fix bitcast of single-element SIMD vectors

[folkertdev]

in effect this reverts #142768 and adds additional tests. That PR relaxed the conditions on an early return in an incorrect way that would create broken LLVM IR.

https://godbolt.org/z/PaaGWTv5a

#![feature(repr_simd)]

#[repr(simd)]
#[derive(Clone, Copy)]
struct S([i64; 1]);

#[no_mangle]
pub extern "C" fn single_element_simd(b: S) -> i64 {
    unsafe { std::mem::transmute(b) }
}

at the time of writing generates this LLVM IR, where the type of the return is different from the function's return type.

define noundef i64 @single_element_simd(<1 x i64> %b) unnamed_addr {
start:
  ret <1 x i64> %b
}

The test output is actually the same for the existing tests, showing that the change didn't actually matter for any tested behavior. It is probably a bit faster to do the early return, but, well, it's incorrect in general.

zullip thread: #t-compiler > Is transmuting a `T` to `Tx1` (one-element SIMD vector) UB?
cc @sayantn
r? @scottmcm

[rustbot]

Some changes occurred in compiler/rustc_codegen_ssa

cc @WaffleLapkin

[workingjubilee]

Thanks.

@bors r+ rollup

[bors]

📌 Commit 51e1666 has been approved by workingjubilee

It is now in the queue for this repository.

[scottmcm]

My bad. Thanks for adding more test coverage.

[a4lg]

Considering the error message in the CI log, do we need to add something like this?

#![cfg_attr(target_family = "wasm", target_feature(enable = "simd128"))]

[matthiaskrgr]

@bors r-
failed here: #143211 (comment)

[folkertdev]

Turns out you need a bunch of cfg_attrs for the various platforms (the current set is based on tests/codegen/simd/extract-insert-dyn.rs). I've also made the test use minicore now, which is much more convenient when cross-compiling.

@rustbot ready