Remove SSE ABI from i586-pc-windows-msvc

[Noratrieb]

As an i586 target, it should not have SSE. This caused the following warning to be emitted:

warning: target feature `sse2` must be enabled to ensure that the ABI of the current target can be implemented correctly
  |
  = note: this was previously accepted by the compiler but is being phased out; it will become a hard error in a future release!
  = note: for more information, see issue #116344 <https://github.com/rust-lang/rust/issues/116344>

warning: 1 warning emitted

see #116344.

r? RalfJung

[rustbot]

These commits modify compiler targets.
(See the Target Tier Policy.)

[Noratrieb]

Before someone questions why I would use this target, don't worry. I was simply running --print target-spec-json on every target :D

[RalfJung]

Ah, I must have missed this one -- having it inherit from the "stronger" target makes this quite error-prone.

@bors r+ rollup

I was simply running --print target-spec-json on every target :D

So you're reasonably sure that this is the only problematic one?

[bors]

📌 Commit 1c66d5b has been approved by RalfJung

It is now in the queue for this repository.

[RalfJung]

Nominating for backporting to the beta; this is a tier 2 target and without this PR everyone using it will get the above warning.

[Noratrieb]

Ah, I must have missed this one -- having it inherit from the "stronger" target makes this quite error-prone.

@bors r+ rollup

I was simply running --print target-spec-json on every target :D

So you're reasonably sure that this is the only problematic one?

Yes, I am sure that this is the only target where this warning is emitted. I also double-checked all the other i585 targets and they either override it when inheriting or don't inherit at all.

[RalfJung]

(To be clear this is meant for backport to the just-branched beta that will be released in 6 weeks, not the about-to-be-released stable.)

[RalfJung]

@Noratrieb

Yes, I am sure that this is the only target where this warning is emitted. I also double-checked all the other i585 targets and they either override it when inheriting or don't inherit at all.

Is there some way we could turn this into an automated test? I was wondering how to best ensure that all our built-in targets pass the target feature ABI check, but so far didn't find anything great.

[Noratrieb]

Having a "execute -vV for all targets" test seems reasonable to me. We could run it with -Dwarnings or even ensure that there's no stderr outout at all (there currently is for some other targets, some MIPS thing saying it's experimental in LLVM and #137150

[apiraino]

Yes, I am sure that this is the only target where this warning is emitted. I also double-checked all the other i585 targets and they either override it when inheriting or don't inherit at all.

@Noratrieb Isn't there a proposal to remove this target completely in MCP#840? Are there actual people using it and thus is a backport interesting? (asking more for curiosity)

[RalfJung]

It seems like a low-risk backport to avoid the situation of everyone using that target getting a bogus warning. That's why I nominated it. I don't think the fact that the target will likely be removed soon changes that calculus.

[Noratrieb]

I wouldn't bother backporting it given that the target useless and will be removed.

[apiraino]

Beta backport accepted as per compiler team on Zulip. A backport PR will be authored by the release team at the end of the current development cycle. Backport labels handled by them.

@rustbot label +beta-accepted