Fix panic message when `RangeFrom` index is out of bounds
Before, the `Range` method was called with `end = slice.len()`. Unfortunately, because `Range::index` first checks the order of the indices (start has to be smaller than end), an out of bounds index leads to `core::slice::slice_index_order_fail` being called. This prints the message 'slice index starts at 27 but ends at 10', which is worse than 'index 27 out of range for slice of length 10'. This is not only useful to normal users reading panic messages, but also for people inspecting assembly and being confused by `slice_index_order_fail` calls.
You can see the produced assembly [here](https://rust.godbolt.org/z/GzMGWf) and try on Playground [here](https://play.rust-lang.org/?version=stable&mode=debug&edition=2018&gist=aada5996b2f3848075a6d02cf4055743). (By the way. this is only about which panic function is called; I'm pretty sure it does not improve anything about performance).
Document the where keyword
Partial fix of #34601 (and last PR for it 🎉).
This documents the `where` keyword.
@rustbot modify labels: T-doc,C-enhancement
Document the ref keyword
Partial fix for #34601.
This documents the `ref` keyword with two examples, one failing to compile because the `ref` keyword is missing, and the same example fixed with the keyword inserted in the correct place.
It also explains (very *very* rapidly) the differences between `&` and `ref`.
I put a link to the best place I could find in the Reference but there may be something better that I didn't find.
Miri: use extern fn to expose interpreter operations to program; fix leak checker on Windows
This PR realizes an idea that @oli-obk has been suggesting for a while: to use Miri-specific `extern` functions to provide some extra capabilities to the program. Initially, we have two of these methods, which libstd itself needs:
* `miri_start_panic`, which replaces the intrinsic of the same name (mostly for consistency, to avoid having multiple mechanisms for Miri-specific functionality).
* `miri_static_root`, which adds an allocation to a list of static "roots" that Miri considers as not having leaked (including all memory reachable through them). This is needed for https://github.com/rust-lang/miri/issues/1302.
We use `extern` functions instead of intrinsics for this so that user code can more easily call these Miri hoolks -- e.g. `miri_static_root` should be useful for https://github.com/rust-lang/miri/issues/1318.
The Miri side of this is at https://github.com/rust-lang/miri/pull/1485.
r? @oli-obk
Rollup of 8 pull requests
Successful merges:
- #72954 (revise RwLock for HermitCore)
- #74367 (Rearrange the pipeline of `pow` to gain efficiency)
- #74491 (Optimize away BitAnd and BitOr when possible)
- #74639 (Downgrade glibc to 2.11.1 for ppc, ppc64 and s390x)
- #74661 (Refactor `region_name`: add `RegionNameHighlight`)
- #74692 (delay_span_bug instead of silent ignore)
- #74698 (fixed error reporting for mismatched traits)
- #74715 (Add a system for creating diffs across multiple mir optimizations.)
Failed merges:
r? @ghost
Refactor `region_name`: add `RegionNameHighlight`
This PR does not change any diagnostics itself, rather it enables further code changes, but I would like to get approval for the refactoring first before making use of it.
In `rustc_mir::borrow_check::diagnostics::region_name`, there is code that allows for, when giving a synthesized name like `'1` to an anonymous lifetime, pointing at e.g. the exact '`&`' that introduces the lifetime.
This PR decouples that code from the specific case of arguments, adding a new enum `RegionNameHighlight`, enabling future changes to use it in other places.
This allows:
* We could change the other `AnonRegionFrom*` variants to use `RegionNameHighlight` to precisely point at where lifetimes are introduced in other locations when they have type annotations, e.g. a closure return `|...| -> &i32`.
* Because of how async functions are lowered this affects async functions as well, see #74072
* for #74597, we could add a second, optional `RegionNameHighlight` to the `AnonRegionFromArgument` variant that highlights a lifetime in the return type of a function when, due to elision, this is the same as the argument lifetime.
* in https://github.com/rust-lang/rust/issues/74497#issuecomment-6606229707 I noticed that a diagnostic was trying to introduce a lifetime `'2` in the opaque type `impl std::future::Future`. The code for the case of arguments has [code to handle cases like this](bbebe7351f/src/librustc_mir/borrow_check/diagnostics/region_name.rs (L365)) but not the others. This refactoring would allow the same code path to handle this.
* It might be appropriate to add another variant of `RegionNameHighlight` to say something like `lifetime '1 appears in the opaque type impl std::future::Future`.
These are quite a few changes so I thought I would make sure the refactoring is OK before I start making changes that rely on it. :)
Downgrade glibc to 2.11.1 for ppc, ppc64 and s390x
As discussed in #73782
I've tested these changes on rust 1.43.0 for all the specified archs and used the resulting binaries to bootstrap building rust 1.43.1.
I've also shortly tested these changes on master on ppc64.
Optimize away BitAnd and BitOr when possible
This PR lets `const_prop` optimize away `a | true == true` , `a & false == false` and `a * 0 = 0`. While I was writing this I've realized that constant propagation misses a lot of opportunities. For example: https://play.rust-lang.org/?version=stable&mode=debug&edition=2018&gist=2a4b45e772f214210a36749b27223bb0
Constant propagation doesn't seem to... propagate constants, additionally the way constant propagation is currently setup makes it tricky to add cases like `a | false == a`.
I tried to organize `eval_rvalue_with_identities` to make the pattern of the optimizations easier to see but it still obscurs what should be a simple peephole optmization.
cc @oli-obk
revise RwLock for HermitCore
- current version is derived from the wasm implementation
- increasing the readability of `Condvar`
- simplify the interface to the libos
correctly deal with unsorted generic parameters
We now stop sorting generic params and instead correctly handle unsorted params in the rest of the compiler.
We still restrict const params to come after type params though, so this PR does not change anything which
is visible to users.
This might slightly influence perf, so let's prevent any unintentional rollups. @bors rollup=never
r? @varkor
Fix ICE while building MIR with type errors
See https://github.com/rust-lang/rust/issues/74047#issuecomment-663290913 for background. Replacing a binding with `PatKind::Wild` (introduced in #51789 and later refactored in #67439) caused an ICE downstream while building MIR.
I noticed that taking this code out no longer triggers the ICEs it was added to prevent. I'm not sure what else changed, or if this change is _correct_, but it does seem to be passing ui tests at least.
r? @oli-obk
cc @estebank
Fixes#74047.
Remove needless unsafety from BTreeMap::drain_filter
Remove one piece of unsafe code in the iteration over the iterator returned by BTreeMap::drain_filter.
- Changes an explicitly unspecified part of the API: when the user-supplied predicate (or some of BTreeMap's code) panicked, and the caller tries to use the iterator again, we no longer offer the same key/value pair to the predicate again but pretend the iterator has finished. Note that Miri does not find UB in the test case added here with the unsafe code (or without).
- Makes the code a little easier on the eyes.
- Makes the code a little harder on the CPU:
```
benchcmp c0 c2 --threshold 3
name c0 ns/iter c2 ns/iter diff ns/iter diff % speedup
btree::set::clone_100_and_drain_all 2,794 2,900 106 3.79% x 0.96
btree::set::clone_100_and_drain_half 2,604 2,964 360 13.82% x 0.88
btree::set::clone_10k_and_drain_half 287,770 322,755 34,985 12.16% x 0.89
```
r? @Amanieu
More BTreeMap test cases, some exposing undefined behaviour
Gathered from other ongoing PRs and all either blessed or ignored by Miri
r? @Mark-Simulacrum
Internally unify rustc_deprecated and deprecated
This PR intentionally tries to be "featureless" in that the behavior is not altered for either attribute, though it more clearly exposes cases where that is the case in the code.
