Use unused_generic_params from crate metadata

[compiler-errors]

Due to the way that separate_provide_extern interacted with the implementation of <ty::InstanceDef<'tcx> as Key>::query_crate_is_local, we actually never hit the foreign provider for unused_generic_params.

Additionally, since the local provider of unused_generic_params calls should_polymorphize, which always returns false if the def-id is foreign, this means that we never actually polymorphize monomorphic instances originating from foreign crates.

We don't actually encode unused_generic_params for items where all generics are used, so I had to tweak the foreign provider to fall back to ty::UnusedGenericParams::new_all_used() to avoid more ICEs when the above bugs were fixed.

[rustbot]

r? @Nilstrieb

(rustbot has picked a reviewer for you, use r? to override)

[compiler-errors]

cc @davidtwco

[Noratrieb]

You had a FIXME in the other PR, could you add this here as well? Would be nice to just default

[Noratrieb]

Oh also, is there a reason why the type doesn't implement that trait? Could you just add the impl? Or is there something to it

[compiler-errors]

For us to have a defaulted table, the value must implement FixedSizeEncoding and IsDefault. Lazy<T> can be FixedSizeEncoding, but can't be checked for defaultness because they're just a file offset.

I could instead encode the table value as Option<LazyValue<UnusedGenericParams>>, but then ProcessQueryValue<T> for Option<T> panics if the value decoded is None -- we could use specializtion to do:

impl ProcessQueryValue<'_, UnusedGenericParams> for Option<UnusedGenericParams> {
    #[inline(always)]
    fn process_decoded(self, _tcx: TyCtxt<'_>, _err: impl Fn() -> !) -> UnusedGenericParams {
        self.unwrap_or_else(|| UnusedGenericParams::new_all_used())
    }
}

but I don't feel like using specialization here is compelling enough to save a few lines in the decoder rather than providing an explicit body here.

[Noratrieb]

haha, yeah, it's fine

[Noratrieb]

r=me

[Noratrieb]

@bors r+

[bors]

📌 Commit ee2d428 has been approved by Nilstrieb

It is now in the queue for this repository.

[davidtwco]

Thanks for catching this - it seems like it could make a significant impact on the effectiveness of polymorphization in cross-crate scenarios.