Disable llvm-bitcode-linker in the default bootstrap profiles

[Kobzol]

I don't think that we really need to enable llvm-bitcode-linker in the default bootstrap profiles, since it seems that it is only useful for running nvptx tests. It should be enabled on CI, which it is, and that should be enough. People can enable it easily locally, if they want.

The linker causes occasionally some rebuild issues (#122491, #126464), but more importantly it is just needless work to build it locally.

I kept it enabled for dist, because it is distributed as a rustup component (for some reason it's not included in extended? not sure).

Fixes: #126464

[Noratrieb]

oh, it's only required for nvptx tests? very good to delete that default then, completely wasted time

[Kobzol]

r? @onur-ozkan

[onur-ozkan]

Wouldn't this cause some (nvptx) tests to fail?

[Kobzol]

If yes, then we should configure CI to either enable it for that runner, or skip that test if the linker isn't available.

There are a bunch of tests that need special conditions on CI. nvptx tests should be running on a few dedicated jobs, not necessarily everywhere, and not necessarily locally by default.

PR CI is green, we'd have to take a look at what happens on an auto build. I'll also try to run the tests locally.

[Noratrieb]

I'd rather have the tests (that I have never run) fail locally (with an error message telling me what to do) than wait the few seconds every time.

[onur-ozkan]

If yes, then we should configure CI to either enable it for that runner, or skip that test if the linker isn't available.

Yeah, running those tests only if the linker is available would be good solution.

[Kobzol]

Yeah, running those tests only if the linker is available would be good solution.

That kind of already happens, because you explicitly need to compile LLVM with support for nvptx for these tests to not be ignored, IIRC.

Unless you have objections, I'd r+ this to check if full CI passes.

@bors rollup=never

[onur-ozkan]

you explicitly need to compile LLVM with support for nvptx for these tests to not be ignored, IIRC.

I think we do that by default when compiling LLVM

rust/config.example.toml

Line 99 in 92af831

#targets = "AArch64;ARM;BPF;Hexagon;LoongArch;MSP430;Mips;NVPTX;PowerPC;RISCV;Sparc;SystemZ;WebAssembly;X86"

[Kobzol]

If I run the nvpts UI/assembly tests locally, they are ignored though (even if I enable the LLVM bitcode linker) 🤷‍♂️ When using LLVM from CI. I'm not sure what causes them to be ignored then.

In any case, the linker does not seem to be required to run the "default" test suite, therefore I don't see a reason why it should be enabled and compiled by default.

[onur-ozkan]

If I run the nvpts UI/assembly tests locally, they are ignored though (even if I enable the LLVM bitcode linker) 🤷‍♂️ When using LLVM from CI. I'm not sure what causes them to be ignored then.

Sounds totally broken 😄. Not a blocker for this PR anyway, so let's ship this.

@bors r+

[bors]

📌 Commit b4df72a has been approved by onur-ozkan

It is now in the queue for this repository.

[bors]

⌛ Testing commit b4df72a with merge bc3618f...

[bors]

☀️ Test successful - checks-actions
Approved by: onur-ozkan
Pushing bc3618f to master...

[rust-timer]

Finished benchmarking commit (bc3618f): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

This benchmark run did not return any relevant results for this metric.

Cycles

This benchmark run did not return any relevant results for this metric.

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 671.095s -> 670.967s (-0.02%)
Artifact size: 320.43 MiB -> 320.36 MiB (-0.02%)