Rust+WebAssembly Working Group Meeting #5

[fitzgen]

Our next meeting will be Thursday, May 3^rd at 16:00 UTC.

Add agenda items here: https://paper.dropbox.com/doc/RustWebAssembly-WG-dcXi5F7Bqlvg10zd7dKve

I will post a hangouts link to this thread and on IRC ~15 minutes before we begin.

Anyone interested in Rust and WebAssembly is invited to join! Yes, that includes you!

[fitzgen]

Reminder that this meeting is happening tomorrow! Add your agenda items :)

[xtuc]

The time is perfect, i'll join this meeting.

I can answer this question I found in the agenda:

Wepback fixes are almost here, anything we can do to help them along?

First any feedback on the approach used would be great. I described the transformation here webpack/webpack#6433 (comment).

Currently we have a blocking issue, i'm working on https://github.com/xtuc/webassemblyjs/tree/master/packages/wasm-edit to improve the robustness of the tool (some binary + transformation are still failling).

[alexcrichton]

Oh awesome thanks for the info @xtuc! Looking forward to seeing you soon :)

FWIW I don't think any of LLD/Clang/Rust have the ability to import a global right now. For example this IR:

@FOO = external global i32

define i32 @foo() {
start:
  %0 = load i32, i32* @FOO, align 4
  ret i32 %0
}

generates this final binary:

  (type (;0;) (func))
  (type (;1;) (func (result i32)))
  (func $__wasm_call_ctors (type 0))
  (func $foo (type 1) (result i32)
    i32.const 0
    i32.load)
  (table (;0;) 1 1 anyfunc)
  (memory (;0;) 2)
  (global (;0;) (mut i32) (i32.const 66560))
  (global (;1;) i32 (i32.const 66560))
  (global (;2;) i32 (i32.const 1024))
  (export "memory" (memory 0))
  (export "__heap_base" (global 1))
  (export "__data_end" (global 2))
  (export "foo" (func $foo)))

In that sense we'd at least be totally workable landing support for the two bugs we have today (not working in chrome and webpack-dev-server caching too aggressively). Once the toolchain has support for importing/exporting globals though I'd love to help out with test cases/fixes where possible!

[xtuc]

@alexcrichton something like that WebAssembly/binaryen#1500?

In my free time I'm working on the support in LLVM and Clang. Rust could probably take advantage of it.

[alexcrichton]

@xtuc I think that binaryen PR is interpreting the "assembly" (is that even the right name for it?) output of LLVM, right? We don't currently use s2wasm from binaryen or the wasm-linker, but rather we use LLVM's native object file format (sort-of wasm binary files) and LLD to link them together. In that sense I think it's different from your PR in that it's primarily adding support for things in binaryen but isn't dealing with LLVM/LLD (which would affect this)

Now it'd be awesome to get this working in LLVM/Clang for sure, this is a definite missing piece of functionality in Rust's wasm support today!

[xtuc]

Yes that's correct, my PR only uses the textual LLVM IR output (.s) mostly because it's easier to test. I'm planning to add it to o2wasm in the future as well.

I haven't looked into the object files for the wasm target yet but we can probably add it to the metadata in order to behave correctly during linking (lld).

[fitzgen]

Hangouts link for this meeting: https://hangouts.google.com/call/vG6rIQOR0lQcHRMgZx5SAAEE

[fitzgen]

Thanks for participating in the meeting, everyone!