what matters is we re doing the right things as doing sizeof, rather than
KINFO_FILE_SIZE (only defined on intel architectures), the kernel
 making sure it matches the expectation in its side.

Co-authored-by: Ralf Jung <post@ralfj.de>

This should fix our precompiled std being unsound in `std::fs` code.

This finally unites TyAndLayout, Layout, and LayoutS into the same crate,
as one might imagine they would be placed. No functional changes.

std::fs::get_path freebsd update.

what matters is we re doing the right things as doing sizeof, rather than passing KINFO_FILE_SIZE (only defined on intel architectures), the kernel
 making sure it matches the expectation in its side.

core/net: add Ipv[46]Addr::from_octets, Ipv6Addr::from_segments.

Adds:

- `Ipv4Address::from_octets([u8;4])`
- `Ipv6Address::from_octets([u8;16])`
- `Ipv6Address::from_segments([u16;8])`

equivalent to the existing `From` impls.

Advantages:

- Consistent with `to_bits, from_bits`.
- More discoverable than the `From` impls.
- Helps with type inference: it's common to want to convert byte slices to IP addrs. If you try this

```rust
fn foo(x: &[u8]) -> Ipv4Addr {
   Ipv4Addr::from(foo.try_into().unwrap())
}
```

it [doesn't work](https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=0e2873312de275a58fa6e33d1b213bec). You have to write `Ipv4Addr::from(<[u8;4]>::try_from(x).unwrap())` instead, which is not great. With `from_octets` it is able to infer the right types.

Found this while porting [smoltcp](https://github.com/smoltcp-rs/smoltcp/) from its own IP address types to the `core::net` types.

~~Tracking issues rust-lang#27709 rust-lang#76205~~
Tracking issue: rust-lang#131360

…hat-got-away, r=scottmcm

library: Const-stabilize `MaybeUninit::assume_init_mut`

FCP completed in rust-lang#86722 (comment)

Also moves const-ness of an unstable fn under the `maybe_uninit_slice` gate, Cc rust-lang#63569

…=saethlin

compiler: `{TyAnd,}Layout` comes home

The `Layout` and `TyAndLayout` types are heavily abstract and have no particular target-specific qualities, though we do use them to answer questions particular to targets. We can keep it that way if we simply move them out of `rustc_target` and into `rustc_abi`. They bring a small entourage of connected types with them, but that's fine.

This will allow us to strengthen a few abstraction barriers over time and thus make the notoriously gnarly layout code easier to refactor. For now, we don't need to worry about that and deliberately use reexports to minimize this particular diff.

…ersion, r=Kobzol

emscripten: Use the latest emsdk 3.1.68

This should fix our precompiled std being unsound in `std::fs` code.

Should resolve rust-lang#131467 I hope.

miri: avoid cloning AllocExtra

We shouldn't be cloning Miri allocations, so make `AllocExtra::clone` panic instead, and adjust the one case where we *do* clone (the leak check) to avoid cloning.

This is in preparation for rust-lang/miri#3966 where I am adding something to `AllocExtra` that cannot (easily) be cloned.

r? ``@saethlin``

merge const_ipv4 / const_ipv6 feature gate into 'ip' feature gate

rust-lang#76205 has been closed a while ago, but there are still some functions that reference it. Those functions are all unstable *and* const-unstable. There's no good reason to use a separate feature gate for their const-stability, so this PR moves their const-stability under the same gate as their regular stability, and therefore removes the remaining references to rust-lang#76205.

…ouxu

Also use outermost const-anon for impl items in `non_local_defs` lint

This PR update the logic for the impl paths (items) in the `non_local_definitions` lint to also consider const-anon in case the impl definition is wrapped inside const-anon it-self wrapped into a const-anon where the items are.

r? `@jieyouxu` *(since you interacted on the issue)*
Fixes *(after beta-backport)* rust-lang#131643