Stream: project-ffi-unwind

Topic: Can C++ catch and swallow a Rust panic

Amanieu (Jan 08 2020 at 21:09, on Zulip):

Basically, C++ can catch a Rust panic with catch (...) and then decide not to rethrow it. #67711 makes sure that the Box<Any> is properly freed in this case, so we don't leak memory.

Amanieu (Jan 08 2020 at 21:10, on Zulip):

However this does affect the way Rust handles double-panics. The problem is that panic_count in TLS is incremented when the panic is thrown, but not decremented when the panic is caught and destroyed by C++.

Amanieu (Jan 08 2020 at 21:10, on Zulip):

This means that the next time a panic occurs, Rust will think that it is a double-panic and abort immediately.

Amanieu (Jan 08 2020 at 21:11, on Zulip):

Now the question is, do we want to allow FFI code to catch and discard Rust panics?

Amanieu (Jan 08 2020 at 21:11, on Zulip):

If yes, then we need to rework panic_count and the way double-panics work.

Amanieu (Jan 08 2020 at 21:12, on Zulip):

If not then we can abort in the exception destructor if it is called outside of catch_unwind, just like pthread_cancel does.

Amanieu (Jan 08 2020 at 21:13, on Zulip):

The main user-visible impact is that backtraces when a double-panic occurs will be slightly less precise: the second panic is treated as a normal panic (backtrace depends on RUST_BACKTRACE), but we will abort a bit later in the landing pad for the destructor which calls abort instead of _Unwind_Resume.

nikomatsakis (Jan 08 2020 at 21:22, on Zulip):

I think we probably don't want to allow that, at least not now, but it's a good issue to raise

Kyle Strand (Jan 08 2020 at 21:50, on Zulip):

Aborting in the exception destructor sounds like the right approach to me. That way C++ can still catch the exception as long as it re-throws it, correct?

Amanieu (Jan 08 2020 at 22:01, on Zulip):

Yes, C++ can catch and rethrow it. Just not catch and ignore it.

Kyle Strand (Jan 08 2020 at 22:23, on Zulip):

Yep, that sounds good to me.

Last update: May 27 2020 at 23:15UTC