do not ICE when return type includes unconstrained anon region
It turns out that this *can* happen after all, if the region is only
used in projections from the input types.
Fixes https://github.com/rust-lang/rust/issues/47511
r? @eddyb
Teach rustc about DW_AT_noreturn and a few more DIFlags
We achieve two small things with this PR:
1. We provide definitions for a few additional llvm debuginfo flags
1. We _use_ one of these new flags, `FlagNoReturn`, and add it to debuginfo for functions with the never return type (`!`).
This commit changes the ABI of SIMD types in the "Rust" ABI to unconditionally
be passed via pointers instead of being passed as immediates. This should fix a
longstanding issue, #44367, where SIMD-using programs ended up showing very odd
behavior at runtime because the ABI between functions was mismatched.
As a bit of a recap, this is sort of an LLVM bug and sort of an LLVM feature
(today's behavior). LLVM will generate code for a function solely looking at the
function it's generating, including calls to other functions. Let's then say
you've got something that looks like:
```llvm
define void @foo() { ; no target features enabled
call void @bar(<i64 x 4> zeroinitializer)
ret void
}
define void @bar(<i64 x 4>) #0 { ; enables the AVX feature
...
}
```
LLVM will codegen the call to `bar` *without* using AVX registers becauase `foo`
doesn't have access to these registers. Instead it's generated with emulation
that uses two 128-bit registers. The `bar` function, on the other hand, will
expect its argument in an AVX register (as it has AVX enabled). This means we've
got a codegen problem!
Comments on #44367 have some more contexutal information but the crux of the
issue is that if we want SIMD to work in general we'll need to ensure that
whenever a function calls another they ABI of the arguments being passed is in
agreement.
One possible solution to this would be to insert "shim functions" where whenever
a `target_feature` mismatch is detected the compiler inserts a shim function
where you pass arguments via memory to the shim and then the shim loads the
values and calls the target function (where the shim and the target have the
same target features enabled). This unfortunately is quite nontrivial to
implement in rustc today (especially when accounting for function pointers and
such).
This commit takes a different solution, *always* passing SIMD arguments through
memory instead of passing as immediates. This strategy solves the problem at the
LLVM layer because the ABI between two functions never uses SIMD registers. This
also shouldn't be a hit to performance because SIMD performance is thought to
often rely on inlining anyway, where a `call` instruction, even if using SIMD
registers, would be disastrous to performance regardless. LLVM should then be
more than capable of fixing all our memory usage to use registers instead after
enough inlining has been performed.
Note that there's a few caveats to this commit though:
* The "platform intrinsic" ABI is omitted from "always pass via memory". This
ABI is used to define intrinsics like `simd_shuffle4` where LLVM and rustc
need to have the arguments as an immediate.
* Additionally this commit does *not* fix the `extern` ("C") ABI. This means
that the bug in #44367 can still happen when using non-Rust-ABI functions. My
hope is that before stabilization we can ban and/or warn about SIMD types in
these functions (as AFAIK there's not much motivation to belong there anyway),
but I'll leave that for a later commit and if this is merged I'll file a
follow-up issue.
All in all this...
Closes#44367
Fix type inhabitedness check for arrays
Arrays of uninhabited types were considered to also be uninhabited if
their length had not been evaluated, causing unsoundness.
Fixes#47563.
r? @eddyb
Fix no_integrated_as option to work with new codegen architecture.
Old implementation called the assembler once per crate, but we need to call
it for each object file instead, because a single crate can now have more
than one object file.
This patch fixes issue #45836 (Can't compile core for msp430 in release mode)
This change can be tested on x86_64 using
```sh
export RUSTFLAGS="-C no_integrated_as -C save_temps"
```
r? @alexcrichton
cc @japaric
rustc_trans: ignore trailing padding larger than 8 bytes.
Fixes#45662 by ignoring a missing second register component, as it could be entirely padding.
rustc_trans: take into account primitives larger than 8 bytes.
Fixes#38763 by marking all "eightbytes" covered by a primitive appropriately, not just the first.
Fix spans in unused import lint for nested groups
This fixes an inconsistency for empty nested groups, and adds a test for all the possible cases of the lint.
```
warning: unused imports: `*`, `Foo`, `baz::{}`, `foobar::*`
--> test.rs:16:11
|
16 | use foo::{Foo, bar::{baz::{}, foobar::*}, *};
| ^^^ ^^^^^^^ ^^^^^^^^^ ^
|
= note: #[warn(unused_imports)] on by default
warning: unused import: `*`
--> test.rs:17:24
|
17 | use foo::bar::baz::{*, *};
| ^
warning: unused import: `use foo::{};`
--> test.rs:18:1
|
18 | use foo::{};
| ^^^^^^^^^^^^
```
cc #44494
Fix into() cast paren check precedence
As discussed in #47699 the logic for determining if an expression needs parenthesis when suggesting an `.into()` cast is incorrect. Two broken examples from nightly are:
```
error[E0308]: mismatched types
--> main.rs:4:10
|
4 | test(foo as i8);
| ^^^^^^^^^ expected i32, found i8
help: you can cast an `i8` to `i32`, which will sign-extend the source value
|
4 | test(foo as i8.into());
|
```
```
error[E0308]: mismatched types
--> main.rs:4:10
|
4 | test(*foo);
| ^^^^ expected i32, found i8
help: you can cast an `i8` to `i32`, which will sign-extend the source value
|
4 | test(*foo.into());
|
```
As suggested by @petrochenkov switch the precedence check to `PREC_POSTFIX`. This catches both `as` and unary operators. Fixes#47699.
r? @petrochenkov
On missing method do not suggest private traits
When encountering a method call for an ADT that doesn't have any
implementation of it, we search for traits that could be implemented
that do have that method. Filter out private non-local traits that would
not be able to be implemented.
This doesn't account for public traits that are in a private scope, but
works as a first approximation and is a more correct behavior than the
current one.
Fix#45781.
As discussed in #47699 the logic for determining if an expression needs
parenthesis when suggesting an `.into()` cast is incorrect. Two broken
examples from nightly are:
```
error[E0308]: mismatched types
--> main.rs:4:10
|
4 | test(foo as i8);
| ^^^^^^^^^ expected i32, found i8
help: you can cast an `i8` to `i32`, which will sign-extend the source value
|
4 | test(foo as i8.into());
|
```
```
error[E0308]: mismatched types
--> main.rs:4:10
|
4 | test(*foo);
| ^^^^ expected i32, found i8
help: you can cast an `i8` to `i32`, which will sign-extend the source value
|
4 | test(*foo.into());
|
```
As suggested by @petrochenkov switch the precedence check to
PREC_POSTFIX. This catches both `as` and unary operators. Fixes#47699.
When E0277's span points at a `for` loop, the actual issue is in the
element being iterated. Instead of pointing at the entire loop, point
only at the first line (when possible) so that the span ends in the
element for which E0277 was triggered.
Immovable generators
This adds support for immovable generators which allow you to borrow local values inside generator across suspension points. These are declared using a `static` keyword:
```rust
let mut generator = static || {
let local = &Vec::new();
yield;
local.push(0i8);
};
generator.resume();
// ERROR moving the generator after it has resumed would invalidate the interior reference
// drop(generator);
```
Region inference is no longer affected by the types stored in generators so the regions inside should be similar to other code (and unaffected by the presence of `yield` expressions). The borrow checker is extended to pick up the slack so interior references still result in errors for movable generators. This fixes#44197, #45259 and #45093.
This PR depends on [PR #44917 (immovable types)](https://github.com/rust-lang/rust/pull/44917), I suggest potential reviewers ignore the first commit as it adds immovable types.
This commit primarily adds the ability to control what kind of LTO happens when
rustc performs LTO, namely allowing values to be specified to the `-C lto`
option, such as `-C lto=thin` and `-C lto=fat`. (where "fat" is the previous
kind of LTO, throw everything in one giant module)
Along the way this also refactors a number of fields which store information
about whether LTO/ThinLTO are enabled to unify them all into one field through
which everything is dispatched, hopefully removing a number of special cases
throughout.
This is intended to help mitigate #47409 but will require a backport as well,
and this would unfortunately need to be an otherwise insta-stable option.
Fix spurious warning on empty proc macro crates
While attempting to reproduce rust-lang/rust#47086 I noticed the following warning:
```shell
> rustc /dev/null --crate-type proc-macro
warning: unused variable: `registrar`
--> /dev/null:0:1
```
As there are no macros to register the automatically generated registrar function for the crate has no body. As a result its `registrar` argument is unused triggering the above warning.
The warning is confusing and not easily actionable by the developer. It could also be triggered legitimately by e.g. having all of the macros in a crate #[cfg]'ed out.
Fix by naming the generated argument `_registrar` inside `mk_registrar()`. This suppresses the unused variable warning.
Change the --unpretty flag to -Z unpretty
First PR 😄 !
-Z unpretty no longer requires -Z unstable-options.
Also, I mildly changed the syntax of the flag to match the other -Z flags. All uses of the flag take the form `unpretty=something` where something can either `string` or `string=string` (see the help messages of the CLI).
Fix#47395
r? @nikomatsakis EDIT: apparently rust-highfive doesn't see edits...
Properly pass down immutability info for thread-locals.
For thread-locals we call into cat_rvalue_node() to create a CMT
(Category, Mutability, Type) that always has McDeclared. This is
incorrect for thread-locals that don't have the 'mut' keyword; we should
use McImmutable there.
Extend cat_rvalue_node() to have an additional mutability parameter. Fix
up all the callers to make use of that function. Also extend one of the
existing unit tests to cover this.
Fixes: #47053
Implement RFC 1946 - intra-rustdoc links
https://github.com/rust-lang/rfcs/pull/1946https://github.com/rust-lang/rust/issues/43466
Note for reviewers: The plain line counts are a little inflated because of how the markdown link parsing was done. [Read the file diff with "whitespace only" changes removed](https://github.com/rust-lang/rust/pull/47046/files?w=1) to get a better view of what actually changed there.
This pulls the name/path resolution mechanisms out of the compiler and runs it on the markdown in a crate's docs, so that links can be made to `SomeStruct` directly rather than finding the folder path to `struct.SomeStruct.html`. Check the `src/test/rustdoc/intra-paths.rs` test in this PR for a demo. The change was... a little invasive, but unlocks a really powerful mechanism for writing documentation that doesn't care about where an item was written to on the hard disk.
Items included:
- [x] Make work with the hoedown renderer
- [x] Handle relative paths
- [x] Parse out the "path ambiguities" qualifiers (`[crate foo]`, `[struct Foo]`, `[foo()]`, `[static FOO]`, `[foo!]`, etc)
- [x] Resolve foreign macros
- [x] Resolve local macros
- [x] Handle the use of inner/outer attributes giving different resolution scopes (handling for non-modules pushed to different PR)
Items not included:
- [ ] Make sure cross-crate inlining works (blocked on refactor described in https://github.com/rust-lang/rust/pull/47046#issuecomment-354824520)
- [ ] Implied Shortcut Reference Links (where just doing `[::std::iter::Iterator][]` without a reference anchor will resolve using the reference name rather than the link target) (requires modifying the markdown parser - blocked on Hoedown/Pulldown switch and https://github.com/google/pulldown-cmark/issues/121)
- [ ] Handle enum variants and UFCS methods (Enum variants link to the enum page, associated methods don't link at all)
- [ ] Emit more warnings/errors when things fail to resolve (linking to a value-namespaced item without a qualifier will emit an error, otherwise the link is just treated as a url, not a rust path)
- [ ] Give better spans for resolution errors (currently the span for the first doc comment is used)
- [ ] Check for inner doc comments on things that aren't modules
I'm making the PR, but it should be noted that most of the work was done by Misdreavus 😄
(Editor's note: This has become a lie, check that commit log, Manish did a ton of work after this PR was opened `>_>`)
