]> git.lizzy.rs Git - rust.git/log
rust.git
19 months agoRollup merge of #104015 - alex:remove-kernel, r=oli-obk
Manish Goregaokar [Wed, 9 Nov 2022 20:39:05 +0000 (15:39 -0500)]
Rollup merge of #104015 - alex:remove-kernel, r=oli-obk

Remove linuxkernel targets

These are not used by the actual Rust-for-Linux project, so they're mostly just confusing.

19 months agoRollup merge of #103929 - BlackHoleFox:apple-targets-cleanup, r=petrochenkov
Manish Goregaokar [Wed, 9 Nov 2022 20:39:04 +0000 (15:39 -0500)]
Rollup merge of #103929 - BlackHoleFox:apple-targets-cleanup, r=petrochenkov

Cleanup Apple-related code in rustc_target

While working on https://github.com/rust-lang/rust/pull/103455, the consistency of the `rustc_target` code for Apple's platforms was "kind of bad." There were two "base" files (`apple_base.rs` and `apple_sdk_base.rs`) that the targets each pulled some parts out of, each and all of them were written slightly differently, and sometimes missed comments other implementations had.

So to hopefully make future maintenance, like implementing https://github.com/rust-lang/compiler-team/issues/556, easier, this makes all of them use similar patterns and the same target base logic everywhere instead of picking bits from both. This also has some other smaller upsides like less stringly-typed functions.

19 months agoRollup merge of #103464 - JakobDegen:mir-parsing, r=oli-obk
Manish Goregaokar [Wed, 9 Nov 2022 20:39:03 +0000 (15:39 -0500)]
Rollup merge of #103464 - JakobDegen:mir-parsing, r=oli-obk

Add support for custom mir

This implements rust-lang/compiler-team#564 . Details about the design, motivation, etc. can be found in there.

r? ```@oli-obk```

19 months agoRollup merge of #103307 - b4den:master, r=estebank
Manish Goregaokar [Wed, 9 Nov 2022 20:39:02 +0000 (15:39 -0500)]
Rollup merge of #103307 - b4den:master, r=estebank

Add context to compiler error message

Changed `creates a temporary which is freed while still in use` to `creates a temporary value which is freed while still in use`.

19 months agoRollup merge of #101005 - SLASHLogin:rustc_codegen_llvm_diagnostics, r=davidtwco
Manish Goregaokar [Wed, 9 Nov 2022 20:39:02 +0000 (15:39 -0500)]
Rollup merge of #101005 - SLASHLogin:rustc_codegen_llvm_diagnostics, r=davidtwco

Migrate rustc_codegen_llvm to SessionDiagnostics

WIP: Port current implementation of diagnostics to the new SessionDiagnostics.

Part of #100717

```@rustbot``` label +A-translation

19 months agoAuto merge of #104192 - Dylan-DPC:rollup-jjo1o80, r=Dylan-DPC
bors [Wed, 9 Nov 2022 16:54:02 +0000 (16:54 +0000)]
Auto merge of #104192 - Dylan-DPC:rollup-jjo1o80, r=Dylan-DPC

Rollup of 7 pull requests

Successful merges:

 - #103570 (Stabilize integer logarithms)
 - #103694 (Add documentation examples for `pointer::mask`)
 - #103919 (Unescaping cleanups)
 - #103933 (Promote {aarch64,i686,x86_64}-unknown-uefi to Tier 2)
 - #103952 (Don't intra linkcheck reference)
 - #104111 (rustdoc: Add mutable to the description)
 - #104125 (Const Compare for Tuples)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

19 months agoAdd missing `emitted_at` to the `Diagnostic`
SLASHLogin [Wed, 9 Nov 2022 14:14:58 +0000 (15:14 +0100)]
Add missing `emitted_at` to the `Diagnostic`

19 months agoAdd constructor for `Diagnostic` that takes `Vec<(DiagnosticMessage, Style)>`
SLASHLogin [Wed, 9 Nov 2022 12:47:46 +0000 (13:47 +0100)]
Add constructor for `Diagnostic` that takes `Vec<(DiagnosticMessage, Style)>`

19 months agoUpdate compiler/rustc_codegen_llvm/src/back/archive.rs
SLASHLogin [Tue, 8 Nov 2022 19:32:09 +0000 (20:32 +0100)]
Update compiler/rustc_codegen_llvm/src/back/archive.rs

Co-authored-by: David Wood <agile.lion3441@fuligin.ink>
19 months agoUse `LayoutError`'s implementation of `IntoDiagnostic`
SLASHLogin [Fri, 4 Nov 2022 19:08:01 +0000 (20:08 +0100)]
Use `LayoutError`'s implementation of `IntoDiagnostic`

19 months agoFormatting
SLASHLogin [Thu, 3 Nov 2022 12:41:54 +0000 (13:41 +0100)]
Formatting

19 months agoAdd `replace_args` method for `rustc_errors::diagnostic::Diagnostic`
SLASHLogin [Thu, 3 Nov 2022 12:09:25 +0000 (13:09 +0100)]
Add `replace_args` method for `rustc_errors::diagnostic::Diagnostic`

19 months agoDelay diagnostic translation in `rustc_codegen_ssa`
SLASHLogin [Wed, 2 Nov 2022 18:23:05 +0000 (19:23 +0100)]
Delay diagnostic translation in `rustc_codegen_ssa`

19 months agoSimplify existing Diagnostic implementations
SLASHLogin [Sun, 30 Oct 2022 18:26:12 +0000 (19:26 +0100)]
Simplify existing Diagnostic implementations

19 months agoPort diagnostics created by `Handler`
SLASHLogin [Sun, 30 Oct 2022 15:07:04 +0000 (16:07 +0100)]
Port diagnostics created by `Handler`

19 months agoFlatten diagnostic structs
SLASHLogin [Sun, 30 Oct 2022 13:33:27 +0000 (14:33 +0100)]
Flatten diagnostic structs

19 months agoCorrect tests to match errors
SLASHLogin [Tue, 30 Aug 2022 10:22:08 +0000 (12:22 +0200)]
Correct tests to match errors

19 months agoFix CI
SLASHLogin [Sat, 27 Aug 2022 08:05:50 +0000 (10:05 +0200)]
Fix CI
Add missing 'the' to the error en-US translation

19 months agoPort `MissingFeatures` and `TargetFeatureDisableOrEnable`
SLASHLogin [Fri, 26 Aug 2022 19:27:17 +0000 (21:27 +0200)]
Port `MissingFeatures` and `TargetFeatureDisableOrEnable`

19 months agoPort `UnknownArchiveKind`
SLASHLogin [Fri, 26 Aug 2022 17:42:29 +0000 (19:42 +0200)]
Port `UnknownArchiveKind`

19 months agoPort `DlltoolFailImportLibrary` and implement `IntoDiagnosticArg` for `Cow<'a, str>`
SLASHLogin [Fri, 26 Aug 2022 17:01:22 +0000 (19:01 +0200)]
Port `DlltoolFailImportLibrary` and implement `IntoDiagnosticArg` for `Cow<'a, str>`

19 months agoPort ErrorCallingDllTool
SLASHLogin [Fri, 26 Aug 2022 12:29:33 +0000 (14:29 +0200)]
Port ErrorCallingDllTool

19 months agoImport ErrorWritingDEFFile
SLASHLogin [Fri, 26 Aug 2022 12:17:15 +0000 (14:17 +0200)]
Import ErrorWritingDEFFile

19 months agoPort ArchiveBuildFailure
SLASHLogin [Fri, 26 Aug 2022 12:11:47 +0000 (14:11 +0200)]
Port ArchiveBuildFailure

19 months agoPort SanitizerMemtagRequiresMte
SLASHLogin [Fri, 26 Aug 2022 10:19:10 +0000 (12:19 +0200)]
Port SanitizerMemtagRequiresMte

19 months agoReuse SymbolAlreadyDefined
SLASHLogin [Fri, 26 Aug 2022 09:48:35 +0000 (11:48 +0200)]
Reuse SymbolAlreadyDefined

19 months agoPort LinkageConstOrMutType error
SLASHLogin [Fri, 26 Aug 2022 08:40:48 +0000 (10:40 +0200)]
Port LinkageConstOrMutType error

19 months agoPort InvalidMinimumAlignment
SLASHLogin [Fri, 26 Aug 2022 08:30:43 +0000 (10:30 +0200)]
Port InvalidMinimumAlignment

19 months agoPort layout size overflow
SLASHLogin [Fri, 26 Aug 2022 08:14:15 +0000 (10:14 +0200)]
Port layout size overflow

19 months agoFormatting
SLASHLogin [Thu, 25 Aug 2022 22:08:05 +0000 (00:08 +0200)]
Formatting

19 months agoPort branch protection on aarch64
SLASHLogin [Thu, 25 Aug 2022 22:03:53 +0000 (00:03 +0200)]
Port branch protection on aarch64

19 months agoFix diag() and formatting
SLASHLogin [Thu, 25 Aug 2022 21:47:11 +0000 (23:47 +0200)]
Fix diag() and formatting

19 months agoPort `symbol_already_defined` error
SLASHLogin [Thu, 25 Aug 2022 21:30:17 +0000 (23:30 +0200)]
Port `symbol_already_defined` error

19 months agoPort Instrument coverage requires llvm 12 to the new struct
SLASHLogin [Thu, 25 Aug 2022 21:08:18 +0000 (23:08 +0200)]
Port Instrument coverage requires llvm 12 to the new struct

19 months agoImport `error creating import library`
SLASHLogin [Thu, 25 Aug 2022 19:01:36 +0000 (21:01 +0200)]
Import `error creating import library`

19 months agoMissing lifetime parameter and formatting
SLASHLogin [Thu, 25 Aug 2022 17:23:55 +0000 (19:23 +0200)]
Missing lifetime parameter and formatting

19 months agoChange String in structs to &'a str
SLASHLogin [Thu, 25 Aug 2022 16:36:15 +0000 (18:36 +0200)]
Change String in structs to &'a str

19 months agoTrailing whitespaces
SLASHLogin [Thu, 25 Aug 2022 14:18:06 +0000 (16:18 +0200)]
Trailing whitespaces

19 months agolocales formatting
SLASHLogin [Thu, 25 Aug 2022 14:15:03 +0000 (16:15 +0200)]
locales formatting

19 months agoFormatting
SLASHLogin [Thu, 25 Aug 2022 13:42:20 +0000 (15:42 +0200)]
Formatting

19 months agoPort unknown feature diagnostic to the new framework
SLASHLogin [Thu, 25 Aug 2022 13:34:30 +0000 (15:34 +0200)]
Port unknown feature diagnostic to the new framework

19 months agoRollup merge of #104125 - ink-feather-org:const_cmp_tuples, r=fee1-dead
Dylan DPC [Wed, 9 Nov 2022 13:51:25 +0000 (19:21 +0530)]
Rollup merge of #104125 - ink-feather-org:const_cmp_tuples, r=fee1-dead

Const Compare for Tuples

Makes the impls for Tuples of ~const `PartialEq` types also `PartialEq`, impls for Tuples of ~const `PartialOrd` types also `PartialOrd`, for Tuples of ~const `Ord` types also `Ord`.

behind the `#![feature(const_cmp)]` gate.

~~Do not merge before #104113 is merged because I want to use this feature to clean up the new test that I added there.~~

r? ``@fee1-dead``

19 months agoRollup merge of #104111 - yancyribbens:add-mutable-to-the-description-for-as-simd...
Dylan DPC [Wed, 9 Nov 2022 13:51:24 +0000 (19:21 +0530)]
Rollup merge of #104111 - yancyribbens:add-mutable-to-the-description-for-as-simd-mut, r=scottmcm

rustdoc: Add mutable to the description

Add mutable the description to differentiate [as_simd](https://github.com/rust-lang/rust/blob/master/library/core/src/slice/mod.rs#L3654) from [as_simd_mut](https://github.com/rust-lang/rust/blob/master/library/core/src/slice/mod.rs#L3654).

19 months agoRollup merge of #103952 - ehuss:dont-intra-linkcheck-reference, r=Mark-Simulacrum
Dylan DPC [Wed, 9 Nov 2022 13:51:24 +0000 (19:21 +0530)]
Rollup merge of #103952 - ehuss:dont-intra-linkcheck-reference, r=Mark-Simulacrum

Don't intra linkcheck reference

This removes the reference from the intra-doc link checks. This causes problems if any of the reference content needs to change, it causes the linkchecker to break. The reference has its own broken link check (https://github.com/rust-lang/reference/tree/master/style-check) which uses pulldown-cmark on the source to find actual broken links (instead of false-positives like this regex does).

I think the intra-doc link check could potentially be removed completely, since I think rustdoc is now checking for them well enough. However, it may serve as a decent regression check.

19 months agoRollup merge of #103933 - nicholasbishop:bishop-uefi-tier-2, r=JohnTitor
Dylan DPC [Wed, 9 Nov 2022 13:51:23 +0000 (19:21 +0530)]
Rollup merge of #103933 - nicholasbishop:bishop-uefi-tier-2, r=JohnTitor

Promote {aarch64,i686,x86_64}-unknown-uefi to Tier 2

MCP: https://github.com/rust-lang/compiler-team/issues/555

CC `@dvdhrm`

19 months agoRollup merge of #103919 - nnethercote:unescaping-cleanups, r=matklad
Dylan DPC [Wed, 9 Nov 2022 13:51:22 +0000 (19:21 +0530)]
Rollup merge of #103919 - nnethercote:unescaping-cleanups, r=matklad

Unescaping cleanups

Some code improvements, and some error message improvements.

Best reviewed one commit at a time.

r? ````@matklad````

19 months agoRollup merge of #103694 - WaffleLapkin:mask_doc_example, r=scottmcm
Dylan DPC [Wed, 9 Nov 2022 13:51:22 +0000 (19:21 +0530)]
Rollup merge of #103694 - WaffleLapkin:mask_doc_example, r=scottmcm

Add documentation examples for `pointer::mask`

The examples are somewhat convoluted, but I don't know how to make this better :(

19 months agoRollup merge of #103570 - lukas-code:stabilize-ilog, r=scottmcm
Dylan DPC [Wed, 9 Nov 2022 13:51:21 +0000 (19:21 +0530)]
Rollup merge of #103570 - lukas-code:stabilize-ilog, r=scottmcm

Stabilize integer logarithms

Stabilizes feature `int_log`.

I've also made the functions const stable, because they don't depend on any unstable const features. `rustc_allow_const_fn_unstable` is just there for `Option::expect`, which could be replaced with a `match` and `panic!`. cc ``@rust-lang/wg-const-eval``

closes https://github.com/rust-lang/rust/issues/70887 (tracking issue)

~~blocked on FCP finishing: https://github.com/rust-lang/rust/issues/70887#issuecomment-1289028216~~
FCP finished: https://github.com/rust-lang/rust/issues/70887#issuecomment-1302121266

19 months agoAuto merge of #103723 - CastilloDel:master, r=jackh726
bors [Wed, 9 Nov 2022 13:45:27 +0000 (13:45 +0000)]
Auto merge of #103723 - CastilloDel:master, r=jackh726

Remove allow(rustc::potential_query_instability) in rustc_trait_selection

Related to https://github.com/rust-lang/rust/issues/84447

This PR needs to be benchmarked to check for regressions.

19 months agoCleanup fn trait ref test
onestacked [Wed, 9 Nov 2022 10:35:28 +0000 (11:35 +0100)]
Cleanup fn trait ref test

19 months agoAuto merge of #102565 - jyn514:refactor-build-manifest, r=Mark-Simulacrum
bors [Wed, 9 Nov 2022 10:24:27 +0000 (10:24 +0000)]
Auto merge of #102565 - jyn514:refactor-build-manifest, r=Mark-Simulacrum

Refactor build-manifest to minimize the number of changes needed to add a new component

- Add all components to `PkgType`
- Automate functionality wherever possible, so functions often don't have to be manually edited
- Where that's not possible, use exhaustive matches on `PkgType` instead of adding individual strings.
- Add documentation for how to add a component. Improve the existing documentation for how to test changes.

I tested locally that this generates an identical manifest before and after my change, as follows:
```sh
git checkout d44e14225ab00e164aa9ea9e8d9e1bee40f96b3e
cargo +nightly run --manifest-path src/tools/build-manifest/Cargo.toml build/dist build/manifest-before 1970-01-01 http://example.com nightly
git checkout refactor-build-manifest
cargo +nightly run --manifest-path src/tools/build-manifest/Cargo.toml build/dist build/manifest-before 1970-01-01 http://example.com nightly
sort -u build/manifest-before/channel-rust-nightly.toml | diff - <(sort -u build/manifest-after/channel-rust-nightly.toml)
```
I then verified by hand that the differences before sorting are inconsequential (mostly targets being slightly reordered).

The only change in behavior is that `llvm-tools` is now properly renamed to `llvm-tools-preview`:
```
; sort -u build/manifest-before/channel-rust-nightly.toml | diff - <(sort -u build/manifest-after/channel-rust-nightly.toml)
784a785
> [renames.llvm-tools]
894a896
> to = "llvm-tools-preview"
```

This is based on https://github.com/rust-lang/rust/pull/102241 and should not be merged before.

19 months agoconst Compare Tuples
onestacked [Mon, 7 Nov 2022 20:47:46 +0000 (21:47 +0100)]
const Compare Tuples

19 months agoAuto merge of #104180 - fee1-dead-contrib:fix-wf-fndef, r=oli-obk
bors [Wed, 9 Nov 2022 07:37:49 +0000 (07:37 +0000)]
Auto merge of #104180 - fee1-dead-contrib:fix-wf-fndef, r=oli-obk

Use `nominal_obligations_without_const` in wf for FnDef

Fixes #104155.

19 months agoAdd support for custom MIR parsing
Jakob Degen [Wed, 3 Aug 2022 11:30:13 +0000 (04:30 -0700)]
Add support for custom MIR parsing

19 months agoAuto merge of #104179 - Manishearth:rollup-yvsx5hh, r=Manishearth
bors [Wed, 9 Nov 2022 04:43:43 +0000 (04:43 +0000)]
Auto merge of #104179 - Manishearth:rollup-yvsx5hh, r=Manishearth

Rollup of 7 pull requests

Successful merges:

 - #100508 (avoid making substs of type aliases late bound when used as fn args)
 - #101381 (Test that target feature mix up with homogeneous floats is sound)
 - #103353 (Fix Access Violation when using lld & ThinLTO on windows-msvc)
 - #103521 (Avoid possible infinite  loop when next_point reaching the end of file)
 - #103559 (first move on a nested span_label)
 - #103778 (Update several crates for improved support of the new targets)
 - #103827 (Properly remap and check for substs compatibility in `confirm_impl_trait_in_trait_candidate`)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

19 months agoUse `nominal_obligations_without_const` in wf for FnDef
Deadbeef [Wed, 9 Nov 2022 02:08:22 +0000 (02:08 +0000)]
Use `nominal_obligations_without_const` in wf for FnDef

19 months agoRollup merge of #103827 - compiler-errors:rpitit-substs-compat, r=wesleywiser
Manish Goregaokar [Wed, 9 Nov 2022 02:03:54 +0000 (21:03 -0500)]
Rollup merge of #103827 - compiler-errors:rpitit-substs-compat, r=wesleywiser

Properly remap and check for substs compatibility in `confirm_impl_trait_in_trait_candidate`

Fixes #103824

19 months agoRollup merge of #103778 - mati865:update-deps, r=Mark-Simulacrum
Manish Goregaokar [Wed, 9 Nov 2022 02:03:54 +0000 (21:03 -0500)]
Rollup merge of #103778 - mati865:update-deps, r=Mark-Simulacrum

Update several crates for improved support of the new targets

This helps with `*-windows-gnullvm` targets by reducing amount of patching.

19 months agoRollup merge of #103559 - AndyJado:var_span_label, r=davidtwco
Manish Goregaokar [Wed, 9 Nov 2022 02:03:53 +0000 (21:03 -0500)]
Rollup merge of #103559 - AndyJado:var_span_label, r=davidtwco

first move on a nested span_label

trying not to be smart this time.

19 months agoRollup merge of #103521 - chenyukang:yukang/fix-103451-avoid-hang, r=jackh726,wesleywiser
Manish Goregaokar [Wed, 9 Nov 2022 02:03:53 +0000 (21:03 -0500)]
Rollup merge of #103521 - chenyukang:yukang/fix-103451-avoid-hang, r=jackh726,wesleywiser

Avoid possible infinite  loop when next_point reaching the end of file

Fixes #103451
If we return a span with `lo` = `hi`, `span_to_snippet` will always get `Ok("")`, which may introduce infinite loop if we don't care.

This PR make `find_width_of_character_at_span` return `width` with 1, so that `span_to_snippet` will get an `Err`.

19 months agoRollup merge of #103353 - wesleywiser:fix_lld_thinlto_msvc, r=michaelwoerister
Manish Goregaokar [Wed, 9 Nov 2022 02:03:52 +0000 (21:03 -0500)]
Rollup merge of #103353 - wesleywiser:fix_lld_thinlto_msvc, r=michaelwoerister

Fix Access Violation when using lld & ThinLTO on windows-msvc

Users report an AV at runtime of the compiled binary when using lld and ThinLTO on windows-msvc. The AV occurs when accessing a static value which is defined in one crate but used in another. Based on the disassembly of the cross-crate use, it appears that the use is not correctly linked with the definition and is instead assigned a garbage pointer value.

If we look at the symbol tables for each crates' obj file, we can see what is happening:

*lib.obj*:

```
COFF SYMBOL TABLE
...
00E 00000000 SECT2  notype       External     | _ZN10reproducer7memrchr2FN17h612b61ca0e168901E
...
```

*bin.obj*:

```
COFF SYMBOL TABLE
...
010 00000000 UNDEF  notype       External     | __imp__ZN10reproducer7memrchr2FN17h612b61ca0e168901E
...
```

The use of the symbol has the "import" style symbol name but the declaration doesn't generate any symbol with the same name. As a result, linking the files generates a warning from lld:

> rust-lld: warning: bin.obj: locally defined symbol imported: reproducer::memrchr::FN::h612b61ca0e168901 (defined in lib.obj) [LNK4217]

and the symbol reference remains undefined at runtime leading to the AV.

To fix this, we just need to detect that we are performing ThinLTO (and thus, static linking) and omit the `dllimport` attribute on the extern item in LLVM IR.

Fixes #81408

19 months agoRollup merge of #101381 - Urgau:target-mixup-homogenous-floats, r=Amanieu
Manish Goregaokar [Wed, 9 Nov 2022 02:03:52 +0000 (21:03 -0500)]
Rollup merge of #101381 - Urgau:target-mixup-homogenous-floats, r=Amanieu

Test that target feature mix up with homogeneous floats is sound

This pull-request adds a test in `src/test/abi/` that test that target feature mix up with homogeneous floats is sound.

This is basically is ripoff of [src/test/ui/simd/target-feature-mixup.rs](https://github.com/rust-lang/rust/blob/47d1cdb0bcac8e417071ce1929d261efe2399ae2/src/test/ui/simd/target-feature-mixup.rs) but for floats and without `#[repr(simd)]`.

*Extracted from https://github.com/rust-lang/rust/pull/97559 since I don't yet know what to do with that PR.*

19 months agoRollup merge of #100508 - BoxyUwU:make_less_things_late_bound, r=nikomatsakis
Manish Goregaokar [Wed, 9 Nov 2022 02:03:51 +0000 (21:03 -0500)]
Rollup merge of #100508 - BoxyUwU:make_less_things_late_bound, r=nikomatsakis

avoid making substs of type aliases late bound when used as fn args

fixes #47511
fixes #85533
(although I did not know theses issues existed when i was working on this :upside_down_face:)

currently `Alias<...>` is treated the same as `Struct<...>` when deciding if generics should be late bound or early bound but this is not correct as `Alias` might normalize to a projection which does not constrain the generics.

I think this needs more tests before merging
more explanation of PR [here](https://hackmd.io/v44a-QVjTIqqhK9uretyQg?view)

Hackmd inline for future readers:
---

This assumes reader is familiar with the concept of early/late bound lifetimes. There's a section on rustc-dev-guide if not (although i think some details are a bit out of date)

## problem & background

Not all lifetimes on a fn can be late bound:
```rust
fn foo<'a>() -> &'a ();
impl<'a> Fn<()> for FooFnDef {
    type Output = &'a (); // uh oh unconstrained lifetime
}
```
so we make make them early bound
```rust
fn foo<'a>() -> &'a ();
impl<'a> Fn<()> for FooFnDef<'a> {// wow look at all that lifetimey
     type Output = &'a ();
}
```
(Closures have the same constraint however it is not enforced leading to soundness bugs, [#84385](https://github.com/rust-lang/rust/pull/84385) implements this "downgrading late bound to early bound" for closures)

lifetimes on fn items are only late bound when they are "constrained" by the fn args:
```rust
fn foo<'a>(_: &'a ()) -> &'a ();
//               late bound, not present on `FooFnItem`
//               vv
impl<'a> Trait<(&'a (),)> for FooFnItem {
    type Output = &'a ();
}

// projections do not constrain inputs
fn bar<'a, T: Trait>(_: <T as Trait<'a>>::Assoc) -> &'a (); //  early bound
                                                            //  vv
impl<'a, T: Trait> Fn<(<T as Trait<'a>>::Assoc,)> for BarFnItem<'a, T> {
    type Output = &'a ();
}
```

current logic for determining if inputs "constrain" a lifetime works off of HIR so does not normalize aliases. It also assumes that any path with no self type constrains all its substs (i.e. `Foo<'a, u32>` has no self type but `T::Assoc` does). This falls apart for top level type aliases (see linked issues):

```rust
type Alias<'a, T> = <T as Trait<'a>>::Assoc;
//                      wow look its a path with no self type uwu
//                      i bet that constrains `'a` so it should be latebound
//                      vvvvvvvvvvv
fn foo<'a, T: Trait>(_: Alias<'a, T>) -> &'a ();
//                     `Alias` normalized to make things clearer
//                     vvvvvvvvvvvvvvvvvvvvvvv
impl<'a, T: Trait> Fn<(<T as Trait<'a>>::Assoc,)> for FooFnDef<T> {
    type Output = &'a ();
    // oh no `'a` isnt constrained wah wah waaaah *trumbone noises*
    // i think, idk what musical instrument that is
}
```

## solution

The PR solves this by having the hir visitor that checks for lifetimes in constraining uses check if the path is a `DefKind::Alias`. If it is we ""normalize"" it by calling `type_of` and walking the returned type. This is a bit hacky as it requires a mapping between the substs on the path in hir, and the generics of the `type Alias<...>` which is on the ty layer.

Alternative solutions may involve calculating the "late boundness" of lifetimes after/during astconv rather than relying on hir at all. We already have code to determine whether a lifetime SHOULD be late bound or not as this is currently how the error for `fn foo<'a, T: Trait>(_: Alias<'a, T>) -> &'a ();` gets emitted.

It is probably not possible to do this right now, late boundness is used by `generics_of` and `gather_explicit_predicates_of` as we currently do not put late bound lifetimes in `Generics`. Although this seems sus to me as the long term goal is to make all generics late bound which would result in `generics_of(function)` being empty? [#103448](https://github.com/rust-lang/rust/pull/103448) places all lifetimes in `Generics` regardless of late boundness so that may be a good step towards making this possible.

19 months agoAuto merge of #103171 - jackh726:gen-interior-hrtb-error, r=cjgillot
bors [Wed, 9 Nov 2022 02:02:28 +0000 (02:02 +0000)]
Auto merge of #103171 - jackh726:gen-interior-hrtb-error, r=cjgillot

Better error for HRTB error from generator interior

cc #100013

This is just a first pass at an error. It could be better, and shouldn't really be emitted in the first place. But this is better than what was being emitted before.

19 months agoAuto merge of #104168 - GuillaumeGomez:rollup-tf4edqc, r=GuillaumeGomez
bors [Tue, 8 Nov 2022 22:50:12 +0000 (22:50 +0000)]
Auto merge of #104168 - GuillaumeGomez:rollup-tf4edqc, r=GuillaumeGomez

Rollup of 12 pull requests

Successful merges:

 - #103928 (Add 'ty_error_with_guaranteed' and 'const_error_with_guaranteed')
 - #104027 (Place config.toml in current working directory if config not found)
 - #104093 (disable btree size tests on Miri)
 - #104097 (run alloc benchmarks in Miri and fix UB)
 - #104104 (Add split-debuginfo print option)
 - #104109 (rustdoc: Add mutable to the description)
 - #104113 (Fix `const_fn_trait_ref_impl`, add test for it)
 - #104114 (Fix invalid background-image file name)
 - #104132 (fix: lint against lint functions)
 - #104139 (Clarify licensing situation of MPSC and SPSC queue)
 - #104147 (Remove an address comparison from the parser)
 - #104165 (Add llvm-main to triagebot.toml)

Failed merges:

 - #104115 (Migrate crate-search element to CSS variables)

r? `@ghost`
`@rustbot` modify labels: rollup

19 months agotests
Boxy [Tue, 8 Nov 2022 22:15:40 +0000 (22:15 +0000)]
tests

19 months agocomment
Boxy [Tue, 8 Nov 2022 21:59:58 +0000 (21:59 +0000)]
comment

19 months agoRollup merge of #104165 - Manishearth:triagebot-llvm, r=Mark-Simulacrum
Guillaume Gomez [Tue, 8 Nov 2022 19:40:54 +0000 (20:40 +0100)]
Rollup merge of #104165 - Manishearth:triagebot-llvm, r=Mark-Simulacrum

Add llvm-main to triagebot.toml

19 months agoRollup merge of #104147 - WaffleLapkin:don't-compare-ptr-addresses-in-parser-, r...
Guillaume Gomez [Tue, 8 Nov 2022 19:40:53 +0000 (20:40 +0100)]
Rollup merge of #104147 - WaffleLapkin:don't-compare-ptr-addresses-in-parser-, r=compiler-errors

Remove an address comparison from the parser

Originally this check was added in #68985, as suggested by https://github.com/rust-lang/rust/pull/68985/commits/940f65782cc5df7fecad27b38cc25b6d1eeaf2e8#r376850175. I don't think that this address check is a robust way of making parser more robust.

This code is also extensively tested by [`ui/parser/issues/issue-35813-postfix-after-cast.rs`](https://github.com/rust-lang/rust/blob/57d3c58ed6e0faf89a62411f96c000ffc9fd3937/src/test/ui/parser/issues/issue-35813-postfix-after-cast.rs).

_Replaces #103700_

r? `@compiler-errors`

19 months agoRollup merge of #104139 - ferrocene:pa-channel-licensing, r=pnkfelix
Guillaume Gomez [Tue, 8 Nov 2022 19:40:53 +0000 (20:40 +0100)]
Rollup merge of #104139 - ferrocene:pa-channel-licensing, r=pnkfelix

Clarify licensing situation of MPSC and SPSC queue

Originally, these two files were licensed under the `BSD-2-Clause` license, as they were based off sample code on a blog licensing those snippets under that license:

* `library/std/src/sync/mpsc/mpsc_queue.rs`
* `library/std/src/sync/mpsc/spsc_queue.rs`

In 2017 though, the author of that blog agreed to relicense their code under the standard `MIT OR Apache-2.0` license in https://github.com/rust-lang/rust/pull/42149. This PR clarifies the situation in the files by expanding the comment at the top of the file.

r? ``@pnkfelix``

19 months agoRollup merge of #104132 - Rejyr:rustc_lint-function-lints, r=davidtwco
Guillaume Gomez [Tue, 8 Nov 2022 19:40:52 +0000 (20:40 +0100)]
Rollup merge of #104132 - Rejyr:rustc_lint-function-lints, r=davidtwco

fix: lint against lint functions

Add `#[rustc_lint_diagnostics]` to lint functions missing them. From [this discussion](https://github.com/rust-lang/rust/pull/101138#issuecomment-1306379999).

r? ```@davidtwco```

19 months agoRollup merge of #104114 - GuillaumeGomez:background-image-path, r=notriddle
Guillaume Gomez [Tue, 8 Nov 2022 19:40:52 +0000 (20:40 +0100)]
Rollup merge of #104114 - GuillaumeGomez:background-image-path, r=notriddle

Fix invalid background-image file name

This is a follow-up of https://github.com/rust-lang/rust/pull/101702.

Apparently the image hash was the wrong one. You can see the error in https://doc.rust-lang.org/nightly/core/primitive.u16.html?search=hello too.

I really need to check if I can adds check for resources load errors in `browser-ui-test`.

cc ``````@jsha``````
r? ``````@notriddle``````

19 months agoRollup merge of #104113 - ink-feather-org:fix_const_fn_ref_impls, r=compiler-errors
Guillaume Gomez [Tue, 8 Nov 2022 19:40:51 +0000 (20:40 +0100)]
Rollup merge of #104113 - ink-feather-org:fix_const_fn_ref_impls, r=compiler-errors

Fix `const_fn_trait_ref_impl`, add test for it

#99943 broke `#[feature(const_fn_trait_ref_impl)]`, this PR fixes this and adds a test for it.

r? ````@fee1-dead````

19 months agoRollup merge of #104109 - yancyribbens:add-mutable-to-comment-for-align-to-mut, r...
Guillaume Gomez [Tue, 8 Nov 2022 19:40:51 +0000 (20:40 +0100)]
Rollup merge of #104109 - yancyribbens:add-mutable-to-comment-for-align-to-mut, r=thomcc

rustdoc: Add mutable to the description

`mutable` is missing from the description.  Currently the description for [align_to](https://github.com/rust-lang/rust/blob/master/library/core/src/slice/mod.rs#L3498) is the same as [align_to_mut](https://github.com/rust-lang/rust/blob/master/library/core/src/slice/mod.rs#L3559)

19 months agoRollup merge of #104104 - kamirr:master, r=lcnr
Guillaume Gomez [Tue, 8 Nov 2022 19:40:50 +0000 (20:40 +0100)]
Rollup merge of #104104 - kamirr:master, r=lcnr

Add split-debuginfo print option

This option prints all supported values for `-Csplit-debuginfo=..`, i.e. only stable ones on stable/beta and all of them on nightly/dev.

Motivated by 1.65.0 regression causing builds with the following entry in `Cargo.toml` to fail on Windows:
```toml
[profile.dev]
split-debuginfo = "unpacked"
```

See https://github.com/rust-lang/cargo/pull/11347 for details.

This will lead to closing https://github.com/rust-lang/rust/issues/103976.

19 months agoRollup merge of #104097 - RalfJung:miri-alloc-benches, r=thomcc
Guillaume Gomez [Tue, 8 Nov 2022 19:40:50 +0000 (20:40 +0100)]
Rollup merge of #104097 - RalfJung:miri-alloc-benches, r=thomcc

run alloc benchmarks in Miri and fix UB

Miri since recently has a "fake monotonic clock" that works even with isolation. Its measurements are not very meaningful but it means we can run these benches and check them for UB.

And that's a good thing since there was UB here: fixes https://github.com/rust-lang/rust/issues/104096.

r? ``@thomcc``

19 months agoRollup merge of #104093 - RalfJung:test-sizes, r=thomcc
Guillaume Gomez [Tue, 8 Nov 2022 19:40:49 +0000 (20:40 +0100)]
Rollup merge of #104093 - RalfJung:test-sizes, r=thomcc

disable btree size tests on Miri

Seems fine not to run these in Miri, they can't have UB anyway. And this lets us do layout randomization in Miri.

r? ``@thomcc``

19 months agoRollup merge of #104027 - ted-tanner:issue-103697-fix, r=jyn514
Guillaume Gomez [Tue, 8 Nov 2022 19:40:49 +0000 (20:40 +0100)]
Rollup merge of #104027 - ted-tanner:issue-103697-fix, r=jyn514

Place config.toml in current working directory if config not found

Fixes an issue where bootsrapping a Rust build would place `config.toml` in `{src_root}` rather than the current working directory

#103697

19 months agoRollup merge of #103928 - chenyukang:yukang/fix-103874-add-ty_error_with_guaranteed...
Guillaume Gomez [Tue, 8 Nov 2022 19:40:48 +0000 (20:40 +0100)]
Rollup merge of #103928 - chenyukang:yukang/fix-103874-add-ty_error_with_guaranteed, r=lcnr

Add 'ty_error_with_guaranteed' and 'const_error_with_guaranteed'

Part of #103874

19 months agoAuto merge of #103252 - lcnr:recompute_applicable_impls, r=jackh726
bors [Tue, 8 Nov 2022 19:35:08 +0000 (19:35 +0000)]
Auto merge of #103252 - lcnr:recompute_applicable_impls, r=jackh726

selection failure: recompute applicable impls

The way we currently skip errors for ambiguous trait obligations seems pretty fragile so we get some duplicate errors because of this.

Removing this info from selection errors changes this system to be closer to my image of our new trait solver and is also making it far easier to change overflow errors to be non-fatal :sparkles:

r? types cc `@estebank`

19 months agoAdd llvm-main to triagebot.toml
Manish Goregaokar [Tue, 8 Nov 2022 19:29:58 +0000 (11:29 -0800)]
Add llvm-main to triagebot.toml

19 months agoReduce the scope of allow(rustc::potential_query_instability) in rustc_trait_selection
CastilloDel [Fri, 28 Oct 2022 21:32:41 +0000 (23:32 +0200)]
Reduce the scope of allow(rustc::potential_query_instability) in rustc_trait_selection

Make InferCtxtExt use a FxIndexMap

This should be faster, because the map is only being used to iterate,
which is supposed to be faster with the IndexMap

Make the user_computed_preds use an IndexMap

It is being used mostly for iteration, so the change shouldn't result in
a perf hit

Make the RegionDeps fields use an IndexMap

This change could be a perf hit. Both `larger` and `smaller` are used
for iteration, but they are also used for insertions.

Make types_without_default_bounds use an IndexMap

It uses extend, but it also iterates and removes items. Not sure if
this will be a perf hit.

Make InferTtxt.reported_trait_errors use an IndexMap

This change brought a lot of other changes. The map seems to have been
mostly used for iteration, so the performance shouldn't suffer.

Add FIXME to change ProvisionalEvaluationCache.map to use an IndexMap

Right now this results in a perf hit. IndexMap doesn't have
the `drain_filter` API, so in `on_completion` we now need to iterate two
times over the map.

19 months agoFix outdated comment
Waffle Maybe [Tue, 8 Nov 2022 17:49:58 +0000 (21:49 +0400)]
Fix outdated comment

Co-authored-by: clubby789 <jamie@hill-daniel.co.uk>
19 months agoAuto merge of #103965 - petrochenkov:effvisperf3, r=oli-obk
bors [Tue, 8 Nov 2022 14:37:40 +0000 (14:37 +0000)]
Auto merge of #103965 - petrochenkov:effvisperf3, r=oli-obk

resolve: More detailed effective visibility tracking for imports

Per-`DefId` tracking is not enough, due to glob imports in particular, which have a single `DefId` for the whole glob import item.
We need to track this stuff per every introduced name (`NameBinding`).

Also drop `extern` blocks from the effective visibility table, they are nominally private and it doesn't make sense to keep them there.

Later commits add some debug-only invariant checking and optimiaztions to mitigate regressions in https://github.com/rust-lang/rust/pull/103965#issuecomment-1304256445.

This is a bugfix and continuation of https://github.com/rust-lang/rust/pull/102026.

19 months agoignore wasm in test
lcnr [Tue, 8 Nov 2022 13:16:49 +0000 (14:16 +0100)]
ignore wasm in test

19 months agodelay errors as bug
lcnr [Mon, 31 Oct 2022 09:27:51 +0000 (10:27 +0100)]
delay errors as bug

19 months agoselection failure: recompute applicable impls
lcnr [Wed, 19 Oct 2022 15:17:19 +0000 (17:17 +0200)]
selection failure: recompute applicable impls

19 months agoRemove an address comparison from the parser
Maybe Waffle [Tue, 8 Nov 2022 11:51:10 +0000 (11:51 +0000)]
Remove an address comparison from the parser

19 months agoAuto merge of #104063 - compiler-errors:ct-norm-unless, r=jackh726
bors [Tue, 8 Nov 2022 10:02:11 +0000 (10:02 +0000)]
Auto merge of #104063 - compiler-errors:ct-norm-unless, r=jackh726

Don't normalize constants unless they need normalization

Maybe makes normalization a bit faster when we have many constants in a type

r? `@ghost`

19 months agoclarify licensing situation of mpsc and spsc queue
Pietro Albini [Tue, 8 Nov 2022 08:36:08 +0000 (09:36 +0100)]
clarify licensing situation of mpsc and spsc queue

19 months agoAuto merge of #104138 - Dylan-DPC:rollup-m3ojpjg, r=Dylan-DPC
bors [Tue, 8 Nov 2022 06:49:52 +0000 (06:49 +0000)]
Auto merge of #104138 - Dylan-DPC:rollup-m3ojpjg, r=Dylan-DPC

Rollup of 7 pull requests

Successful merges:

 - #103446 (Specialize `iter::ArrayChunks::fold` for TrustedRandomAccess iterators)
 - #103651 (Fix `rustc_parse_format` spans following escaped utf-8 multibyte chars)
 - #103865 (Move `fallback_has_occurred` state tracking to `FnCtxt`)
 - #103955 (Update linker-plugin-lto.md to contain up to Rust 1.65)
 - #103987 (Remove `in_tail_expr` from FnCtxt)
 - #104067 (fix debuginfo for windows_gnullvm_base.rs)
 - #104094 (fully move `on_unimplemented` to `error_reporting`)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup

19 months agoRollup merge of #104094 - lcnr:on_unimplemented-move, r=wesleywiser
Dylan DPC [Tue, 8 Nov 2022 05:53:53 +0000 (11:23 +0530)]
Rollup merge of #104094 - lcnr:on_unimplemented-move, r=wesleywiser

fully move `on_unimplemented` to `error_reporting`

the `traits` module has a few too many submodules in my opinion.

19 months agoRollup merge of #104067 - jeremyd2019:patch-1, r=davidtwco
Dylan DPC [Tue, 8 Nov 2022 05:53:53 +0000 (11:23 +0530)]
Rollup merge of #104067 - jeremyd2019:patch-1, r=davidtwco

fix debuginfo for windows_gnullvm_base.rs

These lines (including the FIXME comment) were added to windows_gnu_base.rs in cf2c492ef8c87c049b4e3a62f43c841aafc88cba but windows_gnullvm_base.rs was not updated.  This resulted in an error `LLVM ERROR: dwo only supported with ELF and Wasm` attempting to build on aarch64-pc-windows-gnullvm.

See also https://github.com/msys2/MINGW-packages/pull/13921#issuecomment-1304391707

/cc ```@mati865``` ```@davidtwco```

r? ```@davidtwco```

19 months agoRollup merge of #103987 - compiler-errors:no-in_tail_expr, r=eholk
Dylan DPC [Tue, 8 Nov 2022 05:53:52 +0000 (11:23 +0530)]
Rollup merge of #103987 - compiler-errors:no-in_tail_expr, r=eholk

Remove `in_tail_expr` from FnCtxt

Cleans up yet another unneeded member from `FnCtxt`. The `in_tail_expr` condition wasn't even correct -- it was set for true while typechecking the whole fn body.

19 months agoRollup merge of #103955 - str4d:update-lto-doc-1.65, r=ehuss
Dylan DPC [Tue, 8 Nov 2022 05:53:52 +0000 (11:23 +0530)]
Rollup merge of #103955 - str4d:update-lto-doc-1.65, r=ehuss

Update linker-plugin-lto.md to contain up to Rust 1.65

The table rows were obtained via the script embedded in the page.

19 months agoRollup merge of #103865 - compiler-errors:fallback-has-occurred-tracking, r=eholk
Dylan DPC [Tue, 8 Nov 2022 05:53:51 +0000 (11:23 +0530)]
Rollup merge of #103865 - compiler-errors:fallback-has-occurred-tracking, r=eholk

Move `fallback_has_occurred` state tracking to `FnCtxt`

Removes a ton of callsites that defaulted to `false`

19 months agoRollup merge of #103651 - Alexendoo:parse-format-unicode-escapes, r=wesleywiser
Dylan DPC [Tue, 8 Nov 2022 05:53:51 +0000 (11:23 +0530)]
Rollup merge of #103651 - Alexendoo:parse-format-unicode-escapes, r=wesleywiser

Fix `rustc_parse_format` spans following escaped utf-8 multibyte chars

Currently too many skips are created for char escapes that are larger than 1 byte when encoded in UTF-8, [playground:](https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=c77a9dc669b69b167271b59ed2c8d88c)

```rust
fn main() {
    format!("\u{df}{a}");
    format!("\u{211d}{a}");
    format!("\u{1f4a3}{a}");
}
```
```
error[[E0425]](https://doc.rust-lang.org/stable/error-index.html#E0425): cannot find value `a` in this scope
 --> src/main.rs:2:22
  |
2 |     format!("\u{df}{a}");
  |                      ^ not found in this scope

error[[E0425]](https://doc.rust-lang.org/stable/error-index.html#E0425): cannot find value `a` in this scope
 --> src/main.rs:3:25
  |
3 |     format!("\u{211d}{a}");
  |                         ^ not found in this scope

error[[E0425]](https://doc.rust-lang.org/stable/error-index.html#E0425): cannot find value `a` in this scope
 --> src/main.rs:4:27
  |
4 |     format!("\u{1f4a3}{a}");
  |                           ^ not found in this scope
```

This reduces the number of skips to account for that

Fixes https://github.com/rust-lang/rust-clippy/issues/9727

19 months agoRollup merge of #103446 - the8472:tra-array-chunks, r=Mark-Simulacrum
Dylan DPC [Tue, 8 Nov 2022 05:53:50 +0000 (11:23 +0530)]
Rollup merge of #103446 - the8472:tra-array-chunks, r=Mark-Simulacrum

Specialize `iter::ArrayChunks::fold` for TrustedRandomAccess iterators

```
OLD:
test iter::bench_trusted_random_access_chunks                      ... bench:         368 ns/iter (+/- 4)
NEW:
test iter::bench_trusted_random_access_chunks                      ... bench:          30 ns/iter (+/- 0)
```

The resulting assembly is similar to #103166 but the specialization kicks in under different (partially overlapping) conditions compared to that PR. They're complementary.

In principle a TRA-based specialization could be applied to all `ArrayChunks` methods, including `next()` as we do for `Zip` but that would have all the same hazards as the Zip specialization. Only doing it for `fold` is far less hazardous. The downside is that it only helps with internal, exhaustive iteration. I.e. `for _ in` or `try_fold` will not benefit.

Note that the regular, `try_fold`-based and the specialized `fold()` impl have observably slightly different behavior. Namely the specialized variant does not fetch the remainder elements from the underlying iterator. We do have a few other places in the standard library where beyond-the-end-of-iteration side-effects are being elided under some circumstances but not others.

Inspired by https://old.reddit.com/r/rust/comments/yaft60/zerocost_iterator_abstractionsnot_so_zerocost/

19 months agoSimplify `unescape_{char,byte}`.
Nicholas Nethercote [Tue, 8 Nov 2022 04:59:19 +0000 (15:59 +1100)]
Simplify `unescape_{char,byte}`.

The `usize` isn't needed in the error case.

19 months agocode cleanup with err.emit_unless
yukang [Fri, 4 Nov 2022 08:10:32 +0000 (16:10 +0800)]
code cleanup with err.emit_unless