Ignore unwinding edges when checking for unconditional recursion

The unconditional recursion lint determines if all execution paths
eventually lead to a self-recursive call.

The implementation always follows unwinding edges which limits its
practical utility. For example, it would not lint function `f` because a
call to `g` might unwind. It also wouldn't lint function `h` because an
overflow check preceding the self-recursive call might unwind:

```rust
pub fn f() {
    g();
    f();
}

pub fn g() { /* ... */ }

pub fn h(a: usize) {
  h(a + 1);
}
```

To avoid the issue, assume that terminators that might continue
execution along non-unwinding edges do so.

Fixes #78474.

Update LLVM submodule

Update LLVM to fix #92786.

Rollup of 7 pull requests

Successful merges:

 - #90247 (Improve Duration::try_from_secs_f32/64 accuracy by directly processing exponent and mantissa)
 - #91861 (Replace iterator-based construction of collections by `Into<T>`)
 - #92098 (add OpenBSD platform-support page)
 - #92134 (Add x86_64-pc-windows-msvc linker-plugin-lto instructions)
 - #92256 (Improve selection errors for `~const` trait bounds)
 - #92778 (fs: Use readdir() instead of readdir_r() on Linux and Android)
 - #93338 (Update minifier crate version to 0.0.42)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

Update minifier crate version to 0.0.42

Some issues on the CSS minification.

r? `@notriddle`

fs: Use readdir() instead of readdir_r() on Linux and Android

See #40021 for more details.  Fixes #86649.  Fixes #34668.

Improve selection errors for `~const` trait bounds

Add x86_64-pc-windows-msvc linker-plugin-lto instructions

I had some trouble getting cross language LTO working for this target, in part because the very few links of documentation I could find were linux-centric and because of a few very specific errors I ran into. I'm not sure if this is the correct place to document this, but this is one of the first links I found when looking for documentation so it might be the best place for it.

add OpenBSD platform-support page

It mentions x86_64, i686, aarch64 and sparc64 which are actively maintained and used on OpenBSD (binaries provided by standard package distribution on OpenBSD).

I volontary kept `powerpc-unknown-openbsd` unmentioned as it was added by `@Yn0ga` in #82733, and I am unaware if it is functional or not (I doubt as I added libc support only few days ago, and std `c_char` signess was wrong). `@Yn0ga` maybe you comment on your `powerpc-unknown-openbsd` usage ?

Replace iterator-based construction of collections by `Into<T>`

Just a few quality of life improvements in the doc examples. I also removed some `Vec`s in favor of arrays.

Improve Duration::try_from_secs_f32/64 accuracy by directly processing exponent and mantissa

Closes: #90225

The methods now implement direct processing of exponent and mantissa, which should result in the best possible conversion accuracy (modulo truncation, i.e. float value of 19.995 ns will be represented as 19 ns).

Store hir_id_to_def_id in OwnerInfo.

This is for perf test purposes only. Related to #89278

The unconditional recursion lint determines if all execution paths
eventually lead to a self-recursive call.

The implementation always follows unwinding edges which limits its
practical utility. For example, it would not lint function `f` because a
call to `g` might unwind. It also wouldn't lint function `h` because an
overflow check preceding the self-recursive call might unwind:

```rust
pub fn f() {
    g();
    f();
}

pub fn g() { /* ... */ }

pub fn h(a: usize) {
  h(a + 1);
}
```

To avoid the issue, assume that terminators that might continue
execution along non-unwinding edges do so.

Fix the unsoundness in the `early_otherwise_branch` mir opt pass

Closes #78496 .

This change is a significant rewrite of much of the pass. Exactly what it does is documented in the source file (with ascii art!), and all the changes that are made to the MIR that are not trivially sound are carefully documented. That being said, this is my first time touching MIR, so there are probably some invariants I did not know about that I broke.

This version of the optimization is also somewhat more flexible than the original; for example, we do not care how or where the value on which the parent is switching is computed. There is no requirement that any types be the same. This could be made even more flexible in the future by allowing a wider range of statements in the bodies of `BBC, BBD` (as long as they are all the same of course). This should be a good first step though.

Probably needs a perf run.

r? `@oli-obk` who reviewed things the last time this was touched

rustdoc: Pre-calculate traits that are in scope for doc links

This eliminates one more late use of resolver (part of #83761).
At early doc link resolution time we go through parent modules of items from the current crate, reexports of items from other crates, trait items, and impl items collected by `collect-intra-doc-links` pass, determine traits that are in scope in each such module, and put those traits into a map used by later rustdoc passes.
r? `@jyn514`

Update cargo

9 commits in 95bb3c92bf516017e812e7f1c14c2dea3845b30e..1c034752de0df744fcd7788fcbca158830b8bf85
2022-01-18 17:39:35 +0000 to 2022-01-25 22:36:53 +0000
- Sync toml_edit versions (rust-lang/cargo#10329)
- Check --config for dotted keys only (rust-lang/cargo#10176)
- Remove deprecated --host arg for search and publish cmds (rust-lang/cargo#10327)
- doc: it's valid to use OUT_DIR for intermediate artifacts (rust-lang/cargo#10326)
- Use local git info for version. (rust-lang/cargo#10323)
- Fix documenting with undocumented dependencies. (rust-lang/cargo#10324)
- do not compile test for bins flagged as `test = false` (rust-lang/cargo#10305)
- Port cargo from toml-rs to toml_edit (rust-lang/cargo#10086)
- Fix new::git_default_branch with different default (rust-lang/cargo#10306)

Co-authored-by: NNoah Lev <camelidcamel@gmail.com>

Rollup of 5 pull requests

Successful merges:

 - #93250 (Remove deduplication of early lints)
 - #93286 (Add white-space: nowrap to links in the sidebar)
 - #93291 (minor fix for #93231)
 - #93300 (make Windows abort_internal Miri-compatible)
 - #93303 (Fix ICE when parsing bad turbofish with lifetime argument)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

Fix ICE when parsing bad turbofish with lifetime argument

Generalize conditions where we suggest adding the turbofish operator, so we don't ICE during code like

```rust
fn foo() {
  A<'a,>
}
```

but instead suggest adding a turbofish.

Fixes #93282

make Windows abort_internal Miri-compatible

https://github.com/rust-lang/rust/pull/92828 started calling `abort_internal` on double-panics, uncovering that on Windows this function does not work in Miri because of its use of inline assembly.

Cc `@Amanieu`

minor fix for #93231

In #93231 I introduced the new sidebar colours to make the contrast more balanced and easier to read, but it seems I made a copy-paste error in the light theme, resulting in functions appearing green.

This one line change replaces that colour with it's corrected orange/brown colour.

I have double checked the rest of the colours and they seem ok. Sorry for the inconvenience

Add white-space: nowrap to links in the sidebar

We already have overflow: hidden on these links, but if there is a possibility to wrap, they will wrap. This happens in particular for trait implementations because the punctuation (`<>, `) introduces opportunities for breaks. That produces inconsistent UI. Fix it by forcing them not to wrap.

Demo: https://rustdoc.crud.net/jsha/sidebar-nowrap/std/string/struct.String.html

To see the effect, scroll down to the "Trait Implementations" portion of the sidebar and look at IndexMut. Compare vs:

https://doc.rust-lang.org/std/string/struct.String.html
https://doc.rust-lang.org/nightly/std/string/struct.String.html

r? `@camelid`

Remove deduplication of early lints

We already have a general mechanism for deduplicating reported
lints, so there's no need to have an additional one for early lints
specifically. This allows us to remove some `PartialEq` impls.

Store a `Symbol` instead of an `Ident` in `AssocItem`

This is the same idea as #92533, but for `AssocItem` instead
of `VariantDef`/`FieldDef`.

With this change, we no longer have any uses of
`#[stable_hasher(project(...))]`

Rustdoc: remove ListAttributesIter and use impl Iterator instead

This is a continuation of https://github.com/rust-lang/rust/pull/92227.

I found that `ListAttributesIter` did not optimize well and replacing it with a simple `impl Iterator` resulted in 1-3 % instruction count wins locally.

Because I needed to use `impl Iterator` on a slice of AST attributes, I had to implement it using GAT + impl trait. I also have a version without GAT [here](https://github.com/Kobzol/rust/commit/5470e2a65cbd3086d19f0847f44ca9cbbc049689), if GATs are not welcome in rustdoc :D Locally it resulted in equal performance numbers.

Can I ask for a perf. run? Thanks.

r? rust-lang/rustdoc

This eliminates one more late use of resolver

Remove ordering traits from `OutlivesConstraint`

In two cases where this ordering was used, I've replaced the sorting to use a key that does not rely on `DefId` being `Ord`. This is part of #90317. If I understand correctly, whether this is correct depends on whether the `RegionVid`s are tracked during incremental compilation. But I might be mistaken in this approach. cc `@cjgillot`