rustdoc-json: Better Header Type
- Make ABI an enum, instead of being stringly typed
- Replace Qualifier HashSet with 3 bools
- Merge ABI field into header, as they always occor together
r? ``@CraftSpider``
``@rustbot`` modify labels: +A-rustdoc-json +T-rustdoc
Display for `ParamTy` and `ParamConst` is implemented in terms of print.
Using print avoids creating a new `FmtPrinter` just to display the
parameter name.
Introduce `ChunkedBitSet` and use it for some dataflow analyses.
This reduces peak memory usage significantly for some programs with very
large functions.
r? `@ghost`
This reduces peak memory usage significantly for some programs with very
large functions, such as:
- `keccak`, `unicode_normalization`, and `match-stress-enum`, from
the `rustc-perf` benchmark suite;
- `http-0.2.6` from crates.io.
The new type is used in the analyses where the bitsets can get huge
(e.g. 10s of thousands of bits): `MaybeInitializedPlaces`,
`MaybeUninitializedPlaces`, and `EverInitializedPlaces`.
Some refactoring was required in `rustc_mir_dataflow`. All existing
analysis domains are either `BitSet` or a trivial wrapper around
`BitSet`, and access in a few places is done via `Borrow<BitSet>` or
`BorrowMut<BitSet>`. Now that some of these domains are `ClusterBitSet`,
that no longer works. So this commit replaces the `Borrow`/`BorrowMut`
usage with a new trait `BitSetExt` containing the needed bitset
operations. The impls just forward these to the underlying bitset type.
This required fiddling with trait bounds in a few places.
The commit also:
- Moves `static_assert_size` from `rustc_data_structures` to
`rustc_index` so it can be used in the latter; the former now
re-exports it so existing users are unaffected.
- Factors out some common "clear excess bits in the final word"
functionality in `bit_set.rs`.
- Uses `fill` in a few places instead of loops.
tests: default to more threads for ui-tests
Benchmarks (tested on i5-7200U, 2 cores, 4 threads)
```
master branch:
cargo test // prime caches
cargo --color=always test 70,39s user 21,91s system 180% cpu 51,035 total
cargo --color=always test 70,77s user 22,13s system 180% cpu 51,579 total
cargo --color=always test 70,97s user 22,12s system 180% cpu 51,673 total
cargo --color=always nextest run 78,74s user 22,27s system 220% cpu 45,829 total
cargo --color=always nextest run 78,46s user 21,92s system 224% cpu 44,674 total
cargo --color=always nextest run 78,31s user 22,21s system 228% cpu 43,909 total
Patched (ui_speedup branch):
cargo test // prime cache
cargo --color=always test 97,51s user 32,02s system 288% cpu 44,905 total
cargo --color=always test 99,19s user 31,91s system 276% cpu 47,436 total
cargo --color=always test 98,47s user 31,84s system 284% cpu 45,744 total
cargo --color=always nextest run 102,18s user 30,80s system 350% cpu 37,902 total
cargo --color=always nextest run 99,75s user 29,86s system 350% cpu 36,935 total
cargo --color=always nextest run 100,36s user 29,93s system 351% cpu 37,061 total
```
changelog: use more threads for running clippys ui-tests for ~10% walltime speedup
Fix a layout possible miscalculation in `alloc::RawVec`
A layout miscalculation could happen in `RawVec` when used with a type whose size isn't a multiple of its alignment. I don't know if such type can exist in Rust, but the Layout API provides ways to manipulate such types. Anyway, it is better to calculate memory size in a consistent way.
Rollup of 7 pull requests
Successful merges:
- #94169 (Fix several asm! related issues)
- #94178 (tidy: fire less "ignoring file length unneccessarily" warnings)
- #94179 (solarish current_exe using libc call directly)
- #94196 (compiletest: Print process output info with less whitespace)
- #94208 (Add the let else tests found missing in the stabilization report)
- #94237 (Do not suggest wrapping an item if it has ambiguous un-imported methods)
- #94246 (ScalarMaybeUninit is explicitly hexadecimal in its formatting)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
ScalarMaybeUninit is explicitly hexadecimal in its formatting
This makes `ScalarMaybeUninit` consistent with `Scalar` after the changes in https://github.com/rust-lang/rust/pull/94189.
r? ``@oli-obk``
Do not suggest wrapping an item if it has ambiguous un-imported methods
If the method is defined for the receiver we have, but is ambiguous during probe, then it probably comes from one of several traits that just weren't `use`d. Don't suggest wrapping the receiver in `Box`/etc., even if that makes the method probe unambiguous.
Fixes#94218
Add the let else tests found missing in the stabilization report
In the stabilization report of `let else`, in #93628, I found various cases which weren't tested. This PR adds them.
tidy: fire less "ignoring file length unneccessarily" warnings
This avoids a situation where a file is at the border of the limit,
and alternates between hitting the limit and not hitting it, causing
a back and forth of addition of the ignore-tidy-linelength directive.
As an example, consider the ignore-tidy-filelength of compiler/rustc_typeck/src/collect.rs.
It was added in 2ca4964db5, removed in
37354ebc97 (a revert of the earlier commit), added again in 448d07683a,
removed in 3171bd5bf5, added in 438826fd1a,
and removed in bb0a2f985c.
To avoid this back and forth, we exempt files from the unneccessary
ignoring warning that have length of at least 70% of the limit.
