syntax: ABI-oblivious grammar
This PR has the following effects:
1. `extern $lit` is now legal where `$lit:literal` and `$lit` is substituted for a string literal.
2. `extern "abi_that_does_not_exist"` is now *syntactically* legal whereas before, the set of ABI strings was hard-coded into the grammar of the language. With this PR, the set of ABIs are instead validated and translated during lowering. That seems more appropriate.
3. `ast::FloatTy` is now distinct from `rustc_target::abi::FloatTy`. The former is used substantially more and the translation between them is only necessary in a single place.
4. As a result of 2-3, libsyntax no longer depends on librustc_target, which should improve pipe-lining somewhat.
cc @rust-lang/lang -- the points 1-2 slightly change the definition of the language but in a way which seems consistent with our general principles (in particular wrt. the discussions of turning things into semantic errors). I expect this to be uncontroversial but it's worth letting y'all know. :)
r? @varkor
Use structured suggestions for missing associated items
When encountering an `impl` that is missing associated items required by its `trait`, use structured suggestions at an appropriate place in the `impl`.
We also sever syntax's dependency on rustc_target as a result.
This should slightly improve pipe-lining.
Moreover, some cleanup is done in related code.
Miri: Refactor to_scalar_ptr out of existence
`to_scalar_ptr` is somewhat subtle as it just throws away the 2nd component of a `ScalarPair` if there is one -- without any check if this is truly a pointer or so. And indeed we used it wrong on two occasions!
So I fixed those two, and then refactored things such that everyone calls `ref_to_mplace` instead (which they did anyway, I just moved up the calls), which is the only place that should interpret a `ScalarPair` as a wide ptr -- and it checks the type first. Thus we can remove `to_scalar_ptr` and `to_meta`.
r? @oli-obk
Fixed PhantomData markers in Arc and Rc
Include owned internal structs in `PhantomData` markers in `Arc` (`PhantomData<T>` => `PhantomData<ArcInner<T>>`) and `Rc` (`PhantomData<T>` => `PhantomData<RcBox<T>>`).
Improve std:🧵:Result documentation
Thanks to @dtolnay for pointing out the different premise of the contents of the `Err` variant in `std:🧵:Result` WRT normal error handling.
improve from_raw_parts docs
Triggered by https://github.com/rust-lang/rfcs/pull/2806. Hopefully this helps clarify that joining slices across allocations is not possible in Rust currently.
r? @Centril
Do not require extra LLVM backends for `x.py test` to pass
For long time our testing passed with a partially built LLVM
```
[llvm]
targets = "X86;ARM"
```
, a [recent PR](https://github.com/rust-lang/rust/pull/65809) changed that.
Improve uninit/zeroed lint
* Also warn when creating a raw pointer with a NULL vtable.
* Also identify `MaybeUninit::uninit().assume_init()` and `MaybeUninit::zeroed().assume_init()` as dangerous.
Optimize long-linker-command-line test
Replace O(n^3) text matching with inexpensive hash set lookups.
On my machine this reduces the total runtime of complete
run-make-fulldeps suite from roughly 75 seconds to 45 seconds.
gate rustc_on_unimplemented under rustc_attrs
Move `rustc_on_implemented` from the `on_implemented` gate to `rustc_attrs` as it is internal.
Closes#29628
r? @varkor
Cheaper doc comments
This PR implements the idea from #60935: represent doc comments more cheaply, rather than converting them into `#[doc="..."]` attribute form. Unlike #60936 (which is about coalescing doc comments to reduce their number), this approach does not have any backwards compatibility concerns, and it eliminates about 80-90% of the current cost of doc comments (as estimated using the numbers in #60930, which eliminated the cost of doc comments entirely by treating them as normal comments).
r? @petrochenkov
Fix promotion in a `const` when projections are present
Resolves#65727.
This marks the entire local as "needs promotion" when only a projection of that local appears in a promotable context. This should only affect promotion in a `const` or `static`, not in a `fn` or `const fn`, which is handled in `promote_consts.rs`.
r? @eddyb
`AttrKind` is a new type with two variants, `Normal` and `DocComment`. It's a
big performance win (over 10% in some cases) because `DocComment` lets doc
comments (which are common) be represented very cheaply.
`Attribute` gets some new helper methods to ease the transition:
- `has_name()`: check if the attribute name matches a single `Symbol`; for
`DocComment` variants it succeeds if the symbol is `sym::doc`.
- `is_doc_comment()`: check if it has a `DocComment` kind.
- `{get,unwrap}_normal_item()`: extract the item from a `Normal` variant;
panic otherwise.
Fixes#60935.
Use ident.span instead of def_span in dead-code pass
Hello! First time contributor! :)
This should fix#58729.
According to @estebank in the duplicate #63064, def_span scans forward on the line until it finds a {,
and if it can't find one, falls back to the span for the whole item. This
was apparently written before the identifier span was explicitly tracked on
each node.
This means that if an unused function signature spans multiple lines, the
entire function (potentially hundreds of lines) gets flagged as dead code.
This could, for example, cause IDEs to add error squiggly's to the whole
function.
By using the span from the ident instead, we narrow the scope of this in
most cases. In a wider sense, it's probably safe to use ident.span
instead of def_span in most locations throughout the whole code base,
but since this is my first contribution, I kept it small.
Some interesting points that came up while I was working on this:
- I reorganized the tests a bit to bring some of the dead code ones all
into the same location
- A few tests were for things unrelated to dead code (like the
path-lookahead for parens), so I added #![allow(dead_code)] and
cleaned up the stderr file to reduce noise in the future
- The same fix doesn't apply to const and static declarations. I tried
adding these cases to the match expression, but that created a much
wider change to tests and error messages, so I left it off until I
could get some code review to validate the approach.
Rollup of 9 pull requests
Successful merges:
- #65776 (Rename `LocalInternedString` and more)
- #65973 (caller_location: point to macro invocation sites, like file!/line!, and use in core::panic!.)
- #66015 (librustc_lexer: Refactor the module)
- #66062 (Configure LLVM module PIC level)
- #66086 (bump smallvec to 1.0)
- #66092 (Use KERN_ARND syscall for random numbers on NetBSD, same as FreeBSD.)
- #66103 (Add target thumbv7neon-unknown-linux-musleabihf)
- #66133 (Update the bundled `wasi-libc` repository)
- #66139 (use American spelling for `pluralize!`)
Failed merges:
r? @ghost
Update the bundled `wasi-libc` repository
This updates the libc that the `wasm32-wasi` target links against to the
latest revision, mostly just bringing in minor bug fixes and minor wasm
size improvements.
Add target thumbv7neon-unknown-linux-musleabihf
This is a copy of thumbv7neon-unknown-linux-gnueabihf with musl changes
merged from armv7-unknown-linux-musleabihf. This appears to have been
missed when adding the other ARMv7-A thumb targets.
Use KERN_ARND syscall for random numbers on NetBSD, same as FreeBSD.
This system call is present on all supported NetBSD versions and provides an endless stream of non-blocking random data from the kernel's ChaCha20-based CSPRNG. It doesn't require a file like `/dev/urandom` to be opened.
The system call is documented here (under kern.arandom):
https://netbsd.gw.com/cgi-bin/man-cgi?sysctl+7+NetBSD-7.0
And defined here:
https://nxr.netbsd.org/xref/src/sys/sys/sysctl.h#273
The semantics are the same as FreeBSD so reading 256 bytes per call is fine.
Similar change for getrandom crate: rust-random/getrandom#115
Configure LLVM module PIC level
As of LLVM 9, this is required for 32-bit PowerPC to properly generate
PLT references. Previously, only BigPIC was supported; now LLVM supports
both BigPIC and SmallPIC, and there is no default value provided.
librustc_lexer: Refactor the module
This PR introduces a refactoring of the `librustc_lexer` in order to improve readability.
All the changes performed are only cosmetic and do not introduce any changes the lexer logic or performance.
Newly introduced modules `literal`, `token` and `utils` are just copy-pasted from the `lib.rs` and do not contain even cosmetic changes (I decided to do so so it'll be easier to review changes looking only on diff).
r? @petrochenkov
cc @Centril @matklad
