…checks in the query engine

… allow for recursive invocations.

Previously, "_1" was not marked as "definitely uninitialized" after a "switchInt(move _1)" terminator.

Related discussion: https://internals.rust-lang.org/t/why-is-2-definitely-initialized-after-switchint-move-2/6760

Closes rust-lang#47981

This is pretty unsatisfying since it is working around a span bug. However, I can't track down the span bug and it could be in the parser, proc macro expansion, the user macro, or Syn (or any other library that can manipulate spans). Given that user code can cause this error, I think we need to be more robust here.

…rt statics.

Extend `Termination` trait with a method to determine what happens
with a unit test.

This commit incorporates work by Bastian Köcher <git@kchr.de>.

Also make `std::termination` module public and rename feature.

The lib feature needs a different name from the language feature.

src/rfc-1937-termination-trait/termination-trait-for-result-box-error_ok.rs

The type checker invokes the borrow checker for closures it finds, so
removing the NLL type checker affects ordering of errors somewhat.

Remove "static item recursion checking" in favor of relying on cycle checks in the query engine

Tests are changed to use the cycle check error message instead. Some duplicate tests are removed.

r? @eddyb

…atsakis

Error on nested impl Trait and path projections from impl Trait

cc rust-lang#34511

r? @nikomatsakis

…ests, r=eddyb

Termination trait in tests

Support the `Termination` trait in unit tests (cc rust-lang#43301)

Also, a drive-by fix for rust-lang#47075.

This is joint work with @bkchr.

…coding, r=nikomatsakis

incr.comp.: Don't keep RefCells in on-disk-cache borrowed in order to allow for recursive invocations.

Fixes rust-lang#47972.

r? @nikomatsakis

…komatsakis

incr.comp.: Store DepNode colors in a dense array instead of a hashmap.

Implements half of rust-lang#47293.

r? @nikomatsakis

mir: Gather move at SwitchInt, Assert terminators

Previously, `_1` was not marked as "definitely uninitialized" after a `switchInt(move _1)` terminator.

I think the same goes for the `assert` terminator.

Related discussion: https://internals.rust-lang.org/t/why-is-2-definitely-initialized-after-switchint-move-2/6760

Avoid ICE in arg mistmatch error for tuple variants

Fix rust-lang#47706.

save-analysis: power through bracket mis-counts

Closes rust-lang#47981

This is pretty unsatisfying since it is working around a span bug. However, I can't track down the span bug and it could be in the parser, proc macro expansion, the user macro, or Syn (or any other library that can manipulate spans). Given that user code can cause this error, I think we need to be more robust here.

r? @eddyb

…ebank

unused_unsafe: don't label irrelevant fns

Fixes rust-lang#48131

Diagnostic bugfix to remove an errant note.  Stops the search for an enclosing unsafe scope at the first safe fn encountered.

```rust
pub unsafe fn outer() {
    fn inner() {
        unsafe { /* unnecessary */ }
    }

    inner()
}
```

**Before:**

```
warning: unnecessary `unsafe` block
 --> src/main.rs:3:9
  |
1 | pub unsafe fn outer() {
  | --------------------- because it's nested under this `unsafe` fn
2 |     fn inner() {
3 |         unsafe { /* unnecessary */ }
  |         ^^^^^^ unnecessary `unsafe` block
  |
  = note: #[warn(unused_unsafe)] on by default
```

**After:**

```
warning: unnecessary `unsafe` block
 --> src/main.rs:3:9
  |
3 |         unsafe { /* unnecessary */ }
  |         ^^^^^^ unnecessary `unsafe` block
  |
  = note: #[warn(unused_unsafe)] on by default
```

…fid, r=eddyb

Allow for instantiating statics from upstream crates

This PR makes the infrastructure around translating statics a bit more flexible so that it can also instantiate statics from upstream crates if the need arises. This is preparatory work for a MIR-only RLIBs prototype, where the instantiation of a `static` may be deferred until a leaf crate.

r? @eddyb (feel free to assign to someone else if you're busy)

…ikomatsakis

When encountering invalid token after `unsafe`, mention `{`

Fix rust-lang#37158.

rustc_data_structures: add missing #[inline].

r? @nikomatsakis

…eck-twice, r=eddyb

Fixes rust-lang#47311.
r? @nrc