This allows limiting the number of linker jobs to avoid swapping when
linking LLVM with debug info.

I can't find anything that'd cause unexpected EOF in the source, so let's try
taking a look at the error logs on failures.

When declaring nested unsafe blocks (`unsafe {unsafe {}}`) that trigger
the "unnecessary `unsafe` block" error, point out the enclosing `unsafe
block` or `unsafe fn` that makes it unnecessary.

Some annotations like the "test" annotations might be of interest for
other projects, especially rls. Export all attributes in a new
attributes item.

Remove the AST structure

In recent months there have been a few different people investigating how to make a plugin that
registers a MIR-pass – one that isn’t intended to be eventually merged into rustc proper.

The interface to register MIR passes was added primarily for miri (& later was
found to make prototyping of rustc-proper MIR passes a tiny bit faster). Since miri does not use
this interface anymore it seems like a good time to remove this "feature".

For prototyping purposes a similar interface can be added by developers themselves in their custom
rustc build.

They report their `uname -m` as armv8l rather than aarch64.

Patch originally by Matthias Klose <doko@debian.org>

…r proc macros

This commit alters the compiler to exit quickly if the only output being emitted
is `dep-info`, which doesn't need a lot of other information to generate.

Closes rust-lang#40328

This change introduces a Dockerfile and script which builds a complete
Fuchsia toolchain which can be used to build Rust distribution for
Fuchsia. We only support cross-compiling at the moment, hence only
setting the target.

I don't believe that we need this any more now that `cargo-vendor` isn't
installed to create a source tarball (that only happens on Linux)

This causes problems when first cloning and bootstrapping the repository
unfortunately, so let's ensure that Cargo sticks around in its own workspace.
Because Cargo is a submodule it's not available by default on the inital clone
of the rust-lang/rust repository. Normally it's the responsibility of the
rustbuild to take care of this, but unfortunately to build rustbuild itself we
need to resolve the workspace conflicts.

To deal with this we'll just have to ensure that all submodules are in their own
workspace, which sort of makes sense anyway as updates to dependencies as
bugfixes to Cargo should go to rust-lang/cargo instead of rust-lang/rust. In any
case this commit removes Cargo from the global workspace which should resolve
the issues that we've been seeing.

To actually perform this the `cargo` submodule has been moved to the top
directory to ensure it's outside the scope of `src/Cargo.toml` as a workspace.

…turner

Point to enclosing block/fn on nested unsafe

When declaring nested unsafe blocks (`unsafe {unsafe {}}`) that trigger
the "unnecessary `unsafe` block" error, point out the enclosing `unsafe
block` or `unsafe fn` that makes it unnecessary.

<img width="621" alt="" src="https://cloud.githubusercontent.com/assets/1606434/22139922/26ad468a-de9e-11e6-8884-2945be882ea8.png">

Fixes rust-lang#39144.

Export attributes in save-analysis data

Since this is my first pull-request to rust, I would like to get some feedback about obvious errors in this implementation.

I would like to change the save-analysis data to include arbitrary attribute data.
A use-case I have in mind for this is identifying functions with `#[test]` annotations such that tools like rls can offer a test-runner feature. I described my idea here [rls#173](rust-lang/rls#173).

My changes contain:

1. track a vector of attributes in the various `*Data` types in `data.rs` and `external_data.rs`
2. implement lowering for `Attribute` and `MetaItem`
3. adjust `JsonDumper` to print the attributes

In the lowering of `Attribute` I remove the distinction between `MetaItem` and `NestedMetaItem`. I did this because this distinction is somewhat confusing. For example, `NestedMetaItemKind::Literal` has two identical spans, because both `NestedMetaItem` and `Lit` are defined as `Spanned<_>`.
My model is strictly more general, as it allows an `LitKind` instead of a `Symbol` for `MetaItem` and `Symbol`s are converted into a cooked string. As a consumer of the save-analysis data this shouldn't affect you much.

Example json output of `#[test]` annotation:
```
"attributes": [
  {
    "value": {
      "name": {
        "variant": "Str",
        "fields": [
          "test",
          "Cooked"
        ]
      },
      "kind": "Literal",
      "span": {
        "file_name": "test.rs",
        "byte_start": 2,
        "byte_end": 6,
        "line_start": 1,
        "line_end": 1,
        "column_start": 3,
        "column_end": 7
      }
    },
    "span": {
      "file_name": "test.rs",
      "byte_start": 0,
      "byte_end": 7,
      "line_start": 1,
      "line_end": 1,
      "column_start": 1,
      "column_end": 8
    }
  }
]
```

travis: Fuchsia builder

This change introduces a Dockerfile and script which builds a complete
Fuchsia toolchain which can be used to build Rust distribution for
Fuchsia. We only support cross-compiling at the moment, hence only
setting the target.

…elix

Fix suggestion span error with a line containing multibyte characters

This PR fixes broken suggestions caused by multibyte characters.

e.g. for this code, rustc provides a broken suggestion ([playground](https://is.gd/DWGLu7)):

```rust
fn main() {
    let tup = (1,);
    println!("☃{}", tup[0]);
}
```

```
error: cannot index a value of type `({integer},)`
 --> <anon>:3:21
  |
3 |     println!("☃{}", tup[0]);
  |                     ^^^^^^
  |
help: to access tuple elements, use tuple indexing syntax as shown
  |     println!("☃{}"tup.00]);

error: aborting due to previous error
```

`CodeSuggestion::splice_lines` is misusing `Loc.col` (`CharPos`) as a byte offset when slicing source.

… r=alexcrichton

Restore creating the channel-rust-$channel-date.txt files

I have **not** run this (because I don't know how to 😇), but it *does* compile.

r? @alexcrichton

…akis

Remove ability for plugins to register a MIR pass

In recent months there have been a few different people investigating how to make a plugin that
registers a MIR-pass – one that isn’t intended to be eventually merged into rustc proper.

The interface to register MIR passes was added primarily for miri (& later was
found to make prototyping of rustc-proper MIR passes a tiny bit faster). Since miri does not use
this interface anymore it seems like a good time to remove this "feature".

For prototyping purposes a similar interface can be added by developers themselves in their custom
rustc build.

cc @nikomatsakis

…richton

Fix emscripten test detection

Without this change `rustbuild` will attempt to run `.js.map` files (if they exist) resulting in lots of sadness.

r? @alexcrichton

Support armhf abi on 64-bit ARM cpus

They report their `uname -m` as armv8l rather than aarch64.

Patch originally by Matthias Klose <doko@debian.org>

…alexcrichton

rustbuild: expose LLVM_PARALLEL_LINK_JOBS

This allows limiting the number of linker jobs to avoid swapping when linking LLVM with debug info.

Clean up rustdoc css

r? @rust-lang/docs

Fix incorrect span label formatting

Fix rust-lang#40157.

Don't put Cargo into the rustc workspace

This causes problems when first cloning and bootstrapping the repository
unfortunately, so let's ensure that Cargo sticks around in its own workspace.
Because Cargo is a submodule it's not available by default on the inital clone
of the rust-lang/rust repository. Normally it's the responsibility of the
rustbuild to take care of this, but unfortunately to build rustbuild itself we
need to resolve the workspace conflicts.

To deal with this we'll just have to ensure that all submodules are in their own
workspace, which sort of makes sense anyway as updates to dependencies as
bugfixes to Cargo should go to rust-lang/cargo instead of rust-lang/rust. In any
case this commit removes Cargo from the global workspace which should resolve
the issues that we've been seeing.

To actually perform this the `cargo` submodule has been moved to a new `vendor`
directory to ensure it's outside the scope of `src/Cargo.toml` as a workspace.

Closes rust-lang#40284

Expect macro defs in save-analysis and add expn info to spans for att…

…r proc macros

r? @jseyfried

travis: Attempt to debug sccache failures

I can't find anything that'd cause unexpected EOF in the source, so let's try
taking a look at the error logs on failures.

rustc: Exit quickly on only `--emit dep-info`

This commit alters the compiler to exit quickly if the only output being emitted
is `dep-info`, which doesn't need a lot of other information to generate.

Closes rust-lang#40328

Box docs: no allocation is done for ZSTs.

Updated to add a small bit saying that ZSTs don't actually allocate on `Box::new`.

…or, r=alexcrichton

Distinguish the ways `CStr::from_bytes_with_nul` can fail

Implement placement-in protocol for `VecDeque`

CC rust-lang#30172

r? @nagisa

…r=alexcrichton

OsString::shrink_to_fit.

Considering how the other capacity-related methods are there, I found it odd that this one wasn't included.

Will create a tracking issue once I get an OK on this.

Clarify docs in `VecDeque::resize`

…rson

travis: Remove compiling OpenSSL through homebrew

I don't believe that we need this any more now that `cargo-vendor` isn't
installed to create a source tarball (that only happens on Linux)