Ensure copy* intrinsics also perform the static self-init checks

[oli-obk]

fixes #142532

r? @RalfJung

[rustbot]

Some changes occurred to the CTFE machinery

cc @RalfJung, @oli-obk, @lcnr

Some changes occurred to the CTFE / Miri interpreter

cc @rust-lang/miri

[oli-obk]

Adding relnotes as this will break stable code if someone wrote such code in the last year, but there is absolutely no reason to do so, so not even gonna crater it.

[workingjubilee]

It's a soundness fix and is only reachable by having incorrect unsafe code that should be rejected by const eval anyway, right? Then we're justified in "break first, figure out how to migrate smoothly later".

[theemathas]

In theory, this could also affect code where there's correct unsafe code in one crate, and there's incorrect safe code in another crate misusing the first crate.

[RalfJung]

r=me with comment tweaks

[RalfJung]

Suggested change

	// Do our own "do not read from static items while initializing them" checks
	if let Ok((alloc_id, ..)) = self.ptr_try_get_alloc_id(src, size.bytes() as i64) {
	M::before_alloc_read(self, alloc_id)?;
	}
	// For the overlapping case, it is crucial that we trigger the read hook
	// before the write hook -- the aliasing model cares about the order.
	// Trigger read hooks.
	// For the overlapping case, it is crucial that we trigger the read hooks
	// before the write hook -- the aliasing model cares about the order.
	if let Ok((alloc_id, ..)) = self.ptr_try_get_alloc_id(src, size.bytes() as i64) {
	M::before_alloc_read(self, alloc_id)?;
	}

The core interpreter doesn't know what the hook does, so it seems odd to refer to to the static item recursion check.

[oli-obk]

@bors r=RalfJung

[bors]

📌 Commit cfc22cf has been approved by RalfJung

It is now in the queue for this repository.