This brings the `alloc` crate in-tree.

The code comes from Rust 1.54.0-beta.1, i.e. commit `bf62f4de3`.

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

So that we do not trigger `missing_docs`.

Signed-off-by: Miguel Ojeda <[email protected]>

So that we can enable `-Wmissing_docs` globally.

Signed-off-by: Miguel Ojeda <[email protected]>

The warning is now in the `style` group.

Signed-off-by: Miguel Ojeda <[email protected]>

The warning is now in the `style` group.

Signed-off-by: Miguel Ojeda <[email protected]>

Since we are going to use stable releases [*], we need to enable
the bootstrapping mode to use the unstable features.

[*] For the moment, it is a beta due to a fix we need for LLVM 12,
but the same logic applies.

Signed-off-by: Miguel Ojeda <[email protected]>

Includes:

  - Cross-building a new sysroot with our custom `alloc` in order
    to support `alloc` tests.
  - Documentation generation for in-tree `alloc`.
  - Split `RUSTDOC` command in two for clarity, since now we have
    to also handle `sysroot` differently.
  - Proper handling of both silent and verbose mode for tests
    and `cargo`.
  - Deny missing docs globally `alloc`.
  - Allow broken intra docs for `alloc`.
  - Do not format `rust/alloc` nor `rust/test`, including handling
    the case where `srctree` is a prent folder.
  - `make clean` now cleans `rust/doc` and `rust/test`.
  - Move allocs out of the second expansion section since it is
    in-tree now.
  - Other small cleanups.

I want to clean this up further and factor a few things,
but let's move ahead.

Signed-off-by: Miguel Ojeda <[email protected]>

They are infallible, and could not be actually used because
they will trigger an error when monomorphized, but it is better
to just remove them.

Suggested-by: Gary Guo <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

In preparation for enabling `no_global_oom_handling` for `alloc`,
we need to add some new methods.

They are all marked as:

    #[stable(feature = "kernel", since = "1.0.0")]

for easy identification.

Signed-off-by: Miguel Ojeda <[email protected]>

In preparation for enabling `no_global_oom_handling` for `alloc`,
we need to stop using methods that will disappear when enabling
the configuration option. Instead, we use the new methods
we just added.

Signed-off-by: Miguel Ojeda <[email protected]>

In preparation for enabling `no_global_oom_handling` for `alloc`,
we need to stop using methods that will disappear when enabling
the configuration option. Instead, we use the new methods
we just added.

Signed-off-by: Miguel Ojeda <[email protected]>

Now both `alloc` and users are prepared to be compiled with
`no_global_oom_handling`.

Signed-off-by: Miguel Ojeda <[email protected]>

Sadly, `rustfmt` does not respect `RUSTC_BOOTSTRAP=1` yet,
so we will have to avoid them for the moment.

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>

Now it is needed since we use `objtree` etc. in the target.

Signed-off-by: Miguel Ojeda <[email protected]>

Signed-off-by: Miguel Ojeda <[email protected]>