SIDE_MODULE with WASM seems to have no static initializer

[mdboom]

Thanks for helping me resolve #6365. Behind that, I found another issue using SIDE_MODULE with WASM.

It appears that SIDE_MODULEs don't have static initialization data, and I'm seeing zero'd data where I would expect static values to be. (This is somewhat related to my PR #6355, but that was without WASM -- as far as I can tell that code path is irrelevant when WASM is turned on).

The resulting .wasm files for the SIDE_MODULE don't seem to contain any of the static strings I would expect. Is there maybe just some flag I need to set to output the initialization data?

I have a minimal standalone here: https://github.com/mdboom/emscripten-side-module-no-static

[dschuff]

Possibly related: I discovered that #5844 broke things so that static initializers are missing when using wasm+pthreads, so it might be the case that it broke side modules too. I'm working on a fix for pthreads, I can look at side modules after I get that in.

[kripken]

I can't reproduce this problem (using latest incoming - what version are you on? any local changes?), I see Found static string printed and I do see Message in the .so file properly.

This would surprise me if it were broken, as we have tests for it. Does ./tests/runner.py binaryen0.test_dlfcn* binaryen0.test_dylink* pass for you locally?

[dschuff]

I built a side module with #6370 applied and verified that it at least does have a data section (that was the intent of that PR).

[mdboom]

Sorry for wasting time, folks. I think I had emscripten and binaryen out of sync (emscripten had been updated to a recent incoming and binaryen to an older master). Updating both and rebuilding seems to have resolved the issue. Closing.