Cleanup and document `-C relocation-model`
As the title says, this is mostly a refactoring and documentation.
One potentially observable change here is that `-C relocation-model=default` now takes the default from the Rust target, rather than from the underlying LLVM target. In other words, `-C relocation-model=default` is now equivalent to not specifying the relocation model on command line at all.
Apparently no one used that option because it has other bugs as well, e.g. PIC `default` wasn't treated as PIC in some places.
When working with an arbitrary reader or writer, code that uses vectored
operations may end up being slower than code that copies into a single
buffer when the underlying reader or writer doesn't actually support
vectored operations. These new methods allow you to ask the reader or
witer up front if vectored operations are efficiently supported.
Currently, you have to use some heuristics to guess by e.g. checking if
the read or write only accessed the first buffer. Hyper is one concrete
example of a library that has to do this dynamically:
0eaf304644/src/proto/h1/io.rs (L582-L594)
The referenced `sanitizer-address/Makefile` no longer exists, so perhaps these options are no longer necessary as well.
Even if they are still necessary, they should use `-C relocation-model=static` instead.
If an extern C function is defined as
```
extern "C" {
fn malloc(size: u32) -> *mut std::ffi::c_void;
}
```
on a 64-bit machine(ie. pointer sizes don't match), throw an undefined
behaviour.
Replace thread_local with generator resume arguments in box_region.
Fixes#68922.
Continuation of #70622. Added a short doc, hope it makes sense.
r? @jonas-schievink
Don't hold the predecessor cache lock longer than necessary
#71044 returns a `LockGuard` with the predecessor cache to callers of `Body::predecessors`. As a result, the lock around the predecessor cache could be held for an arbitrarily long time. This PR uses reference counting for ownership of the predecessor cache, meaning the lock is only ever held within `PredecessorCache::compute`. Checking this API for potential sources of deadlock is much easier now, since we no longer have to consider its consumers, only its internals.
This required removing `predecessors_for`, since there is no equivalent to `LockGuard::map` for `Arc` and `Rc`. I believe this could be emulated with `owning_ref::{Arc,Rc}Ref`, but I don't think it's necessary. Also, we continue to return an opaque type from `Body::predecessors` with the lifetime of the `Body`, not `'static`.
This depends on #71044. Only the last two commits are new.
r? @nikomatsakis
[breaking change] Disallow statics initializing themselves
fixes#71078
Self-initialization is unsound because it breaks privacy assumptions that unsafe code can make. In
```rust
pub mod foo {
#[derive(Debug, Copy, Clone)]
pub struct Foo {
x: (),
}
}
pub static FOO: foo::Foo = FOO;
```
unsafe could could expect that ony functions inside the `foo` module were able to create a value of type `Foo`.
Add all remaining `DefKind`s.
r? @eddyb or @Centril
~~I'm not sure if this is what you were thinking of. There are also a few places where I'm not sure what the correct choice is because I don't fully understand the meaning of some variants.~~
~~In general, it feels a bit odd to add some of these as `DefKind`s (e.g. `Arm`) because they don't feel like definitions. Are there things that it makes sense not to add?~~
Rollup of 7 pull requests
Successful merges:
- #69041 (proc_macro: Stabilize `Span::resolved_at` and `Span::located_at`)
- #69813 (Implement BitOr and BitOrAssign for the NonZero integer types)
- #70712 (stabilize BTreeMap::remove_entry)
- #71168 (Deprecate `{Box,Rc,Arc}::into_raw_non_null`)
- #71544 (Replace filter_map().next() calls with find_map())
- #71545 (Fix comment in docstring example for Error::kind)
- #71548 (Add missing Send and Sync impls for linked list Cursor and CursorMut.)
Failed merges:
r? @ghost
Add missing Send and Sync impls for linked list Cursor and CursorMut.
Someone pointed out these to me, and i think it's indeed reasonable to add those impl.
r? @Amanieu
Implement BitOr and BitOrAssign for the NonZero integer types
This provides overloaded operators for `NonZero$Int | NonZero$Int`, `NonZero$Int | $Int`, and `$Int | NonZero$Int`. It also provides `BitOrAssign` where `self` is `NonZero$Int`, for symmetry.
It's a pretty small conceptual addition, but is good becasue but avoids a case where the operation is obviously sound, but you'd otherwise need unsafe to do it.
In crates trying to minimize `unsafe` usage, this is unfortunate and makes working with `NonZero` types often not worth it, even if the operations you're doing are clearly sound.
I've marked these as stable as I've been told in the past that trait impls are automatically stable. I'm happy to change it to unstable if this wasn't correct information.
I'm not entirely confident what version I should have put down, so I followed https://www.whatrustisit.com. Hopefully it's correct for this.
Apologies in advance if this has come up before, but I couldn't find it.
This was only used for linkage test cases, which is already covered by
the run-make-fulldeps/symbol-visibility test -- which fairly extensively makes
sure we're correctly exporting the right symbols at the right visibility (for
various Rust crate types).
Rollup of 5 pull requests
Successful merges:
- #71364 (Ignore -Zprofile when building compiler_builtins)
- #71494 (Fix span of while (let) expressions after lowering)
- #71517 ( Quick and dirty fix of the unused_braces lint)
- #71523 (Take a single root node in range_search)
- #71533 (Revert PR 70566 for const validation fix)
Failed merges:
r? @ghost
Take a single root node in range_search
The unsafe code can be justified within range_search, as it makes sure to not
overlap the returned references, but from the callers perspective it's an
entirely safe algorithm and there's no need for the caller to know about the
duplication.
cc @ssomers
r? @Amanieu
Quick and dirty fix of the unused_braces lint
cc @lcnr
Adresses #70814
This at least prevents lint output, if no span is available. Even though this also prevents the `unused_parens` lint from emitting, when the `DUMMY_SP` is used there, but I think that should be ok, since error messages without a span are quite useless anyway.
Clippy CI is currently blocked on this bug. If this quick and dirty fix should be rejected, I could try to work around this in Clippy.
r? @shepmaster
Fix span of while (let) expressions after lowering
Credit goes to @alex-700 who found this while trying to fix a suggestion in Clippy.
While `if`, `try`, `for` and `await` expressions get the span of the original expression when desugared, `while` loops got the span of the scrutinee, which lead to weird code, when building the suggestion, that randomly worked: https://github.com/rust-lang/rust-clippy/pull/5511/files#diff-df4e9d2bf840a5f2e3b580bef73da3bcR106-R108
I'm wondering, if `DesugaringKind` should get a variant `WhileLoop` and instead of using the span of the `ast::ExprKind::While` expr directly, a new span with `self.mark_span_with_reason` should be used, like it is done with `for` loops.
There was some fallout, but I think that is acceptable. If not, I need some help to find out where this can be fixed.
Ignore -Zprofile when building compiler_builtins
#70846 made the `compiler_builtins` crate ignore the default codegen-units setting and instead always split each function into a different codegen unit.
This unfortunately breaks `-Zprofile` which requires a single codegen unit per crate (see #71283). You can notice this when building with `cargo -Zbuild-std` and `RUSTFLAGS` containing `-Zprofile`.
This PR works around this issue by just ignoring `-Zprofile` for the `compiler-builtins` crate.
Set `--cfg bootstrap` for stage0 rustdoc
Resolves#71455.
With this patch, running `./x.py doc --stage 0 src/libstd` with a clean `build` dir successfully outputs docs for `core`, `alloc` and `std` in under a minute. This kind of turnaround for viewing small changes to the standard library documentation is quite nice, and I think we should endeavour to keep it working. I'm not sure how involved that would be though.
r? @Mark-Simulacrum
