Shortcuts for min/max on double-ended BTreeMap/BTreeSet iterators
Closes#59947: a performance tweak that might benefit some. Optimizes `min` and `max ` on all btree double-ended iterators that do not drop, i.e. the iterators created by:
- `BTreeMap::iter`
- `BTreeMap::iter_mut`
- `BTreeMap::keys` and `BTreeSet::iter`
- `BTreeMap::range` and `BTreeSet::range`
- `BTreeMap::range_mut`
Also in these (currently) single-ended iterators, but obviously for `min` only:
- `BTreeSet::difference`
- `BTreeSet::intersection`
- `BTreeSet::symmetric_difference`
- `BTreeSet::union`
Did not do this in iterators created by `into_iter` to preserve drop order, as outlined in #62316.
Did not do this in iterators created by `drain_filter`, possibly to preserve drop order, possibly to preserve predicate invocation, mostly to not have to think about it too hard (I guess maybe it wouldn't be a change for `min`, which is the only shortcut possible in this single-ended iterator).
Fortanix SGX target libunwind build process changes
Ticket: https://github.com/fortanix/rust-sgx/issues/174
LLVM related changes (merged): https://github.com/rust-lang/llvm-project/pull/57
Description: libunwind changes needed to run code in sgx environment via rust-sgx.
Target that uses this in rust: x86_64-fortanix-unknown-sgx.
Without this change, rust std for this toolchain is forced to use a precompiled library loaded via environment variable.
With this change we act the same as musl target.
This is a re-attempt of #72389 (which was reverted in #73594)
Instead of using `ExpnKind::Desugaring` to represent operators, this PR
checks the lang item directly.
Show the values and computation that would overflow a const evaluation or propagation
Fixes#71134
In contrast to the example in the issue it doesn't use individual spans for each operand. The effort required to implement that is quite high compared to the little (if at all) benefit it would bring to diagnostics.
cc @shepmaster
The way this is implemented it is also fairly easy to do the same for overflow panics at runtime, but that should be done in a separate PR since it may have runtime performance implications.
- Pass around document_private a lot more
- Add tests
+ Add tests for intra-doc links to private items
+ Add ignored tests for warnings in reference links
This commit replaces the remaining uses of `ParamEnv::reveal_all` with
`LateContext`'s `param_env` (normally `Reveal::UserFacing`).
Signed-off-by: David Wood <david@davidtw.co>
Rollup of 13 pull requests
Successful merges:
- #72620 (Omit DW_AT_linkage_name when it is the same as DW_AT_name)
- #72967 (Don't move cursor in search box when using arrows to navigate results)
- #73102 (proc_macro: Stop flattening groups with dummy spans)
- #73297 (Support configurable deny-warnings for all in-tree crates.)
- #73507 (Cleanup MinGW LLVM linkage workaround)
- #73588 (Fix handling of reserved registers for ARM inline asm)
- #73597 (Record span of `const` kw in GenericParamKind)
- #73629 (Make AssocOp Copy)
- #73681 (Update Chalk to 0.14)
- #73707 (Fix links in `SliceIndex` documentation)
- #73719 (emitter: column width defaults to 140)
- #73729 (disable collectionbenches for android)
- #73748 (Add code block to code in documentation of `List::rebase_onto`)
Failed merges:
r? @ghost
emitter: column width defaults to 140
Fixes#72509.
This PR modifies the column width computation in the emitter when `termize::dimensions` returns `None` so that it uses the default value of 140 (which is used in UI testing currently) instead of `usize::MAX` which just ends up causing overflows in later computations.
I also tried changing the computations which used `column_width` with their saturating equivalent, but the output appeared the same - so I decided to go with this approach because I feel like it's less likely to accidentally re-introduce an ICE like this in future (e.g. adding a non-saturating operation on `column_width` in future).
I haven't added a test because I couldn't come up with a MCVE. I stumbled upon this running rustc-perf with the `piston-image` benchmark (running in tmux; it only happened with stage two builds only; and only when running through Cargo, not rustc directly with the same flags). In addition, given the nature of the issue, I don't know that we *could* write a UI test for this. Open to suggestions here though.
r? @estebank
Update Chalk to 0.14
Not a ton here. Notable changes:
- Update to `0.14.0`
- New dependency on `tracing`, in `librustc_traits` only
- `FnAbi` from Chalk is `rustc_target::spec::abi::Abi`
- `Dynamic` actually lowers region
- Actually lower closures, with some tests. This doesn't 100% work, but can't confirm that's *only* because of closure lowering.
- Use `FxIndexSet` instead of `FxHashSet` in `chalk_fulfill`, which seems to have fixed the non-deterministic test error ordering. Guess we'll see on CI
- Actually implement `opaque_ty_data`, though I don't think this is sufficient for tests for them (I haven't added any)
- Uncomment some of the chalk tests that now work
r? @nikomatsakis
Record span of `const` kw in GenericParamKind
Context: this is needed for a fix of https://github.com/rust-lang/rustfmt/issues/4263,
which currently records the span of a const generic param incorrectly
because the location of the `const` kw is not known.
I am not sure how to add tests for this; any guidance in how to do so
would be appreciated 🙂
Fix handling of reserved registers for ARM inline asm
`r6` is now disallowed as an operand since LLVM sometimes uses it as a base pointer.
The check against using the frame pointer as an operand now takes the platform into account and will block either `r7` or `r11` as appropriate.
Fixes#73450
cc @cbiffle
Cleanup MinGW LLVM linkage workaround
LLVM correctly passes `uuid` for a long time, GCC unwinding library (either static or dynamic) comes from [windows_gnu_base](06e47688bf/src/librustc_target/spec/windows_gnu_base.rs).
Support configurable deny-warnings for all in-tree crates.
This removes the hard-coded `deny(warnings)` on all in-tree tools, and allows it to be configured from the config. This is just a personal preference, as I find `deny(warnings)` frustrating during development or doing small tests.
This also fixes some regressions in terms of warning handling. Warnings used to be dependent on `SourceType`, but in #64316 it was changed to be based on `Mode`. This means tools like rustdoc no longer used the same settings as the rest of the tree. It also made `SourceType` useless since the only thing it was used for was warnings. I think it would be better for everything in the tree to use the same settings.
Fixes#64523
proc_macro: Stop flattening groups with dummy spans
Reduce the scope of the hack described in https://github.com/rust-lang/rust/issues/72545#issuecomment-640276791.
We still pass AST fragments to attribute and derive macros as single nonterminal tokens rather than as tokens streams, but now use a precise flag instead of the span-based heuristic that could do lead to incorrect behavior in unrelated cases.
https://github.com/rust-lang/rust/pull/73345 attempts to fully resolve this issue, but there are some compatibility issues to be addressed.
Don't move cursor in search box when using arrows to navigate results
## What happens
- Go to https://doc.rust-lang.org/stable/std/index.html
- Press 's' to focus the search box
- Type a query like 'test'
- Press the down arrow one or more times to change which search result is highlighted
- Press the up arrow once to go up one search result
- Notice the cursor in the search box is now at the beginning of your query, such that if you now typed 'a' the search box would contain 'atest', when it would be expected that the cursor would have remained where it was and if you typed 'a' at this point it would result in 'testa'
- Press the down arrow once to go down one search result
- Now notice the cursor is at the end of your query again
## What I expected
I expected that changing which search result was highlighted using the up and down arrows would have no effect on where the cursor was in the search box.
## The fix
This PR prevents the default action of the up and down arrows when the custom keydown events are happening during a search.
Omit DW_AT_linkage_name when it is the same as DW_AT_name
The DWARF standard suggests that it might be useful to include
`DW_AT_linkage_name` when it is *distinct* from the identifier name.
Fixes#46487.
Fixes#59422.
rustc_lint: only query `typeck_tables_of` when a lint needs it.
This was prompted by @jyn514 running into a situation where `rustdoc` wants to run the `unused_doc` lint without triggering type-checking (as an alternative to the "everybody loops" approach - type-checking may error/ICE because of the `rustdoc` feature of allowing multi-platform docs where the actual bodies of functions may refer to APIs for different platforms).
There was also this comment in the source:
```rust
// FIXME: Make this lazy to avoid running the TypeckTables query?
```
The main effect of this is for lint authors, who now need to use `cx.tables()` to get `&TypeckTables`, as opposed to having them always available in `cx.tables`.
r? @oli-obk or @Manishearth
