Co-authored-by: Felix S Klock II <pnkfelix@pnkfx.org>

…ly contains file name

Technically, wiping bootstrap builds can increase the build time.
But in practice, trying to manually resolve post-bump issues and
even accidentally removing the entire build directory will result
in a much greater loss of time. After all, the bootstrap build process
is not a particularly lengthy operation.

Signed-off-by: onur-ozkan <work@onurozkan.dev>

This is the counterpart of `Span::trim_start`.

This will allow the span extractor to produce multiple separate buckets,
instead of just one flat list of spans.

This is less elegant than returning an iterator, but more flexible.

…kets

This performs the same task as the hole-carving code in the main span refiner,
but in a separate earlier pass.

Now that hole spans are handled by a separate earlier pass, this code never
sees hole spans, and therefore doesn't need to deal with them.

on netbsd, procfs is not as central as on linux/solaris thus
can be perfectly not mounted.
Thus using fcntl with F_GETPATH, the kernel deals with MAXPATHLEN
internally too.

…r (can) use it for local items

`ct_infer` and `lower_ty` will correctly result in an error constant or type respectively, as they go through a `HirTyLowerer` method (just like `HirTyLowerer::allow_infer` is a method implemented by both implementors

…y forbidden

…Lowerer`

…n enum that exhaustively supports all call sites

…types-improvement, r=pnkfelix

`rustc --explain E0582` additional example

## Context
*From rust-lang#124744*

Expands the example for E0582, an error ensuring that lifetime in a function's return type is sufficiently constrained (e.g. actually tied to some input type), to show an additional example where one sees the lifetime occurring syntactically among the relevant function input types, but is nonetheless rejected by rustc because a syntactic occurrence is not always sufficient.

…elix

Detect when user is trying to create a lending `Iterator` and give a custom explanation

The scope for this diagnostic is to detect lending iterators specifically and it's main goal is to help beginners to understand that what they are trying to implement might not be possible for `Iterator` trait specifically.

I ended up to changing the wording from originally proposed in the ticket because it might be misleading otherwise: `Data` might have a lifetime parameter but it can be unrelated to items user is planning to return.

Fixes rust-lang#125337

Add intra-doc-links to rustc_middle crate-level docs.

Makes it slightly faster to find these modules, as you don't need to hunt for them in the big list.

…and, r=lcnr

Don't drop `Unsize` candidate in intercrate mode

Fixes rust-lang#125767

Various `HirTyLowerer` cleanups

Previously there was some ad-hoc specialization going on, because you could call `allows_infer`, which basically was deciding between whether the trait object was backed by `FnCtxt` or by `ItemCtxt`. I moved all the different logic into dedicated methods on `HirTyLowerer` and removed `allows_infer`

best reviewed commit-by-commit

…ichaelwoerister

rustc_codegen_ssa: fix `get_rpath_relative_to_output` panic when lib only contains file name

<!--
If this PR is related to an unstable feature or an otherwise tracked effort,
please link to the relevant tracking issue here. If you don't know of a related
tracking issue or there are none, feel free to ignore this.

This PR will get automatically assigned to a reviewer. In case you would like
a specific user to review your work, you can assign it to them by using

    r​? <reviewer name>
-->

When compiles program with `-C rpath=yes` but with no output filename specified, or with filename ONLY, we will get an ICE for now. Fix it by treat empty `output` path in `get_rpath_relative_to_output`  as current dir.

Before this patch:

```bash
rustc -C prefer_dynamic=yes -C rpath=yes -O h.rs  # ICE, no output filename specified
rustc -o hello -C prefer_dynamic=yes -C rpath=yes -O h.rs # ICE, output filename has no path
rustc -o ./hello -C prefer_dynamic=yes -C rpath=yes -O h.rs # Works
```

All those examples work after the patch.

Close rust-lang#119571.
Close rust-lang#125785.

…bertlarsan68

delete bootstrap build before switching to bumped rustc

Technically, wiping bootstrap builds can increase the build time. But in practice, trying to manually resolve post-bump issues and even accidentally removing the entire build directory will result in a much greater loss of time. After all, the bootstrap build process is not a particularly lengthy operation.

Workaround for rust-lang#125578

coverage: Carve out hole spans in a separate early pass

When extracting spans from MIR for use in coverage instrumentation, we sometimes need to identify *hole spans* (currently just closures), and carve up the other spans so that they don't overlap with holes.

This PR simplifies the main coverage-span-refiner by extracting the hole-carving process into a separate early pass. That pass produces a series of independent buckets, and we run the span-refiner on each bucket separately.

There is almost no difference in the resulting mappings, other than in some edge cases involving macros.

…r=cuviper

std::unix::fs::get_path: using fcntl codepath for netbsd instead.

on netbsd, procfs is not as central as on linux/solaris thus can be perfectly not mounted.
Thus using fcntl with F_GETPATH, the kernel deals with MAXPATHLEN internally too.

…=compiler-errors

set `has_unconstrained_ty_var` when generalizing aliases in bivariant contexts

this previously prevented the `regression-31157` benchmark from building

r? `@compiler-errors`