Add APIs for uninitialized Box, Rc, and Arc. (Plus get_mut_unchecked)
Assigning `MaybeUninit::<Foo>::uninit()` to a local variable is usually free, even when `size_of::<Foo>()` is large. However, passing it for example to `Arc::new` [causes at least one copy](https://youtu.be/F1AquroPfcI?t=4116) (from the stack to the newly allocated heap memory) even though there is no meaningful data. It is theoretically possible that a Sufficiently Advanced Compiler could optimize this copy away, but this is [reportedly unlikely to happen soon in LLVM](https://youtu.be/F1AquroPfcI?t=5431).
This PR proposes two sets of features:
* Constructors for containers (`Box`, `Rc`, `Arc`) of `MaybeUninit<T>` or `[MaybeUninit<T>]` that do not initialized the data, and unsafe conversions to the known-initialized types (without `MaybeUninit`). The constructors are guaranteed not to make unnecessary copies.
* On `Rc` and `Arc`, an unsafe `get_mut_unchecked` method that provides `&mut T` access without checking the reference count. `Arc::get_mut` involves multiple atomic operations whose cost can be non-trivial. `Rc::get_mut` is less costly, but we add `Rc::get_mut_unchecked` anyway for symmetry with `Arc`.
These can be useful independently, but they will presumably be typical when the new constructors of `Rc` and `Arc` are used.
An alternative with a safe API would be to introduce `UniqueRc` and `UniqueArc` types that have the same memory layout as `Rc` and `Arc` (and so zero-cost conversion to them) but are guaranteed to have only one reference. But introducing entire new types feels “heavier” than new constructors on existing types, and initialization of `MaybeUninit<T>` typically requires unsafe code anyway.
Summary of new APIs (all unstable in this PR):
```rust
impl<T> Box<T> { pub fn new_uninit() -> Box<MaybeUninit<T>> {…} }
impl<T> Box<MaybeUninit<T>> { pub unsafe fn assume_init(self) -> Box<T> {…} }
impl<T> Box<[T]> { pub fn new_uninit_slice(len: usize) -> Box<[MaybeUninit<T>]> {…} }
impl<T> Box<[MaybeUninit<T>]> { pub unsafe fn assume_init(self) -> Box<[T]> {…} }
impl<T> Rc<T> { pub fn new_uninit() -> Rc<MaybeUninit<T>> {…} }
impl<T> Rc<MaybeUninit<T>> { pub unsafe fn assume_init(self) -> Rc<T> {…} }
impl<T> Rc<[T]> { pub fn new_uninit_slice(len: usize) -> Rc<[MaybeUninit<T>]> {…} }
impl<T> Rc<[MaybeUninit<T>]> { pub unsafe fn assume_init(self) -> Rc<[T]> {…} }
impl<T> Arc<T> { pub fn new_uninit() -> Arc<MaybeUninit<T>> {…} }
impl<T> Arc<MaybeUninit<T>> { pub unsafe fn assume_init(self) -> Arc<T> {…} }
impl<T> Arc<[T]> { pub fn new_uninit_slice(len: usize) -> Arc<[MaybeUninit<T>]> {…} }
impl<T> Arc<[MaybeUninit<T>]> { pub unsafe fn assume_init(self) -> Arc<[T]> {…} }
impl<T: ?Sized> Rc<T> { pub unsafe fn get_mut_unchecked(this: &mut Self) -> &mut T {…} }
impl<T: ?Sized> Arc<T> { pub unsafe fn get_mut_unchecked(this: &mut Self) -> &mut T {…} }
```
Opaque builtin derive macros
* Buiilt-in derives are now opaque macros
* This required limiting the visibility of some previously unexposed functions in `core`.
* This also required the change to `Ident` serialization.
* All gensyms are replaced with hygienic identifiers
* Use hygiene to avoid most other name-resolution issues with buiilt-in derives.
* As far as I know the only remaining case that breaks is an ADT that has the same name as one of its parameters. Fixing this completely seemed to be more effort than it's worth.
* Remove gensym in `Ident::decode`, which lead to linker errors due to `inline` being gensymmed.
* `Ident`now panics if incremental compilation tries to serialize it (it currently doesn't).
* `Ident` no longer uses `gensym` to emulate cross-crate hygiene. It only applied to reexports.
* `SyntaxContext` is no longer serializable.
* The long-term fix for this is to properly implement cross-crate hygiene, but this seemed to be acceptable for now.
* Move type/const parameter shadowing checks to `resolve`
* This was previously split between resolve and type checking. The type checking pass compared `InternedString`s, not Identifiers.
* Removed the `SyntaxContext` from `{ast, hir}::{InlineAsm, GlobalAsm}`
cc #60869
r? @petrochenkov
Modify librustc_llvm to pass -DNDEBUG while compiling.
Currently, librustc_llvm builds are not reproducible because the LLVM files it compiles use the debug version of llvm_unreachable, which uses __FILE__. To fix this, we propagate NDEBUG from bootstrap if applicable and use it when compiling librustc_llvm.
r? @alexcrichton
rustc_codegen_utils: account for 1-indexed anonymous lifetimes in v0 mangling.
I don't really understand why `anonymize_late_bound_regions` starts with `BrAnon(1)` instead of `BrAnon(0)`, but it does (maybe @nikomatsakis knows?): c43d03a19f/src/librustc/ty/fold.rs (L696-L712)
Thankfully, the mangling format and demangler implementations are fine, and I just needed to offset the anonymized lifetime indices by `1` to get the correct mangling.
cc @alexcrichton @michaelwoerister
Hash the remapped sysroot instead of the original.
One of the reasons that rustc builds are not reproducible is because the --sysroot path is dependent on the current directory. We can fix this by hashing the remapped sysroot instead of the original when applicable.
Note that with this patch, the hash will stay the same if both the sysroot and the remapped path change. However, given that if the contents of the sysroot change the hash will also stay the same, this might be acceptable. I would appreciate feedback on the best way to do this.
This helps #34902, although it does not fix it by itself.
Override Cycle::try_fold
It's not very pretty, but I believe this is the simplest way to correctly implement `Cycle::try_fold`. The following may seem correct:
```rust
loop {
acc = self.iter.try_fold(acc, &mut f)?;
self.iter = self.orig.clone();
}
```
...but this loops infinitely in case `self.orig` is empty, as opposed to returning `acc`. So we first have to fully iterate `self.orig` to check whether it is empty or not, and before _that_, we have to iterate the remaining elements of `self.iter`.
This should always call `self.orig.clone()` the same amount of times as repeated `next()` calls would.
r? @scottmcm
Most `Ident`s are serialized as `InternedString`s the exceptions are:
* Reexports
* Attributes
* Idents in macro definitions
Using gensyms helped reexports emulate hygiene. However, the actual item
wouldn't have a gensymmed name so would be usable cross-crate. So
removing this case until we have proper cross-crate hygiene seems
sensible.
Codegen attributes (`inline`, `export_name`) are resolved by their
`Symbol`. This meant that opaque macro-expanded codegen attributes could
cause linker errors. This prevented making built-in derives hygienic.
Update RLS
This fixes handling default configuration for the `crate_blacklist`
RLS configuration.
Technically this isn't needed, as the VS Code extension can be
configured to accept a predefined blacklist that's equal to the default
one but it's best that it also lands so that we don't need to work
around that.
Without this, manually passing a `null` value as the configuration
unfortunately crashes the RLS. This is the last fix related to configuration.
cc https://github.com/rust-lang/rust/pull/63472
r? @Mark-Simulacrum
Rename overflowing_{add,sub,mul} intrinsics to wrapping_{add,sub,mul}.
These confused @Gankra, and then, also me, especially since `overflowing_*` *methods* also exist, but they map to `*_with_overflow` intrinsics!
r? @oli-obk / @nikomatsakis cc @Mark-Simulacrum (on the rustbuild workaround)
add git keyword to submodule comments in config.example.toml
I searched over config.example.toml file looking for a place to disable git submodules from being updated, and missed the two options related to this because they did not include the keyword git. This pr simply adds git to the relevant comments so hopefully others won't also miss that these options exist.
Remap paths for proc-macro crates.
The remap-debuginfo config option remaps paths in most crates, but it does not apply to proc-macros, so they are still non-reproducible. This patch fixes that.
I'm not completely sure if this is the best way to do this, but to get reproducible builds we need librustc_macros to be built with --remap-path-prefix. I was previously modifying Cargo to pass that argument to all child crates, so this seems simpler and more correct.
I did not add a test since there do not seem to be any existing tests for RUSTC_DEBUGINFO_MAP.
r? @alexcrichton
resolve: Populate external modules in more automatic and lazy way
So, resolve had this function `populate_module_if_necessary` for loading module children from other crates from metadata.
I never really understood when it should've been called and when not.
This PR removes the function and loads the module children automatically on the first access instead.
r? @eddyb
Rollup of 10 pull requests
Successful merges:
- #60492 (Add custom nth_back for Chain)
- #61780 (Finalize the error type for `try_reserve`)
- #63495 ( Remove redundant `ty` fields from `mir::Constant` and `hair::pattern::PatternRange`.)
- #63525 (Make sure that all file loading happens via SourceMap)
- #63595 (add sparc64-unknown-openbsd target)
- #63604 (Some update for vxWorks)
- #63613 (Hygienize use of built-in macros in the standard library)
- #63632 (A couple of comment fixes.)
- #63634 (ci: properly set the job name in CPU stats)
- #63636 (ci: move linkcheck from mingw-2 to mingw-1)
Failed merges:
r? @ghost
Move some code into `build_reduced_graph.rs` to keep `BuildReducedGraphVisitor` it private
Also move the def collector call to `build_reduced_graph.rs`, it belongs there.
ci: move linkcheck from mingw-2 to mingw-1
Running UI tests now takes a huge amount of time on mingw builders
(between 40 and 50 minutes), with mingw-1 builders taking even an hour
less to finish than mingw-2. This PR moves linkcheck from mingw-2 to
mingw-1, removing between 10 and 20 minutes of runtime on the -2
builders.
r? @alexcrichton
A couple of comment fixes.
This change fixes a couple of comments in the compiler code.
The first change was discussed with @eddyb. This one confused me because I thought only *public* defs were in the table (not so).
The second was a typo I noticed yesterday.
OK to go in?
Some update for vxWorks
1. support crt-static
2. change armv7_wrs_vxworks to armv7_wrs_vxworks_eabihf.
3. change vx-cxx to wr-c++, vx-ar to wr-ar and vx-run to wr-run.
4. code cleanup
r? @alexcrichton
add sparc64-unknown-openbsd target
on OpenBSD, some architectures relies on libc++ (from LLVM) and some
others on libestdc++ (particular version of libstdc++ from GCC).
sparc64-unknown-openbsd needs libestdc++ and libgcc (as x86_64 some
years ago). Reintroduce the support of them for openbsd, only for
sparc64 arch. Some others architectures on OpenBSD could use them too.
Make sure that all file loading happens via SourceMap
That way, callers don't need to repeat "let's add this to sm manually
for tracking dependencies" trick.
It should make it easier to switch to using `FileLoader` for binary
files in the future as well
cc #62948
r? @petrochenkov
Remove redundant `ty` fields from `mir::Constant` and `hair::pattern::PatternRange`.
Fixes#56137.
As a side-effect, associated const literals have the correct type now, which should make things easier for #61041.
r? @oli-obk / @matthewjasper cc @davidtwco @varkor
Finalize the error type for `try_reserve`
See tracking issue comments from https://github.com/rust-lang/rust/issues/48043#issuecomment-500828346.
It is now:
```rust
/// The error type for `try_reserve` methods.
#[derive(Clone, PartialEq, Eq, Debug)]
#[unstable(feature = "try_reserve", reason = "new API", issue="48043")]
pub enum TryReserveError {
/// Error due to the computed capacity exceeding the collection's maximum
/// (usually `isize::MAX` bytes).
CapacityOverflow,
/// The memory allocator returned an error
AllocError {
/// The layout of allocation request that failed
layout: Layout,
#[doc(hidden)]
#[unstable(feature = "container_error_extra", issue = "0", reason = "\
Enable exposing the allocator’s custom error value \
if an associated type is added in the future: \
https://github.com/rust-lang/wg-allocators/issues/23")]
non_exhaustive: (),
},
}
#[unstable(feature = "try_reserve", reason = "new API", issue="48043")]
impl From<LayoutErr> for TryReserveError {
#[inline]
fn from(_: LayoutErr) -> Self {
TryReserveError::CapacityOverflow
}
}
```
Changes:
* A `Layout` is included. Firefox wants to log the size of failed allocations. If this were not part of the return value of e.g. `HashMap::try_reserve`, users would only be able to estimate based on `HashMap::capacity` and assumptions about the allocation strategy of `HashMap`.
* There’s a dummy field that can stay unstable when `try_reserve` and the rest of this enum are stabilized. This forces non-exhaustive matching ~(https://github.com/rust-lang/rust/issues/44109 is not implemented yet for variants)~ and allows adding another field in the future if we want to expose custom error values from the allocator. See https://github.com/rust-lang/wg-allocators/issues/23.
- If the `Alloc` trait is stabilized without an associated error type and with a zero-size `AllocErr` type, we can simply remove this dummy field.
- If an associated type is added, we can add a default type parameter to `ContainerError` and a generic field to the `AllocError` variant.
* ~Moved from the `collections` module to the `alloc` module, and replaced `Collection` in the enum name with `Container`. The wold collection implies a multiplicity of items which is not relevant to this type. For example we may want to use this error type in a future `Box::try_new` method.~
- Renamed to `TryReserveError`, after the methods that involve this type: https://github.com/rust-lang/rust/pull/61780#issuecomment-501392487
* Replaced `Err` with `Error` in the enum and variant names. There is more precedent for this in https://doc.rust-lang.org/std/error/trait.Error.html#implementors, `AllocErr` and `LayoutErr` are the odd ones.
* ~Dropped `Alloc` in the enum name. `ContainerAllocError` with a mouthful, and being in the `alloc` module already provides the same indication.~
… and add a separately-unstable field to force non-exhaustive matching
(`#[non_exhaustive]` is no implemented yet on enum variants)
so that we have the option to later expose the allocator’s error value.
CC https://github.com/rust-lang/wg-allocators/issues/23
