Report coverage `0` of dead blocks
Fixes: #84018
With `-Z instrument-coverage`, coverage reporting of dead blocks
(for example, blocks dropped because a conditional branch is dropped,
based on const evaluation) is now supported.
If `instrument-coverage` is enabled, `simplify::remove_dead_blocks()`
finds all dropped coverage `Statement`s and adds their `code_region`s as
`Unreachable` coverage `Statement`s to the `START_BLOCK`, so they are
still included in the coverage map.
Check out the resulting changes in the test coverage reports in this PR (in [commit 1](0b0d293c7c)).
r? `@tmandry`
cc: `@wesleywiser`
Fixes: #84884
This solution might be considered a compromise, but I think it is the
better choice.
The results in the `closure.rs` test correctly resolve all test cases
broken as described in #84884.
One test pattern (in both `closure_macro.rs` and
`closure_macro_async.rs`) was also affected, and removes coverage
statistics for the lines inside the closure, because the closure
includes a macro. (The coverage remains at the callsite of the macro, so
we lose some detail, but there isn't a perfect choice with macros.
Often macro implementations are split across the macro and the callsite,
and there doesn't appear to be a single "right choice" for which body
should be covered. For the current implementation, we can't do both.
The callsite is most likely to be the preferred site for coverage.
I applied this fix to all `MacroKinds`, not just `Bang`.
I'm trying to resolve an issue of lost coverage in a
`MacroKind::Attr`-based, function-scoped macro. Instead of only
searching for a body_span that is "not a function-like macro" (that is,
MacroKind::Bang), I'm expanding this to all `MacroKind`s. Maybe I should
expand this to `ExpnKind::Desugaring` and `ExpnKind::AstPass` (or
subsets, depending on their sub-kinds) as well, but I'm not sure that's
a good idea.
I'd like to add a test of the `Attr` macro on functions, but I need time
to figure out how to constract a good, simple example without external
crate dependencies. For the moment, all tests still work as expected (no
change), this new commit shouldn't have a negative affect, and more
importantly, I believe it will have a positive effect. I will try to
confirm this.
Fixes: #84018
With `-Z instrument-coverage`, coverage reporting of dead blocks
(for example, blocks dropped because a conditional branch is dropped,
based on const evaluation) is now supported.
If `instrument-coverage` is enabled, `simplify::remove_dead_blocks()`
finds all dropped coverage `Statement`s and adds their `code_region`s as
`Unreachable` coverage `Statement`s to the `START_BLOCK`, so they are
still included in the coverage map.
Check out the resulting changes in the test coverage reports in this PR.
Vastly improves coverage spans for macros
Fixes: #84561
This resolves problems where macros like `trace!(...)` would show zero coverage if tracing was disabled, and `assert_eq!(...)` would show zero coverage if the assertion did not fail, because only one coverage span was generated, for the branch.
This PR started with an idea that I could just drop branching blocks with same span as expanded macro. (See the fixed issue for more details.)
That did help, but it didn't resolve everything.
I also needed to add a span specifically for the macro name (plus `!`) to ensure the macro gets coverage even if it's internal expansion adds conditional branching blocks that are retained, and would otherwise drop the outer span. Now that outer span is _only_ the `(argument, list)`, which can safely be dropped now), because the macro name has its own span.
While testing, I also noticed the spanview debug output can cause an ICE on a function with no body. The
workaround for this is included in this PR (separate commit).
r? `@tmandry`
cc? `@wesleywiser`
Adds feature-gated `#[no_coverage]` function attribute, to fix derived Eq `0` coverage issue #83601
Derived Eq no longer shows uncovered
The Eq trait has a special hidden function. MIR `InstrumentCoverage`
would add this function to the coverage map, but it is never called, so
the `Eq` trait would always appear uncovered.
Fixes: #83601
The fix required creating a new function attribute `no_coverage` to mark
functions that should be ignored by `InstrumentCoverage` and the
coverage `mapgen` (during codegen).
Adding a `no_coverage` feature gate with tracking issue #84605.
r? `@tmandry`
cc: `@wesleywiser`
Improve coverage spans for chained function calls
Fixes: #84180
For chained function calls separated by the `?` try operator, the
function call following the try operator produced a MIR `Call` span that
matched the span of the first call. The `?` try operator started a new
span, so the second call got no span.
It turns out the MIR `Call` terminator has a `func` `Operand`
for the `Constant` representing the function name, and the function
name's Span can be used to reset the starting position of the span.
r? `@tmandry`
cc: `@wesleywiser`
The Eq trait has a special hidden function. MIR `InstrumentCoverage`
would add this function to the coverage map, but it is never called, so
the `Eq` trait would always appear uncovered.
Fixes: #83601
The fix required creating a new function attribute `no_coverage` to mark
functions that should be ignored by `InstrumentCoverage` and the
coverage `mapgen` (during codegen).
While testing, I also noticed two other issues:
* spanview debug file output ICEd on a function with no body. The
workaround for this is included in this PR.
* `assert_*!()` macro coverage can appear covered if followed by another
`assert_*!()` macro. Normally they appear uncovered. I submitted a new
Issue #84561, and added a coverage test to demonstrate this issue.
Fix coverage ICE because fn_sig can have a span that crosses file bou…
Fixes: #83792
MIR `InstrumentCoverage` assumed the `FnSig` span was contained within a
single file, but this is not always the case. Some macro constructions
can result in a span that starts in one `SourceFile` and ends in a
different one.
The `FnSig` span is included in coverage results as long as that span is
in the same `SourceFile` and the same macro context, but by assuming the
`FnSig` span's `hi()` and `lo()` were in the same file, I took this for
granted, and checked only that the `FnSig` `hi()` was in the same
`SourceFile` as the `body_span`.
I actually drop the `hi()` though, and extend the `FnSig` span to the
`body_span.lo()`, so I really should have simply checked that the
`FnSig` span's `lo()` was in the `SourceFile` of the `body_span`.
r? `@tmandry`
cc: `@wesleywiser`
Fixes: #84180
For chained function calls separated by the `?` try operator, the
function call following the try operator produced a MIR `Call` span that
matched the span of the first call. The `?` try operator started a new
span, so the second call got no span.
It turns out the MIR `Call` terminator has a `func` `Operand`
for the `Constant` representing the function name, and the function
name's Span can be used to reset the starting position of the span.
further split up const_fn feature flag
This continues the work on splitting up `const_fn` into separate feature flags:
* `const_fn_trait_bound` for `const fn` with trait bounds
* `const_fn_unsize` for unsizing coercions in `const fn` (looks like only `dyn` unsizing is still guarded here)
I don't know if there are even any things left that `const_fn` guards... at least libcore and liballoc do not need it any more.
`@oli-obk` are you currently able to do reviews?
Fixes: #83792
MIR `InstrumentCoverage` assumed the `FnSig` span was contained within a
single file, but this is not always the case. Some macro constructions
can result in a span that starts in one `SourceFile` and ends in a
different one.
The `FnSig` span is included in coverage results as long as that span is
in the same `SourceFile` and the same macro context, but by assuming the
`FnSig` span's `hi()` and `lo()` were in the same file, I took this for
granted, and checked only that the `FnSig` `hi()` was in the same
`SourceFile` as the `body_span`.
I actually drop the `hi()` though, and extend the `FnSig` span to the
`body_span.lo()`, so I really should have simply checked that the
`FnSig` span's `lo()` was in the `SourceFile` of the `body_span`.
Implement a lint that highlights all moves larger than a configured limit
Tracking issue: #83518
[MCP 420](https://github.com/rust-lang/compiler-team/issues/420) still ~blazing~ in progress
r? ```@pnkfelix```
The main open issue I see with this minimal impl of the feature is that the lint is immediately "stable" (so it can be named on stable), even if it is never executed on stable. I don't think we have the concept of unstable lint names or hiding lint names without an active feature gate, so that would be a bigger change.
coverage of async function bodies should match non-async
This fixes some missing coverage within async function bodies.
Commit 1 demonstrates the problem in the fixed issue, and commit 2 corrects it.
Fixes: #83985
Don't concatenate binders across types
Partially addresses #83737
There's actually two issues that I uncovered in #83737. The first is that we are concatenating bound vars across types, i.e. in
```
F: Fn(&()) -> &mut (dyn Future<Output = ()> + Unpin)
```
the bound vars on `Future` get set as `for<anon>` since those are the binders on `Fn(&()`. This is obviously wrong, since we should only concatenate directly nested trait refs. This is solved here by introducing a new `TraitRefBoundary` scope, that we put around the "syntactical" trait refs and basically don't allow concatenation across.
Now, this alone *shouldn't* be a super terrible problem. At least not until you consider the other issue, which is a much more elusive and harder to design a "perfect" fix. A repro can be seen in:
```
use core::future::Future;
async fn handle<F>(slf: &F)
where
F: Fn(&()) -> &mut (dyn for<'a> Future<Output = ()> + Unpin),
{
(slf)(&()).await;
}
```
Notice the `for<'a>` around `Future`. Here, `'a` is unused, so the `for<'a>` Binder gets changed to a `for<>` Binder in the generator witness, but the "local decl" still has it. This has heavy intersections with region anonymization and erasing. Luckily, it's not *super* common to find this unique set of circumstances. It only became apparently because of the first issue mentioned here. However, this *is* still a problem, so I'm leaving #83737 open.
r? `@nikomatsakis`
Use AnonConst for asm! constants
This replaces the old system which used explicit promotion. See #83169 for more background.
The syntax for `const` operands is still the same as before: `const <expr>`.
Fixes#83169
Because the implementation is heavily based on inline consts, we suffer from the same issues:
- We lose the ability to use expressions derived from generics. See the deleted tests in `src/test/ui/asm/const.rs`.
- We are hitting the same ICEs as inline consts, for example #78174. It is unlikely that we will be able to stabilize this before inline consts are stabilized.
Add function core::iter::zip
This makes it a little easier to `zip` iterators:
```rust
for (x, y) in zip(xs, ys) {}
// vs.
for (x, y) in xs.into_iter().zip(ys) {}
```
You can `zip(&mut xs, &ys)` for the conventional `iter_mut()` and
`iter()`, respectively. This can also support arbitrary nesting, where
it's easier to see the item layout than with arbitrary `zip` chains:
```rust
for ((x, y), z) in zip(zip(xs, ys), zs) {}
for (x, (y, z)) in zip(xs, zip(ys, zs)) {}
// vs.
for ((x, y), z) in xs.into_iter().zip(ys).zip(xz) {}
for (x, (y, z)) in xs.into_iter().zip((ys.into_iter().zip(xz)) {}
```
It may also format more nicely, especially when the first iterator is a
longer chain of methods -- for example:
```rust
iter::zip(
trait_ref.substs.types().skip(1),
impl_trait_ref.substs.types().skip(1),
)
// vs.
trait_ref
.substs
.types()
.skip(1)
.zip(impl_trait_ref.substs.types().skip(1))
```
This replaces the tuple-pair `IntoIterator` in #78204.
There is prior art for the utility of this in [`itertools::zip`].
[`itertools::zip`]: https://docs.rs/itertools/0.10.0/itertools/fn.zip.html
coverage bug fixes and optimization support
Adjusted LLVM codegen for code compiled with `-Zinstrument-coverage` to
address multiple, somewhat related issues.
Fixed a significant flaw in prior coverage solution: Every counter
generated a new counter variable, but there should have only been one
counter variable per function. This appears to have bloated .profraw
files significantly. (For a small program, it increased the size by
about 40%. I have not tested large programs, but there is anecdotal
evidence that profraw files were way too large. This is a good fix,
regardless, but hopefully it also addresses related issues.
Fixes: #82144
Invalid LLVM coverage data produced when compiled with -C opt-level=1
Existing tests now work up to at least `opt-level=3`. This required a
detailed analysis of the LLVM IR, comparisons with Clang C++ LLVM IR
when compiled with coverage, and a lot of trial and error with codegen
adjustments.
The biggest hurdle was figuring out how to continue to support coverage
results for unused functions and generics. Rust's coverage results have
three advantages over Clang's coverage results:
1. Rust's coverage map does not include any overlapping code regions,
making coverage counting unambiguous.
2. Rust generates coverage results (showing zero counts) for all unused
functions, including generics. (Clang does not generate coverage for
uninstantiated template functions.)
3. Rust's unused functions produce minimal stubbed functions in LLVM IR,
sufficient for including in the coverage results; while Clang must
generate the complete LLVM IR for each unused function, even though
it will never be called.
This PR removes the previous hack of attempting to inject coverage into
some other existing function instance, and generates dedicated instances
for each unused function. This change, and a few other adjustments
(similar to what is required for `-C link-dead-code`, but with lower
impact), makes it possible to support LLVM optimizations.
Fixes: #79651
Coverage report: "Unexecuted instantiation:..." for a generic function
from multiple crates
Fixed by removing the aforementioned hack. Some "Unexecuted
instantiation" notices are unavoidable, as explained in the
`used_crate.rs` test, but `-Zinstrument-coverage` has new options to
back off support for either unused generics, or all unused functions,
which avoids the notice, at the cost of less coverage of unused
functions.
Fixes: #82875
Invalid LLVM coverage data produced with crate brotli_decompressor
Fixed by disabling the LLVM function attribute that forces inlining, if
`-Z instrument-coverage` is enabled. This attribute is applied to
Rust functions with `#[inline(always)], and in some cases, the forced
inlining breaks coverage instrumentation and reports.
FYI: `@wesleywiser`
r? `@tmandry`
Adjusted LLVM codegen for code compiled with `-Zinstrument-coverage` to
address multiple, somewhat related issues.
Fixed a significant flaw in prior coverage solution: Every counter
generated a new counter variable, but there should have only been one
counter variable per function. This appears to have bloated .profraw
files significantly. (For a small program, it increased the size by
about 40%. I have not tested large programs, but there is anecdotal
evidence that profraw files were way too large. This is a good fix,
regardless, but hopefully it also addresses related issues.
Fixes: #82144
Invalid LLVM coverage data produced when compiled with -C opt-level=1
Existing tests now work up to at least `opt-level=3`. This required a
detailed analysis of the LLVM IR, comparisons with Clang C++ LLVM IR
when compiled with coverage, and a lot of trial and error with codegen
adjustments.
The biggest hurdle was figuring out how to continue to support coverage
results for unused functions and generics. Rust's coverage results have
three advantages over Clang's coverage results:
1. Rust's coverage map does not include any overlapping code regions,
making coverage counting unambiguous.
2. Rust generates coverage results (showing zero counts) for all unused
functions, including generics. (Clang does not generate coverage for
uninstantiated template functions.)
3. Rust's unused functions produce minimal stubbed functions in LLVM IR,
sufficient for including in the coverage results; while Clang must
generate the complete LLVM IR for each unused function, even though
it will never be called.
This PR removes the previous hack of attempting to inject coverage into
some other existing function instance, and generates dedicated instances
for each unused function. This change, and a few other adjustments
(similar to what is required for `-C link-dead-code`, but with lower
impact), makes it possible to support LLVM optimizations.
Fixes: #79651
Coverage report: "Unexecuted instantiation:..." for a generic function
from multiple crates
Fixed by removing the aforementioned hack. Some "Unexecuted
instantiation" notices are unavoidable, as explained in the
`used_crate.rs` test, but `-Zinstrument-coverage` has new options to
back off support for either unused generics, or all unused functions,
which avoids the notice, at the cost of less coverage of unused
functions.
Fixes: #82875
Invalid LLVM coverage data produced with crate brotli_decompressor
Fixed by disabling the LLVM function attribute that forces inlining, if
`-Z instrument-coverage` is enabled. This attribute is applied to
Rust functions with `#[inline(always)], and in some cases, the forced
inlining breaks coverage instrumentation and reports.
Mark early otherwise optimization unsound
r? `@oli-obk`
cc `@tmiasko`
Related to #78496 and #82905
Should I also bump this one to level 3 or 4 or given that is unsound it doesn't matter?.
Probably need to adjust some tests.
Make source-based code coverage compatible with MIR inlining
When codegenning code coverage use the instance that coverage data was
originally generated for, to ensure basic level of compatibility with
MIR inlining.
Fixes#83061
Implement (but don't use) valtree and refactor in preparation of use
This PR does not cause any functional change. It refactors various things that are needed to make valtrees possible. This refactoring got big enough that I decided I'd want it reviewed as a PR instead of trying to make one huge PR with all the changes.
cc `@rust-lang/wg-const-eval` on the following commits:
* 2027184 implement valtree
* eeecea9 fallible Scalar -> ScalarInt
* 042f663 ScalarInt convenience methods
cc `@eddyb` on ef04a6d
cc `@rust-lang/wg-mir-opt` for cf1700c (`mir::Constant` can now represent either a `ConstValue` or a `ty::Const`, and it is totally possible to have two different representations for the same value)
When codegenning code coverage use the instance that coverage data was
originally generated for, to ensure basic level of compatibility with
MIR inlining.
Add Option::get_or_default
Tracking issue: #82901
The original issue is #55042, which was closed, but for an invalid reason (see discussion there). Opening this to reconsider (I hope that's okay). It seems like the only gap for `Option` being "entry-like".
I ran into a need for this method where I had a `Vec<Option<MyData>>` and wanted to do `vec[n].get_or_default().my_data_method()`. Using an `Option` as an inner component of a data structure is probably where the need for this will normally arise.
Stabilize `unsafe_op_in_unsafe_fn` lint
This makes it possible to override the level of the `unsafe_op_in_unsafe_fn`, as proposed in https://github.com/rust-lang/rust/issues/71668#issuecomment-729770896.
Tracking issue: #71668
r? ```@nikomatsakis``` cc ```@SimonSapin``` ```@RalfJung```
# Stabilization report
This is a stabilization report for `#![feature(unsafe_block_in_unsafe_fn)]`.
## Summary
Currently, the body of unsafe functions is an unsafe block, i.e. you can perform unsafe operations inside.
The `unsafe_op_in_unsafe_fn` lint, stabilized here, can be used to change this behavior, so performing unsafe operations in unsafe functions requires an unsafe block.
For now, the lint is allow-by-default, which means that this PR does not change anything without overriding the lint level.
For more information, see [RFC 2585](https://github.com/rust-lang/rfcs/blob/master/text/2585-unsafe-block-in-unsafe-fn.md)
### Example
```rust
// An `unsafe fn` for demonstration purposes.
// Calling this is an unsafe operation.
unsafe fn unsf() {}
// #[allow(unsafe_op_in_unsafe_fn)] by default,
// the behavior of `unsafe fn` is unchanged
unsafe fn allowed() {
// Here, no `unsafe` block is needed to
// perform unsafe operations...
unsf();
// ...and any `unsafe` block is considered
// unused and is warned on by the compiler.
unsafe {
unsf();
}
}
#[warn(unsafe_op_in_unsafe_fn)]
unsafe fn warned() {
// Removing this `unsafe` block will
// cause the compiler to emit a warning.
// (Also, no "unused unsafe" warning will be emitted here.)
unsafe {
unsf();
}
}
#[deny(unsafe_op_in_unsafe_fn)]
unsafe fn denied() {
// Removing this `unsafe` block will
// cause a compilation error.
// (Also, no "unused unsafe" warning will be emitted here.)
unsafe {
unsf();
}
}
```
This updates all places where match branches check on StatementKind or UseContext.
This doesn't properly implement them, but adds TODOs where they are, and also adds some best
guesses to what they should be in some cases.
Disable destination propagation on all mir-opt-levels
The new `// compile-flags: -Zunsound-mir-opts` are inserted without an extra newline to avoid introducing a large mir-opt diff.
Add option to enable MIR inlining independently of mir-opt-level
Add `-Zinline-mir` option that enables MIR inlining independently of the
current MIR opt level. The primary use-case is enabling MIR inlining on the
default MIR opt level.
Turn inlining thresholds into optional values to make it possible to configure
different defaults depending on the current mir-opt-level (although thresholds
are yet to be used in such a manner).
The storage markers constitute a substantial portion of all MIR
statements. At the same time, for builds without any optimizations,
the storage markers have no further use during and after MIR
optimization phase.
If storage markers are not necessary for code generation, remove them.
Use small hash set in `mir_inliner_callees`
Use small hash set in `mir_inliner_callees` to avoid temporary
allocation when possible and quadratic behaviour for large number of
callees.
Consider inexpensive inlining criteria first
Refactor inlining decisions so that inexpensive criteria are considered first:
1. Based on code generation attributes.
2. Based on MIR availability (examines call graph).
3. Based on MIR body.
Refactor inlining decisions so that inexpensive criteria are considered first:
1. Based on code generation attributes.
2. Based on MIR availability (examines call graph).
3. Based on MIR body.
