Miri on Windows: run .CRT$XLB linker section on thread-end

[RalfJung]

Hopefully fixes #123583

First commit is originally by @bjorn3

r? @oli-obk
Cc @ChrisDenton

[rustbot]

The Miri subtree was changed

cc @rust-lang/miri

[ChrisDenton]

Rather than #[cfg_attr(miri, used)], could we maybe have like a #[rustc_windows_run_tls_dtors] attribute that backends can choose what to do with? Currently it would be ignored by most backends but that could change in the future.

[RalfJung]

I don't have the time to design and implement such a feature, sorry.

[bjorn3]

With the first commit you should be able to remove the miri special case at

rust/library/std/src/sys/pal/unix/args.rs

Line 105 in 5934aaa

#[cfg(any(miri, not(all(target_os = "linux", target_env = "gnu"))))]

[RalfJung]

I haven't taken your entire commit, only the part that turns link_section into an array. So I can't remove that special case, no -- that could be done in a future PR.

This is intended to be a targeted fix for the Windows issue that makes rustc CI spuriously fail.

[oli-obk]

@bors r+ rollup

[bors]

📌 Commit 5934aaa has been approved by oli-obk

It is now in the queue for this repository.

[RalfJung]

@bors p=1
hopefully fixes CI failures that break unrelated PRs

[bors]

⌛ Testing commit 5934aaa with merge 0230848...

[bors]

☀️ Test successful - checks-actions
Approved by: oli-obk
Pushing 0230848 to master...

[rust-timer]

Finished benchmarking commit (0230848): 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)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-4.2%	[-4.2%, -4.2%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-4.2%	[-4.2%, -4.2%]	1

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: 677.082s -> 676.684s (-0.06%)
Artifact size: 316.00 MiB -> 316.01 MiB (0.00%)