This allows printing them in the Debug impl as well as getting them
again using the get_args() method. This allows programs that would
normally spawn another process to more easily show which program they
would have spawned if not for the fact that the target doesn't support
spawning child processes without requiring intrusive changes to keep the
args. For example rustc compiled to wasi will show the full linker
invocation that would have been done.

Fixes cross-language LTO on RISC-V targets (Fixes rust-lang#121924)

Instead of creating a cstring.

Co-authored-by: LoveSy <shana@zju.edu.cn>

…er the loop

…nt captures to child captures

fix: restore issues_entry_limit

Port exit-code run-make test to use rust

As part of rust-lang#121876

~~As draft because formatting will fail because `x fmt` isn't working for me for some reason, I'll debug that later, just opening this now for review, will mark as ready when formatting is fixed~~ (misleading message from x fmt)

cc `@jieyouxu`

…ate-hint, r=estebank

Unconditionally show update nightly hint on ICE

Instead of trying to guess if a update nightly hint should be shown (by checking for system time, querying version and channel info etc.), just show the update nightly hint for nightly compilers. This avoids breaking tests that match on ICE test outputs on nightly/dev channels.

> Another issue is that the outdated nightly hint triggers for ICE tests, causing a mismatch with the test expectation. There doesn't seem to be any env var to suppress this.

See <https://rust-lang.zulipchat.com/#narrow/stream/326414-t-infra.2Fbootstrap/topic/stage0.20compiletest.20broken/near/425543681> for context.

Clean up tests/ui by removing `does-nothing.rs`

In [a previous PR](rust-lang#123297 (comment)), it was suggested that this test be removed:

> it's testing a basic diagnostic for an unknown variable (added over a decade ago for rust-lang#154) that is already covered by probably dozens or hundreds of other tests.

It was then suggested that [opening a new PR](rust-lang#123563 (comment)) for this would be more organized.

I'm setting this as a draft, as:

1. The tests/ui directory is rather disorganized, a large quantity of tests are not even contained inside their own directories. This PR could turn into "clean up the UI tests directory", if I were to place everything into categories (for example, everything related to CLI flags could get placed in a cli directory).
2. This will have a merge conflict with rust-lang#123563 should that get merged. I trust that _this time_, I won't run into [The Incident](rust-lang#123297 (comment)) while rebasing. Edit: Yay, I did it properly!

…k, r=jieyouxu

Don't use bytepos offsets when computing semicolon span for removal

Causes problems when we recover confusable characters w/ a different byte width

Fixes rust-lang#123607

…nikic,DianQK

Set target-abi module flag for RISC-V targets

Fixes cross-language LTO on RISC-V targets (Fixes rust-lang#121924)

…=jhpratt

Store all args in the unsupported Command implementation

This allows printing them in the Debug impl as well as getting them again using the get_args() method. This allows programs that would normally spawn another process to more easily show which program they would have spawned if not for the fact that the target doesn't support spawning child processes without requiring intrusive changes to keep the args. For example rustc compiled to wasi will show the full linker invocation that would have been done.

…mpiler-errors

async closure coroutine by move body MirPass refactoring

Unsure about the last commit, but I think the other changes help in simplifying the control flow