The binding coroutines in libretro builds are constructed on the VM
stack, so reallocating the VM memory would corrupt the memory of any
currently existing coroutines.
I've changed it so that the coroutines are no longer constructed on the
VM stack so that they're unaffected by VM memory reallocations, and
added a "slot" mechanism for storing variables on the VM stack. (Any
Ruby `VALUE`s used by a coroutine have to be stored on the VM stack so
that the Ruby garbage collector doesn't free them while they're being
used, which is why the slot mechanism is necessary.)
Okay, the coroutine implementation of `sandbox_malloc` is clearly
broken. It would be working if Asyncify instrumented the `memory.grow`
WebAssembly instruction, but it doesn't instrument it.
This commit reverts commit 42c4ff9497 and
also increases the default VM memory allocation from 64 MiB to 96 MiB to
account for the lack of ability to increase the memory allocation at run
time. I'll find some new way to implement increasing the memory
allocation later.
According to AddressSanitizer, when `sandbox_malloc` causes the
WebAssembly memory to grow in size, every single coroutine on the
sandbox stack gets corrupted. So if `sandbox_malloc` is going to cause
the memory to grow in size, we need to yield so that there are no
coroutines on the sandbox stack while the reallocation occurs.
Somehow, using `rb_class_new_instance` causes use-after-free errors in
the subsequent call to `set_private_data`.
Well, `rb_obj_alloc` is what's used in the normal MRI bindings. I just
replaced the `rb_obj_alloc` calls with `rb_class_new_instance`
at some point in binding-sandbox. Guess that wasn't a good idea.
I've made it so that `Graphics.update` pauses the Ruby VM and returns to
the libretro frontend. Once the libretro frontend calls `retro_run()`
again, the Ruby VM resumes. This allows the libretro frontend to control
the rendering loop.
