core: Remove panics from some `Layout` methods
`Layout` is often used at the core of allocation APIs and is as a result pretty
sensitive to codegen in various circumstances. I was profiling `-C opt-level=z`
with a wasm project recently and noticed that the `unwrap()` wasn't removed
inside of `Layout`, causing the program to be much larger than it otherwise
would be. If inlining were more aggressive LLVM would have figured out that the
panic could be eliminated, but in general the methods here can't panic in the
first place!
As a result this commit makes the following tweaks:
* Removes `unwrap()` and replaces it with `unsafe` in `Layout::new` and
`Layout::for_value`. For posterity though a debug assertion was left behind.
* Removes an `unwrap()` in favor of `?` in the `repeat` method. The comment
indicating that the function call couldn't panic wasn't quite right in that if
`alloc_size` becomes too large and if `align` is high enough it could indeed
cause a panic.
This'll hopefully mean that panics never get introduced into code in the first
place, ensuring that `opt-level=z` is closer to `opt-level=s` in this regard.
Add error codes for libsyntax_ext
I intend to add error codes for `libsyntax_ext` as well. However, they cannot be used at stage 0 directly so I thought it might be possible to enable them at the stage 1 only so we can have access to the macros. However, the error code registration seems to not work this way. Currently I get the following error:
```
error: used diagnostic code E0660 not registered
--> libsyntax_ext/asm.rs:93:25
|
93 | span_err!(cx, sp, E0660, "malformed inline assembly");
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
= note: this error originates in a macro outside of the current crate (in Nightly builds, run with -Z external-macro-backtrace for more info)
error: used diagnostic code E0661 not registered
--> libsyntax_ext/asm.rs:151:33
|
151 | / span_err!(cx, sp, E0661,
152 | | "output operand constraint lacks '=' or '+'");
| |________________________________________________________________________________________^
|
= note: this error originates in a macro outside of the current crate (in Nightly builds, run with -Z external-macro-backtrace for more info)
error: aborting due to 2 previous errors
error: Could not compile `syntax_ext`.
```
If anyone has an idea, I'd gladly take it. I'm trying to figure this out on my side as well. I also opened this PR to know if it was worth it to continue (maybe we don't want this?).
Anyway, any answer for both questions is very welcome!
cc @rust-lang/compiler
add target features when extracting and running doctests
When rendering documentation, rustdoc will happily load target features into the cfg environment from the current target, but fails to do this when doing anything with doctests. This would lead to situations where, thanks to https://github.com/rust-lang/rust/pull/48759, functions tagged with `#[target_feature]` couldn't run doctests, thanks to the automatic `#[doc(cfg(target_feature = "..."))]`.
Currently, there's no way to pass codegen options to rustdoc that will affect its rustc sessions, but for now this will let you use target features that come default on the platform you're targeting.
Fixes https://github.com/rust-lang/rust/issues/49723
Cleanup liballoc use statements
Some modules were still using the deprecated `allocator` module, use the
`alloc` module instead.
Some modules were using `super` while it's not needed.
Some modules were more or less ordering them, and other not, so the
latter have been modified to match the others.
Make --emit=metadata output metadata regardless of link
Fixes#40109. I'm not sure whether this condition was important here or not, but I can't see why it is required (removing it doesn't cause the error the comment warns about, so I'm assuming it's safe). If this is too heavy-handed, I can special-case on `OutputType::Metadata`.
r? @nrc
improve Atomic*::fetch_update docs
This clarifies that fetch_update *always* returns the previous value, either as `Ok(_)` or `Err(_)`, depending on whether the supplied update function returned `Some(_)` or `None`.
[doc] note the special type inference handling for shift ops
This adds a note to the docs about the difference between the shift ops and the corresponding trait methods when it comes to type inference.
Ignore copyright year when generating deriving span tests
Previously, generate-deriving-span-tests.py would regenerate all the tests anew, even if they hadn't changed. This creates unnecessary diffs that only change the copyright year. Now we check to see if any of the content of the test has changed before generating the new one.
Add docs for the test crate with the std docs
If the compiler docs aren't going to include the test crate then it may as well be included with std.
Fixes#49388
Cross-compile builder to Windows for PRs on Travis
I chose a completely arbitrary windows target here (I have no idea what's best, we could do multiple -- they are relatively fast).
proc_macro: Avoid cached TokenStream more often
This commit adds even more pessimization to use the cached `TokenStream` inside
of an AST node. As a reminder the `proc_macro` API requires taking an arbitrary
AST node and transforming it back into a `TokenStream` to hand off to a
procedural macro. Such functionality isn't actually implemented in rustc today,
so the way `proc_macro` works today is that it stringifies an AST node and then
reparses for a list of tokens.
This strategy unfortunately loses all span information, so we try to avoid it
whenever possible. Implemented in #43230 some AST nodes have a `TokenStream`
cache representing the tokens they were originally parsed from. This
`TokenStream` cache, however, has turned out to not always reflect the current
state of the item when it's being tokenized. For example `#[cfg]` processing or
macro expansion could modify the state of an item. Consequently we've seen a
number of bugs (#48644 and #49846) related to using this stale cache.
This commit tweaks the usage of the cached `TokenStream` to compare it to our
lossy stringification of the token stream. If the tokens that make up the cache
and the stringified token stream are the same then we return the cached version
(which has correct span information). If they differ, however, then we will
return the stringified version as the cache has been invalidated and we just
haven't figured that out.
Closes#48644Closes#49846
Make OnDiskCache thread-safer
I'm not sure if `synthetic_expansion_infos` is handled correctly.
`interpret_alloc_cache` and `interpret_alloc_size` seems to be wrong though, since the code may now decode two `AllocId`s in parallel. I'd like some input on how to fix that.
cc @oli-obk
r? @michaelwoerister
macros: Remove matching on "complex" nonterminals requiring AST comparisons
So, you can actually use nonterminals from outer macros in left hand side of nested macros and invocations of nested macros will try to match passed arguments to them.
```rust
macro outer($nt_item: item) {
macro inner($nt_item) {
struct S;
}
inner!($nt_item); // OK, `$nt_item` matches `$nt_item`
}
```
Why this is bad:
- We can't do this matching correctly. When two nonterminals are compared, the original tokens are lost and we have to compare AST fragments instead. Right now the comparison is done by `PartialEq` impls derived on AST structures.
- On one hand, AST loses information compared to original tokens (e.g. trailing separators and other simplifications done during parsing to AST), so we can produce matches that are not actually correct.
- On another hand derived `PartialEq` impls for AST structures don't make much sense in general and compare various auxiliary garbage like spans. For the argument nonterminal to match we should use literally the same token (possibly cloned) as was used in the macro LHS (as in the example above). So we can reject matches that are actually correct.
- Support for nonterminal matching is the only thing that forces us to derive `PartialEq` for all (!) AST structures. As I mentioned these impls are also mostly nonsensical.
This PR removes support for matching on all nonterminals except for "simple" ones like `ident`, `lifetime` and `tt` for which we have original tokens that can be compared.
After this is done I'll submit another PR removing huge number of `PartialEq` impls from AST and HIR structures.
This is an arcane feature and I don't personally know why would anyone use it, but the change should ideally go through crater.
We'll be able to support this feature again in the future when all nonterminals have original token streams attached to them in addition to (or instead of) AST fragments.
Some modules were still using the deprecated `allocator` module, use the
`alloc` module instead.
Some modules were using `super` while it's not needed.
Some modules were more or less ordering them, and other not, so the
latter have been modified to match the others.
`Layout` is often used at the core of allocation APIs and is as a result pretty
sensitive to codegen in various circumstances. I was profiling `-C opt-level=z`
with a wasm project recently and noticed that the `unwrap()` wasn't removed
inside of `Layout`, causing the program to be much larger than it otherwise
would be. If inlining were more aggressive LLVM would have figured out that the
panic could be eliminated, but in general the methods here can't panic in the
first place!
As a result this commit makes the following tweaks:
* Removes `unwrap()` and replaces it with `unsafe` in `Layout::new` and
`Layout::for_value`. For posterity though a debug assertion was left behind.
* Removes an `unwrap()` in favor of `?` in the `repeat` method. The comment
indicating that the function call couldn't panic wasn't quite right in that if
`alloc_size` becomes too large and if `align` is high enough it could indeed
cause a panic.
This'll hopefully mean that panics never get introduced into code in the first
place, ensuring that `opt-level=z` is closer to `opt-level=s` in this regard.
traits: Implement interning for Goal and Clause
r? @nikomatsakis
Close#49054
Contains some refactoring for the interning mechanism, mainly aimed at reducing pain when changing types of interning map.
This should be mostly good, although I'm not sure with the naming of `Goal::from_poly_domain_goal`.
Add GlobalAlloc trait + tweaks for initial stabilization
This is the outcome of discussion at the Rust All Hands in Berlin. The high-level goal is stabilizing sooner rather than later the ability to [change the global allocator](https://github.com/rust-lang/rust/issues/27389), as well as allocating memory without abusing `Vec::with_capacity` + `mem::forget`.
Since we’re not ready to settle every detail of the `Alloc` trait for the purpose of collections that are generic over the allocator type (for example the possibility of a separate trait for deallocation only, and what that would look like exactly), we propose introducing separately **a new `GlobalAlloc` trait**, for use with the `#[global_allocator]` attribute.
We also propose a number of changes to existing APIs. They are batched in this one PR in order to minimize disruption to Nightly users.
The plan for initial stabilization is detailed in the tracking issue https://github.com/rust-lang/rust/issues/49668.
CC @rust-lang/libs, @glandium
## Immediate breaking changes to unstable features
* For pointers to allocated memory, change the pointed type from `u8` to `Opaque`, a new public [extern type](https://github.com/rust-lang/rust/issues/43467). Since extern types are not `Sized`, `<*mut _>::offset` cannot be used without first casting to another pointer type. (We hope that extern types can also be stabilized soon.)
* In the `Alloc` trait, change these pointers to `ptr::NonNull` and change the `AllocErr` type to a zero-size struct. This makes return types `Result<ptr::NonNull<Opaque>, AllocErr>` be pointer-sized.
* Instead of a new `Layout`, `realloc` takes only a new size (in addition to the pointer and old `Layout`). Changing the alignment is not supported with `realloc`.
* Change the return type of `Layout::from_size_align` from `Option<Self>` to `Result<Self, LayoutErr>`, with `LayoutErr` a new opaque struct.
* A `static` item registered as the global allocator with the `#[global_allocator]` **must now implement the new `GlobalAlloc` trait** instead of `Alloc`.
## Eventually-breaking changes to unstable features, with a deprecation period
* Rename the respective `heap` modules to `alloc` in the `core`, `alloc`, and `std` crates. (Yes, this does mean that `::alloc::alloc::Alloc::alloc` is a valid path to a trait method if you have `exetrn crate alloc;`)
* Rename the the `Heap` type to `Global`, since it is the entry point for what’s registered with `#[global_allocator]`.
Old names remain available for now, as deprecated `pub use` reexports.
## Backward-compatible changes
* Add a new [extern type](https://github.com/rust-lang/rust/issues/43467) `Opaque`, for use in pointers to allocated memory.
* Add a new `GlobalAlloc` trait shown below. Unlike `Alloc`, it uses bare `*mut Opaque` without `NonNull` or `Result`. NULL in return values indicates an error (of unspecified nature). This is easier to implement on top of `malloc`-like APIs.
* Add impls of `GlobalAlloc` for both the `Global` and `System` types, in addition to existing impls of `Alloc`. This enables calling `GlobalAlloc` methods on the stable channel before `Alloc` is stable. Implementing two traits with identical method names can make some calls ambiguous, but most code is expected to have no more than one of the two traits in scope. Erroneous code like `use std::alloc::Global; #[global_allocator] static A: Global = Global;` (where `Global` is defined to call itself, causing infinite recursion) is not statically prevented by the type system, but we count on it being hard enough to do accidentally and easy enough to diagnose.
```rust
extern {
pub type Opaque;
}
pub unsafe trait GlobalAlloc {
unsafe fn alloc(&self, layout: Layout) -> *mut Opaque;
unsafe fn dealloc(&self, ptr: *mut Opaque, layout: Layout);
unsafe fn alloc_zeroed(&self, layout: Layout) -> *mut Opaque {
// Default impl: self.alloc() and ptr::write_bytes()
}
unsafe fn realloc(&self, ptr: *mut Opaque, old_layout: Layout, new_size: usize) -> *mut Opaque {
// Default impl: self.alloc() and ptr::copy_nonoverlapping() and self.dealloc()
}
fn oom(&self) -> ! {
// intrinsics::abort
}
// More methods with default impls may be added in the future
}
```
## Bikeshed
The tracking issue https://github.com/rust-lang/rust/issues/49668 lists some open questions. If consensus is reached before this PR is merged, changes can be integrated.
Hygiene 2.0: Avoid comparing fields by name
There are two separate commits here (not counting tests):
- The first one unifies named (`obj.name`) and numeric (`obj.0`) field access expressions in AST and HIR. Before field references in these expressions are resolved it doesn't matter whether the field is named or numeric (it's just a symbol) and 99% of code is common. After field references are resolved we work with
them by index for all fields (see the second commit), so it's again not important whether the field was named or numeric (this includes MIR where all fields were already by index).
(This refactoring actually fixed some bugs in HIR-based borrow checker where borrows through names (`S {
0: ref x }`) and indices (`&s.0`) weren't considered overlapping.)
- The second commit removes all by-name field comparison and instead resolves field references to their indices once, and then uses those resolutions. (There are still a few name comparisons in save-analysis, because save-analysis is weird, but they are made correctly hygienic).
Thus we are fixing a bunch of "secondary" field hygiene bugs (in borrow checker, lints).
Fixes https://github.com/rust-lang/rust/issues/46314
Ak 44493 infer predicate
**WIP** Implements #44493
Things to do:
- [x] add feature gate and appropriate tests (see [forge](https://forge.rust-lang.org/feature-guide.html) for some details)
- [x] add a unit testing system similar to `#[rustc_variance]`
- [x] to see how, maybe `rg rustc_variance` and take some notes
- [ ] add more tests:
- [x] we need to decide how to handle `struct Foo<'a, T> { x: &'a T::Item }`
- [x] handle explicit predicates on types
- [ ] handle explicit predicates on `dyn Trait` (this could be put off to a follow-up PR)
- [ ] handle explicit predicates on projections (this could be put off to a follow-up PR)
