]> git.lizzy.rs Git - rust.git/log
rust.git
3 years agoEncode `ScalarInt::bytes` as `u128` instead of `[u8; 16]` to see if that caused the...
Oliver Scherer [Thu, 1 Oct 2020 07:47:36 +0000 (09:47 +0200)]
Encode `ScalarInt::bytes` as `u128` instead of `[u8; 16]` to see if that caused the performance regression

3 years agoSplit the "raw integer bytes" part out of `Scalar`
Oliver Scherer [Sat, 26 Sep 2020 13:15:35 +0000 (15:15 +0200)]
Split the "raw integer bytes" part out of `Scalar`

3 years agoAuto merge of #78711 - m-ou-se:rollup-pxqnny7, r=m-ou-se
bors [Tue, 3 Nov 2020 18:58:06 +0000 (18:58 +0000)]
Auto merge of #78711 - m-ou-se:rollup-pxqnny7, r=m-ou-se

Rollup of 7 pull requests

Successful merges:

 - #77950 (Add support for SHA256 source file hashing)
 - #78624 (Sync rustc_codegen_cranelift)
 - #78626 (Improve errors about #[deprecated] attribute)
 - #78659 (Corrected suggestion for generic parameters in `function_item_references` lint)
 - #78687 (Suggest library/std when running all stage 0 tests)
 - #78699 (Show more error information in lldb_batchmode)
 - #78709 (Fix panic in bootstrap for non-workspace path dependencies.)

Failed merges:

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

3 years agoRollup merge of #78709 - ehuss:fix-in_tree_crates-non-member, r=Mark-Simulacrum
Mara Bos [Tue, 3 Nov 2020 18:32:44 +0000 (19:32 +0100)]
Rollup merge of #78709 - ehuss:fix-in_tree_crates-non-member, r=Mark-Simulacrum

Fix panic in bootstrap for non-workspace path dependencies.

If you add a `path` dependency to a `Cargo.toml` that is located outside of the workspace, then the `in_tree_crates` function can panic because it finds a path dependency that is not defined (since it uses `cargo metadata --no-deps`).  This fixes it by skipping over those entries, which are usually not things you select on the command-line.

Fixes #78617

3 years agoRollup merge of #78699 - pietroalbini:lldb-error, r=jyn514
Mara Bos [Tue, 3 Nov 2020 18:32:42 +0000 (19:32 +0100)]
Rollup merge of #78699 - pietroalbini:lldb-error, r=jyn514

Show more error information in lldb_batchmode

Even more information to try and debug #78665.

3 years agoRollup merge of #78687 - jyn514:bootstrap-help, r=Mark-Simulacrum
Mara Bos [Tue, 3 Nov 2020 18:32:40 +0000 (19:32 +0100)]
Rollup merge of #78687 - jyn514:bootstrap-help, r=Mark-Simulacrum

Suggest library/std when running all stage 0 tests

r? ``@Mark-Simulacrum``
cc ``@ijackson``

For context, this came out of a discord conversation where ``@ijackson`` was running `test --stage 1` when they were only adding doc-tests to the standard library.

3 years agoRollup merge of #78659 - ayrtonm:fn-ref-lint-fix, r=oli-obk
Mara Bos [Tue, 3 Nov 2020 18:32:38 +0000 (19:32 +0100)]
Rollup merge of #78659 - ayrtonm:fn-ref-lint-fix, r=oli-obk

Corrected suggestion for generic parameters in `function_item_references` lint

This commit handles functions with generic type parameters like you pointed out as well as const generics. Also this is probably a minor thing, but the type alias you used in the example doesn't show up so the suggestion right now would be `size_of::<[u8; 16]> as fn() ->`. This is because the lint checker works with MIR instead of HIR. I don't think we can get the alias at that point, but let me know if I'm wrong and there's a way to fix this. Also I put you as the reviewer, but I'm not sure if you want to review it or if it makes more sense to ask one of the original reviewers of this lint.
closes #78571

3 years agoRollup merge of #78626 - fusion-engineering-forks:deprecated-trait-impl, r=estebank
Mara Bos [Tue, 3 Nov 2020 18:32:36 +0000 (19:32 +0100)]
Rollup merge of #78626 - fusion-engineering-forks:deprecated-trait-impl, r=estebank

Improve errors about #[deprecated] attribute

This change:

1. Turns `#[deprecated]` on a trait impl block into an error, which fixes #78625;
2. Changes these and other errors about `#[deprecated]` to use the span of the attribute instead of the item; and
3. Turns this error into a lint, to make sure it can be capped with `--cap-lints` and doesn't break any existing dependencies.

Can be reviewed per commit.

---
Example:
```rust
struct X;

#[deprecated = "a"]
impl Default for X {
    #[deprecated = "b"]
    fn default() -> Self {
        X
    }
}
```

Before:
```
error: This deprecation annotation is useless
 --> src/main.rs:6:5
  |
6 | /     fn default() -> Self {
7 | |         X
8 | |     }
  | |_____^
```

After:
```
error: this `#[deprecated]' annotation has no effect
 --> src/main.rs:3:1
  |
3 | #[deprecated = "a"]
  | ^^^^^^^^^^^^^^^^^^^ help: try removing the deprecation attribute
  |
  = note: `#[deny(useless_deprecated)]` on by default

error: this `#[deprecated]' annotation has no effect
 --> src/main.rs:5:5
  |
5 |     #[deprecated = "b"]
  |     ^^^^^^^^^^^^^^^^^^^ help: try removing the deprecation attribute
```

3 years agoRollup merge of #78624 - bjorn3:update_cg_clif-2020-11-01, r=jyn514
Mara Bos [Tue, 3 Nov 2020 18:32:34 +0000 (19:32 +0100)]
Rollup merge of #78624 - bjorn3:update_cg_clif-2020-11-01, r=jyn514

Sync rustc_codegen_cranelift

This fixes bootstrapping of rustc using cg_clif again. It regressed a while before #77975 got merged.

Fixes https://github.com/bjorn3/rustc_codegen_cranelift/issues/743

3 years agoRollup merge of #77950 - arlosi:sha256, r=eddyb
Mara Bos [Tue, 3 Nov 2020 18:32:26 +0000 (19:32 +0100)]
Rollup merge of #77950 - arlosi:sha256, r=eddyb

Add support for SHA256 source file hashing

Adds support for `-Z src-hash-algorithm sha256`, which became available in LLVM 11.

Using an older version of LLVM will cause an error `invalid checksum kind` if the hash algorithm is set to sha256.

r? `@eddyb`
cc #70401 `@est31`

3 years agoFix panic in bootstrap for non-workspace path dependencies.
Eric Huss [Tue, 3 Nov 2020 17:54:11 +0000 (09:54 -0800)]
Fix panic in bootstrap for non-workspace path dependencies.

3 years agoAuto merge of #76931 - oli-obk:const_prop_inline_lint_madness, r=wesleywiser
bors [Tue, 3 Nov 2020 16:32:34 +0000 (16:32 +0000)]
Auto merge of #76931 - oli-obk:const_prop_inline_lint_madness, r=wesleywiser

Properly handle lint spans after MIR inlining

The first commit shows what happens when we apply mir inlining and then cause lints on the inlined MIR.
The second commit fixes that.

r? `@wesleywiser`

3 years agoAuto merge of #78597 - RalfJung:raw-retag, r=oli-obk
bors [Tue, 3 Nov 2020 13:48:54 +0000 (13:48 +0000)]
Auto merge of #78597 - RalfJung:raw-retag, r=oli-obk

Retagging: do not retag 'raw reborrows'

When doing `&raw const (*raw_ptr).field`, we do not want any retagging; the original provenance should be fully preserved.

Fixes https://github.com/rust-lang/miri/issues/1608
Test added by https://github.com/rust-lang/miri/pull/1614

Not sure whom to ask for review on this... `@oli-obk` can you have a look? Or maybe highfive makes a good choice.^^

3 years agolldb_batchmode: show more error information
Pietro Albini [Tue, 3 Nov 2020 11:01:46 +0000 (12:01 +0100)]
lldb_batchmode: show more error information

Even more information to try and debug #78665.

3 years agoMerge commit '03f01bbe901d60b71cf2c5ec766aef5e532ab79d' into update_cg_clif-2020...
bjorn3 [Tue, 3 Nov 2020 10:00:04 +0000 (11:00 +0100)]
Merge commit '03f01bbe901d60b71cf2c5ec766aef5e532ab79d' into update_cg_clif-2020-11-01

3 years agoAuto merge of #78489 - bugadani:array, r=estebank
bors [Tue, 3 Nov 2020 09:19:01 +0000 (09:19 +0000)]
Auto merge of #78489 - bugadani:array, r=estebank

Minor cleanup around incremental compilation

* Remove some short lived vectors
* Fix some typos
* Avoid some reallocations

3 years agoAuto merge of #78697 - JohnTitor:rollup-q0fchpv, r=JohnTitor
bors [Tue, 3 Nov 2020 06:56:46 +0000 (06:56 +0000)]
Auto merge of #78697 - JohnTitor:rollup-q0fchpv, r=JohnTitor

Rollup of 8 pull requests

Successful merges:

 - #78376 (Treat trailing semicolon as a statement in macro call)
 - #78400 (Fix unindent in doc comments)
 - #78575 (Add a test for compiletest rustc-env & unset-rustc-env directives)
 - #78616 (Document -Zinstrument-coverage)
 - #78663 (Fix ICE when a future-incompat-report has its command-line level capped)
 - #78664 (Fix intrinsic size_of stable link)
 - #78668 (inliner: Remove redundant loop)
 - #78676 (add mipsel-unknown-none target)

Failed merges:

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

3 years agoRollup merge of #78676 - kiffie:embedded-bare-mipsr2, r=jonas-schievink
Yuki Okushi [Tue, 3 Nov 2020 06:27:21 +0000 (15:27 +0900)]
Rollup merge of #78676 - kiffie:embedded-bare-mipsr2, r=jonas-schievink

add mipsel-unknown-none target

This adds a target for bare MIPS32r2, little endian, softfloat. This target can be used for PIC32 microcontrollers (or possibly for other devices that have a MIPS MCU core such as the M4K core).

Tried to find a name for the target that is in line with the naming scheme apparently used for the other MIPS targets.

r? `@jonas-schievink`

3 years agoRollup merge of #78668 - tmiasko:inline, r=oli-obk
Yuki Okushi [Tue, 3 Nov 2020 06:27:19 +0000 (15:27 +0900)]
Rollup merge of #78668 - tmiasko:inline, r=oli-obk

inliner: Remove redundant loop

No functional changes intended.

3 years agoRollup merge of #78664 - pickfire:patch-4, r=jonas-schievink
Yuki Okushi [Tue, 3 Nov 2020 06:27:18 +0000 (15:27 +0900)]
Rollup merge of #78664 - pickfire:patch-4, r=jonas-schievink

Fix intrinsic size_of stable link

I noticed that it is pointing to the same link when I was reading
https://github.com/rust-lang/rust-clippy/issues/2997

3 years agoRollup merge of #78663 - Aaron1011:fix/cap-future-compat, r=tmandry
Yuki Okushi [Tue, 3 Nov 2020 06:27:16 +0000 (15:27 +0900)]
Rollup merge of #78663 - Aaron1011:fix/cap-future-compat, r=tmandry

Fix ICE when a future-incompat-report has its command-line level capped

Fixes #78660

With PR https://github.com/rust-lang/rust/pull/75534 merged, we now run
more lint-related code for future-incompat-report, even when their final
level is Allow. Some lint-related code was not expecting `Level::Allow`,
and had an explicit panic.

This PR explicitly tracks the lint level set on the command line before
`--cap-lints` is applied. This is used to emit a more precise error
note (e.g. we don't say that `-W lint-name` was specified on the
command line just because a lint was capped to Warn). As a result, we
can now correctly emit a note that `-A` was used if we got
`Level::Allow` from the command line (before the cap is applied).

3 years agoRollup merge of #78616 - richkadel:unstable-book-instr-cov, r=tmandry
Yuki Okushi [Tue, 3 Nov 2020 06:27:13 +0000 (15:27 +0900)]
Rollup merge of #78616 - richkadel:unstable-book-instr-cov, r=tmandry

Document -Zinstrument-coverage

r? ``@tmandry``
FYI ``@wesleywiser``

Here is my proposed document for LLVM source-based code coverage.

I based it on the `profile.md` page, in the same directory, and on the Clang guide for LLVM source-based coverage.

3 years agoRollup merge of #78575 - tmiasko:compiletest-rustc-env, r=Aaron1011
Yuki Okushi [Tue, 3 Nov 2020 06:27:11 +0000 (15:27 +0900)]
Rollup merge of #78575 - tmiasko:compiletest-rustc-env, r=Aaron1011

Add a test for compiletest rustc-env & unset-rustc-env directives

... and move compiletest meta tests into a separate directory.

3 years agoRollup merge of #78400 - GuillaumeGomez:fix-unindent, r=jyn514
Yuki Okushi [Tue, 3 Nov 2020 06:27:05 +0000 (15:27 +0900)]
Rollup merge of #78400 - GuillaumeGomez:fix-unindent, r=jyn514

Fix unindent in doc comments

Fixes #70732

r? ``@jyn514``

3 years agoRollup merge of #78376 - Aaron1011:feature/consistent-empty-expr, r=petrochenkov
Yuki Okushi [Tue, 3 Nov 2020 06:27:03 +0000 (15:27 +0900)]
Rollup merge of #78376 - Aaron1011:feature/consistent-empty-expr, r=petrochenkov

Treat trailing semicolon as a statement in macro call

See #61733 (comment)

We now preserve the trailing semicolon in a macro invocation, even if
the macro expands to nothing. As a result, the following code no longer
compiles:

```rust
macro_rules! empty {
    () => { }
}

fn foo() -> bool { //~ ERROR mismatched
    { true } //~ ERROR mismatched
    empty!();
}
```

Previously, `{ true }` would be considered the trailing expression, even
though there's a semicolon in `empty!();`

This makes macro expansion more token-based.

3 years agoAuto merge of #78448 - rylev:cache-foreign_modules, r=wesleywiser
bors [Tue, 3 Nov 2020 04:27:27 +0000 (04:27 +0000)]
Auto merge of #78448 - rylev:cache-foreign_modules, r=wesleywiser

foreign_modules query hash table lookups

When compiling a large monolithic crate we're seeing huge times in the `foreign_modules` query due to repeated iteration over foreign modules (in order to find a module by its id). This implements hash table lookups so that which massively reduces time spent in that query in this particular case. We'll need to see if the overhead of creating the hash table has a negative impact on performance in more normal compilation scenarios.

I'm working with `@wesleywiser` on this.

3 years agoAuto merge of #77271 - petrochenkov:notokenexp, r=Aaron1011
bors [Tue, 3 Nov 2020 00:58:24 +0000 (00:58 +0000)]
Auto merge of #77271 - petrochenkov:notokenexp, r=Aaron1011

Expand `NtExpr` tokens only in key-value attributes

Implement the experiment described in https://github.com/rust-lang/rust/issues/55414#issuecomment-554005412

This PR also removes some customization points and token visiting functionality from AST visitors.
Read-only visitor no longer visits tokens, mutable visitor visits tokens only when specifically enabled, mutable token visiting is restricted to its single intended use case.

I haven't changed the representation of `MacArgs::Eq` yet, but it potentially can use a `TokenTree` or a `Token` instead of `TokenStream`.
It's hard to get rid of `Nonterminal::NtExpr` there (and e.g. replace it with `ast::Expr`) due to the dual nature of key-value attributes (the value is both an expression and a token stream, depending on context), and `Nonterminal` has all the machinery for maintaining both representations in sync.

Fixes https://github.com/rust-lang/rust/issues/55414
Fixes https://github.com/rust-lang/rust/issues/43860

3 years agoSuggest library/std when running all stage 0 tests
Joshua Nelson [Mon, 2 Nov 2020 22:34:28 +0000 (17:34 -0500)]
Suggest library/std when running all stage 0 tests

3 years agoadd mipsel-unknown-none target
Stephan [Mon, 2 Nov 2020 22:31:15 +0000 (23:31 +0100)]
add mipsel-unknown-none target

3 years agoindicate calling convention
Stephan [Mon, 2 Nov 2020 22:29:00 +0000 (23:29 +0100)]
indicate calling convention

3 years agoimprove comments
Stephan [Mon, 2 Nov 2020 22:18:32 +0000 (23:18 +0100)]
improve comments

Co-authored-by: Jonas Schievink <jonasschievink@gmail.com>
3 years agoExpand `NtExpr` tokens only in key-value attributes
Vadim Petrochenkov [Sun, 27 Sep 2020 13:52:51 +0000 (16:52 +0300)]
Expand `NtExpr` tokens only in key-value attributes

3 years agoadd blacklist for unsupported calling conventions
Stephan [Mon, 2 Nov 2020 21:52:47 +0000 (22:52 +0100)]
add blacklist for unsupported calling conventions

3 years agoaddressed feedback
Rich Kadel [Mon, 2 Nov 2020 19:59:24 +0000 (11:59 -0800)]
addressed feedback

3 years agoMerge pull request #1105 from bjorn3/test_rustc_bootstrapping
bjorn3 [Mon, 2 Nov 2020 18:14:50 +0000 (19:14 +0100)]
Merge pull request #1105 from bjorn3/test_rustc_bootstrapping

Test bootstrapping of rustc using cg_clif

3 years agoadd mipsel_unknown_none target
Stephan [Mon, 2 Nov 2020 18:11:24 +0000 (19:11 +0100)]
add mipsel_unknown_none target

3 years agoTreat trailing semicolon as a statement in macro call
Aaron Hill [Sun, 25 Oct 2020 21:14:19 +0000 (17:14 -0400)]
Treat trailing semicolon as a statement in macro call

See https://github.com/rust-lang/rust/issues/61733#issuecomment-716188981

We now preserve the trailing semicolon in a macro invocation, even if
the macro expands to nothing. As a result, the following code no longer
compiles:

```rust
macro_rules! empty {
    () => { }
}

fn foo() -> bool { //~ ERROR mismatched
    { true } //~ ERROR mismatched
    empty!();
}
```

Previously, `{ true }` would be considered the trailing expression, even
though there's a semicolon in `empty!();`

This makes macro expansion more token-based.

3 years agoTest bootstrapping of rustc using cg_clif
bjorn3 [Sun, 1 Nov 2020 15:57:14 +0000 (16:57 +0100)]
Test bootstrapping of rustc using cg_clif

3 years agoRevert unintentional change
bjorn3 [Mon, 2 Nov 2020 17:54:10 +0000 (18:54 +0100)]
Revert unintentional change

3 years agoMerge pull request #1104 from bjorn3/build_system_refactor
bjorn3 [Mon, 2 Nov 2020 17:46:36 +0000 (18:46 +0100)]
Merge pull request #1104 from bjorn3/build_system_refactor

Build system refactor

3 years agoUpload prebuilt cg_clif
bjorn3 [Mon, 2 Nov 2020 17:24:21 +0000 (18:24 +0100)]
Upload prebuilt cg_clif

3 years agoRefactor the build system
bjorn3 [Mon, 2 Nov 2020 17:16:57 +0000 (18:16 +0100)]
Refactor the build system

3 years agoAuto merge of #78607 - HeroicKatora:post-order-walk-iter, r=davidtwco
bors [Mon, 2 Nov 2020 16:01:10 +0000 (16:01 +0000)]
Auto merge of #78607 - HeroicKatora:post-order-walk-iter, r=davidtwco

Transform post order walk to an iterative approach

The previous recursive approach might overflow the stack when walking a
particularly deep, list-like, graph. In particular, dominator
calculation for borrow checking does such a traversal and very long
functions might lead to a region dependency graph with in this
problematic structure.

This addresses what appears to be the cause of #78567 (`@SunHao-0` thanks for the stack trace).

3 years agoAuto merge of #78667 - pietroalbini:lldb-unbuffered, r=Mark-Simulacrum
bors [Mon, 2 Nov 2020 13:07:01 +0000 (13:07 +0000)]
Auto merge of #78667 - pietroalbini:lldb-unbuffered, r=Mark-Simulacrum

Try running lldb_batchmode.py with PYTHONUNBUFFERED

When reporting fatal errors, LLVM calls abort() to exit the program. There is a chance that might interfere with Python printing stuff to stdout, as by default it relies on buffering to increase performance.

This commit tries to disable Python buffering, to hopefully get useful logs while debugging #78665.

3 years agoImprove deprecation attribute diagnostic messages.
Mara Bos [Mon, 2 Nov 2020 12:17:14 +0000 (13:17 +0100)]
Improve deprecation attribute diagnostic messages.

(From the PR feedback.)

Co-authored-by: Esteban Küber <esteban@kuber.com.ar>
3 years agocompiletest: try running lldb_batchmode.py with PYTHONUNBUFFERED
Pietro Albini [Mon, 2 Nov 2020 12:20:16 +0000 (13:20 +0100)]
compiletest: try running lldb_batchmode.py with PYTHONUNBUFFERED

When reporting fatal errors, LLVM calls abort() to exit the program.
There is a chance that might interfere with Python printing stuff to
stdout, as by default it relies on buffering to increase performance.

This commit tries to disable Python buffering, to hopefully get useful
logs while debugging #78665.

3 years agoAuto merge of #78661 - JohnTitor:rollup-er2isja, r=JohnTitor
bors [Mon, 2 Nov 2020 10:42:45 +0000 (10:42 +0000)]
Auto merge of #78661 - JohnTitor:rollup-er2isja, r=JohnTitor

Rollup of 5 pull requests

Successful merges:

 - #78606 (Clarify handling of final line ending in str::lines())
 - #78610 (Do not remove tokens before AST json serialization)
 - #78620 (Trivial fixes to bitwise operator documentation)
 - #78627 (Point out that total_cmp is no strict superset of partial comparison)
 - #78637 (Add fetch_update methods to AtomicBool and AtomicPtr)

Failed merges:

r? `@ghost`

3 years agoAuto merge of #75020 - JohnTitor:fix-multispan, r=estebank,tmandry
bors [Mon, 2 Nov 2020 08:17:36 +0000 (08:17 +0000)]
Auto merge of #75020 - JohnTitor:fix-multispan, r=estebank,tmandry

Avoid complex diagnostics in snippets which contain newlines

Fixes #70935

r? `@estebank` `@tmandry`

3 years agoFix intrinsic size_of stable link
Ivan Tham [Mon, 2 Nov 2020 08:03:23 +0000 (16:03 +0800)]
Fix intrinsic size_of stable link

I noticed that it is pointing to the same link when I was reading
https://github.com/rust-lang/rust-clippy/issues/2997

3 years agoFix format
Yuki Okushi [Mon, 2 Nov 2020 07:59:11 +0000 (16:59 +0900)]
Fix format

3 years agoOnly separate notes if span is multiline
Yuki Okushi [Mon, 2 Nov 2020 07:54:53 +0000 (16:54 +0900)]
Only separate notes if span is multiline

3 years agoAdd "this has type `{}` which {}" note
Yuki Okushi [Thu, 22 Oct 2020 05:15:40 +0000 (14:15 +0900)]
Add "this has type `{}` which {}" note

3 years agoAddress some code reviews
Yuki Okushi [Sat, 10 Oct 2020 19:48:17 +0000 (04:48 +0900)]
Address some code reviews

3 years agoSeparate complex multispan into some notes
Yuki Okushi [Sat, 10 Oct 2020 18:05:57 +0000 (03:05 +0900)]
Separate complex multispan into some notes

3 years agoAvoid complex diagnostics in snippets which contain newlines
Yuki Okushi [Sat, 1 Aug 2020 16:29:21 +0000 (01:29 +0900)]
Avoid complex diagnostics in snippets which contain newlines

3 years agoFix ICE when a future-incompat-report has its command-line level capped
Aaron Hill [Mon, 2 Nov 2020 06:37:26 +0000 (01:37 -0500)]
Fix ICE when a future-incompat-report has its command-line level capped

Fixes #78660

With PR https://github.com/rust-lang/rust/pull/75534 merged, we now run
more lint-related code for future-incompat-report, even when their final
level is Allow. Some lint-related code was not expecting `Level::Allow`,
and had an explicit panic.

This PR explicitly tracks the lint level set on the command line before
`--cap-lints` is applied. This is used to emit a more precise error
note (e.g. we don't say that `-W lint-name` was specified on the
command line just because a lint was capped to Warn). As a result, we
can now correctly emit a note that `-A` was used if we got
`Level::Allow` from the command line (before the cap is applied).

3 years agoRollup merge of #78637 - mystor:atomic_ptr_bool, r=m-ou-se
Yuki Okushi [Mon, 2 Nov 2020 05:14:41 +0000 (14:14 +0900)]
Rollup merge of #78637 - mystor:atomic_ptr_bool, r=m-ou-se

Add fetch_update methods to AtomicBool and AtomicPtr

These methods were stabilized for the integer atomics in #71843, but the methods were not added for the non-integer atomics `AtomicBool` and `AtomicPtr`.

3 years agoRollup merge of #78627 - est31:total_cmp_no_superset, r=m-ou-se
Yuki Okushi [Mon, 2 Nov 2020 05:14:38 +0000 (14:14 +0900)]
Rollup merge of #78627 - est31:total_cmp_no_superset, r=m-ou-se

Point out that total_cmp is no strict superset of partial comparison

Partial comparison and total_cmp are not equal. This helps
preventing the mistake of creating float wrappers that
base their Ord impl on total_cmp and their PartialOrd impl on
the PartialOrd impl of the float type. PartialOrd and Ord
[are required to agree with each other](https://doc.rust-lang.org/std/cmp/trait.Ord.html#how-can-i-implement-ord).

3 years agoRollup merge of #78620 - gabhijit:bitops_doc_fix, r=m-ou-se
Yuki Okushi [Mon, 2 Nov 2020 05:14:36 +0000 (14:14 +0900)]
Rollup merge of #78620 - gabhijit:bitops_doc_fix, r=m-ou-se

Trivial fixes to bitwise operator documentation

Added fixes to documentation of `BitAnd`, `BitOr`, `BitXor` and
`BitAndAssign`, where the documentation for implementation on
`Vector<bool>` was using logical operators in place of the bitwise
operators.

r? @steveklabnik
Closes #78619

3 years agoRollup merge of #78610 - petrochenkov:nostriptok, r=Aaron1011
Yuki Okushi [Mon, 2 Nov 2020 05:14:34 +0000 (14:14 +0900)]
Rollup merge of #78610 - petrochenkov:nostriptok, r=Aaron1011

Do not remove tokens before AST json serialization

`TokenStripper` is error-prone and introduces one more use of `MutVisitor`.
It's much simpler to treat serialization as just one more place that wants lazy token stream to turn into a real token stream.
Also, no code is better than more code, in general.
r? @Aaron1011

(I also merged tests for `TokenStripper` ICEs into one.)

3 years agoRollup merge of #78606 - autarch:patch-1, r=m-ou-se
Yuki Okushi [Mon, 2 Nov 2020 05:14:33 +0000 (14:14 +0900)]
Rollup merge of #78606 - autarch:patch-1, r=m-ou-se

Clarify handling of final line ending in str::lines()

I found the description as it stands a bit confusing. I've added a bit more explanation to make it clear that a trailing line ending does not produce a final empty line.

3 years agoAuto merge of #78592 - fpoli:nll-facts-dir, r=matthewjasper
bors [Mon, 2 Nov 2020 04:39:05 +0000 (04:39 +0000)]
Auto merge of #78592 - fpoli:nll-facts-dir, r=matthewjasper

Add option to customize the nll-facts' folder location

This PR adds a `nll-facts-dir` option to specify the location of the directory in which NLL facts are dumped into. It works the same way `dump-mir-dir` controls the location used by the `dump-mir` option.

3 years agoDocument -Zinstrument-coverage
Rich Kadel [Sat, 31 Oct 2020 23:39:09 +0000 (16:39 -0700)]
Document -Zinstrument-coverage

3 years agoCorrected suggestion for generic parameters in `function_item_references` lint
Ayrton [Mon, 2 Nov 2020 03:14:21 +0000 (23:14 -0400)]
Corrected suggestion for generic parameters in `function_item_references` lint

This lint was incorrectly suggesting casting a function to a pointer without
specifying generic type parameters or const generics. This would cause a
compiler error since the missing parameters couldn't be inferred. This commit
fixed the suggestion and added a few tests with generics.

3 years agoAuto merge of #78645 - JulianKnodt:i78622, r=estebank
bors [Mon, 2 Nov 2020 02:18:36 +0000 (02:18 +0000)]
Auto merge of #78645 - JulianKnodt:i78622, r=estebank

Add delay_span_bug to no longer ICE

Fixes #78622

r? `@matthewjasper`

3 years agoAuto merge of #78605 - nox:relax-elf-relocations, r=nagisa
bors [Mon, 2 Nov 2020 00:12:32 +0000 (00:12 +0000)]
Auto merge of #78605 - nox:relax-elf-relocations, r=nagisa

Implement -Z relax-elf-relocations=yes|no

This lets rustc users tweak whether the linker should relax ELF relocations without recompiling a whole new target with its own libcore etc.

3 years agoinliner: Remove redundant loop
Tomasz Miąsko [Mon, 2 Nov 2020 00:00:00 +0000 (00:00 +0000)]
inliner: Remove redundant loop

No functional changes intended.

3 years agoAdd delay_span_bug to no longer ICE
kadmin [Sun, 1 Nov 2020 22:31:19 +0000 (22:31 +0000)]
Add delay_span_bug to no longer ICE

3 years agoImprove code in unindent_comment a bit more
Guillaume Gomez [Sun, 1 Nov 2020 15:40:48 +0000 (16:40 +0100)]
Improve code in unindent_comment a bit more

3 years agoAdd test for #[deprecated] attribute on trait impl block.
Mara Bos [Sun, 1 Nov 2020 12:09:03 +0000 (13:09 +0100)]
Add test for #[deprecated] attribute on trait impl block.

3 years agoTurn 'useless #[deprecated]' error into a lint.
Mara Bos [Sun, 1 Nov 2020 11:58:47 +0000 (12:58 +0100)]
Turn 'useless #[deprecated]' error into a lint.

3 years agoUse the right span for errors about #[deprecated] attributes.
Mara Bos [Sun, 1 Nov 2020 11:29:57 +0000 (12:29 +0100)]
Use the right span for errors about #[deprecated] attributes.

3 years agoDeny #[deprecated] on trait impl blocks.
Mara Bos [Sat, 31 Oct 2020 14:34:10 +0000 (15:34 +0100)]
Deny #[deprecated] on trait impl blocks.

They have no effect there, but were silently accepted.

3 years agoAdd fetch_update methods to AtomicBool and AtomicPtr
Nika Layzell [Sun, 1 Nov 2020 18:24:22 +0000 (13:24 -0500)]
Add fetch_update methods to AtomicBool and AtomicPtr

These methods were stabilized for the integer atomics in #71843, but the methods
were not added for the non-integer atomics `AtomicBool` and `AtomicPtr`.

3 years agoUpdate build instructions
bjorn3 [Sun, 1 Nov 2020 18:51:35 +0000 (19:51 +0100)]
Update build instructions

3 years agoMake it easier to use build_sysroot.sh
bjorn3 [Sun, 1 Nov 2020 18:47:14 +0000 (19:47 +0100)]
Make it easier to use build_sysroot.sh

3 years agoSplit the actual tests out into scripts/tests.sh
bjorn3 [Sun, 1 Nov 2020 18:39:44 +0000 (19:39 +0100)]
Split the actual tests out into scripts/tests.sh

3 years agoEnsure that sysroot build works with CARGO_TARGET_DIR set
bjorn3 [Sun, 1 Nov 2020 18:38:21 +0000 (19:38 +0100)]
Ensure that sysroot build works with CARGO_TARGET_DIR set

3 years agoPoint out that total_cmp is no strict superset of partial comparison
est31 [Sun, 1 Nov 2020 13:27:36 +0000 (14:27 +0100)]
Point out that total_cmp is no strict superset of partial comparison

Partial comparison and total_cmp are not equal. This helps
preventing the mistake of creating float wrappers that
base their Ord impl on total_cmp and their PartialOrd impl on
the PartialOrd impl of the float type. PartialOrd and Ord
are required to agree with each other.

3 years agoFix transmutes between vectors and integers
bjorn3 [Sun, 1 Nov 2020 17:35:19 +0000 (18:35 +0100)]
Fix transmutes between vectors and integers

Fixes #1102

3 years agoAuto merge of #75534 - Aaron1011:feature/new-future-breakage, r=pnkfelix
bors [Sun, 1 Nov 2020 16:52:28 +0000 (16:52 +0000)]
Auto merge of #75534 - Aaron1011:feature/new-future-breakage, r=pnkfelix

Implement rustc side of report-future-incompat

cc https://github.com/rust-lang/rust/issues/71249

This is an alternative to `@pnkfelix's` initial implementation in https://github.com/pnkfelix/rust/commits/prototype-rustc-side-of-report-future-incompat (mainly because I started working before seeing that branch :smile: ).

My approach outputs the entire original `Diagnostic`, in a way that is compatible with incremental compilation. This is not yet integrated with compiletest, but can be used manually by passing `-Z emit-future-incompat-report` to `rustc`.

Several changes are made to support this feature:
* The `librustc_session/lint` module is moved to a new crate `librustc_lint_defs` (name bikesheddable). This allows accessing lint definitions from `librustc_errors`.
* The `Lint` struct is extended with an `Option<FutureBreakage>`. When present, it indicates that we should display a lint in the future-compat report. `FutureBreakage` contains additional information that we may want to display in the report (currently, a `date` field indicating when the crate will stop compiling).
* A new variant `rustc_error::Level::Allow` is added. This is used when constructing a diagnostic for a future-breakage lint that is marked as allowed (via `#[allow]` or `--cap-lints`). This allows us to capture any future-breakage diagnostics in one place, while still discarding them before they are passed to the `Emitter`.
* `DiagnosticId::Lint` is extended with a `has_future_breakage` field, indicating whether or not the `Lint` has future breakage information (and should therefore show up in the report).
* `Session` is given access to the `LintStore` via a new `SessionLintStore` trait (since `librustc_session` cannot directly reference `LintStore` without a cyclic dependency). We use this to turn a string `DiagnosticId::Lint` back into a `Lint`, to retrieve the `FutureBreakage` data.

Currently, `FutureBreakage.date` is always set to `None`. However, this could potentially be interpreted by Cargo in the future.

I've enabled the future-breakage report for the `ARRAY_INTO_ITER` lint, which can be used to test out this PR. The intent is to use the field to allow Cargo to determine the date of future breakage (as described in [RFC 2834](https://github.com/rust-lang/rfcs/blob/master/text/2834-cargo-report-future-incompat.md)) without needing to parse the diagnostic itself.

cc `@pnkfelix`

3 years agoRemove incorrect statement about line ending content in lines doc change
Dave Rolsky [Sun, 1 Nov 2020 15:11:20 +0000 (09:11 -0600)]
Remove incorrect statement about line ending content in lines doc change

3 years agoAuto merge of #78553 - Nadrieril:fix-78549, r=varkor
bors [Sun, 1 Nov 2020 14:37:50 +0000 (14:37 +0000)]
Auto merge of #78553 - Nadrieril:fix-78549, r=varkor

Fix #78549

Before #78430, this worked because `specialize_constructor` didn't actually care too much which constructor was passed to it unless needed. That PR however handles `&str` as a special case, and I did not anticipate patterns for the `&str` type other than string literals.
I am not very confident there are not other similar oversights left, but hopefully only `&str` was different enough to break my assumptions.

Fixes https://github.com/rust-lang/rust/issues/78549

3 years agoHide anonymous allocations from linked artifact
bjorn3 [Sun, 1 Nov 2020 13:45:41 +0000 (14:45 +0100)]
Hide anonymous allocations from linked artifact

3 years agoMerge pull request #1100 from mominul/trans->codegen
bjorn3 [Sun, 1 Nov 2020 13:44:25 +0000 (14:44 +0100)]
Merge pull request #1100 from mominul/trans->codegen

Rename trans to codegen

3 years agoRename trans to codegen
Muhammad Mominul Huque [Sun, 1 Nov 2020 13:24:30 +0000 (19:24 +0600)]
Rename trans to codegen

3 years agodocumentation examples fixes in rustfmt convention
Abhijit Gadgil [Sun, 1 Nov 2020 13:23:22 +0000 (18:53 +0530)]
documentation examples fixes in rustfmt convention

3 years agoAuto merge of #78623 - m-ou-se:rollup-m6y5j0m, r=m-ou-se
bors [Sun, 1 Nov 2020 11:54:58 +0000 (11:54 +0000)]
Auto merge of #78623 - m-ou-se:rollup-m6y5j0m, r=m-ou-se

Rollup of 6 pull requests

Successful merges:

 - #78073 (Add #[inline] to some functions in core::str.)
 - #78596 (Fix doc links to std::fmt)
 - #78599 (Add note to process::arg[s] that args shouldn't be escaped or quoted)
 - #78602 (fix various aliasing issues in the standard library)
 - #78603 (expand: Tweak a comment in implementation of `macro_rules`)
 - #78621 (Inline Default::default() for atomics)

Failed merges:

r? `@ghost`

3 years agoRollup merge of #78621 - solson:inline, r=m-ou-se
Mara Bos [Sun, 1 Nov 2020 10:53:39 +0000 (11:53 +0100)]
Rollup merge of #78621 - solson:inline, r=m-ou-se

Inline Default::default() for atomics

Functions like `AtomicUsize::default()` are not cross-crate inlineable before this PR ([see assembly output here](https://play.rust-lang.org/?version=stable&mode=release&edition=2018&gist=e353321766418f759c69fb141d3732f8)), which can lead to unexpected performance issues when initializing a large array using this function, e.g. as seen [here](https://github.com/spacejam/sled/blob/d513996a85875be8c813fd0e30a548b89682289a/src/histogram.rs#L53) which should turn into a simple loop writing zeroes but doesn't.

r? @m-ou-se

3 years agoRollup merge of #78603 - petrochenkov:fourdigits, r=matthewjasper
Mara Bos [Sun, 1 Nov 2020 10:53:37 +0000 (11:53 +0100)]
Rollup merge of #78603 - petrochenkov:fourdigits, r=matthewjasper

expand: Tweak a comment in implementation of `macro_rules`

The answer to the removed FIXME is that we don't apply mark to the span `sp` just because that span is no longer used. We could apply it, but that would just be unnecessary extra work.

The comments in code tell why the span is unused, it's a span of `$var` literally, which is lost for `tt` variables because their tokens are outputted directly, but kept for other variables which are outputted as [groups](https://doc.rust-lang.org/nightly/proc_macro/struct.Group.html) and `sp` is kept as the group's span.

Closes https://github.com/rust-lang/rust/issues/2887

3 years agoRollup merge of #78602 - RalfJung:raw-ptr-aliasing-issues, r=m-ou-se
Mara Bos [Sun, 1 Nov 2020 10:53:36 +0000 (11:53 +0100)]
Rollup merge of #78602 - RalfJung:raw-ptr-aliasing-issues, r=m-ou-se

fix various aliasing issues in the standard library

This fixes various cases where the standard library either used raw pointers after they were already invalidated by using the original reference again, or created raw pointers for one element of a slice and used it to access neighboring elements.

3 years agoRollup merge of #78599 - panstromek:master, r=m-ou-se
Mara Bos [Sun, 1 Nov 2020 10:53:34 +0000 (11:53 +0100)]
Rollup merge of #78599 - panstromek:master, r=m-ou-se

Add note to process::arg[s] that args shouldn't be escaped or quoted

This came out of discussion on [forum](https://users.rust-lang.org/t/how-to-get-full-output-from-command/50626), where I recently asked a question and it turned out that the problem was redundant quotation:

```rust
 Command::new("rg")
        .arg("\"pattern\"") // this will look for "pattern" with quotes included
```

This is something that has bitten me few times already (in multiple languages actually), so It'd be grateful to have it in the docs, even though it's not sctrictly Rust specific problem. Other users also agreed.

This can be really annoying to debug, because in many cases (inluding mine), quotes can be legal part of the argument, so the command doesn't fail, it just behaves unexpectedly. Not everybody (including me) knows that quotes around arguments are part of the shell and not part of the called program. Coincidentally, somoene had the same problem [yesterday](https://www.reddit.com/r/rust/comments/jkxelc/going_crazy_over_running_a_curl_process_from_rust/) on reddit.

I am not a native speaker, so I welcome any corrections or better formulation, I don't expect this to be merged as is. I was also reminded that this is platform/shell specific behaviour, but I didn't find a good way to formulate that briefly, any ideas welcome.

 It's also my first PR here, so I am not sure I did everything correctly, I did this just from Github UI.

3 years agoRollup merge of #78596 - pavlukivan:master, r=m-ou-se
Mara Bos [Sun, 1 Nov 2020 10:53:33 +0000 (11:53 +0100)]
Rollup merge of #78596 - pavlukivan:master, r=m-ou-se

Fix doc links to std::fmt

`std::format` and `core::write` macros' docs linked to `core::fmt` for format string reference, even though only `std::fmt` has format string documentation (and the link titles were `std::fmt`)

3 years agoRollup merge of #78073 - fusion-engineering-forks:inline, r=eddyb
Mara Bos [Sun, 1 Nov 2020 10:53:29 +0000 (11:53 +0100)]
Rollup merge of #78073 - fusion-engineering-forks:inline, r=eddyb

Add #[inline] to some functions in core::str.

Almost all str functions already had #[inline].

3 years agoAuto merge of #78531 - cuviper:unwrap-metadata, r=tmandry
bors [Sun, 1 Nov 2020 09:30:11 +0000 (09:30 +0000)]
Auto merge of #78531 - cuviper:unwrap-metadata, r=tmandry

rustc_llvm: unwrap LLVMMetadataRef before casting

Directly casting the opaque pointer was [reported] to cause an
"incomplete type" error with GCC 9.3:

```
llvm-wrapper/RustWrapper.cpp:939:31:   required from here
/usr/include/c++/9.3/type_traits:1301:12: error: invalid use of incomplete type 'struct LLVMOpaqueMetadata'
 1301 |     struct is_base_of
      |            ^~~~~~~~~~
In file included from [...]/rust/src/llvm-project/llvm/include/llvm-c/BitReader.h:23,
                 from llvm-wrapper/LLVMWrapper.h:1,
                 from llvm-wrapper/RustWrapper.cpp:1:
[...]/rust/src/llvm-project/llvm/include/llvm-c/Types.h:89:16: note: forward declaration of 'struct LLVMOpaqueMetadata'
   89 | typedef struct LLVMOpaqueMetadata *LLVMMetadataRef;
      |                ^~~~~~~~~~~~~~~~~~
```

[reported]: https://zulip-archive.rust-lang.org/182449tcompilerhelp/12215halprustcllvmbuildfail.html#214915124

A simple `unwrap` fixes the issue.

r? `@eddyb`

3 years agoMerge pull request #1099 from bjorn3/fix_rustc_bootstrap
bjorn3 [Sun, 1 Nov 2020 09:10:23 +0000 (10:10 +0100)]
Merge pull request #1099 from bjorn3/fix_rustc_bootstrap

Update Cranelift

3 years agoUpdate Cranelift
bjorn3 [Sun, 1 Nov 2020 08:50:33 +0000 (09:50 +0100)]
Update Cranelift

Fixes bootstrapping of rustc using cg_clif

Fixes #1097

3 years agoAuto merge of #78420 - estebank:suggest-assoc-fn, r=petrochenkov
bors [Sun, 1 Nov 2020 06:49:16 +0000 (06:49 +0000)]
Auto merge of #78420 - estebank:suggest-assoc-fn, r=petrochenkov

Suggest calling associated `fn` inside `trait`s

When calling a function that doesn't exist inside of a trait's
associated `fn`, and another associated `fn` in that trait has that
name, suggest calling it with the appropriate fully-qualified path.

Expand the label to be more descriptive.

Prompted by the following user experience:
https://users.rust-lang.org/t/cannot-find-function/50663

3 years agoInline Default::default() for atomics
Scott Olson [Sun, 1 Nov 2020 04:38:41 +0000 (04:38 +0000)]
Inline Default::default() for atomics

3 years agoFixes incorrect paranthesis.
Abhijit Gadgil [Sun, 1 Nov 2020 03:38:19 +0000 (09:08 +0530)]
Fixes incorrect paranthesis.