GH-130397: use __stack_high and __stack_low LLVM WASM attributes

[FFY00]

• Issue: Implement stack overflow protection for webassembly #130397

[hoodmane]

This same approach would also work on emscripten but instead I think we're going the route of making pthread_getattr_np work so it can use the usual Linux code path.

[hoodmane]

This should be size_t, not char right?

[brettcannon]

CI suggests this isn't enough to fix the issue.

Do we need to manually set the stack size, roll back the stack checking changes that are causing the issue, or something else?

[hoodmane]

I was just looking into this today and I think a big part of the problem is that _Py_get_machine_stack_pointer(); does not seem to grow very fast at all in the stack overflow tests. I adjusted _Py_ReachedRecursionLimit() to occasionally log the value it got from _Py_get_machine_stack_pointer() and then ran test_ast_recursion_limit to segfault.

Not only does it get nowhere close to exhausting the 5mb of shadow stack, the shadow stack pointer only seems to move up/down by about ~4000. I am not sure if you'll get the same results on WASI, but if this is true and not just some measurement artifact, the whole approach to stack checks might not be workable at all.

         soft_limit: 3807680
calls: 62 here_addr: 4305520
calls: 63 here_addr: 4305920
calls: 64 here_addr: 4305008
calls: 65 here_addr: 4303776
calls: 66 here_addr: 4302336
calls: 67 here_addr: 4301776
calls: 68 here_addr: 4302768
calls: 69 here_addr: 4303200
calls: 70 here_addr: 4303776
calls: 71 here_addr: 4305024
calls: 72 here_addr: 4305024
calls: 73 here_addr: 4305024
calls: 74 here_addr: 4305024
calls: 75 here_addr: 4305920
calls: 76 here_addr: 4305920
RangeError: Maximum call stack size exceeded

[python-cla-bot]

All commit authors signed the Contributor License Agreement.