]> git.lizzy.rs Git - rust.git/log
rust.git
4 years agoRollup merge of #68858 - ljedrz:collapse_stable_hash_foos, r=michaelwoerister
Dylan DPC [Wed, 5 Feb 2020 12:14:37 +0000 (13:14 +0100)]
Rollup merge of #68858 - ljedrz:collapse_stable_hash_foos, r=michaelwoerister

Merge item id stable hashing functions

Supersedes https://github.com/rust-lang/rust/pull/67999 splitting out the pure cleanup bits, i.e. merging `hash_item_id`, `hash_impl_item_id` and `hash_trait_item_id` into a common `hash_reference_to_item`.

r? @michaelwoerister

4 years agoRollup merge of #68851 - JohnTitor:fix-issue-number, r=Centril
Dylan DPC [Wed, 5 Feb 2020 12:14:34 +0000 (13:14 +0100)]
Rollup merge of #68851 - JohnTitor:fix-issue-number, r=Centril

Fix issue number of `capacity` method

Follow-up of https://github.com/rust-lang/rust/pull/68558#issuecomment-582117131

r? @alexcrichton

4 years agoRollup merge of #68846 - king6cong:doc-fix, r=GuillaumeGomez
Dylan DPC [Wed, 5 Feb 2020 12:14:33 +0000 (13:14 +0100)]
Rollup merge of #68846 - king6cong:doc-fix, r=GuillaumeGomez

doc fix on doc attribute

None

4 years agoRollup merge of #68840 - Centril:rec-lim-curr-crate, r=estebank
Dylan DPC [Wed, 5 Feb 2020 12:14:31 +0000 (13:14 +0100)]
Rollup merge of #68840 - Centril:rec-lim-curr-crate, r=estebank

On suggesting `#![recursion_limit = "X"]`, note current crate name

This would have saved me much confusion e.g. when reading the log output in https://github.com/rust-lang/rust/pull/68788#issuecomment-581852191.

r? @estebank

4 years agoRollup merge of #68832 - GuillaumeGomez:clean-up-3-err-codes, r=estebank
Dylan DPC [Wed, 5 Feb 2020 12:14:30 +0000 (13:14 +0100)]
Rollup merge of #68832 - GuillaumeGomez:clean-up-3-err-codes, r=estebank

Clean up E0264, E0267 and E0268 explanations

r? @Dylan-DPC

4 years agoRollup merge of #68809 - ecstatic-morse:const-int-functions, r=oli-obk
Dylan DPC [Wed, 5 Feb 2020 12:14:28 +0000 (13:14 +0100)]
Rollup merge of #68809 - ecstatic-morse:const-int-functions, r=oli-obk

Make more arithmetic functions unstably const

This is a smaller version of #66884 (thanks @9999years) that constifies many of the arithmetic functions on integer primitives from #53718 that were blocked on #49146.

This makes the following things unstably const.

- `feature = const_int_unchecked_arith`
  - `intrinsics::unchecked_add`
  - `intrinsics::unchecked_sub`
  - `intrinsics::unchecked_mul`
  - `intrinsics::unchecked_div`
  - `intrinsics::unchecked_rem`

- `feature = const_checked_int_methods`
  - `checked_add`
  - `checked_sub`
  - `checked_mul`
  - `checked_div` (Uses `intrinsics::unchecked_div` internally)
  - `checked_rem` (Uses `intrinsics::unchecked_rem` internally)
  - `checked_neg`
  - `checked_shl`
  - `checked_shr`
  - `checked_abs`

- `feature = const_saturating_int_methods`
  - `saturating_mul`
  - `saturating_neg`  (Uses `intrinsics::unchecked_sub` internally)
  - `saturating_abs` (Uses `intrinsics::unchecked_sub` internally)

- `feature = const_wrapping_int_methods`
  - `wrapping_div`
  - `wrapping_rem`

- `feature = const_overflowing_int_methods`
  - `overflowing_div`
  - `overflowing_rem`

- `feature = const_euclidean_int_methods`
  - `checked_div_euclid`
  - `checked_rem_euclid`
  - `wrapping_div_euclid`
  - `wrapping_rem_euclid`
  - `overflowing_div_euclid`
  - `overflowing_rem_euclid`

Exponentiation and operations on the `NonZero` types are left to a later PR.

r? @oli-obk
cc @rust-lang/wg-const-eval @rust-lang/libs

4 years agoRollup merge of #68790 - nnethercote:improve-merge_from_succ, r=nikomatsakis
Dylan DPC [Wed, 5 Feb 2020 12:14:25 +0000 (13:14 +0100)]
Rollup merge of #68790 - nnethercote:improve-merge_from_succ, r=nikomatsakis

Improve `merge_from_succ`

A couple of small performance wins.

r? @nikomatsakis

4 years agoRollup merge of #68762 - ForNeVeR:patch-1, r=alexcrichton
Dylan DPC [Wed, 5 Feb 2020 12:14:22 +0000 (13:14 +0100)]
Rollup merge of #68762 - ForNeVeR:patch-1, r=alexcrichton

Strip unnecessary subexpression

It became unnecessary since a06baa56b95674fc626b3c3fd680d6a65357fe60 reformatted the file. The comment is currently a bit misleading.

4 years agomove item reference comment
ljedrz [Wed, 5 Feb 2020 10:29:07 +0000 (11:29 +0100)]
move item reference comment

4 years agomerge item id stable hashing functions
ljedrz [Wed, 5 Feb 2020 10:11:34 +0000 (11:11 +0100)]
merge item id stable hashing functions

4 years agoFix issue number of `capacity` method
Yuki Okushi [Wed, 5 Feb 2020 06:34:33 +0000 (15:34 +0900)]
Fix issue number of `capacity` method

4 years agodoc fix on doc attribute
king6cong [Wed, 5 Feb 2020 04:56:24 +0000 (12:56 +0800)]
doc fix on doc attribute

4 years agoImplement remaining `unchecked` arithmetic intrinsics
Dylan MacKenzie [Wed, 5 Feb 2020 01:09:22 +0000 (17:09 -0800)]
Implement remaining `unchecked` arithmetic intrinsics

4 years agoFix test
Dylan MacKenzie [Tue, 4 Feb 2020 19:46:13 +0000 (11:46 -0800)]
Fix test

4 years agoUse consistent feature naming
Dylan MacKenzie [Tue, 4 Feb 2020 19:46:03 +0000 (11:46 -0800)]
Use consistent feature naming

4 years agoAuto merge of #68831 - Dylan-DPC:rollup-j6x15y9, r=Dylan-DPC
bors [Wed, 5 Feb 2020 03:20:43 +0000 (03:20 +0000)]
Auto merge of #68831 - Dylan-DPC:rollup-j6x15y9, r=Dylan-DPC

Rollup of 7 pull requests

Successful merges:

 - #68282 (Instrument C / C++ in MemorySanitizer example)
 - #68758 (Fix 59191 - ICE when macro replaces crate root with non-module item)
 - #68805 (bootstrap: fix clippy warnings)
 - #68810 (Remove Copy impl from OnceWith)
 - #68815 (remove redundant imports (clippy::single_component_path_imports))
 - #68818 (fix couple of perf related clippy warnings)
 - #68819 (Suggest `split_at_mut` on multiple mutable index access)

Failed merges:

r? @ghost

4 years ago`#![recursion_limit = "X"]`: note current crate name.
Mazdak Farrokhzad [Wed, 5 Feb 2020 01:57:30 +0000 (02:57 +0100)]
`#![recursion_limit = "X"]`: note current crate name.

4 years agoAuto merge of #68544 - Aaron1011:remove-overlapping-traits, r=estebank
bors [Tue, 4 Feb 2020 23:56:49 +0000 (23:56 +0000)]
Auto merge of #68544 - Aaron1011:remove-overlapping-traits, r=estebank

Remove the `overlapping_marker_traits` feature

See #29864

This has been replaced by `#[feature(marker_trait_attr)]`

A few notes:

* Due to PR #68057 not yet being in the bootstrap compiler, it's
  necessary to continue using `#![feature(overlapping_marker_traits)]`
  under `#[cfg(bootstrap)]` to work around type inference issues.
* I've updated tests that used `overlapping_marker_traits` to now use
  `marker_trait_attr` where applicable

The test `src/test/ui/overlap-marker-trait.rs` doesn't make any sense
now that `overlapping_marker_traits`, so I removed it.

The test `src/test/ui/traits/overlap-permitted-for-marker-traits-neg.rs`
now fails, since it's no longer possible to have multiple overlapping
negative impls of `Send`. I believe that this is the behavior we want
(assuming that `Send` is not going to become a `#[marker]` trait, so I
renamed the test to `overlap-permitted-for-marker-traits-neg`

4 years agoClean up E0264, E0267 and E0268 explanations
Guillaume Gomez [Tue, 4 Feb 2020 21:03:54 +0000 (22:03 +0100)]
Clean up E0264, E0267 and E0268 explanations

4 years agoRollup merge of #68819 - estebank:split_at_mut, r=oli-obk
Dylan DPC [Tue, 4 Feb 2020 20:51:58 +0000 (21:51 +0100)]
Rollup merge of #68819 - estebank:split_at_mut, r=oli-obk

Suggest `split_at_mut` on multiple mutable index access

cc #58792.

4 years agoRollup merge of #68818 - matthiaskrgr:misc_perf, r=Mark-Simulacrum
Dylan DPC [Tue, 4 Feb 2020 20:51:56 +0000 (21:51 +0100)]
Rollup merge of #68818 - matthiaskrgr:misc_perf, r=Mark-Simulacrum

fix couple of perf related clippy warnings

librustc: don't clone a type that is copy
librustc_incremental: use faster vector initialization
librustc_typeck: don't clone a type that is copy
librustdoc: don't create a vector where a slice will do

4 years agoRollup merge of #68815 - matthiaskrgr:redundant_imports, r=alexcrichton
Dylan DPC [Tue, 4 Feb 2020 20:51:55 +0000 (21:51 +0100)]
Rollup merge of #68815 - matthiaskrgr:redundant_imports, r=alexcrichton

remove redundant imports (clippy::single_component_path_imports)

4 years agoRollup merge of #68810 - ollie27:once_with_copy, r=Dylan-DPC
Dylan DPC [Tue, 4 Feb 2020 20:51:53 +0000 (21:51 +0100)]
Rollup merge of #68810 - ollie27:once_with_copy, r=Dylan-DPC

Remove Copy impl from OnceWith

Iterators typically don't implement `Copy` and this shouldn't be an exception.

4 years agoRollup merge of #68805 - matthiaskrgr:cleanup_bootstrap, r=Mark-Simulacrum
Dylan DPC [Tue, 4 Feb 2020 20:51:52 +0000 (21:51 +0100)]
Rollup merge of #68805 - matthiaskrgr:cleanup_bootstrap, r=Mark-Simulacrum

bootstrap: fix clippy warnings

r? @oli-obk

4 years agoRollup merge of #68758 - daboross:fix-59191, r=petrochenkov
Dylan DPC [Tue, 4 Feb 2020 20:51:50 +0000 (21:51 +0100)]
Rollup merge of #68758 - daboross:fix-59191, r=petrochenkov

Fix 59191 - ICE when macro replaces crate root with non-module item

Hi,

This should fix #59191! My friend and I are working on learning the rustc codebase through contributions, so please feel free to mention anything amiss or that could be done better.

The code adds an explicit case for when a macro applied to the crate root (via an inner attribute) replaces it with something nonsensical, like a function. The crate root must be a module, and the error message reflects this.

---

I should note that there are a few other weird edge cases here, like if they do output a module, it succeeds but uses that module's name as a prefix for all names in the crate. I'm assuming that's an issue for stabilizing #54726, though.

4 years agoRollup merge of #68282 - tmiasko:sanitizer-example, r=steveklabnik
Dylan DPC [Tue, 4 Feb 2020 20:51:48 +0000 (21:51 +0100)]
Rollup merge of #68282 - tmiasko:sanitizer-example, r=steveklabnik

Instrument C / C++ in MemorySanitizer example

Modify the example to instrument C / C++ in addition to Rust, since it
will be generally required (e.g., when using libbacktrace for symbolication).

4 years agoAuto merge of #68558 - HeroicKatora:buf-writer-capacity, r=alexcrichton
bors [Tue, 4 Feb 2020 20:30:53 +0000 (20:30 +0000)]
Auto merge of #68558 - HeroicKatora:buf-writer-capacity, r=alexcrichton

Add a method to query the capacity of a BufWriter and BufReader

Besides the obvious of retrieving the parameter used to construct the writer, this method allows consumers to control the number of `flush` calls during write operations. For `BufReader` it gives an upper bound on the returned buffer in `fill_buf` which might influence the allocation behaviour of a consumer.

4 years agoAdd tests for newly const arithmetic fns
Dylan MacKenzie [Mon, 3 Feb 2020 22:23:21 +0000 (14:23 -0800)]
Add tests for newly const arithmetic fns

Co-Authored-By: 9999years <rbt@sent.as>
4 years agoMake saturating arithmetic using intrinsics `const`
Dylan MacKenzie [Mon, 3 Feb 2020 21:17:13 +0000 (13:17 -0800)]
Make saturating arithmetic using intrinsics `const`

4 years agoMake checked division `const`
Dylan MacKenzie [Mon, 3 Feb 2020 21:19:32 +0000 (13:19 -0800)]
Make checked division `const`

4 years agoConst-stabilize some arithmetic intrinsics
Dylan MacKenzie [Mon, 3 Feb 2020 20:49:31 +0000 (12:49 -0800)]
Const-stabilize some arithmetic intrinsics

4 years agoMake wrapping arithmetic `const`
Dylan MacKenzie [Mon, 3 Feb 2020 20:53:34 +0000 (12:53 -0800)]
Make wrapping arithmetic `const`

Co-Authored-By: 9999years <rbt@sent.as>
4 years agoMake `saturating_mul` a `const fn`
Dylan MacKenzie [Mon, 3 Feb 2020 21:16:38 +0000 (13:16 -0800)]
Make `saturating_mul` a `const fn`

Co-Authored-By: 9999years <rbt@sent.as>
4 years agoMake overflowing arithmetic `const`
Dylan MacKenzie [Mon, 3 Feb 2020 20:45:51 +0000 (12:45 -0800)]
Make overflowing arithmetic `const`

Co-Authored-By: 9999years <rbt@sent.as>
4 years agoMake checked arithmetic besides division `const`
Dylan MacKenzie [Mon, 3 Feb 2020 21:19:07 +0000 (13:19 -0800)]
Make checked arithmetic besides division `const`

Co-Authored-By: 9999years <rbt@sent.as>
4 years agoMake euclidean division `const`
Dylan MacKenzie [Mon, 3 Feb 2020 20:31:39 +0000 (12:31 -0800)]
Make euclidean division `const`

Co-Authored-By: 9999years <rbt@sent.as>
4 years agoRemove the `overlapping_marker_traits` feature
Aaron Hill [Sat, 25 Jan 2020 20:30:19 +0000 (15:30 -0500)]
Remove the `overlapping_marker_traits` feature

See #29864

This has been replaced by `#[feature(marker_trait_attr)]`

A few notes:

* Due to PR #68057 not yet being in the bootstrap compiler, it's
  necessary to continue using `#![feature(overlapping_marker_traits)]`
  under `#[cfg(bootstrap)]` to work around type inference issues.
* I've updated tests that used `overlapping_marker_traits` to now use
  `marker_trait_attr` where applicable

The test `src/test/ui/overlap-marker-trait.rs` doesn't make any sense
now that `overlapping_marker_traits`, so I removed it.

The test `src/test/ui/traits/overlap-permitted-for-marker-traits-neg.rs`
now fails, since it's no longer possible to have multiple overlapping
negative impls of `Send`. I believe that this is the behavior we want
(assuming that `Send` is not going to become a `#[marker]` trait, so I
renamed the test to `overlap-permitted-for-marker-traits-neg`

4 years agoAuto merge of #68377 - estebank:fn-obligations-spans, r=oli-obk
bors [Tue, 4 Feb 2020 17:17:55 +0000 (17:17 +0000)]
Auto merge of #68377 - estebank:fn-obligations-spans, r=oli-obk

Tweak obligation error output

- Point at arguments or output when fn obligations come from them, or ident when they don't
- Point at `Sized` bound (fix #47990)
- When object unsafe trait uses itself in associated item suggest using `Self` (fix #66424, fix #33375, partially address #38376, cc #61525)
-  Point at reason in object unsafe trait with `Self` in supertraits or `where`-clause (cc #40533, cc #68377)
- On implicit type parameter `Sized` obligations, suggest `?Sized` (fix #57744, fix #46683)

4 years agoAuto merge of #68708 - Mark-Simulacrum:stage0-step, r=pietroalbini
bors [Tue, 4 Feb 2020 14:16:18 +0000 (14:16 +0000)]
Auto merge of #68708 - Mark-Simulacrum:stage0-step, r=pietroalbini

Step stage0 to bootstrap from 1.42

This also includes a commit which fixes the rustfmt downloading logic to redownload when the rustfmt channel changes, and bumps rustfmt to a more recent version.

4 years agoDrop unused extern crates
Mark Rousskov [Tue, 4 Feb 2020 13:14:08 +0000 (08:14 -0500)]
Drop unused extern crates

4 years agoAuto merge of #68804 - ecstatic-morse:qualif-cursor-lazy, r=estebank
bors [Tue, 4 Feb 2020 10:58:45 +0000 (10:58 +0000)]
Auto merge of #68804 - ecstatic-morse:qualif-cursor-lazy, r=estebank

Always use lazy qualif getters during const-checking

`has_mut_interior_eager_seek` was needed to work around an overly restrictive bound on the `per_local` argument to the `Qualif` trait. This PR makes that bound `FnMut` instead of `Fn` so we can seek cursors inside of it, resolving a FIXME in the const-checking code.

4 years agoAuto merge of #68601 - 0dvictor:split, r=tmandry
bors [Tue, 4 Feb 2020 05:48:54 +0000 (05:48 +0000)]
Auto merge of #68601 - 0dvictor:split, r=tmandry

Split `join_codegen_and_link()` into two steps

`join_codegen_and_link()` is split to `join_codegen()` and `link()`.

4 years agoSuggest `split_at_mut` on multiple mutable index access
Esteban Küber [Tue, 4 Feb 2020 01:58:28 +0000 (17:58 -0800)]
Suggest `split_at_mut` on multiple mutable index access

4 years agofix couple of perf related clipyp warnings
Matthias Krüger [Tue, 4 Feb 2020 01:28:11 +0000 (02:28 +0100)]
fix couple of perf related clipyp warnings

librustc: don't clone a type that is copy
librustc_incremental: use faster vector initialization
librustc_typeck: don't clone a type that is copy
librustdoc: don't create a vector where a slice will do

4 years agoSplit `join_codegen_and_link()` into two steps
Victor Ding [Tue, 28 Jan 2020 13:16:14 +0000 (00:16 +1100)]
Split `join_codegen_and_link()` into two steps

`join_codegen_and_link()` is split to `join_codegen()` and `link()`.

4 years agoremove redundant imports (clippy::single_component_path_imports)
Matthias Krüger [Tue, 4 Feb 2020 00:05:45 +0000 (01:05 +0100)]
remove redundant imports (clippy::single_component_path_imports)

4 years agoRemove Copy impl from OnceWith
Oliver Middleton [Mon, 3 Feb 2020 22:30:17 +0000 (22:30 +0000)]
Remove Copy impl from OnceWith

Iterators typically don't implement `Copy` and this shouldn't be an exception.

4 years agoAuto merge of #67668 - matthewjasper:or-patterns, r=pnkfelix
bors [Mon, 3 Feb 2020 22:02:26 +0000 (22:02 +0000)]
Auto merge of #67668 - matthewjasper:or-patterns, r=pnkfelix

Implement MIR lowering for or-patterns

This is the last thing needed to get meaningful run-pass tests for or-patterns. There probably need to be more tests before stabilizing this, but the most important cases should have been covered.

Note: we can generate exponentially large MIR CFGs when using or-patterns containing bindings, type ascriptions, or that are for a match arm with a guard. `src/test/mir-opt/exponential-or.rs` shows the best case for what we currently do.

cc #54883
closes #60350
closes #67514

cc @Centril
r? @pnkfelix

4 years agoAvoid scheduling repeated `StorageDead`s
Matthew Jasper [Sat, 25 Jan 2020 00:10:40 +0000 (00:10 +0000)]
Avoid scheduling repeated `StorageDead`s

Also add some comments

4 years agoMake use of `Place: Copy`
Matthew Jasper [Sat, 11 Jan 2020 20:30:42 +0000 (20:30 +0000)]
Make use of `Place: Copy`

4 years agoApply suggestions from code review
matthewjasper [Mon, 30 Dec 2019 17:47:10 +0000 (17:47 +0000)]
Apply suggestions from code review

Co-Authored-By: Mazdak Farrokhzad <twingoow@gmail.com>
4 years agoAddress review comments
Matthew Jasper [Sun, 29 Dec 2019 17:28:41 +0000 (17:28 +0000)]
Address review comments

4 years agoMake `Candidate` private
Matthew Jasper [Sun, 29 Dec 2019 17:00:23 +0000 (17:00 +0000)]
Make `Candidate` private

4 years agoAdd more tests for or-patterns
Matthew Jasper [Fri, 27 Dec 2019 18:11:20 +0000 (18:11 +0000)]
Add more tests for or-patterns

4 years agobootstrap: fix clippy warnings
Matthias Krüger [Mon, 3 Feb 2020 19:13:30 +0000 (20:13 +0100)]
bootstrap: fix clippy warnings

4 years agoChange wording for object unsafe because of assoc const
Esteban Küber [Mon, 3 Feb 2020 18:43:45 +0000 (10:43 -0800)]
Change wording for object unsafe because of assoc const

4 years agoAuto merge of #68803 - Dylan-DPC:rollup-b4x6ghj, r=Dylan-DPC
bors [Mon, 3 Feb 2020 18:40:54 +0000 (18:40 +0000)]
Auto merge of #68803 - Dylan-DPC:rollup-b4x6ghj, r=Dylan-DPC

Rollup of 8 pull requests

Successful merges:

 - #68678 (Install robots.txt into rust-docs tarballs)
 - #68711 (Added upper bound of what vecs and boxes can allocate)
 - #68744 (Do not ICE in `type-alias-impl-trait` with save-analysis)
 - #68777 (Clean up E0263 explanation)
 - #68787 (Optimize core::ptr::align_offset (part 1))
 - #68797 (Fix links to types instead of modules)
 - #68798 (Test that `#[track_caller]` as `fn()` respects RT / CTFE equivalence)
 - #68800 (Stabilize `core::iter::once_with()`)

Failed merges:

r? @ghost

4 years agoTreat `Rvalue::AddressOf` the same as `Rvalue::Ref`
Dylan MacKenzie [Mon, 3 Feb 2020 18:37:36 +0000 (10:37 -0800)]
Treat `Rvalue::AddressOf` the same as `Rvalue::Ref`

4 years agoEliminate "eager" qualif getter
Dylan MacKenzie [Mon, 3 Feb 2020 18:33:29 +0000 (10:33 -0800)]
Eliminate "eager" qualif getter

All qualif getters are now lazy

4 years agoPass correct closure type to `Qualif` methods
Dylan MacKenzie [Mon, 3 Feb 2020 18:32:54 +0000 (10:32 -0800)]
Pass correct closure type to `Qualif` methods

4 years agoTake `FnMut` instead of `Fn` in `Qualif` methods
Dylan MacKenzie [Mon, 3 Feb 2020 18:27:38 +0000 (10:27 -0800)]
Take `FnMut` instead of `Fn` in `Qualif` methods

4 years agoRollup merge of #68800 - JohnTitor:stabilize-once-with, r=Centril
Dylan DPC [Mon, 3 Feb 2020 17:58:37 +0000 (18:58 +0100)]
Rollup merge of #68800 - JohnTitor:stabilize-once-with, r=Centril

Stabilize `core::iter::once_with()`

Fixes #57581

FCP: https://github.com/rust-lang/rust/issues/57581#issuecomment-576178031

r? @SimonSapin

4 years agoRollup merge of #68798 - Centril:caller-loc-ctfe-rt-equiv, r=RalfJung
Dylan DPC [Mon, 3 Feb 2020 17:58:36 +0000 (18:58 +0100)]
Rollup merge of #68798 - Centril:caller-loc-ctfe-rt-equiv, r=RalfJung

Test that `#[track_caller]` as `fn()` respects RT / CTFE equivalence

r? @RalfJung cc @anp @eddyb

4 years agoRollup merge of #68797 - GuillaumeGomez:link-to-types, r=Dylan-DPC
Dylan DPC [Mon, 3 Feb 2020 17:58:35 +0000 (18:58 +0100)]
Rollup merge of #68797 - GuillaumeGomez:link-to-types, r=Dylan-DPC

Fix links to types instead of modules

r? @Dylan-DPC

4 years agoRollup merge of #68787 - amosonn:patch-1, r=nagisa
Dylan DPC [Mon, 3 Feb 2020 17:58:33 +0000 (18:58 +0100)]
Rollup merge of #68787 - amosonn:patch-1, r=nagisa

Optimize core::ptr::align_offset (part 1)

r? @nagisa
See #68616 for main discussion.

4 years agoRollup merge of #68777 - GuillaumeGomez:clean-up-e0263, r=Dylan-DPC
Dylan DPC [Mon, 3 Feb 2020 17:58:32 +0000 (18:58 +0100)]
Rollup merge of #68777 - GuillaumeGomez:clean-up-e0263, r=Dylan-DPC

Clean up E0263 explanation

r? @Dylan-DPC

4 years agoRollup merge of #68744 - JohnTitor:fix-ice-save-analysis, r=cramertj
Dylan DPC [Mon, 3 Feb 2020 17:58:30 +0000 (18:58 +0100)]
Rollup merge of #68744 - JohnTitor:fix-ice-save-analysis, r=cramertj

Do not ICE in `type-alias-impl-trait` with save-analysis

FIxes #68621
Fixes #68750

4 years agoRollup merge of #68711 - hman523:fix-68593, r=Dylan-DPC
Dylan DPC [Mon, 3 Feb 2020 17:58:27 +0000 (18:58 +0100)]
Rollup merge of #68711 - hman523:fix-68593, r=Dylan-DPC

Added upper bound of what vecs and boxes can allocate

Fixed issue #68593
I added a line of documentation to these two files to reflect that vectors and boxes ensure that they never allocate more than `isize::MAX` bytes.
r? @steveklabnik

4 years agoRollup merge of #68678 - Mark-Simulacrum:robots, r=pietroalbini
Dylan DPC [Mon, 3 Feb 2020 17:58:23 +0000 (18:58 +0100)]
Rollup merge of #68678 - Mark-Simulacrum:robots, r=pietroalbini

Install robots.txt into rust-docs tarballs

Fixes #68677.

I believe this might just work out from the central-station perspective, but even if it doesn't, this is a prerequisite step anyway.

4 years agoStabilize `core::iter::once_with()`
Yuki Okushi [Mon, 3 Feb 2020 15:47:04 +0000 (00:47 +0900)]
Stabilize `core::iter::once_with()`

4 years agotrack_caller test caller_location ctfe/rt equivalence wrt. fnptrs
Mazdak Farrokhzad [Mon, 3 Feb 2020 15:15:30 +0000 (16:15 +0100)]
track_caller test caller_location ctfe/rt equivalence wrt. fnptrs

4 years agoFix links to types instead of modules
Guillaume Gomez [Mon, 3 Feb 2020 15:12:35 +0000 (16:12 +0100)]
Fix links to types instead of modules

4 years agoAuto merge of #68665 - eddyb:debuginfo-early-create-var, r=nagisa
bors [Mon, 3 Feb 2020 13:06:44 +0000 (13:06 +0000)]
Auto merge of #68665 - eddyb:debuginfo-early-create-var, r=nagisa

codegen: create DIVariables ahead of using them with llvm.dbg.declare.

Instead of having `rustc_codegen_llvm` bundle creation of a `DIVariable` and `llvm.dbg.declare` into a single operation, they are now two separate methods, and the `DIVariable` is created earlier, specifically when `mir::VarDebugInfo`s are being partitioned into locals.

While this isn't currently needed, it's a prerequisite for #48300, which adds fragmented debuginfo, where one `mir::VarDebugInfo` has multiple parts of itself mapped to different `mir::Place`s.
For debuggers to see one composite variable instead of several ones with the same name, we need to create a single `DIVariable` and share it between multiple `llvm.dbg.declare` calls, which are likely pointing to different MIR locals.
That makes the `per_local_var_debug_info` partitioning a good spot to do this in, as we can create *exactly* one `DIVariable` per `mir::VarDebugInfo`, and refer to it as many things as needed.

I'm opening this PR separately because I want to test its perf impact in isolation (see https://github.com/rust-lang/rust/pull/48300#issuecomment-580121438).

r? @nagisa or @oli-obk cc @michaelwoerister @nikomatsakis

4 years agorustc_codegen_llvm: avoid redundant calls to span_start.
Eduard-Mihai Burtescu [Mon, 3 Feb 2020 10:14:05 +0000 (12:14 +0200)]
rustc_codegen_llvm: avoid redundant calls to span_start.

4 years agorustc_codegen_ssa: convert mir::VarDebugInfo into a custom PerLocalVarDebugInfo.
Eduard-Mihai Burtescu [Wed, 22 Jan 2020 17:45:22 +0000 (19:45 +0200)]
rustc_codegen_ssa: convert mir::VarDebugInfo into a custom PerLocalVarDebugInfo.

4 years agorustc_codegen_ssa: split declare_local into create_dbg_var and dbg_var_addr.
Eduard-Mihai Burtescu [Sun, 26 Jan 2020 16:50:13 +0000 (18:50 +0200)]
rustc_codegen_ssa: split declare_local into create_dbg_var and dbg_var_addr.

4 years agoAuto merge of #68778 - RalfJung:raw-addr-of, r=eddyb
bors [Mon, 3 Feb 2020 09:54:09 +0000 (09:54 +0000)]
Auto merge of #68778 - RalfJung:raw-addr-of, r=eddyb

add raw-addr-of variant to mir_raw_fat_ptr

As suggested at https://github.com/rust-lang/rust/pull/48300#discussion_r372520388

r? @eddyb

4 years agoAuto merge of #68772 - matthewjasper:relate-opt, r=davidtwco
bors [Mon, 3 Feb 2020 06:38:34 +0000 (06:38 +0000)]
Auto merge of #68772 - matthewjasper:relate-opt, r=davidtwco

Avoid exponential behaviour when relating types

When equating bound types we check subtyping in both directions. Since closures are invariant in their substs, we end up comparing the two types an exponential number of times. If there are no bound variables this isn't needed.

Closes #68061

4 years agoChange expansion error to be non-fatal
David [Sun, 2 Feb 2020 19:28:32 +0000 (11:28 -0800)]
Change expansion error to be non-fatal

Changes the error handler for inner attributes that replace the root
with a non-module. Previously it would emit a fatal error. It now emits
an empty expasion and a non-fatal error like the existing handler for a
failed expansion.

4 years agoAuto merge of #68756 - JohnTitor:fix-ice-save-analysis-2, r=davidtwco
bors [Mon, 3 Feb 2020 03:28:08 +0000 (03:28 +0000)]
Auto merge of #68756 - JohnTitor:fix-ice-save-analysis-2, r=davidtwco

Fix ICE with save-analysis

Fixes #68749

It should be fine since it's the same way as `visit_expr`.

4 years agoPull out a special case in `merge_from_succ`.
Nicholas Nethercote [Mon, 3 Feb 2020 02:32:59 +0000 (13:32 +1100)]
Pull out a special case in `merge_from_succ`.

This is a small perf win.

4 years agoFix up `merge_from_succ`.
Nicholas Nethercote [Mon, 3 Feb 2020 01:11:11 +0000 (12:11 +1100)]
Fix up `merge_from_succ`.

This function has a variable `changed` that is erroneously used for two
related-but-different purpose:
- to detect if the current element has changed;
- to detect if any elements have changed.

As a result, its use for the first purpose is broken, because if any
prior element changed then the code always thinks the current element
has changed. This is only a performance bug, not a correctness bug,
because we frequently end up calling `assign_unpacked` unnecessarily to
overwrite the element with itself.

This commit adds `any_changed` to correctly distinguish between the two
purposes. This is a small perf win for some benchmarks.

4 years agoAuto merge of #68735 - JohnTitor:fix-ice-0202, r=estebank
bors [Mon, 3 Feb 2020 00:04:16 +0000 (00:04 +0000)]
Auto merge of #68735 - JohnTitor:fix-ice-0202, r=estebank

Use `next_point` to avoid ICE

Fixes #68730

r? @estebank (I think you're familiar with that)

4 years agoOptimize core::ptr::align_offset
Amos Onn [Sun, 2 Feb 2020 01:18:33 +0000 (02:18 +0100)]
Optimize core::ptr::align_offset

- As explained in the comment inside mod_inv, it is valid to work mod
  `usize::max_value()` right until the end.

4 years agoOptimize core::ptr::align_offset
Amos Onn [Tue, 28 Jan 2020 21:14:04 +0000 (22:14 +0100)]
Optimize core::ptr::align_offset

- When calculating the inverse, it's enough to work `mod a/g` instead
  of `mod a`.

4 years agoOptimize core::ptr::align_offset
Amos Onn [Thu, 30 Jan 2020 19:04:24 +0000 (20:04 +0100)]
Optimize core::ptr::align_offset

- Stopping condition inside mod_inv can be >= instead of >
- Remove intrinsics::unchecked_rem, we are working modulu powers-of-2 so
we can simply mask

4 years agoDeal with spans showing `std` lib
Esteban Küber [Sun, 2 Feb 2020 21:51:23 +0000 (13:51 -0800)]
Deal with spans showing `std` lib

Address #53081

4 years agoPoint at reason in object unsafe trait with `Self` in supertraits or `where`-clause
Esteban Küber [Sun, 2 Feb 2020 20:51:30 +0000 (12:51 -0800)]
Point at reason in object unsafe trait with `Self` in supertraits or `where`-clause

4 years agofix rebase
Esteban Küber [Sun, 2 Feb 2020 20:50:07 +0000 (12:50 -0800)]
fix rebase

4 years agoAuto merge of #68720 - wesleywiser:llvm_time_trace, r=davidtwco
bors [Sun, 2 Feb 2020 20:33:47 +0000 (20:33 +0000)]
Auto merge of #68720 - wesleywiser:llvm_time_trace, r=davidtwco

Add support for enabling the LLVM time-trace feature

I found this helpful while investigating an LLVM performance issue.
Passing `-Z llvm-time-trace` causes a `llvm_timings.json` file to be
created. This file can be inspected in with the Chrome Profiler
tools or with any other compatible tool like SpeedScope.

More information on the LLVM feature:

- https://aras-p.info/blog/2019/01/16/time-trace-timeline-flame-chart-profiler-for-Clang/

- https://reviews.llvm.org/rL357340

4 years agoAccount for `?Sized` type parameter bounds
Esteban Küber [Sun, 2 Feb 2020 10:03:54 +0000 (02:03 -0800)]
Account for `?Sized` type parameter bounds

4 years agoRemove duplicated code
Esteban Küber [Sun, 2 Feb 2020 09:38:12 +0000 (01:38 -0800)]
Remove duplicated code

4 years agoSuggest `?Sized` on type parameters
Esteban Küber [Sun, 2 Feb 2020 08:54:13 +0000 (00:54 -0800)]
Suggest `?Sized` on type parameters

4 years agoadd tests for structured suggestion
Esteban Küber [Sat, 1 Feb 2020 03:04:58 +0000 (19:04 -0800)]
add tests for structured suggestion

4 years agoUse more appropriate spans on object unsafe traits and provide structured suggestions...
Esteban Küber [Sat, 1 Feb 2020 02:48:35 +0000 (18:48 -0800)]
Use more appropriate spans on object unsafe traits and provide structured suggestions when possible

4 years agoWording changes to object unsafe trait errors
Esteban Küber [Sat, 1 Feb 2020 00:47:00 +0000 (16:47 -0800)]
Wording changes to object unsafe trait errors

Stemming from the thread at https://twitter.com/indygreg/status/1223279056398929920

4 years agoPoint at `Sized` requirements
Esteban Küber [Fri, 31 Jan 2020 04:12:46 +0000 (20:12 -0800)]
Point at `Sized` requirements

Make #47990 easier to understand

4 years agoreview comments
Esteban Küber [Fri, 31 Jan 2020 03:35:48 +0000 (19:35 -0800)]
review comments

4 years agoRemove duplicated code
Esteban Küber [Thu, 30 Jan 2020 18:34:38 +0000 (10:34 -0800)]
Remove duplicated code

4 years agoSlight rewording of diagnostic message
Esteban Küber [Thu, 30 Jan 2020 18:20:47 +0000 (10:20 -0800)]
Slight rewording of diagnostic message