These assertions detect situations where a BCB node would have both a physical
counter and one or more in-edge counters/expressions.
For most BCBs that situation would indicate an implementation bug. However,
it's perfectly fine in the case of a BCB having an edge that loops back to
itself.
Given the complexity and risk involved in fixing the assertions, and the fact
that nothing relies on them actually being true, this patch just removes them
instead.
When we try to extract coverage-relevant spans from MIR, sometimes we see MIR
statements/terminators whose spans cover the entire function body. Those spans
tend to be unhelpful for coverage purposes, because they often represent
compiler-inserted code, e.g. the implicit return value of `()`.
Some of these tests were originally written as part of a custom `run-make`
test, so at that time they weren't able to use the normal compiletest header
directive parser.
Now that they're properly integrated, there's no need for them to use
`compile-flags` to specify the edition, since they can use `edition` instead.
coverage: Don't instrument `#[automatically_derived]` functions
This PR makes the coverage instrumentor detect and skip functions that have [`#[automatically_derived]`](https://doc.rust-lang.org/reference/attributes/derive.html#the-automatically_derived-attribute) on their enclosing impl block.
Most notably, this means that methods generated by built-in derives (e.g. `Clone`, `Debug`, `PartialEq`) are now ignored by coverage instrumentation, and won't appear as executed or not-executed in coverage reports.
This is a noticeable change in user-visible behaviour, but overall I think it's a net improvement. For example, we've had a few user requests for this sort of change (e.g. #105055, https://github.com/rust-lang/rust/issues/84605#issuecomment-1902069040), and I believe it's the behaviour that most users will expect/prefer by default.
It's possible to imagine situations where users would want to instrument these derived implementations, but I think it's OK to treat that as an opportunity to consider adding more fine-grained option flags to control the details of coverage instrumentation, while leaving this new behaviour as the default.
(Also note that while `-Cinstrument-coverage` is a stable feature, the exact details of coverage instrumentation are allowed to change. So we *can* make this change; the main question is whether we *should*.)
Fixes#105055.
coverage: Format all coverage tests with `rustfmt`
As suggested by <https://github.com/rust-lang/rust/pull/119984#discussion_r1452856806>.
Test files in `tests/` are normally ignored by `x fmt`, but sometimes those files end up being run through `rustfmt` anyway, either by `rust-analyzer` or by hand.
When that happens, it's annoying to have to manually revert formatting changes that are unrelated to the actual changes being made. So it's helpful for the tests in the repository to already have standard formatting beforehand.
However, there are several coverage tests that deliberately use non-standard formatting, so that line counts reveal more information about where code regions begin and end. In those cases, we can use `#[rustfmt::skip]` to prevent that code from being disturbed.
``@rustbot`` label +A-code-coverage
Some of these tests use non-standard formatting that we can simulate by
strategically adding `//` line comments.
One contains `where` clauses that would be split across multiple lines, which
we can keep on one line by moving the bounds to the generic type instead.
These tests deliberately use non-standard formatting, so that the line
execution counts reported by `llvm-cov` reveal additional information about
where code regions begin and end.
coverage: Hoist some complex code out of the main span refinement loop
The span refinement loop in `spans.rs` takes the spans that have been extracted from MIR, and modifies them to produce more helpful output in coverage reports.
It is also one of the most complicated pieces of code in the coverage instrumentor. It has an abundance of moving pieces that make it difficult to understand, and most attempts to modify it end up accidentally changing its behaviour in unacceptable ways.
This PR nevertheless tries to make a dent in it by hoisting two pieces of special-case logic out of the main loop, and into separate preprocessing passes. Coverage tests show that the resulting mappings are *almost* identical, with all known differences being unimportant.
This should hopefully unlock further simplifications to the refinement loop, since it now has fewer edge cases to worry about.
Normally, each test in `tests/coverage` is automatically run in both
`coverage-map` mode and `coverage-run` mode.
This new family of directives allows an individual test to specify that it
should not be run in a particular mode.
coverage: Skip instrumenting a function if no spans were extracted from MIR
The immediate symptoms of #118643 were fixed by #118666, but some users reported that their builds now encounter another coverage-related ICE:
```
error: internal compiler error: compiler/rustc_codegen_llvm/src/coverageinfo/mapgen.rs:98:17: A used function should have had coverage mapping data but did not: (...)
```
I was able to reproduce at least one cause of this error: if no relevant spans could be extracted from a function, but the function contains `CoverageKind::SpanMarker` statements, then codegen still thinks the function is instrumented and complains about the fact that it has no coverage spans.
This PR prevents that from happening in two ways:
- If we didn't extract any relevant spans from MIR, skip instrumenting the entire function and don't create a `FunctionCoverateInfo` for it.
- If coverage codegen sees a `CoverageKind::SpanMarker` statement, skip it early and avoid creating `func_coverage`.
---
Fixes#118850.
coverage: Simplify the heuristic for ignoring `async fn` return spans
The code for extracting coverage spans from MIR has a special heuristic for dealing with `async fn`, so that the function's closing brace does not have a confusing double count.
The code implementing that heuristic is currently mixed in with the code for flushing remaining spans after the main refinement loop, making the refinement code harder to understand.
We can solve that by hoisting the heuristic to an earlier stage, after the spans have been extracted and sorted but before they have been processed by the refinement loop.
The coverage tests verify that the heuristic is still effective, so coverage mappings/reports for `async fn` have not changed.
---
This PR also has the side-effect of fixing the `None some_prev` panic that started appearing after #118525.
The old code assumed that `prev` would always be present after the refinement loop. That was only true if the list of collected spans was non-empty, but prior to #118525 that didn't seem to come up in practice. After that change, the list of collected spans could be empty in some specific circumstances, leading to panics.
The new code uses an `if let` to inspect `prev`, which correctly does nothing if there is no span present.
When MIR is built for an if-not expression, the `!` part of the condition
doesn't correspond to any MIR statement, so coverage instrumentation normally
can't see it.
We can fix that by deliberately injecting a dummy statement whose sole purpose
is to associate that span with its enclosing block.
When we extract coverage spans from MIR, we try to "un-expand" them back to
spans that are inside the function's body span.
In cases where that doesn't succeed, the current code just swaps in the entire
body span instead. But that tends to result in coverage spans that are
completely unrelated to the control flow of the affected code, so it's better
to just discard those spans.
Add thinlto support to codegen, assembly and coverage tests
Using `--emit=llvm-ir` with thinlto usually result in multiple IR files.
Resolve test case failure issue reported in #113923.
In some cases we need to prepare a coverage expression that is the sum of an
arbitrary number of other terms. This patch simplifies the code paths that
build those sums.
This causes some churn in the mappings, because the previous code was building
its sums in a somewhat idiosyncratic order.
Without the workaround applied, this test will produce malformed mappings that
cause `llvm-cov` to fail.
(And if it does emit well-formed mappings, they should be obviously incorrect.)
