Currently a test without a `failure-status` directive is treated as having an
expected failure-status of 1, but `run-coverage` tests will want to treat those
tests as expecting success instead.
This will allow the `run-coverage` mode to easily set environment variable
`LLVM_PROFILE_FILE`, and to prevent the executable from being deleted after a
successful run.
Make compiletest aware of targets without dynamic linking
Some parts of the compiletest internals and some tests require dynamic linking to work, which is not supported by all targets. Before this PR, this was handled by if branches matching on the target name.
This PR loads whether a target supports dynamic linking or not from the target spec, and adds a `// needs-dynamic-linking` attribute for tests that require it. Note that I was not able to replace all the old conditions based on the target name, as some targets have `dynamic_linking: true` in their spec but pretend they don't have it in compiletest.
Also, to get this to work I had to *partially* revert #111472 (cc `@djkoloski` `@tmandry` `@bjorn3).` On one hand, only the target spec contains whether a target supports dynamic linking, but on the other hand a subset of the fields can be overridden through `-C` flags (as far as I'm aware only `-C panic=$strategy`). The solution I came up with is to take the target spec as the base, and then override the panic strategy based on `--print=cfg`. Hopefully that should not break y'all again.
`hir`: Add `Become` expression kind (explicit tail calls experiment)
This adds `hir::ExprKind::Become` alongside ast lowering. During hir-thir lowering we currently lower `become` as `return`, so that we can partially test `become` without ICEing.
cc `@scottmcm`
r? `@Nilstrieb`
rust-installer: migrate to clap 4.2, change to 2021 edition and fix few clippy lints
Updated rust-installer to clap 4.2, dropping last user of clap v3; changes to 2021 edition, fixes few clippy warns.
Move windows-sys arm32 shim to c.rs
This moves the arm32 shim in to c.rs instead of appending to the generated file itself.
This makes it simpler to change these workarounds if/when needed. The downside is we need to exclude a couple of functions from being generated (see the comment). A metadata solution could help here but they'll be easy enough to add back if that happens.
Fix rustdoc gui tester
Problem was that the `main` was always exiting with `0`, whether or not there was an error. It led to failing GUI tests being ignored in the CI since no one saw them.
r? ````@klensy````
Syntactically accept `become` expressions (explicit tail calls experiment)
This adds `ast::ExprKind::Become`, implements parsing and properly gates the feature.
cc `@scottmcm`
Add a fully fledged `Clause` type, rename old `Clause` to `ClauseKind`
Does two basic things before I put up a more delicate set of PRs (along the lines of #112714, but hopefully much cleaner) that migrate existing usages of `ty::Predicate` to `ty::Clause` (`predicates_of`/`item_bounds`/`ParamEnv::caller_bounds`).
1. Rename `Clause` to `ClauseKind`, so it's parallel with `PredicateKind`.
2. Add a new `Clause` type which is parallel to `Predicate`.
* This type exposes `Clause::kind(self) -> Binder<'tcx, ClauseKind<'tcx>>` which is parallel to `Predicate::kind` 😸
The new `Clause` type essentially acts as a newtype wrapper around `Predicate` that asserts that it is specifically a `PredicateKind::Clause`. Turns out from experimentation[^1] that this is not negative performance-wise, which is wonderful, since this a much simpler design than something that requires encoding the discriminant into the alignment bits of a predicate kind, or something else like that...
r? ``@lcnr`` or ``@oli-obk``
[^1]: https://github.com/rust-lang/rust/pull/112714#issuecomment-1595653910
mmap/munmap/mremamp shims
This adds basic support for `mmap`/`mremap`/`munmap`, with the specific goal of testing allocators targeting Linux under Miri.
This supports `mmap` with `MAP_PRIVATE|MAP_ANONYMOUS`, and `PROT_READ|PROT_WRITE`, and explicitly does not support `MAP_SHARED` (because that's asking for MMIO) as well as any kind of file mapping (because it seems like nobody does `MAP_PRIVATE` on files even though that would be very sensible). And (officially) we don't support `MAP_FIXED`, so we always ignore the `addr` argument.
This supports `mremap` only when the implementation is allowed to move the mapping (so no `MREMAP_FIXED`, no `MREMAP_DONTUNMAP`, and required `MREMAP_MAYMOVE`), and also when the entirety of a region previously mapped by `mmap` is being remapped.
This supports `munmap` but only when the entirety of a region previously mapped by `mmap` is unmapped.
