Commit Graph

155007 Commits

Author SHA1 Message Date
bors
63cc2bb3d0 Auto merge of #88243 - nikic:newpm-2, r=nagisa
Enable new pass manager with LLVM 13

The new pass manager is enabled by default in clang since Clang/LLVM 13. Per the recent discussion on llvm-dev (https://lists.llvm.org/pipermail/llvm-dev/2021-August/152305.html) the legacy pass manager will be unmaintained in LLVM 14 and removed entirely in LLVM 15.

This switches us to use the new pass manager if LLVM >= 13 is used. It's possible to still use the old pass manager using `-Z new-llvm-pass-manager=no`.
2021-09-25 13:43:52 +00:00
bors
e9f29a8519 Auto merge of #89030 - nbdd0121:box2, r=jonas-schievink
Introduce `Rvalue::ShallowInitBox`

Polished version of #88700.

Implements MCP rust-lang/compiler-team#460, and should allow #43596 to go forward.

In short, creating an empty box is split from a nullary-op `NullOp::Box` into two steps, first a call to `exchange_malloc`, then a `Rvalue::ShallowInitBox` which transmutes `*mut u8` to a shallow-initialized `Box<T>`. This allows the `exchange_malloc` call to unwind. Details can be found in the MCP.

`NullOp::Box` is not yet removed, purely to make reverting easier in case anything goes wrong as the result of this PR. If revert is needed a reversion of "Use Rvalue::ShallowInitBox for box expression" commit followed by a test bless should be sufficient.

Experiments in #88700 showed a very slight compile-time perf regression due to (supposedly) slightly more time spent in LLVM. We could omit unwind edge generation (in non-`oom=panic` case) in box expression MIR construction to restore perf; but I don't think it's necessary since runtime perf isn't affected and perf difference is rather small.
2021-09-25 11:01:13 +00:00
Nikita Popov
51203dc1c4 Pin panic-in-drop=abort test to old pass manager 2021-09-25 12:40:16 +02:00
Nikita Popov
31834434a0 Make expectation in panic-in-drop-abort.rs test more precise
Check whether a call/invoke of the function exists, but don't
match a leftover function declaration.

Also remove the CHECK-LABELs: In panic-in-drop=unwind mode the
call will not actually be in either of those functions, so
remove the restriction and look for any calls.
2021-09-25 11:24:23 +02:00
Nikita Popov
074bbc6b87 Use correct pipeline for LTO at O0
Unlike the pre-link piplines, the LTO pipelines do support O0,
and using them is required to avoid leaving behind undefined
references for the linker.
2021-09-25 11:24:23 +02:00
Nikita Popov
be01f42f73 Enable new pass manager on LLVM 13
The new pass manager is enabled by default in clang since
Clang/LLVM 13. While the discussion about this is still ongoing
(https://lists.llvm.org/pipermail/llvm-dev/2021-August/152305.html)
it's expected that support for the legacy pass manager will be
dropped either in LLVM 14 or 15.

This switches us to use the new pass manager if LLVM >= 13 is used.
2021-09-25 11:24:23 +02:00
bors
218a96cae0 Auto merge of #89237 - BoxyUwU:trackersMcCaller, r=eddyb
make `#[track_caller]` actually do stuff in `Steal::borrow`

makes this ICE message useful:
``thread 'rustc' panicked at 'attempted to read from stolen value', /rustc/ac2d9fc509e36d1b32513744adf58c34bcc4f43c\compiler\rustc_data_structures\src\steal.rs:37:21``
2021-09-25 07:33:25 +00:00
bors
60fe8b3a65 Auto merge of #87220 - petrochenkov:derivecfglimit2, r=Aaron1011
Make `#[derive(A, B, ...)]` cfg-eval its input only for `A, B, ...` and stabilize `feature(macro_attributes_in_derive_output)`

Stabilization report: https://github.com/rust-lang/rust/pull/87220#issuecomment-881923657

Closes #81119
r? `@Aaron1011`
2021-09-25 04:35:59 +00:00
bors
6867dd2c01 Auto merge of #89230 - workingjubilee:rollup-1swktdq, r=workingjubilee
Rollup of 8 pull requests

Successful merges:

 - #88893 (Add 1.56.0 release notes)
 - #89001 (Be explicit about using Binder::dummy)
 - #89072 (Avoid a couple of Symbol::as_str calls in cg_llvm )
 - #89104 (Simplify scoped_thread)
 - #89208 ([rfc 2229] Drop fully captured upvars in the same order as the regular drop code)
 - #89210 (Add missing time complexities to linked_list.rs)
 - #89217 (Enable "generate-link-to-definition" option on rust tools docs as well)
 - #89221 (Give better error for `macro_rules! name!`)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup
2021-09-25 00:20:28 +00:00
Gary Guo
ab64580977 Bless tests 2021-09-25 01:08:41 +01:00
Gary Guo
19eaee2c32 Report heap allocation instead of non-const fn for exchange_malloc call 2021-09-25 01:08:41 +01:00
Gary Guo
511333fcc4 Use Rvalue::ShallowInitBox for box expression 2021-09-25 01:08:41 +01:00
Gary Guo
c38da2e0a3 Introduce Rvalue::ShallowInitBox 2021-09-25 01:08:41 +01:00
Ellen
f1e71a5b41 arrr caught ya caller
awd
2021-09-25 00:34:02 +01:00
bors
043972f152 Auto merge of #86246 - FabianWolff:issue-83471, r=estebank
Add basic checks for well-formedness of `fn`/`fn_mut` lang items

This pull request fixes #83471. Lang items are never actually checked for well-formedness (#9307). This means that one can get an ICE quite easily, e.g. as follows:
```rust
#![feature(lang_items)]
#[lang = "fn"]
trait MyFn {
    const call: i32 = 42;
}

fn main() {
    (|| 42)();
}
```
or this:
```rust
#![feature(lang_items)]
#[lang = "fn"]
trait MyFn {
    fn call(i: i32, j: i32);
}

fn main() {
    (|| 42)();
}
```
Ideally, there should probably be a more comprehensive strategy for checking lang items for well-formedness, but for the time being, I have added some rudimentary well-formedness checks that prevent #83471 and similar issues.
2021-09-24 21:30:35 +00:00
Vadim Petrochenkov
85f02901a3 Stabilize feature(macro_attributes_in_derive_output) 2021-09-24 21:48:30 +03:00
Vadim Petrochenkov
92804cd490 derive: Do not configure or clone items unless necessary 2021-09-24 21:45:51 +03:00
Vadim Petrochenkov
c993984e4d builtin_macros: Make #[derive(A, B, ...)] cfg-eval its input only for A, B, ... 2021-09-24 21:45:51 +03:00
Jubilee
6f31fa58fd
Rollup merge of #89221 - aDotInTheVoid:macro-error-1, r=estebank
Give better error for `macro_rules! name!`

r? ``@estebank``

``@rustbot`` modify labels: +A-diagnostics +A-parser
2021-09-24 11:40:17 -07:00
Jubilee
0cd9dd3b3c
Rollup merge of #89217 - GuillaumeGomez:generate-link-to-def-rust-tools, r=jyn514
Enable "generate-link-to-definition" option on rust tools docs as well

Just realized that we enable the option for the compiler crates, but we don't have it for rustdoc and the other tools documentation...

Part of https://github.com/rust-lang/rust/issues/89095.

cc ``@rust-lang/rustdoc``
r? ``@Mark-Simulacrum``
2021-09-24 11:40:16 -07:00
Jubilee
0fa43494bd
Rollup merge of #89210 - Takashiidobe:master, r=kennytm
Add missing time complexities to linked_list.rs

Most functions in LinkedList have time complexities in their description:
Like push front:

```
Adds an element first in the list.

This operation should compute in O(1) time.
```

Time complexities were missing for the following, so I've added them in this PR:

contains: O(n)
front: O(1)
front_mut: O(1)
back: O(1)
back_mut: O(1)
2021-09-24 11:40:15 -07:00
Jubilee
7ade6ed48e
Rollup merge of #89208 - wesleywiser:rfc_2229_droporder, r=nikomatsakis
[rfc 2229] Drop fully captured upvars in the same order as the regular drop code

Currently, with the new 2021 edition, if a closure captures all of the
fields of an upvar, we'll drop those fields in the order they are used
within the closure instead of the normal drop order (the definition
order of the fields in the type).

This changes that so we sort the captured fields by the definition order
which causes them to drop in that same order as well.

Fixes rust-lang/project-rfc-2229#42

r? `@nikomatsakis`
2021-09-24 11:40:14 -07:00
Jubilee
80d9886e4c
Rollup merge of #89104 - Mark-Simulacrum:spawn-unchecked, r=nagisa,bjorn3
Simplify scoped_thread

Avoids a bunch of manual pointer manipulation in favor of using the appropriate std API.
2021-09-24 11:40:13 -07:00
Jubilee
8a454f8101
Rollup merge of #89072 - bjorn3:less_symbol_as_str, r=michaelwoerister
Avoid a couple of Symbol::as_str calls in cg_llvm

This should improve performance a tiny bit. Also remove `Symbol::len` and make `SymbolIndex` private.
2021-09-24 11:40:12 -07:00
Jubilee
ee2e97c416
Rollup merge of #89001 - jackh726:binder-cleanup, r=nikomatsakis
Be explicit about using Binder::dummy

This is somewhat of a late followup to the binder refactor PR. It removes `ToPredicate` and `ToPolyTraitImpls` that hide the use of `Binder::dummy`. While this does make code a bit more verbose, it allows us be more careful about where we create binders.

Another alternative here might be to add a new trait `ToBinder` or something with a `dummy()` fn. Which could still allow grepping but allows doing something like `trait_ref.dummy()` (but I also wonder if longer-term, it would be better to be even more explicit with a `bind_with_vars(ty::List::empty())` *but* that's not clear yet.

r? ``@nikomatsakis``
2021-09-24 11:40:11 -07:00
Jubilee
2de7f1075f
Rollup merge of #88893 - Mark-Simulacrum:relnotes, r=pietroalbini
Add 1.56.0 release notes

Notable things:

* Rustdoc section currently absent, for lack of things to call out that I could find. It seems like https://github.com/rust-lang/rust/pull/87451 is the only potential candidate, but that seems like more of a bugfix and doesn't seem to warrant inclusion to me. But we can add it if desired.

As with the 1.55.0 release notes, my intent is to leave this open for approximately 1-2 weeks at the early part of the cycle, and then merge it in (after fixing any feedback). Further iteration can happen in subsequent issues (and PRs).

[Rendered](https://github.com/Mark-Simulacrum/rust/blob/relnotes/RELEASES.md)
2021-09-24 11:40:11 -07:00
Fabian Wolff
cb6c13977a Make error message for malformed fn/fn_mut lang item more specific 2021-09-24 20:15:19 +02:00
bors
73422130ee Auto merge of #88872 - petrochenkov:getmod2, r=cjgillot
resolve: Some module-related refactorings

See the individual commits.
2021-09-24 18:12:33 +00:00
Vadim Petrochenkov
8f035b87e0 Disable a spuriously failing test 2021-09-24 20:28:47 +03:00
Wesley Wiser
dd918048a5 Update and add more tests 2021-09-24 12:27:42 -04:00
Fabian Wolff
4337089098 Add basic checks for well-formedness of fn/fn_mut lang items 2021-09-24 18:26:27 +02:00
Vadim Petrochenkov
ab834e5ea9 resolve: Refactor obtaining Module from its DefId
The `Option<Module>` version is supported for the case where we don't know whether the `DefId` refers to a module or not.
Non-local traits and enums are also correctly found now.
2021-09-24 18:57:15 +03:00
Vadim Petrochenkov
a8021888c8 resolve: Use a single common map for local and foreign modules 2021-09-24 18:57:15 +03:00
Vadim Petrochenkov
1a23858983 resolve: Rename some expansion def scope methods 2021-09-24 18:57:15 +03:00
Vadim Petrochenkov
6e9adcbd82 resolve: Cleanup module allocation
Construction of all modules is now centralized and performed by `fn new_module`.
2021-09-24 18:57:15 +03:00
Vadim Petrochenkov
fd58eea4e1 resolve: Do not cache nearest parent mod in ModuleData 2021-09-24 18:57:14 +03:00
Mark Rousskov
a307dcc962 Add 1.56.0 release notes
Co-authored-by: Josh Triplett <josh@joshtriplett.org>
2021-09-24 11:24:00 -04:00
bors
f06f9bbd3a Auto merge of #88999 - Migi:master, r=oli-obk
Make `Duration` respect `width` when formatting using `Debug`

When printing or writing a `std::time::Duration` using `Debug` formatting, it previously completely ignored any specified `width`. This is unlike types like integers and floats, which do pad to `width`, for both `Display` and `Debug`, though not all types consider `width` in their `Debug` output (see e.g. #30164). Curiously, `Duration`'s `Debug` formatting *did* consider `precision`.

This PR makes `Duration` pad to `width` just like integers and floats, so that
```rust
format!("|{:8?}|", Duration::from_millis(1234))
```
returns
```
|1.234s  |
```

Before you ask "who formats `Debug` output?", note that `Duration` doesn't actually implement `Display`, so `Debug` is currently the only way to format `Duration`s. I think that's wrong, and `Duration` should get a `Display` implementation, but in the meantime there's no harm in making the `Debug` formatting respect `width` rather than ignore it.

I chose the default alignment to be left-aligned. The general rule Rust uses is: numeric types are right-aligned by default, non-numeric types left-aligned. It wasn't clear to me whether `Duration` is a numeric type or not. The fact that a formatted `Duration` can end with suffixes of variable length (`"s"`, `"ms"`, `"µs"`, etc.) made me lean towards left-alignment, but it would be trivial to change it.

Fixes issue #88059.
2021-09-24 15:22:26 +00:00
Nixon Enraght-Moony
ed3b751799 Give better error for macro_rules! name! 2021-09-24 16:21:34 +01:00
Takashi Idobe
cebba31d4a
unitalicize O(1) complexities 2021-09-24 08:33:49 -05:00
Takashi Idobe
cb1c06fdd8
Merge branch 'rust-lang:master' into master 2021-09-24 08:31:03 -05:00
bors
a0648eab36 Auto merge of #88835 - FabianWolff:issue-88770, r=petrochenkov
Fix error recovery in format macro parsing

Fixes #88770. Basically, the assumption in the following comment is incorrect:
b69fe57261/compiler/rustc_builtin_macros/src/format.rs (L167-L172)

This is only true in the first iteration of the loop, when [`p.clear_expected_tokens()`](b69fe57261/compiler/rustc_builtin_macros/src/format.rs (L164)) is called. In subsequent iterations, `p.expected_tokens` won't be empty, so `p.expect()` won't actually call `unexpected_try_recover()`:
b69fe57261/compiler/rustc_parse/src/parser/mod.rs (L487-L498)

Instead, it will call `expect_one_of()`, which _can_ recover and return `Ok()`. This PR handles this case to fix the ICE in #88770.
2021-09-24 11:54:29 +00:00
Guillaume Gomez
9fa59e19e5 Enable "generate-link-to-definition" option on rust tools docs as well 2021-09-24 12:05:35 +02:00
bors
91d8da1f4b Auto merge of #89130 - nikic:update-llvm-2, r=cuviper
Update LLVM submodule

This merges the upstream `release/13.x` branch to pull in the second fix for #88769.
2021-09-24 08:12:15 +00:00
bors
197fc8591e Auto merge of #89120 - In-line:remove_unneded_visible_parents_map, r=estebank
Disable visible path calculation for PrettyPrinter in Ok path of compiler
2021-09-24 05:29:49 +00:00
Wesley Wiser
389365631d Fix tidy and respond to some feedback 2021-09-23 22:15:12 -04:00
bors
293b8f2c11 Auto merge of #89211 - workingjubilee:rollup-fj4eduk, r=workingjubilee
Rollup of 7 pull requests

Successful merges:

 - #88612 (Add a better error message for #39364)
 - #89023 (Resolve issue : Somewhat confusing error with extended_key_value_attributes)
 - #89148 (Suggest `_` in turbofish if param will be inferred from fn argument)
 - #89171 (Run `no_core` rustdoc tests only on Linux)
 - #89176 (Change singular to plural)
 - #89184 (Temporarily rename int_roundings functions to avoid conflicts)
 - #89200 (Fix typo)

Failed merges:

r? `@ghost`
`@rustbot` modify labels: rollup
2021-09-24 01:48:02 +00:00
Jubilee
1875cecfd4
Rollup merge of #89200 - hkmatsumoto:patch, r=workingjubilee
Fix typo
2021-09-23 17:31:47 -07:00
Jubilee
384dd53641
Rollup merge of #89184 - joshtriplett:master, r=estebank
Temporarily rename int_roundings functions to avoid conflicts

These functions are unstable, but because they're inherent they still
introduce conflicts with stable trait functions in crates. Temporarily
rename them to fix these conflicts, until we can resolve those conflicts
in a better way.
2021-09-23 17:31:46 -07:00
Jubilee
a071193826
Rollup merge of #89176 - TaKO8Ki:change-singular-to-plural, r=lcnr
Change singular to plural
2021-09-23 17:31:45 -07:00