SIGBUS: access to undefined memory

[seanlinsley]

When using grow, if the stack is exhausted you run into an unusual error:

signal: 10, SIGBUS: access to undefined memory

Maybe #59 could change the error type? Otherwise it would help to document this.

Some more background info: I'm trying to use grow as a way to support a higher recursion limit in the prost crate without needing upstream changes (so grow is a better fit than maybe_grow) pganalyze/pg_query.rs#17

[nagisa]

SIGBUS and SIGSEGV are both common signals raised when an unreadable page is accessed. Which one it’ll be is architecture and probably OS-dependent.

#59 could help, by printing additional messages about stack exhaustion, but I’m pretty confident it would still re-raise SIGBUS if SIGBUS was originally signalled. There's no value in changing the type of the signal. I don't expect adding additional messaging to be of high priority to anybody, and it is also pretty difficult thing to implement correctly, due to it having to run inside a signal handler.

[nagisa]

Triage: closing as this is unlikely to be acted upon.