[browser][coreCLR] Export __stack_pointer global

[pavelsavara]

Problem

The __stack_pointer WebAssembly global export was being removed from the final .wasm binary despite being explicitly requested via -sEXPORTED_FUNCTIONS=___stack_pointer and -Wl,--export=__stack_pointer.

This global is needed by libCorerun.js to share the stack pointer with dynamically loaded Webcil sub-modules (wasmExports.__stack_pointer is passed as an import to child wasm instances).

Root Cause

Emscripten's wasm-metadce pass (dead code elimination) unconditionally excludes all wasm global exports from its reachability graph:

# building.py line 766
exports = sorted(set(settings.WASM_EXPORTS) - set(settings.WASM_GLOBAL_EXPORTS))

Since __stack_pointer is a wasm global (not a function), it's never included in the DCE graph regardless of EXPORTED_FUNCTIONS. The wasm-metadce tool then considers it unreachable and removes it.

Metadce runs when OPT_LEVEL >= 3 or SHRINK_LEVEL >= 1 (i.e., -O3, -Os, or -Oz).

Fix

1. Set -O2 for Browser Release builds in configureoptimization.cmake — this prevents metadce from running while still enabling wasm-opt optimizations.
2. Add -Wl,--export=__stack_pointer to all Browser link targets (corerun, browserhost, BrowserWasmApp) to ensure wasm-ld emits the export.
3. Add ___stack_pointer to EmccExportedFunction in eng/native.wasm.targets

Follow-up

We should re-visit this after we upgrade emscripten, we could have more efficient way how to do it

[dotnet-policy-service]

Tagging subscribers to 'arch-wasm': @lewing, @pavelsavara
See info in area-owners.md if you want to be subscribed.

[Copilot]

Pull request overview

This PR updates the browser-wasm (CoreCLR) build/link configuration to ensure the __stack_pointer WebAssembly global remains exported in optimized builds, so JS code can access wasmExports.__stack_pointer when instantiating related modules.

Changes:

• Adds explicit linker export of the __stack_pointer global (-Wl,--export=__stack_pointer) and ensures exports.js is linked where needed.
• Adjusts CoreCLR browser-wasm relink arguments to include the new export flag and reposition -lexports.js.
• Updates shared wasm export lists and attempts to change browser Release optimization behavior to avoid toolchain DCE removing the global.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
src/native/corehost/browserhost/CMakeLists.txt	Adds -lexports.js and linker export for __stack_pointer to the browserhost link.
src/mono/browser/build/BrowserWasmApp.CoreCLR.targets	Updates CoreCLR relink args to include -lexports.js and -Wl,--export=__stack_pointer.
src/coreclr/hosts/corerun/CMakeLists.txt	Adds -lexports.js and linker export for __stack_pointer; removes explicit ___stack_pointer from EXPORTED_FUNCTIONS list.
eng/native/configureoptimization.cmake	Changes browser Release optimization intent to -O2 (to avoid metadce).
eng/native.wasm.targets	Adds ___stack_pointer to the exported function list for browser targets.

[kg]

Thank you for getting to the bottom of this. I'll test it out soon.

[kg]

I did gh pr checkout 128567, then rmdir artifacts, then did a release mode test run with dotnet run wasm-ryujit-runner.cs -- --auto-build --config Release --checkout Z:\runtime --assembly Z:\wasm-test\bin\Debug\net10.0\wasm-test.dll --corerun.

It failed:

/// Run '"node"  ./corerun.js -c /Users/kg/AppData/Local/Temp/wasm-ryujit-runner/sandbox/IL /Users/kg/AppData/Local/Temp/wasm-ryujit-runner/sandbox/IL/test-module.wasm' in cwd 'C:\Users\kg\AppData\Local\Temp\wasm-ryujit-runner\sandbox'...
Failed to construct WebAssembly instance for Webcil image: {
  wasmPath: '/Users/kg/AppData/Local/Temp/wasm-ryujit-runner/sandbox/IL/test-module.wasm',
  errorMessage: '__stack_pointer was not preserved by the linker or optimizer'
}
Unhandled exception. System.BadImageFormatException: Bad IL format. The format of the file '/Users/kg/AppData/Local/Temp/wasm-ryujit-runner/sandbox/IL/test-module.wasm' is invalid.
Aborted()

[pavelsavara]

Thanks @kg !

I made another change here and then used your tool

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated 1 comment.

[kg]

Helpers appear to be broken in Release builds atm but this indeed fixes running Hello World by preserving the export, thank you!

EDIT: This is just because your branch is based on a point in time before i fixed write barriers, so everything looks good.

[pavelsavara]

/ba-g unrelated CI failures