bootstrap: validate rust.codegen-backends & target.<triple>.codegen-backends

[GrigorenkoPV]

As per #142184 (comment).

Closes #142184.

[rustbot]

r? @albertlarsan68

rustbot has assigned @albertlarsan68.
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]

This PR modifies src/bootstrap/src/core/config.

If appropriate, please update CONFIG_CHANGE_HISTORY in src/bootstrap/src/utils/change_tracker.rs.

[workingjubilee]

mh, I remember a PR deliberately introduce the fallthrough to the previous else if this condition fails to obtain? it was about using an out-of-tree codegen backend, I believe.

cc @FractalFir @WaffleLapkin

[bjorn3]

I suspect the intent was to panic for codegen backend names starting with rustc_codegen_, while only warning for unknown backends. As written before this PR however it would only panic/warn when the name starts with rustc_codegen_ and don't check the name otherwise however.

[FractalFir]

I am not very familiar with this part of the compiler, but, to be honest, I don't see why we should panic on an unknown backend. If I am understanding things correctly, then this will warn people about unknown codegen backends anyway.

My backend in particular does not strictly need this to function(I can override the backend using RUSTFLAGS), but I still think a warning may be better than an outright error here.

[bjorn3]

The panic is for when we know for sure that the backend is specified wrong. Backends should never be specified with the rustc_codegen_ prefix. That is already implicitly added, so you would end up with a path like compiler/rustc_codegen_rustc_codegen_foo if you specified rustc_codegen_foo. Do note that we already panic on other invalid config options.

[WaffleLapkin]

I agree with everything above, I think. Panic on rustc_codegen_ prefix makes sense, but checking for known values (llvm/cranelift/gcc) is annoying for out-of-tree backends like my employer's.