unlike the other Parse*Error types, ParseCharError didn't have these implemented for whatever reason

Currently if `--out-dir` is set to a non-existent directory, the
compiler will throw unfriendly messages like `error: could not write
output to subdir/example.crate.allocator.rcgu.o: No such file or
directory`, which, while not completely unreadable, isn’t very
user-friendly either. This change creates the directory automatically
if it does not yet exist.

Commit 7ed00ca improved our error reporting by including the target
function in our error messages when there is an argument count mismatch.
A simple example from the UI tests is:

```
error[E0593]: function is expected to take a single 2-tuple as argument, but it takes 0 arguments
  --> $DIR/closure-arg-count.rs:32:53
   |
32 |     let _it = vec![1, 2, 3].into_iter().enumerate().map(foo);
   |                                                     ^^^ expected function that takes a single 2-tuple as argument
...
44 | fn foo() {}
   | -------- takes 0 arguments
```

However, this assumed the target span was always available. This does
not hold true if the target function is in `std` or another crate. A
simple example from rust-lang#48046 is assigning `str::split` to a function type
with a different number of arguments.

Fix by removing all of the labels and suggestions related to the target
span when it's not found.

Fixes rust-lang#48046

Unfortunately left out it means that when the `#![feature(proc_macro)]` flag is
in effect it fails to find `rustc_args_required_const` for expansion. This
version, however, is verified to work with stdsimd's requirements!

This commit disallows acquiring a function pointer to functions tagged as
`#[rustc_args_required_const]`. This is intended to be used as future-proofing
for the stdsimd crate to avoid taking a function pointer to any intrinsic which
has a hard requirement that one of the arguments is a constant value.

…(in mobile too)

Right now the ccache setting is only used for LLVM, but this tweaks it to also
be used for build scripts so C++ builds like `librustc_llvm` can be a bit
speedier.

A bug was recently fixed in dlmalloc which meant that released memory to the
system accidentally wasn't getting reused, causing programs to be far slower
than they should be!

rust-lang#48048

Various versions of LLVM do various things with debuginfo but we're most curious
that our own code is working so let's just disable optimizations in this test to
ensure we have as much pristine debuginfo as possible.

Eventually the LLVM version of Emscripten and the main LLVM will diverge, and we
can't cross-compile an older version of LLVM using a newer version of LLVM's
`llvm-config`.

Right now the `--build` option was accidentally omitted, so we're bootstraping
from `x86_64` to `i686`. In addition to being slower (more compiles) that's not
actually bootstrapping!

derive PartialEq and Eq for `ParseCharError`

unlike the other Parse*Error types, ParseCharError didn't have these implemented for whatever reason

…, r=nikomatsakis

Remove unused data structures

Cleanup; as far as I can tell the compiler no longer uses these.

Create a directory for --out-dir if it does not already exist

Currently if `--out-dir` is set to a non-existent directory, the compiler will throw unfriendly messages like `error: could not write output to subdir/example.crate.allocator.rcgu.o: No such file or
directory`, which, while not completely unreadable, isn’t very user-friendly either. This change creates the directory automatically if it does not yet exist.

…lexcrichton

Customizable extended tools

This PR adds `build.tools` option to manage installation of extended rust tools.

By default it doesn't change installation. All tools are built and `rls` and `rustfmt` allowed to fail installation.

If some set of tools chosen only those tools are built and installed without any fails allowed.

It solves some slotting issues with extended build enabled: https://bugs.gentoo.org/show_bug.cgi?id=645498

…-on-target-without-span, r=estebank

Fix ICE for mismatched args on target without span

Commit 7ed00ca improved our error reporting by including the target function in our error messages when there is an argument count mismatch. A simple example from the UI tests is:

```
error[E0593]: function is expected to take a single 2-tuple as argument, but it takes 0 arguments
  --> $DIR/closure-arg-count.rs:32:53
   |
32 |     let _it = vec![1, 2, 3].into_iter().enumerate().map(foo);
   |                                                     ^^^ expected function that takes a single 2-tuple as argument
...
44 | fn foo() {}
   | -------- takes 0 arguments
```

However, this assumed the target span was always available. This does not hold true if the target function is in `std` or another crate. A simple example from rust-lang#48046 is assigning `str::split` to a function type with a different number of arguments.

Fix by omitting all of the labels and suggestions related to the target span when it's not found.

Fixes rust-lang#48046

r? @estebank

…uietMisdreavus

rustdoc: Hide `-> ()` in cross crate inlined Fn* bounds

Update binaryen to fix -Werror with GCC 8

r? @alexcrichton

…lacrum

rustbuild: Pass `ccache` to build scripts

Right now the ccache setting is only used for LLVM, but this tweaks it to also
be used for build scripts so C++ builds like `librustc_llvm` can be a bit
speedier.

…Misdreavus

intra-doc-links: bail early for linky things

r? @QuietMisdreavus

…-proc-macro, r=eddyb

Disallow function pointers to #[rustc_args_required_const]

This commit disallows acquiring a function pointer to functions tagged as
`#[rustc_args_required_const]`. This is intended to be used as future-proofing
for the stdsimd crate to avoid taking a function pointer to any intrinsic which
has a hard requirement that one of the arguments is a constant value.

Note that the first commit here isn't related specifically to this feature, but was necessary to get this working in stdsimd!

…=QuietMisdreavus

Hide theme button under menu in mobile mode and fix top margin issue …

Fixes rust-lang#48060.

r? @QuietMisdreavus

…k-Simulacrum

Update the dlmalloc submodule

A bug was recently fixed in dlmalloc which meant that released memory to the
system accidentally wasn't getting reused, causing programs to be far slower
than they should be!

Fix visitation order of calls so that it matches execution order. Fixes rust-lang#48048

r? @nikomatsakis

…komatsakis

Forcibly disable optimizations in backtrace-debuginfo

Various versions of LLVM do various things with debuginfo but we're most curious
that our own code is working so let's just disable optimizations in this test to
ensure we have as much pristine debuginfo as possible.

This is currently extracted from rust-lang#47828 to land standalone

… r=Mark-Simulacrum

Don't cross-compile Emscripten's LLVM

Eventually the LLVM version of Emscripten and the main LLVM will diverge, and we
can't cross-compile an older version of LLVM using a newer version of LLVM's
`llvm-config`.

This is extracted from rust-lang#47828

…p, r=Mark-Simulacrum

ci: Actually bootstrap on i686 dist

Right now the `--build` option was accidentally omitted, so we're bootstraping
from `x86_64` to `i686`. In addition to being slower (more compiles) that's not
actually bootstrapping!

…labnik Update books for next release