

The ordering of targets in `pal/mod.rs` did not end up using the wasip2 implementation, so after reordering that I've edited the implementation to compile correctly.

This commit provides a component model intrinsic in the standard library by default on the `wasm32-wasip2` target. This intrinsic is not required by the component model itself but is quite common to use, for example it's needed if a wasm module receives a string or a list. The intention of this commit is to provide an overridable definition in the standard library through a weak definition of this function. That means that downstream crates can provide their own customized and more specific versions if they'd like, but the standard library's version should suffice for general-purpose use.

sunfishcode

As I mentioned above, it's inconvenient that this cabi_realloc isn't available for no_std builds, however there are alternatives if we use Nightly Rust, so we can revisit this later if needed.

This commit adds some initial support for `wasm32-wasip2` in `wit-bindgen`. Everything works as-is but one of the benefits of a new target is that we can move runtime bits such as `cabi_realloc` into the standard library rather than in this crate. This PR sets up infrastructure to test the `wasm32-wasip2` target and additionally assumes rust-lang/rust#122411 for providing the implementation of `cabi_realloc`.

jeffparsons · 2024-03-23T01:41:14Z

Is this stuck? Might it need a "bors r+" to get it into the queue or is Bors meant to pick up review approvals now?

alexcrichton · 2024-03-24T16:28:25Z

Ah while @rylev and @sunfishcode and I are all stakeholders on the wasm32-wasip2 target I'm not sure any of us have r+ permissions, so we'll need someone else to tell bors this is ready I think.

rylev · 2024-04-02T16:15:46Z

Looks like we might need @m-ou-se to reroll the review.

m-ou-se · 2024-04-02T17:03:36Z

@bors r+

bors · 2024-04-02T17:03:39Z

📌 Commit 5af8187 has been approved by m-ou-se

It is now in the queue for this repository.

…kingjubilee Rollup of 4 pull requests Successful merges: - rust-lang#122411 ( Provide cabi_realloc on wasm32-wasip2 by default ) - rust-lang#123349 (Fix capture analysis for by-move closure bodies) - rust-lang#123359 (Link against libc++abi and libunwind as well when building LLVM wrappers on AIX) - rust-lang#123388 (use a consistent style for links) r? `@ghost` `@rustbot` modify labels: rollup

Rollup merge of rust-lang#122411 - alexcrichton:wasm32-wasip2-cabi-realloc, r=m-ou-se Provide cabi_realloc on wasm32-wasip2 by default This commit provides a component model intrinsic in the standard library by default on the `wasm32-wasip2` target. This intrinsic is not required by the component model itself but is quite common to use, for example it's needed if a wasm module receives a string or a list. The intention of this commit is to provide an overridable definition in the standard library through a weak definition of this function. That means that downstream crates can provide their own customized and more specific versions if they'd like, but the standard library's version should suffice for general-purpose use.

This commit adds some initial support for `wasm32-wasip2` in `wit-bindgen`. Everything works as-is but one of the benefits of a new target is that we can move runtime bits such as `cabi_realloc` into the standard library rather than in this crate. This PR sets up infrastructure to test the `wasm32-wasip2` target and additionally assumes rust-lang/rust#122411 for providing the implementation of `cabi_realloc`.

rustbot assigned m-ou-se Mar 12, 2024

rustbot added O-wasi Operating system: Wasi, Webassembly System Interface S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Mar 12, 2024

alexcrichton mentioned this pull request Mar 12, 2024

Start supporting the wasm32-wasip2 target bytecodealliance/wit-bindgen#892

Merged

sunfishcode reviewed Mar 12, 2024

View reviewed changes

alexcrichton added 2 commits March 13, 2024 08:15

Get wasm32-wasip2 compiling with its custom pal implementation

3311283

The ordering of targets in `pal/mod.rs` did not end up using the wasip2 implementation, so after reordering that I've edited the implementation to compile correctly.

rylev approved these changes Mar 13, 2024

View reviewed changes

alexcrichton force-pushed the wasm32-wasip2-cabi-realloc branch from 47db843 to 5af8187 Compare March 13, 2024 16:38

sunfishcode approved these changes Mar 14, 2024

View reviewed changes

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Apr 2, 2024

workingjubilee mentioned this pull request Apr 3, 2024

Rollup of 4 pull requests #123402

Merged

bors merged commit abb0393 into rust-lang:master Apr 3, 2024
11 checks passed

rustbot added this to the 1.79.0 milestone Apr 3, 2024

alexcrichton mentioned this pull request May 7, 2024

module does not export a function named cabi_realloc bytecodealliance/wasi-rs#87

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Provide cabi_realloc on wasm32-wasip2 by default #122411

Provide cabi_realloc on wasm32-wasip2 by default #122411

alexcrichton commented Mar 12, 2024

rustbot commented Mar 12, 2024

alexcrichton commented Mar 12, 2024

sunfishcode Mar 12, 2024

alexcrichton Mar 12, 2024

bors commented Mar 13, 2024

sunfishcode left a comment

jeffparsons commented Mar 23, 2024

alexcrichton commented Mar 24, 2024

rylev commented Apr 2, 2024

m-ou-se commented Apr 2, 2024

bors commented Apr 2, 2024

Provide cabi_realloc on wasm32-wasip2 by default #122411

Provide cabi_realloc on wasm32-wasip2 by default #122411

Conversation

alexcrichton commented Mar 12, 2024

rustbot commented Mar 12, 2024

alexcrichton commented Mar 12, 2024

sunfishcode Mar 12, 2024

Choose a reason for hiding this comment

alexcrichton Mar 12, 2024

Choose a reason for hiding this comment

bors commented Mar 13, 2024

sunfishcode left a comment

Choose a reason for hiding this comment

jeffparsons commented Mar 23, 2024

alexcrichton commented Mar 24, 2024

rylev commented Apr 2, 2024

m-ou-se commented Apr 2, 2024

bors commented Apr 2, 2024