]> git.lizzy.rs Git - rust.git/log
rust.git
3 years agoReport an error if resolution of closure call functions failed
Fabian Wolff [Sat, 12 Jun 2021 16:32:25 +0000 (18:32 +0200)]
Report an error if resolution of closure call functions failed

3 years agoAuto merge of #86215 - FabianWolff:unnameable-types, r=jackh726
bors [Sat, 12 Jun 2021 11:12:16 +0000 (11:12 +0000)]
Auto merge of #86215 - FabianWolff:unnameable-types, r=jackh726

Do not suggest to add type annotations for unnameable types

Consider this example:
```rust
const A = || 42;

struct S<T> { t: T }
const B: _ = S { t: || 42 };
```
This currently produces the following output:
```
error: missing type for `const` item
 --> src/lib.rs:1:7
  |
1 | const A = || 42;
  |       ^ help: provide a type for the item: `A: [closure@src/lib.rs:1:11: 1:16]`

error[E0121]: the type placeholder `_` is not allowed within types on item signatures
 --> src/lib.rs:4:10
  |
4 | const B: _ = S { t: || 42 };
  |          ^
  |          |
  |          not allowed in type signatures
  |          help: replace `_` with the correct type: `S<[closure@src/lib.rs:4:21: 4:26]>`

error: aborting due to 2 previous errors
```
However, these suggestions are obviously useless, because the suggested types cannot be written down. With my changes, the suggestion is replaced with a note, because there is no simple fix:
```
error: missing type for `const` item
 --> test.rs:1:7
  |
1 | const A = || 42;
  |       ^
  |
note: however, the inferred type `[closure@test.rs:1:11: 1:16]` cannot be named
 --> test.rs:1:11
  |
1 | const A = || 42;
  |           ^^^^^

error[E0121]: the type placeholder `_` is not allowed within types on item signatures
 --> test.rs:4:10
  |
4 | const B: _ = S { t: || 42 };
  |          ^ not allowed in type signatures
  |
note: however, the inferred type `S<[closure@test.rs:4:21: 4:26]>` cannot be named
 --> test.rs:4:14
  |
4 | const B: _ = S { t: || 42 };
  |              ^^^^^^^^^^^^^^

error: aborting due to 2 previous errors
```

3 years agoAuto merge of #86130 - BoxyUwU:abstract_const_as_cast, r=oli-obk
bors [Sat, 12 Jun 2021 08:50:22 +0000 (08:50 +0000)]
Auto merge of #86130 - BoxyUwU:abstract_const_as_cast, r=oli-obk

const_eval_checked: Support as casts in abstract consts

3 years agoAuto merge of #86180 - cjgillot:defmv, r=petrochenkov
bors [Sat, 12 Jun 2021 06:09:20 +0000 (06:09 +0000)]
Auto merge of #86180 - cjgillot:defmv, r=petrochenkov

Hash DefId in rustc_span.

This is mostly just moving code around. Changes are simplifications of unneeded callbacks from rustc_span to rustc_middle.

r? `@petrochenkov`

3 years agoAuto merge of #86234 - Mark-Simulacrum:new-vers, r=Mark-Simulacrum
bors [Sat, 12 Jun 2021 03:46:33 +0000 (03:46 +0000)]
Auto merge of #86234 - Mark-Simulacrum:new-vers, r=Mark-Simulacrum

Bump to 1.55

r? `@Mark-Simulacrum`

3 years agoAuto merge of #86226 - JohnTitor:rollup-5ubdolf, r=JohnTitor
bors [Sat, 12 Jun 2021 01:21:56 +0000 (01:21 +0000)]
Auto merge of #86226 - JohnTitor:rollup-5ubdolf, r=JohnTitor

Rollup of 7 pull requests

Successful merges:

 - #85800 (Fix some diagnostic issues with const_generics_defaults feature gate)
 - #85823 (Do not suggest ampmut if rhs is already mutable)
 - #86153 (Print dummy spans as `no-location`)
 - #86174 (Detect incorrect vtable alignment during const eval)
 - #86189 (Make `relate_type_and_mut` public)
 - #86205 (Run full const-generics test for issue-72293)
 - #86217 (Remove "generic type" in boxed.rs)

Failed merges:

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

3 years agoBump to 1.54
Mark Rousskov [Fri, 11 Jun 2021 23:19:55 +0000 (19:19 -0400)]
Bump to 1.54

3 years agoSuggest a FnPtr type if a FnDef type is found
Fabian Wolff [Fri, 11 Jun 2021 18:12:40 +0000 (20:12 +0200)]
Suggest a FnPtr type if a FnDef type is found

3 years agoRollup merge of #86217 - fee1-dead:adjust-box-doc, r=m-ou-se
Yuki Okushi [Fri, 11 Jun 2021 16:16:04 +0000 (01:16 +0900)]
Rollup merge of #86217 - fee1-dead:adjust-box-doc, r=m-ou-se

Remove "generic type" in boxed.rs

r? ``@m-ou-se``

3 years agoRemove "generic type" in boxed.rs
Deadbeef [Fri, 11 Jun 2021 20:11:48 +0000 (04:11 +0800)]
Remove "generic type" in boxed.rs

3 years agoRollup merge of #86205 - JohnTitor:full-test-for-72293, r=oli-obk
Yuki Okushi [Fri, 11 Jun 2021 16:16:02 +0000 (01:16 +0900)]
Rollup merge of #86205 - JohnTitor:full-test-for-72293, r=oli-obk

Run full const-generics test for issue-72293

Closes #72293
r? ```@oli-obk```

3 years agoRollup merge of #86189 - JohnTitor:relate-fn-pub, r=Aaron1011
Yuki Okushi [Fri, 11 Jun 2021 16:16:01 +0000 (01:16 +0900)]
Rollup merge of #86189 - JohnTitor:relate-fn-pub, r=Aaron1011

Make `relate_type_and_mut` public

#85343 improved diagnostics around `Relate` impls but made `relate_type_and_mut` private, which was accessible as `relate` previously. This makes it public so that we can use it on rust-semverver.

r? ```@Aaron1011```

3 years agoRollup merge of #86174 - lqd:const-ub-align, r=RalfJung
Yuki Okushi [Fri, 11 Jun 2021 16:16:00 +0000 (01:16 +0900)]
Rollup merge of #86174 - lqd:const-ub-align, r=RalfJung

Detect incorrect vtable alignment during const eval

This PR fixes #86132 by detecting invalid alignment values for trait objects in the interpreter, and emitting an error about this conversion failure, to avoid the ICE.

I've noticed that the error emitted at https://github.com/rust-lang/rust/blob/a50d72158e08e02cfc051b863017bdbd2c45b637/compiler/rustc_mir/src/interpret/traits.rs#L163-L166 doesn't seem to be present in the const-ub tests, so I've tried adding a test that triggers both of these cases: one for the invalid size, and another for the invalid alignment that #86132 tracks (I have found different magic values triggering different `Align::from_bytes` errors than the "power of 2" one, if need be).

However, when doing that, I *cannot* for the life of me figure out the correct incantation to make these 2 errors trigger with the "it is undefined behavior to use this value" message rather than the "any use of this value will cause an error" lint.

I've tried Oli's suggestions of different values, tuples and arrays, using the transparent wrapper trick from the other tests and I was only able to trigger the regular const-ub errors about the size of the vtable, or that the drop pointer was invalid. Maybe these "type validation failed" errors happen before this part of the interpreter is reached and there just needs some magic incorrect values to bypass them, I don't know.

Since this fixes an ICE, and if the constants are indeed used, these 2 tests will turn into a hard error, I thought I'd open the PR anyways. And if ```@RalfJung``` you know of a way I could manage that (if you think that these tests are worth checking that the `throw_ub_format!` does indeed create const-ub errors as we expect) I'd be grateful.

For that reason, r? ```@RalfJung``` and cc ```@oli-obk.```

3 years agoRollup merge of #86153 - tmiasko:dummy-span, r=estebank
Yuki Okushi [Fri, 11 Jun 2021 16:15:59 +0000 (01:15 +0900)]
Rollup merge of #86153 - tmiasko:dummy-span, r=estebank

Print dummy spans as `no-location`

Fixes #58808.

3 years agoRollup merge of #85823 - fee1-dead:borrowck-0, r=jackh726
Yuki Okushi [Fri, 11 Jun 2021 16:15:57 +0000 (01:15 +0900)]
Rollup merge of #85823 - fee1-dead:borrowck-0, r=jackh726

Do not suggest ampmut if rhs is already mutable

Removes invalid suggestion in #85765, although it should highlight the user type instead of the local variable.

Looking at the comments of this line:
https://github.com/rust-lang/rust/blob/84b1005bfd22e2cb2a4c13b0b81958fe72628354/compiler/rustc_mir_build/src/build/matches/mod.rs#L2107

It was intentionally set to `None`, causing it to highlight the local variable instead. I am not sure if I will be able to fix it.

Fixes #85765

3 years agoRollup merge of #85800 - BoxyUwU:const-param-default-diagnostics, r=oli-obk
Yuki Okushi [Fri, 11 Jun 2021 16:15:56 +0000 (01:15 +0900)]
Rollup merge of #85800 - BoxyUwU:const-param-default-diagnostics, r=oli-obk

Fix some diagnostic issues with const_generics_defaults feature gate

This PR makes a few changes:
- print out const param defaults in "lifetime ordering" errors rather than discarding them
- update `is_simple_text` to account for const params when checking if a type has no generics, this was causing a note to be failed to add to an error message
- fixes some diagnostic wording that incorrectly said there was ordering restrictions between type/const params despite the `const_generics_defaults` feature gate is active

3 years agoAuto merge of #85885 - bjorn3:remove_box_region, r=cjgillot
bors [Fri, 11 Jun 2021 16:11:20 +0000 (16:11 +0000)]
Auto merge of #85885 - bjorn3:remove_box_region, r=cjgillot

Don't use a generator for BoxedResolver

The generator is non-trivial and requires unsafe code anyway. Using regular unsafe code without a generator is much easier to follow.

Based on #85810 as it touches rustc_interface too.

3 years agoMake DummyHashStableContext dummier.
Camille GILLOT [Fri, 11 Jun 2021 14:53:27 +0000 (16:53 +0200)]
Make DummyHashStableContext dummier.

3 years agoSprinkle inline.
Camille GILLOT [Fri, 11 Jun 2021 14:48:24 +0000 (16:48 +0200)]
Sprinkle inline.

3 years agoDo not suggest to add type annotations for unnameable types
Fabian Wolff [Fri, 11 Jun 2021 11:30:12 +0000 (13:30 +0200)]
Do not suggest to add type annotations for unnameable types

3 years agoAuto merge of #86116 - FabianWolff:issue-86100, r=varkor
bors [Fri, 11 Jun 2021 10:25:53 +0000 (10:25 +0000)]
Auto merge of #86116 - FabianWolff:issue-86100, r=varkor

Suggest a trailing comma if a 1-tuple is expected and a parenthesized expression is found

This pull request fixes #86100. The following code:
```rust
fn main() {
    let t: (i32,) = (1);
}
```
currently produces:
```
warning: unnecessary parentheses around assigned value
 --> test.rs:2:21
  |
2 |     let t: (i32,) = (1);
  |                     ^^^ help: remove these parentheses
  |
  = note: `#[warn(unused_parens)]` on by default

error[E0308]: mismatched types
 --> test.rs:2:21
  |
2 |     let t: (i32,) = (1);
  |            ------   ^^^ expected tuple, found integer
  |            |
  |            expected due to this
  |
  = note: expected tuple `(i32,)`
              found type `{integer}`

error: aborting due to previous error; 1 warning emitted
```
With my changes, I get the same warning and the following error:
```
error[E0308]: mismatched types
 --> test.rs:2:21
  |
2 |     let t: (i32,) = (1);
  |            ------   ^^^ expected tuple, found integer
  |            |
  |            expected due to this
  |
  = note: expected tuple `(i32,)`
              found type `{integer}`
help: use a trailing comma to create a tuple with one element
  |
2 |     let t: (i32,) = (1,);
  |                     ^^^^
```
i.e. I have added a suggestion to add a trailing comma to create a 1-tuple. This suggestion is only issued if a 1-tuple is expected and the expression (`(1)` in the example above) is surrounded by parentheses and does not already have a tuple type. In this situation, I'd say that it is pretty likely that the user meant to create a tuple.

3 years agoHash DefId in rustc_span.
Camille GILLOT [Mon, 19 Apr 2021 20:27:49 +0000 (22:27 +0200)]
Hash DefId in rustc_span.

3 years agoAuto merge of #85994 - tmiasko:monomorphic-needs-drop, r=RalfJung
bors [Fri, 11 Jun 2021 07:44:58 +0000 (07:44 +0000)]
Auto merge of #85994 - tmiasko:monomorphic-needs-drop, r=RalfJung

Disallow non-monomorphic calls to `needs_drop` in interpreter

otherwise evaluation could change after further substitutions.

3 years agoAuto merge of #86204 - alexcrichton:wasm-simd-stable, r=Amanieu
bors [Fri, 11 Jun 2021 05:02:41 +0000 (05:02 +0000)]
Auto merge of #86204 - alexcrichton:wasm-simd-stable, r=Amanieu

std: Stabilize wasm simd intrinsics

This commit performs two changes to stabilize Rust support for
WebAssembly simd intrinsics:

* The stdarch submodule is updated to pull in rust-lang/stdarch#1179.
* The `wasm_target_feature` feature gate requirement for the `simd128`
  feature has been removed, stabilizing the name `simd128`.

This should conclude the FCP started on #74372 and...

Closes #74372

3 years agostd: Stabilize wasm simd intrinsics
Alex Crichton [Thu, 10 Jun 2021 14:11:23 +0000 (07:11 -0700)]
std: Stabilize wasm simd intrinsics

This commit performs two changes to stabilize Rust support for
WebAssembly simd intrinsics:

* The stdarch submodule is updated to pull in rust-lang/stdarch#1179.
* The `wasm_target_feature` feature gate requirement for the `simd128`
  feature has been removed, stabilizing the name `simd128`.

This should conclude the FCP started on #74372 and...

Closes #74372

3 years agoAuto merge of #85961 - 1000teslas:issue-71519-fix, r=petrochenkov
bors [Fri, 11 Jun 2021 02:21:52 +0000 (02:21 +0000)]
Auto merge of #85961 - 1000teslas:issue-71519-fix, r=petrochenkov

MVP for using rust-lld as part of cc

Will fix #71519. I need to figure out how to write a test showing that lld is used instead of whatever linker cc normally uses. When I manually run rustc using `echo 'fn main() {}' | RUSTC_LOG=rustc_codegen_ssa::back::link=debug ./rustc -Clinker-flavor=gcc-lld --crate-type bin -Clink-arg=-Wl,-v` (thanks to bjorn3 on Zulip), I can see that lld is used, but I'm not sure how to inspect that output in a test.

3 years agoAuto merge of #85630 - gilescope:to_digit_speedup3, r=nagisa
bors [Thu, 10 Jun 2021 23:14:11 +0000 (23:14 +0000)]
Auto merge of #85630 - gilescope:to_digit_speedup3, r=nagisa

to_digit simplification (less jumps)

I just realised we might be able to make use of the fact that changing case in ascii is easy to help simplify to_digit some more.

It looks a bit cleaner and it looks like it's less jumps and there's less instructions in the generated assembly:

https://godbolt.org/z/84Erh5dhz

The benchmarks don't really tell me much. Maybe a slight improvement on the var radix.

Before:
```
test char::methods::bench_to_digit_radix_10                     ... bench:      53,819 ns/iter (+/- 8,314)
test char::methods::bench_to_digit_radix_16                     ... bench:      57,265 ns/iter (+/- 10,730)
test char::methods::bench_to_digit_radix_2                      ... bench:      55,077 ns/iter (+/- 5,431)
test char::methods::bench_to_digit_radix_36                     ... bench:      56,549 ns/iter (+/- 3,248)
test char::methods::bench_to_digit_radix_var                    ... bench:      43,848 ns/iter (+/- 3,189)

test char::methods::bench_to_digit_radix_10                     ... bench:      51,707 ns/iter (+/- 10,946)
test char::methods::bench_to_digit_radix_16                     ... bench:      52,835 ns/iter (+/- 2,689)
test char::methods::bench_to_digit_radix_2                      ... bench:      51,012 ns/iter (+/- 2,746)
test char::methods::bench_to_digit_radix_36                     ... bench:      53,210 ns/iter (+/- 8,645)
test char::methods::bench_to_digit_radix_var                    ... bench:      40,386 ns/iter (+/- 4,711)

test char::methods::bench_to_digit_radix_10                     ... bench:      54,088 ns/iter (+/- 5,677)
test char::methods::bench_to_digit_radix_16                     ... bench:      55,972 ns/iter (+/- 17,229)
test char::methods::bench_to_digit_radix_2                      ... bench:      52,083 ns/iter (+/- 2,425)
test char::methods::bench_to_digit_radix_36                     ... bench:      54,132 ns/iter (+/- 1,548)
test char::methods::bench_to_digit_radix_var                    ... bench:      41,250 ns/iter (+/- 5,299)
```
After:
```
test char::methods::bench_to_digit_radix_10                     ... bench:      48,907 ns/iter (+/- 19,449)
test char::methods::bench_to_digit_radix_16                     ... bench:      52,673 ns/iter (+/- 8,122)
test char::methods::bench_to_digit_radix_2                      ... bench:      48,509 ns/iter (+/- 2,885)
test char::methods::bench_to_digit_radix_36                     ... bench:      50,526 ns/iter (+/- 4,610)
test char::methods::bench_to_digit_radix_var                    ... bench:      38,618 ns/iter (+/- 3,180)

test char::methods::bench_to_digit_radix_10                     ... bench:      54,202 ns/iter (+/- 6,994)
test char::methods::bench_to_digit_radix_16                     ... bench:      56,585 ns/iter (+/- 8,448)
test char::methods::bench_to_digit_radix_2                      ... bench:      50,548 ns/iter (+/- 1,674)
test char::methods::bench_to_digit_radix_36                     ... bench:      52,749 ns/iter (+/- 2,576)
test char::methods::bench_to_digit_radix_var                    ... bench:      40,215 ns/iter (+/- 3,327)

test char::methods::bench_to_digit_radix_10                     ... bench:      50,233 ns/iter (+/- 22,272)
test char::methods::bench_to_digit_radix_16                     ... bench:      50,841 ns/iter (+/- 19,981)
test char::methods::bench_to_digit_radix_2                      ... bench:      50,386 ns/iter (+/- 4,555)
test char::methods::bench_to_digit_radix_36                     ... bench:      52,369 ns/iter (+/- 2,737)
test char::methods::bench_to_digit_radix_var                    ... bench:      40,417 ns/iter (+/- 2,766)
```

I removed the likely as it resulted in a few less instructions. (It's not been in there long - I added it in the last to_digit iteration).

3 years agoRun full const-generics test for issue-72293
Yuki Okushi [Thu, 10 Jun 2021 21:20:54 +0000 (06:20 +0900)]
Run full const-generics test for issue-72293

3 years agoAuto merge of #80080 - rylev:qpath-on-struct, r=petrochenkov
bors [Thu, 10 Jun 2021 20:32:43 +0000 (20:32 +0000)]
Auto merge of #80080 - rylev:qpath-on-struct, r=petrochenkov

Allow qualified paths in struct construction (both expressions and patterns)

Fixes #79658

3 years agoFurther simplification of to_digit
Giles Cope [Thu, 10 Jun 2021 19:16:35 +0000 (20:16 +0100)]
Further simplification of to_digit

3 years agoAuto merge of #86098 - pietroalbini:test-stable, r=Mark-Simulacrum
bors [Thu, 10 Jun 2021 17:51:48 +0000 (17:51 +0000)]
Auto merge of #86098 - pietroalbini:test-stable, r=Mark-Simulacrum

Add the x86_64-gnu-stable builder

During the 1.52 release process we had to deal with some commits that passed the test suite on the nightly branch but failed on the beta or stable branch. In that case it was due to some UI tests including the channel name in the output, but other changes might also be dependent on the channel.

This commit adds a new CI job that runs the Linux x86_64 test suite with the stable branch, ensuring nightly changes also work as stable. To ensure the new job works the following other changes are present:

* The `ui-fulldeps/session-derive-errors.rs` test has been disabled on beta and stable, which required adding support for `// ignore-{channel}` and `// only-{channel}`.
* The `rustdoc/intra-doc/field.rs` has been fixed.

r? `@Mark-Simulacrum`
fixes https://github.com/rust-lang/release-team/issues/11

3 years agoRemove unnecessary transmute
bjorn3 [Thu, 10 Jun 2021 17:36:27 +0000 (19:36 +0200)]
Remove unnecessary transmute

3 years agoAuto merge of #86020 - nagisa:nagisa/outliner, r=pnkfelix
bors [Thu, 10 Jun 2021 15:11:01 +0000 (15:11 +0000)]
Auto merge of #86020 - nagisa:nagisa/outliner, r=pnkfelix

Disable the machine outliner by default

This addresses a codegen-issue that needs to be fixed upstream in LLVM.
While we wait for the fix, we can disable it.

Verified manually that the outliner is no longer run when
`-Copt-level=z` is specified, and also that you can override this with
`-Cllvm-args=-enable-machine-outliner` if you need it anyway.

A regression test is not really feasible in this instance, given that we
do not have any minimal reproducers.

Fixes #85351

cc `@pnkfelix`

3 years agosupport `as _` and add tests
Ellen [Thu, 10 Jun 2021 13:53:38 +0000 (14:53 +0100)]
support `as _` and add tests

3 years agoAuto merge of #82639 - jyn514:stable-options, r=Mark-Simulacrum
bors [Thu, 10 Jun 2021 12:47:54 +0000 (12:47 +0000)]
Auto merge of #82639 - jyn514:stable-options, r=Mark-Simulacrum

Don't pass -Z unstable-options by default for UI tests

Unconditionally passing -Z unstable-options makes it impossible to test whether an option requires unstable-options or not.

This uncovered quite a lot of bugs, I'll open issues for each. These don't strictly need to be fixed before this is merged, it just makes the diff much larger because of the changes to diagnostics.

- https://github.com/rust-lang/rust/issues/82636
- https://github.com/rust-lang/rust/issues/82637
- https://github.com/rust-lang/rust/issues/82638

3 years agoAdd support for using qualified paths with structs in expression and pattern
Ryan Levick [Thu, 10 Dec 2020 12:20:07 +0000 (13:20 +0100)]
Add support for using qualified paths with structs in expression and pattern
position.

3 years agoAuto merge of #85741 - tmiasko:ssa, r=nagisa
bors [Thu, 10 Jun 2021 10:06:58 +0000 (10:06 +0000)]
Auto merge of #85741 - tmiasko:ssa, r=nagisa

Use preorder traversal when checking for SSA locals

Traverse blocks in topological sort of dominance partial order, to ensure that
local analyzer correctly identifies locals that are already in static single
assignment form, while avoiding dependency on implicit numeric order of blocks.

When rebuilding the standard library, this change reduces the number of locals
that require an alloca from 62452 to 62348.

3 years agogcc-lld mvp
1000teslas [Wed, 2 Jun 2021 09:48:33 +0000 (19:48 +1000)]
gcc-lld mvp

ignore test if rust-lld not found

create ld -> rust-lld symlink at build time instead of run time

for testing in ci

copy instead of symlinking

remove linux check

test for linker, suggestions from bjorn3

fix overly restrictive lld matcher

use -Zgcc-ld flag instead of -Clinker-flavor

refactor code adding lld to gcc path

revert ci changes

suggestions from petrochenkov

rename gcc_ld to gcc-ld in dirs

3 years agoAuto merge of #86186 - JohnTitor:rollup-upaw6wx, r=JohnTitor
bors [Thu, 10 Jun 2021 03:11:24 +0000 (03:11 +0000)]
Auto merge of #86186 - JohnTitor:rollup-upaw6wx, r=JohnTitor

Rollup of 7 pull requests

Successful merges:

 - #82037 (Make symbols stripping work on MacOS X)
 - #84687 (Multiple improvements to RwLocks)
 - #85997 (rustdoc: Print a warning if the diff when comparing to old nightlies is empty)
 - #86051 (Updated code examples and wording in move keyword documentation )
 - #86111 (fix off by one in `std::iter::Iterator` documentation)
 - #86113 (build doctests with lld if use-lld = true)
 - #86175 (update Miri)

Failed merges:

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

3 years agoMake `relate_type_and_mut` public
Yuki Okushi [Thu, 10 Jun 2021 02:48:51 +0000 (11:48 +0900)]
Make `relate_type_and_mut` public

3 years agoRollup merge of #86175 - RalfJung:miri, r=RalfJung
Yuki Okushi [Thu, 10 Jun 2021 02:02:16 +0000 (11:02 +0900)]
Rollup merge of #86175 - RalfJung:miri, r=RalfJung

update Miri

We had some nice PRs land today, let's ship those. :)
Cc `@rust-lang/miri` r? `@ghost`

3 years agoRollup merge of #86113 - the8472:doctest-lld, r=Mark-Simulacrum
Yuki Okushi [Thu, 10 Jun 2021 02:02:15 +0000 (11:02 +0900)]
Rollup merge of #86113 - the8472:doctest-lld, r=Mark-Simulacrum

build doctests with lld if use-lld = true

results when running `./x.py test library/core --doc --stage 0`:

```
# OLD
test result: FAILED. 2844 passed; 6 failed; 28 ignored; 0 measured; 0 filtered out; finished in 21.13s

# NEW
test result: FAILED. 2844 passed; 6 failed; 28 ignored; 0 measured; 0 filtered out; finished in 11.92s
```

3 years agoRollup merge of #86111 - spookyvision:master, r=JohnTitor
Yuki Okushi [Thu, 10 Jun 2021 02:02:14 +0000 (11:02 +0900)]
Rollup merge of #86111 - spookyvision:master, r=JohnTitor

fix off by one in `std::iter::Iterator` documentation

the range `(0..10)` is documented as "The even numbers from zero to ten." - should be ".. to nine".

3 years agoRollup merge of #86051 - erer1243:update_move_keyword_docs, r=Mark-Simulacrum
Yuki Okushi [Thu, 10 Jun 2021 02:02:13 +0000 (11:02 +0900)]
Rollup merge of #86051 - erer1243:update_move_keyword_docs, r=Mark-Simulacrum

Updated code examples and wording in move keyword documentation

Had a conversation with someone on the Rust Discord who was confused by the move keyword documentation. Some of the wording is odd sounding ("owned by value" - what else can something be owned by?). Also, some of the examples used Copy types when demonstrating move, leading to variables still being accessible in the outer scope after the move, contradicting the examples' comments.

I changed the move keyword documentation a bit, removing that odd wording and changing all the examples to use non-Copy types

3 years agoRollup merge of #85997 - jyn514:rustdoc-diff, r=Mark-Simulacrum
Yuki Okushi [Thu, 10 Jun 2021 02:02:12 +0000 (11:02 +0900)]
Rollup merge of #85997 - jyn514:rustdoc-diff, r=Mark-Simulacrum

rustdoc: Print a warning if the diff when comparing to old nightlies is empty

This avoids confusing situations where it's unclear whether there's a
bug in the diff tool or not:

```
26: `@has` check failed
        `XPATH PATTERN` did not match
        // `@has` - '//code/a[`@href="{{channel}}/std/primitive.i32.html"]'` 'i32'

Encountered 6 errors

------------------------------------------

info: generating a diff against nightly rustdoc

failures:
    [rustdoc] rustdoc/primitive-reexport.rs
```

3 years agoRollup merge of #84687 - a1phyr:improve_rwlock, r=m-ou-se
Yuki Okushi [Thu, 10 Jun 2021 02:02:10 +0000 (11:02 +0900)]
Rollup merge of #84687 - a1phyr:improve_rwlock, r=m-ou-se

Multiple improvements to RwLocks

This PR replicates #77147, #77380 and #84650 on RWLocks :
- Split `sys_common::RWLock` in `StaticRWLock` and `MovableRWLock`
- Unbox rwlocks on some platforms (Windows, Wasm and unsupported)
- Simplify `RwLock::into_inner`

Notes to reviewers :
- For each target, I copied `MovableMutex` to guess if `MovableRWLock` should be boxed.
- ~A comment says that `StaticMutex` is not re-entrant, I don't understand why and I don't know whether it applies to `StaticRWLock`.~

r? `@m-ou-se`

3 years agoRollup merge of #82037 - calavera:strip_debuginfo_osx, r=petrochenkov
Yuki Okushi [Thu, 10 Jun 2021 02:02:01 +0000 (11:02 +0900)]
Rollup merge of #82037 - calavera:strip_debuginfo_osx, r=petrochenkov

Make symbols stripping work on MacOS X

As reported in the [stabilization issue](https://github.com/rust-lang/rust/issues/72110), MacOS' linker doesn't support the `-s` and `-S` flags to strip symbols anymore. However, the os ships a separated tool to perform these operations.

This change allows the compiler to use that tool after a target has been compiled to strip symbols.

For rationale, see: https://github.com/rust-lang/rust/issues/72110#issuecomment-641169818
For option selection, see: https://www.unix.com/man-page/osx/1/strip/

Signed-off-by: David Calavera <david.calavera@gmail.com>
3 years agoAuto merge of #85910 - cjgillot:no-meta-version, r=Aaron1011
bors [Thu, 10 Jun 2021 00:39:25 +0000 (00:39 +0000)]
Auto merge of #85910 - cjgillot:no-meta-version, r=Aaron1011

Drop metadata_encoding_version.

Part of #85153

r? `@Aaron1011`

3 years agodetect incorrect vtable alignment during const eval instead of ICE-ing
Rémy Rakic [Tue, 8 Jun 2021 21:47:04 +0000 (23:47 +0200)]
detect incorrect vtable alignment during const eval instead of ICE-ing

also add tests for these 2 kinds of errors for size and alignment,
as the existing size check wasn't apparently tested

3 years agoupdate Miri
Ralf Jung [Wed, 9 Jun 2021 19:06:09 +0000 (21:06 +0200)]
update Miri

3 years agoAdd more tests + visit_ty in some places
Ellen [Wed, 9 Jun 2021 18:28:41 +0000 (19:28 +0100)]
Add more tests + visit_ty in some places

3 years agoAuto merge of #86003 - pnkfelix:issue-84297-revert-81238, r=Mark-Simulacrum
bors [Wed, 9 Jun 2021 16:47:05 +0000 (16:47 +0000)]
Auto merge of #86003 - pnkfelix:issue-84297-revert-81238, r=Mark-Simulacrum

Make copy/copy_nonoverlapping fn's again

Make copy/copy_nonoverlapping fn's again, rather than intrinsics.

This a short-term change to address issue #84297.

It effectively reverts PRs #81167 #81238 (and part of #82967), #83091, and parts of #79684.

3 years agoMake symbols stripping work on MacOS X
David Calavera [Fri, 12 Feb 2021 18:24:16 +0000 (10:24 -0800)]
Make symbols stripping work on MacOS X

As reported in the stabilization issue, MacOS' linker doesn't support the `-s` and `-S` flags to strip symbols anymore. However, the os ships a separated tool to perform these operations.

This change allows the compiler to use that tool after a target has been compiled to strip symbols.

For rationale, see: https://github.com/rust-lang/rust/issues/72110#issuecomment-641169818
For option selection, see: https://www.unix.com/man-page/osx/1/strip/

Signed-off-by: David Calavera <david.calavera@gmail.com>
3 years agoAuto merge of #86150 - cjgillot:notable, r=michaelwoerister
bors [Wed, 9 Jun 2021 14:06:10 +0000 (14:06 +0000)]
Auto merge of #86150 - cjgillot:notable, r=michaelwoerister

Do not require the DefPathTable to construct the on-disk cache.

r? `@michaelwoerister`

3 years agoAdd safety comments
bjorn3 [Wed, 9 Jun 2021 12:51:42 +0000 (14:51 +0200)]
Add safety comments

3 years agoUse explicit drop impl
bjorn3 [Wed, 9 Jun 2021 12:47:01 +0000 (14:47 +0200)]
Use explicit drop impl

3 years agoAuto merge of #85975 - the8472:revert-take-tra, r=scottmcm
bors [Wed, 9 Jun 2021 11:24:54 +0000 (11:24 +0000)]
Auto merge of #85975 - the8472:revert-take-tra, r=scottmcm

Revert "implement TrustedRandomAccess for Take iterator adapter"

This reverts commit 37a5b515e9c36ee3f57d9e0d7db7efce2fb02195 (#83990).

The original change unintentionally caused side-effects from certain iterator chains combining `take`, `zip` and `next_back()` to be omitted which is observable by user code and thus likely a breaking change

Technically one could declare it not a breaking change since `Zip`'s API contract is silent about about its backwards iteration behavior but on the other hand there is nothing in the stable Iterator API that could justify the currently observable behavior. And either way, this impact wasn't noticed or discussed in the original PR.

Fixes #85969

3 years agoAuto merge of #86118 - spastorino:tait-soundness-bug, r=nikomatsakis
bors [Wed, 9 Jun 2021 09:00:16 +0000 (09:00 +0000)]
Auto merge of #86118 - spastorino:tait-soundness-bug, r=nikomatsakis

Create different inference variables for different defining uses of TAITs

Fixes #73481

r? `@nikomatsakis`
cc `@oli-obk`

3 years agoAuto merge of #86107 - Smittyvb:peephole-optim-eq-bool, r=wesleywiser
bors [Wed, 9 Jun 2021 06:06:06 +0000 (06:06 +0000)]
Auto merge of #86107 - Smittyvb:peephole-optim-eq-bool, r=wesleywiser

Peephole optimize `x == false` and `x != true`

This adds peephole optimizations to make `x == false`, `false == x`, `x != true`, and `true != x` get optimized to `!x` in the `instcombine` MIR pass. That pass currently handles `x == true` -> `x` already.

3 years agoAuto merge of #86160 - JohnTitor:rollup-8ark9x7, r=JohnTitor
bors [Wed, 9 Jun 2021 03:15:34 +0000 (03:15 +0000)]
Auto merge of #86160 - JohnTitor:rollup-8ark9x7, r=JohnTitor

Rollup of 11 pull requests

Successful merges:

 - #85676 (Fix documentation style inconsistencies for IP addresses)
 - #85715 (Document `From` impls in string.rs)
 - #85791 (Add `Ipv6Addr::is_unicast`)
 - #85957 (Display defaults on const params- rustdoc )
 - #85982 (Enable rustdoc to document safe wasm intrinsics)
 - #86121 (Forwarding implementation for Seek trait's stream_position method)
 - #86124 (Include macro name in 'local ambiguity' error)
 - #86128 (Refactor: Extract render_summary from render_impl.)
 - #86142 (Simplify proc_macro code using Bound::cloned().)
 - #86158 (Update books)
 - #86159 (Update cargo)

Failed merges:

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

3 years agoRollup merge of #86159 - ehuss:update-cargo, r=ehuss
Yuki Okushi [Wed, 9 Jun 2021 03:04:10 +0000 (12:04 +0900)]
Rollup merge of #86159 - ehuss:update-cargo, r=ehuss

Update cargo

11 commits in 0cecbd67323ca14a7eb6505900d0d7307b00355b..aa8b09297bb3156b849e73db48af4cd050492fe6
2021-06-01 20:09:13 +0000 to 2021-06-09 00:28:53 +0000
- Add `--prune` option for cargo-tree (rust-lang/cargo#9520)
- Fix typo in gitignore docs. (rust-lang/cargo#9556)
- Fix typos in command_prelude.rs (rust-lang/cargo#9552)
- Make clippy happy (rust-lang/cargo#9551)
- Remove some dead code. (rust-lang/cargo#9546)
- Add additional test for CJK progress width (rust-lang/cargo#9515)
- Pull in semver 1.0.3 'x' fix (rust-lang/cargo#9544)
- Add some tracking issues to unstable docs. (rust-lang/cargo#9540)
- Update documentation for include/exclude. (rust-lang/cargo#9538)
- Bump mdbook version for contrib guide. (rust-lang/cargo#9534)
- Replace deprecated `[replace]` references with `[patch]` (rust-lang/cargo#9533)

3 years agoRollup merge of #86158 - ehuss:update-books, r=ehuss
Yuki Okushi [Wed, 9 Jun 2021 03:04:09 +0000 (12:04 +0900)]
Rollup merge of #86158 - ehuss:update-books, r=ehuss

Update books

## reference

4 commits in 9c68af3ce6ccca2395e1868addef26a0542e9ddd..8f598e2af6c25b4a7ee88ef6a8196d9b8ea50ca8
2021-05-24 09:53:32 -0700 to 2021-06-01 19:00:46 +0100
- Add crate and module to glossary. (rust-lang-nursery/reference#1016)
- Fix type_length_limit example. (rust-lang-nursery/reference#1026)
- Rearrange HRTB grammar. (rust-lang-nursery/reference#1011)
- Revert "Temporarily remove pat_param." (rust-lang-nursery/reference#1010)

## rustc-dev-guide

6 commits in 50de7f0682adc5d95ce858fe6318d19b4b951553..c8da5bfd1c7c71d90ef1646f5e0a9f6609d5c78a
2021-05-20 15:02:20 +0200 to 2021-06-04 09:08:56 +0200
- Fix some links (rust-lang/rustc-dev-guide#1137)
- explain Miri engine vs Miri-the-tool
- Add more information about no_hash query modifier. (rust-lang/rustc-dev-guide#1133)
- improve section introduction
- not all tools require waiting for a nightly release before they can be fixed
- Describe the difference of rustc_lint vs rustc_lint_defs.

3 years agoRollup merge of #86142 - m-ou-se:proc-macro-subspan-bound-cloned-cleanup, r=petrochenkov
Yuki Okushi [Wed, 9 Jun 2021 03:04:08 +0000 (12:04 +0900)]
Rollup merge of #86142 - m-ou-se:proc-macro-subspan-bound-cloned-cleanup, r=petrochenkov

Simplify proc_macro code using Bound::cloned().

3 years agoRollup merge of #86128 - jsha:render-impl-into-mod, r=GuillaumeGomez
Yuki Okushi [Wed, 9 Jun 2021 03:04:07 +0000 (12:04 +0900)]
Rollup merge of #86128 - jsha:render-impl-into-mod, r=GuillaumeGomez

Refactor: Extract render_summary from render_impl.

This allows for a more readable straight-through logic in render_impl without need for a closure.

I think this will make #85970 a bit more of a straightforward change.

This is a pure refactoring. I've verified that the output of `x.py doc library/std` is byte-for-byte identical.

r? `@GuillaumeGomez`

3 years agoRollup merge of #86124 - Aaron1011:ambig-macro-name, r=varkor
Yuki Okushi [Wed, 9 Jun 2021 03:04:06 +0000 (12:04 +0900)]
Rollup merge of #86124 - Aaron1011:ambig-macro-name, r=varkor

Include macro name in 'local ambiguity' error

Currently, we only point at the span of the macro argument. When the
macro call is itself generated by another macro, this can make it
difficult or impossible to determine which macro is responsible for
producing the error.

3 years agoRollup merge of #86121 - nickshiling:forwarding_impl_for_seek_trait_stream_position...
Yuki Okushi [Wed, 9 Jun 2021 03:04:05 +0000 (12:04 +0900)]
Rollup merge of #86121 - nickshiling:forwarding_impl_for_seek_trait_stream_position, r=dtolnay

Forwarding implementation for Seek trait's stream_position method

Forwarding implementations for `Seek` trait's `stream_position` were missed when it was stabilized in `1.51.0`

3 years agoRollup merge of #85982 - alexcrichton:doc-safe-wasm, r=jyn514
Yuki Okushi [Wed, 9 Jun 2021 03:04:04 +0000 (12:04 +0900)]
Rollup merge of #85982 - alexcrichton:doc-safe-wasm, r=jyn514

Enable rustdoc to document safe wasm intrinsics

This commit fixes an issue not found during #84988 where rustdoc is used
to document cross-platform intrinsics but it was requiring that
functions which use `#[target_feature]` are `unsafe` erroneously, even
if they're WebAssembly specific. Rustdoc today, for example, already has
a special case where it enables annotations like
`#[target_feature(enable = "simd128")]` on platforms other than
WebAssembly. The purpose of this commit is to relax the "require all
`#[target_feature]` functions are `unsafe`" requirement for all targets
whenever rustdoc is running, enabling all targets to fully document
other targets, such as WebAssembly, where intrinsics functions aren't
always `unsafe`.

3 years agoRollup merge of #85957 - BoxyUwU:rustdoc-const-generic-defaults, r=oli-obk
Yuki Okushi [Wed, 9 Jun 2021 03:04:03 +0000 (12:04 +0900)]
Rollup merge of #85957 - BoxyUwU:rustdoc-const-generic-defaults, r=oli-obk

Display defaults on const params- rustdoc

previously rustdoc would render this struct declaration:
`pub struct Foo<const N: usize = 10>;`
as:
`pub struct Foo<const N: usize>;`
this PR changes it to render correctly

3 years agoRollup merge of #85791 - CDirkx:is_unicast, r=joshtriplett
Yuki Okushi [Wed, 9 Jun 2021 03:04:01 +0000 (12:04 +0900)]
Rollup merge of #85791 - CDirkx:is_unicast, r=joshtriplett

Add `Ipv6Addr::is_unicast`

Adds an unstable utility method `Ipv6Addr::is_unicast` under the feature flag `ip` (tracking issue: #27709).

Added for completeness with the other unicast methods (see also https://github.com/rust-lang/rust/issues/85604#issuecomment-848220455) and opposite of `is_multicast`.

3 years agoRollup merge of #85715 - fee1-dead:document-string, r=JohnTitor
Yuki Okushi [Wed, 9 Jun 2021 03:03:59 +0000 (12:03 +0900)]
Rollup merge of #85715 - fee1-dead:document-string, r=JohnTitor

Document `From` impls in string.rs

3 years agoRollup merge of #85676 - CDirkx:ip-style, r=JohnTitor
Yuki Okushi [Wed, 9 Jun 2021 03:03:54 +0000 (12:03 +0900)]
Rollup merge of #85676 - CDirkx:ip-style, r=JohnTitor

Fix documentation style inconsistencies for IP addresses

Pulled out of #85655 as it is unrelated. Fixes some inconsistencies in the docs for IP addresses:
- Currently some addresses are backticked, some are not, this PR backticks everything consistently. (looks better imo)
- Lowercase hex-literals are used when writing addresses.

3 years agoUpdate cargo
Eric Huss [Wed, 9 Jun 2021 02:38:25 +0000 (19:38 -0700)]
Update cargo

3 years agoUpdate books
Eric Huss [Wed, 9 Jun 2021 02:15:11 +0000 (19:15 -0700)]
Update books

3 years agoUpdate library/core/src/iter/traits/iterator.rs
Anatol Ulrich [Wed, 9 Jun 2021 00:51:30 +0000 (02:51 +0200)]
Update library/core/src/iter/traits/iterator.rs

Co-authored-by: Yuki Okushi <jtitor@2k36.org>
3 years agoPrint dummy spans as `no-location`
Tomasz Miąsko [Wed, 9 Jun 2021 00:00:00 +0000 (00:00 +0000)]
Print dummy spans as `no-location`

3 years agoDo not require the DefPathTable to construct the on-disk cache.
Camille GILLOT [Tue, 8 Jun 2021 20:18:53 +0000 (22:18 +0200)]
Do not require the DefPathTable to construct the on-disk cache.

3 years agoadd VecMap docs
Santiago Pastorino [Tue, 8 Jun 2021 20:16:05 +0000 (17:16 -0300)]
add VecMap docs

3 years agoUse impl FnMut directly as predicate type
Santiago Pastorino [Tue, 8 Jun 2021 19:52:33 +0000 (16:52 -0300)]
Use impl FnMut directly as predicate type

3 years agoExplicitly pass find arguments down the predicate so coercions can apply
Santiago Pastorino [Tue, 8 Jun 2021 19:50:33 +0000 (16:50 -0300)]
Explicitly pass find arguments down the predicate so coercions can apply

3 years agoDo not deconstruct OpaqueTypeKey to make the code more clear
Santiago Pastorino [Tue, 8 Jun 2021 19:48:58 +0000 (16:48 -0300)]
Do not deconstruct OpaqueTypeKey to make the code more clear

3 years agodefin_ty_def_id -> definition_ty_def_id
Santiago Pastorino [Tue, 8 Jun 2021 19:45:44 +0000 (16:45 -0300)]
defin_ty_def_id -> definition_ty_def_id

3 years agoAuto merge of #86139 - hyd-dev:miri, r=RalfJung
bors [Tue, 8 Jun 2021 18:33:51 +0000 (18:33 +0000)]
Auto merge of #86139 - hyd-dev:miri, r=RalfJung

Update Miri

Fixes #86133.

r? `@RalfJung`

3 years agoDon't require LintStore to live for 'a in configure_and_expand_inner
bjorn3 [Tue, 1 Jun 2021 09:52:44 +0000 (11:52 +0200)]
Don't require LintStore to live for 'a in configure_and_expand_inner

3 years agoRevert "Let several methods take &Resolver instead of a BoxedResolver wrapper"
bjorn3 [Tue, 1 Jun 2021 09:51:32 +0000 (11:51 +0200)]
Revert "Let several methods take &Resolver instead of a BoxedResolver wrapper"

This reverts commit 5343ec338f72a61e2f51f9d90117092c8e8a725a.

3 years agoUse a submodule as safety boundary for BoxedResolver
bjorn3 [Tue, 1 Jun 2021 08:49:42 +0000 (10:49 +0200)]
Use a submodule as safety boundary for BoxedResolver

3 years agoDon't use a generator for BoxedResolver
bjorn3 [Tue, 1 Jun 2021 08:47:39 +0000 (10:47 +0200)]
Don't use a generator for BoxedResolver

The generator is non-trivial and requires unsafe code anyway. Using
regular unsafe code without a generator is much easier to follow.

3 years agoDon't return a BoxedResolver on errors
bjorn3 [Tue, 1 Jun 2021 08:15:28 +0000 (10:15 +0200)]
Don't return a BoxedResolver on errors

3 years agoInline two more methods
bjorn3 [Tue, 1 Jun 2021 08:04:13 +0000 (10:04 +0200)]
Inline two more methods

3 years agoLet several methods take &Resolver instead of a BoxedResolver wrapper
bjorn3 [Mon, 31 May 2021 17:48:47 +0000 (19:48 +0200)]
Let several methods take &Resolver instead of a BoxedResolver wrapper

3 years agoStore boxed metadata loader in CrateLoader
bjorn3 [Mon, 31 May 2021 15:36:53 +0000 (17:36 +0200)]
Store boxed metadata loader in CrateLoader

3 years agoReplace Pin::new with .as_mut()
bjorn3 [Mon, 31 May 2021 14:33:09 +0000 (16:33 +0200)]
Replace Pin::new with .as_mut()

3 years agoUse more accurate lifetimes
bjorn3 [Mon, 31 May 2021 14:24:09 +0000 (16:24 +0200)]
Use more accurate lifetimes

3 years agoInline PinnedGenerator
bjorn3 [Mon, 31 May 2021 13:28:17 +0000 (15:28 +0200)]
Inline PinnedGenerator

3 years agoInline the rest of box_region
bjorn3 [Mon, 31 May 2021 13:20:50 +0000 (15:20 +0200)]
Inline the rest of box_region

3 years agoInline box_region macro calls
bjorn3 [Mon, 31 May 2021 13:17:04 +0000 (15:17 +0200)]
Inline box_region macro calls

3 years agoSimplify box_region macros
bjorn3 [Mon, 31 May 2021 12:51:31 +0000 (14:51 +0200)]
Simplify box_region macros

3 years agoSimplify proc_macro code using Bound::cloned().
Mara Bos [Tue, 8 Jun 2021 16:09:29 +0000 (18:09 +0200)]
Simplify proc_macro code using Bound::cloned().

3 years agoUpdate Miri
hyd-dev [Tue, 8 Jun 2021 15:33:39 +0000 (23:33 +0800)]
Update Miri

3 years agoDestructure OpaqueTypeKey in certain cases to simplify code
Santiago Pastorino [Tue, 8 Jun 2021 12:50:16 +0000 (09:50 -0300)]
Destructure OpaqueTypeKey in certain cases to simplify code

3 years agoSimplify code by using VecMap::get_by
Santiago Pastorino [Tue, 8 Jun 2021 12:41:26 +0000 (09:41 -0300)]
Simplify code by using VecMap::get_by