std panicking: Provide panic::always_abort

We must change the atomic read on panic entry to `Acquire`, to pick up
a possible an `always_panic` on another thread.

We add `count` to the names of panic_count::get and ::is_zaero,
because now there is another reason why panic ought to maybe abort.
Renaming these ensures that we have checked every call site to ensure
that they don't need further adjustment.

Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
Co-authored-by: Mara Bos <m-ou.se@m-ou.se>

library/std/src/panic.rs

@@ -461,5 +461,41 @@ pub fn resume_unwind(payload: Box<dyn Any + Send>) -> ! {
     panicking::rust_panic_without_hook(payload)
 }
 
+/// Make all future panics abort directly without running the panic hook or unwinding.
+///
+/// There is no way to undo this; the effect lasts until the process exits or
+/// execs (or the equivalent).
+///
+/// # Use after fork
+///
+/// This function is particularly useful for calling after `libc::fork`.  After `fork`, in a
+/// multithreaded program it is (on many platforms) not safe to call the allocator.  It is also
+/// generally highly undesirable for an unwind to unwind past the `fork`, because that results in
+/// the unwind propagating to code that was only ever expecting to run in the parent.
+///
+/// `panic::always_abort()` helps avoid both of these.  It directly avoids any further unwinding,
+/// and if there is a panic, the abort will occur without allocating provided that the arguments to
+/// panic can be formatted without allocating.
+///
+/// Examples
+///
+/// ```no_run
+/// #![feature(panic_always_abort)]
+/// use std::panic;
+///
+/// panic::always_abort();
+///
+/// let _ = panic::catch_unwind(|| {
+///     panic!("inside the catch");
+/// });
+///
+/// // We will have aborted already, due to the panic.
+/// unreachable!();
+/// ```
+#[unstable(feature = "panic_always_abort", issue = "84438")]
+pub fn always_abort() {
+    crate::panicking::panic_count::set_always_abort();
+}
+
 #[cfg(test)]
 mod tests;