investigate "this assumes only one [wasm] can possibly run at a time" comment in context of ROtrx

[spoonincode]

A wasm_runtime_interface exposes

leap/libraries/chain/include/eosio/chain/webassembly/runtime_interface.hpp

Lines 26 to 27 in d382d71

	//immediately exit the currently running wasm_instantiated_module_interface. Yep, this assumes only one can possibly run at a time.
	virtual void immediately_exit_currently_running_module() = 0;

This is pretty interesting as it suggests this function only has defined behavior if a single wasm is executing, which of course is not the case now with ROtrx.

This is called via eosio_exit host function.

leap/libraries/chain/webassembly/cf_system.cpp

Lines 48 to 50 in d382d71

	void interface::eosio_exit( int32_t code ) const {
	context.control.get_wasm_interface().exit();
	}

via

leap/libraries/chain/wasm_interface.cpp

Lines 115 to 117 in d382d71

	void wasm_interface::exit() {
	my->runtime_interface->immediately_exit_currently_running_module();
	}

So it can absolutely be called via multiple ROtrx at a time.

But... Most likely this is a relic of the way WAVM and/or wabt runtimes were integrated. OC stubs this out as a do-nothing, and EOS VM looks like it throws an exception that likely has proper semantics. I think this "singleton" immediately_exit_currently_running_module() should be completely removed.

[linh2931]

Thanks Matt @spoonincode. Should we just take it out?

[spoonincode]

We'll still need to, I guess, throw an exception for EOS VM in eosio_exit impl. Otherwise yeah I think it can all be ripped out.