Do not promote &mut of a non-ZST ever
Since ~pre-1.0~ 1.36, we have accepted code like this:
```rust
static mut TEST: &'static mut [i32] = {
let x = &mut [1,2,3];
x
};
```
I tracked it back to https://github.com/rust-lang/rust/pull/21744, but unfortunately could not find any discussion or RFC that would explain why we thought this was a good idea. And it's not, it breaks all sorts of things -- see https://github.com/rust-lang/rust/issues/75556.
To fix https://github.com/rust-lang/rust/issues/75556, we have to stop promoting non-ZST mutable references no matter the context, which is what this PR does. It's a breaking change.
Notice that this still works, since it does not rely on promotion:
```rust
static mut TEST: &'static mut [i32] = &mut [0,1,2];
```
Cc `@rust-lang/wg-const-eval`
Support dataflow problems on arbitrary lattices
This PR implements last of the proposed extensions I mentioned in the design meeting for the original dataflow refactor. It extends the current dataflow framework to work with arbitrary lattices, not just `BitSet`s. This is a prerequisite for dataflow-enabled MIR const-propagation. Personally, I am skeptical of the usefulness of doing const-propagation pre-monomorphization, since many useful constants only become known after monomorphization (e.g. `size_of::<T>()`) and users have a natural tendency to hand-optimize the rest. It's probably worth exprimenting with, however, and others have shown interest cc `@rust-lang/wg-mir-opt.`
The `Idx` associated type is moved from `AnalysisDomain` to `GenKillAnalysis` and replaced with an associated `Domain` type that must implement `JoinSemiLattice`. Like before, each `Analysis` defines the "bottom value" for its domain, but can no longer override the dataflow join operator. Analyses that want to use set intersection must now use the `lattice::Dual` newtype. `GenKillAnalysis` impls have an additional requirement that `Self::Domain: BorrowMut<BitSet<Self::Idx>>`, which effectively means that they must use `BitSet<Self::Idx>` or `lattice::Dual<BitSet<Self::Idx>>` as their domain.
Most of these changes were mechanical. However, because a `Domain` is no longer always a powerset of some index type, we can no longer use an `IndexVec<BasicBlock, GenKillSet<A::Idx>>>` to store cached block transfer functions. Instead, we use a boxed `dyn Fn` trait object. I discuss a few alternatives to the current approach in a commit message.
The majority of new lines of code are to preserve existing Graphviz diagrams for those unlucky enough to have to debug dataflow analyses. I find these diagrams incredibly useful when things are going wrong and considered regressing them unacceptable, especially the pretty-printing of `MovePathIndex`s, which are used in many dataflow analyses. This required a parallel `fmt` trait used only for printing dataflow domains, as well as a refactoring of the `graphviz` module now that we cannot expect the domain to be a `BitSet`. Some features did have to be removed, such as the gen/kill display mode (which I didn't use but existed to mirror the output of the old dataflow framework) and line wrapping. Since I had to rewrite much of it anyway, I took the opportunity to switch to a `Visitor` for printing dataflow state diffs instead of using cursors, which are error prone for code that must be generic over both forward and backward analyses. As a side-effect of this change, we no longer have quadratic behavior when writing graphviz diagrams for backward dataflow analyses.
r? `@pnkfelix`
inliner: Check for codegen fn attributes compatibility
* Check for target features compatibility
* Check for no_sanitize attribute compatibility
Fixes#76259.
Tools, tests, and experimenting with MIR-derived coverage counters
Leverages the new mir_dump output file in HTML+CSS (from #76074) to visualize coverage code regions
and the MIR features that they came from (including overlapping spans).
See example below.
The `run-make-fulldeps/instrument-coverage` test has been refactored to maximize test coverage and reduce code duplication. The new tests support testing with and without `-Clink-dead-code`, so Rust coverage can be tested on MSVC (which, currently, only works with `link-dead-code` _disabled_).
New tests validate coverage region generation and coverage reports with multiple counters per function. Starting with a simple `if-else` branch tests, coverage tests for each additional syntax type can be added by simply dropping in a new Rust sample program.
Includes a basic, MIR-block-based implementation of coverage injection,
available via `-Zexperimental-coverage`. This implementation has known
flaws and omissions, but is simple enough to validate the new tools and
tests.
The existing `-Zinstrument-coverage` option currently enables
function-level coverage only, which at least appears to generate
accurate coverage reports at that level.
Experimental coverage is not accurate at this time. When branch coverage
works as intended, the `-Zexperimental-coverage` option should be
removed.
This PR replaces the bulk of PR #75828, with the remaining parts of
that PR distributed among other separate and indentpent PRs.
This PR depends on two of those other PRs: #76002, #76003 and #76074
Rust compiler MCP rust-lang/compiler-team#278
Relevant issue: #34701 - Implement support for LLVMs code coverage
instrumentation
r? @tmandry
FYI: @wesleywiser
diagnostics: shorten paths of unique symbols
This is a step towards implementing a fix for #50310, and continuation of the discussion in [Pre-RFC: Nicer Types In Diagnostics - compiler - Rust Internals](https://internals.rust-lang.org/t/pre-rfc-nicer-types-in-diagnostics/11139). Impressed upon me from previous discussion in #21934 that an RFC for this is not needed, and I should just come up with code.
The recent improvements to `use` suggestions that I've contributed have given rise to this implementation. Contrary to previous suggestions, it's rather simple logic, and I believe it only reduces the amount of cognitive load that a developer would need when reading type errors.
-----
If a symbol name can only be imported from one place, and as long as it was not glob-imported anywhere in the current crate, we can trim its printed path to the last component.
This has wide implications on error messages with types, for example, shortening `std::vec::Vec` to just `Vec`, as long as there is no other `Vec` importable from anywhere.
Adds a new mir_dump output file in HTML/CSS to visualize code regions
and the MIR features that they came from (including overlapping spans).
See example below:
Includes a basic, MIR-block-based implementation of coverage injection,
available via `-Zexperimental-coverage`. This implementation has known
flaws and omissions, but is simple enough to validate the new tools and
tests.
The existing `-Zinstrument-coverage` option currently enables
function-level coverage only, which at least appears to generate
accurate coverage reports at that level.
Experimental coverage is not accurate at this time. When branch coverage
works as intended, the `-Zexperimental-coverage` option should be
removed.
This PR replaces the bulk of PR #75828, with the remaining parts of
that PR distributed among other separate and indentpent PRs.
This PR depends on three of those other PRs: #76000, #76002, and
Rust compiler MCP rust-lang/compiler-team#278
Relevant issue: #34701 - Implement support for LLVMs code coverage
instrumentation
If a symbol name can only be imported from one place for a type, and
as long as it was not glob-imported anywhere in the current crate, we
can trim its printed path and print only the name.
This has wide implications on error messages with types, for example,
shortening `std::vec::Vec` to just `Vec`, as long as there is no other
`Vec` importable anywhere.
This adds a new '-Z trim-diagnostic-paths=false' option to control this
feature.
On the good path, with no diagnosis printed, we should try to avoid
issuing this query, so we need to prevent trimmed_def_paths query on
several cases.
This change also relies on a previous commit that differentiates
between `Debug` and `Display` on various rustc types, where the latter
is trimmed and presented to the user and the former is not.
The HIR Id trick is insufficient to prevent query cycles when optimizing
generators, since merely requesting a layout of a generator also
computes its `optimized_mir`.
Make no attempts to inline functions into generators within the same
crate to avoid query cycles.
Similar to `-Z dump-mir-graphviz`, this adds the option to write
HTML+CSS files that allow users to analyze the spans associated with MIR
elements (by individual statement, just terminator, or overall basic
block).
This PR was split out from PR #76004, and exposes an API for spanview
HTML+CSS files that is also used to analyze code regions chosen for
coverage instrumentation (in a follow-on PR).
Rust compiler MCP rust-lang/compiler-team#278
Relevant issue: #34701 - Implement support for LLVMs code coverage
instrumentation
Fix `-Z instrument-coverage` on MSVC
Found that `-C link-dead-code` (which was enabled automatically
under `-Z instrument-coverage`) was causing the linking error that
resulted in segmentation faults in coverage instrumented binaries. Link
dead code is now disabled under MSVC, allowing `-Z instrument-coverage`
to be enabled under MSVC for the first time.
More details are included in Issue #76038 .
Note this PR makes it possible to support `Z instrument-coverage` but
does not enable instrument coverage for MSVC in existing tests. It will be
enabled in another PR to follow this one (both PRs coming from original
PR #75828).
r? @tmandry
FYI: @wesleywiser
Found that -C link-dead-code (which was enabled automatically
under -Z instrument-coverage) was causing the linking error that
resulted in segmentation faults in coverage instrumented binaries. Link
dead code is now disabled under MSVC, allowing `-Z instrument-coverage`
to be enabled under MSVC for the first time.
More details are included in Issue #76038.
(This PR was broken out from PR #75828)
I've tried a few ways of implementing this, but each fell short.
Adding an auxiliary `_Idx` associated type to `Analysis` that defaults
to `!` but is overridden in the blanket impl of `Analysis` for `A:
GenKillAnalysis` to `A::Idx` seems promising, but the trait solver is
unable to prove equivalence between `A::Idx` and `A::_Idx` within the
overridden version of `into_engine`. Without full-featured
specialization, removing `into_engine` or splitting it into a different
trait would have a significant ergonomic penalty.
Alternatively, we could erase the index type and store a
`GenKillSet<u32>` as well as a function pointer for transmuting between
`&mut A::Domain` and `&mut BitSet<u32>` in the hopes that LLVM can
devirtualize a simple function pointer better than the boxed closure.
However, this is brittle, requires `unsafe` code, and doesn't work for
index types that aren't the same size as a `u32` (e.g. `usize`) since
`GenKillSet` stores a `HybridBitSet`, which may be a `Vec<I>`. Perhaps
safe transmute could help here?
This commit removes the obsolete printer and replaces all uses of it
with `FmtPrinter`. Of the replaced uses, all but one use was in `debug!`
logging, two cases were notable:
- `MonoItem::to_string` is used in `-Z print-mono-items` and therefore
affects the output of all codegen-units tests.
- `DefPathBasedNames` was used in `librustc_codegen_llvm/type_of.rs`
with `LLVMStructCreateNamed` and that'll now get different values, but
this should result in no functional change.
Signed-off-by: David Wood <david@davidtw.co>
