Implement Integer funnel shifts

[sayantn]

Parent: #145686
ACP: rust-lang/libs-team#642

This implements funnel shifts on primitive integer types. Implements this for cg_llvm, with a fallback impl for everything else

Thanks @folkertdev for the fixes and tests

cc @rust-lang/libs-api

[rustbot]

r? @tgross35

rustbot has assigned @tgross35.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[rustbot]

Some changes occurred to the intrinsics. Make sure the CTFE / Miri interpreter
gets adapted for the changes, if necessary.

cc @rust-lang/miri, @RalfJung, @oli-obk, @lcnr

Some changes occurred to the CTFE / Miri interpreter

cc @rust-lang/miri

Some changes occurred to the CTFE machinery

cc @RalfJung, @oli-obk, @lcnr

[tgross35]

I see there is a tracking issue, but not an ACP for this. Could you create one? It's an issue template at https://github.com/rust-lang/libs-team/issues.

[rustbot]

This PR was rebased onto a different master commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

[rustbot]

Some changes occurred in compiler/rustc_codegen_gcc

cc @antoyo, @GuillaumeGomez

Some changes occurred in compiler/rustc_codegen_cranelift

cc @bjorn3

[sayantn]

@bjorn3 could you look into this ICE? The clif code looks fine enough, and I don't have enough expertise on clif to see what's going wrong.

[bjorn3]

At the very least there is a bug in the Cranelift IR verifier causing it to panic at https://github.com/bytecodealliance/wasmtime/blob/ecda6e330a946821b5edc0a90bd62288cf7df943/cranelift/codegen/src/verifier/mod.rs#L1907

But the thing you are doing wrong is let zero = fx.bcx.ins().iconst(ty, 0); I think, which is not valid if ty is an i128. You can use the type_zero_value helper function from the common module instead.

Edit: I'm not sure how that even passed the typecheck pass of the verifier.

[tgross35]

Could you update the tracking issue to have signatures matching rust-lang/libs-team#642 (comment)? This should match then.

[sayantn]

Done 👍🏽, could you check if it makes sense (I'm bad at writing docs etc)

[tgross35]

It's easier than that: tracking issues should show function signatures in something like an impl uX { ... } block, no need to write docs :)

[sayantn]

Should I add more variants of the functions? (e.g. wrapping, overflowing, checked, unchecked etc, the default one is the traditionally strict variant)

[tgross35]

I don't think there is any need in the initial PR here. They can be added as a followup

[clarfonthey]

Also worth mentioning that the hardest part here is getting the intrinsic and actual functionality working, since all the other variants, at least for shifts, just require a few explicit checks.

[sayantn]

In that case, I guess this PR is ready. @rustbot ready

[RalfJung]

This attribute means the fallback implementation must check the size to be valid and do unreachable_unchecked (or equivalent) in that case.

Please add Miri tests to ensure this works correctly: all cases of UB must be detected.

[RalfJung]

Looking at the implementation, I don't think you are checking for UB here. Seems like unchecked_funnel_shl(0, 0, n) will just work for any n.

Please don't just blindly add attributes like this, make sure you understand them or ask what to do. Adding this attribute incorrectly introduces hard-to-find bugs into the very tools unsafe code authors trust to find bugs for them.

EDIT: Or, are you relying on the checks in unchecked_shr/unchecked_shl? That's super subtle, definitely needs a comment at least.