coverage: Don't rely on the custom traversal to find enclosing loops
This opens up the possibility of modifying or removing the custom graph traversal used in coverage counter creation, without losing access to the heuristics that care about a node's enclosing loops.
Actually changing the traversal is left for future work, because this PR on its own doesn't change the emitted coverage mappings at all.
Effects cleanup
- removed extra bits from predicates queries that are no longer needed in the new system
- removed the need for `non_erasable_generics` to take in tcx and DefId, removed unused arguments in callers
r? compiler-errors
- removed extra bits from predicates queries that are no longer needed in the new system
- removed the need for `non_erasable_generics` to take in tcx and DefId, removed unused arguments in callers
Then we can rename the _raw functions to drop their suffix, and instead
explicitly use is_stable_const_fn for the few cases where that is really what
you want.
Move `cmp_in_dominator_order` out of graph dominator computation
Dominator-order information is only needed for coverage graphs, and is easy enough to collect by just traversing the graph again.
This avoids wasted work when computing graph dominators for any other purpose.
Dominator-order information is only needed for coverage graphs, and is easy
enough to collect by just traversing the graph again.
This avoids wasted work when computing graph dominators for any other purpose.
coverage: Make counter creation handle node/edge counters more uniformly
Similar to #130380, this is another round of small improvements informed by my ongoing attempts to overhaul coverage counter creation.
One of the big benefits is getting rid of the awkward special-case that would sometimes attach an edge counter to a node instead. That was needed by the code that chooses which out-edge should be given a counter expression, but we can avoid that by making the corresponding check a little smarter.
I've also renamed several things to be simpler and more consistent, which should help with future changes.
Continue to get rid of `ty::Const::{try_}eval*`
This PR mostly does:
* Removes all of the `try_eval_*` and `eval_*` helpers from `ty::Const`, and replace their usages with `try_to_*`.
* Remove `ty::Const::eval`.
* Rename `ty::Const::normalize` to `ty::Const::normalize_internal`. This function is still used in the normalization code itself.
* Fix some weirdness around the `TransmuteFrom` goal.
I'm happy to split it out further; for example, I could probably land the first part which removes the helpers, or the changes to codegen which are more obvious than the changes to tools.
r? BoxyUwU
Part of https://github.com/rust-lang/rust/issues/130704
Dont ICE when computing coverage of synthetic async closure body
I'm not totally certain if this is *right*, but at least it doesn't ICE.
The issue is that we end up generating two MIR bodies for each async closure, since the `FnOnce` and `Fn`/`FnMut` implementations have different borrowing behavior of their captured variables. They should ideally both contribute to the coverage, since those MIR bodies are (*to the user*) the same code and should have no behavioral differences.
This PR at least suppresses the ICEs, and then I guess worst case we can fix this the right way later.
r? Zalathar or re-roll
Fixes#131190
Make destructors on `extern "C"` frames to be executed
This would make the example in #123231 print "Noisy Drop". I didn't mark this as fixing the issue because the behaviour is yet to be spec'ed.
Tracking:
- https://github.com/rust-lang/rust/issues/74990
Don't check unsize goal in MIR validation when opaques remain
Similarly to `mir_assign_valid_types`, let's just skip when there are opaques. Fixes#130921.
coverage: Multiple small tweaks to counter creation
I've been experimenting with some larger changes to how coverage counters are assigned to parts of the control-flow graph, and while none of that is ready yet, along the way I've repeatedly found myself wanting these smaller tweaks as a base.
There are no changes to compiler output.
Don't use Immediate::offset to transmute pointers to integers
This applies the relatively new `assert_matches_abi` check in the `offset` operation on immediates, which makes sure that if offsets are used to alter the layout (which is possible because the field layout is arbitrarily picked by the caller), this is not done in a way that breaks the invariant of the `Immediate` type.
This leads to ICEs in a GVN mir-opt test, so the second commit fixes GVN.
Fixes https://github.com/rust-lang/rust/issues/131064.
This makes it possible for other parts of counter-assignment to check whether a
node is guaranteed to end up with some kind of counter.
Switching from `impl Fn` to a concrete `&BitSet` just avoids the hassle of
trying to store a closure in a struct field, and currently there's no
foreseeable need for this information to not be a bitset.
Disable jump threading `UnOp::Not` for non-bool
Fix#131195, where jumpthreading was optimizing `!a == b` into `a != b` for non-bool, where this is definitely not true.
This code can sometimes witness malformed coverage attributes in builds that
are going to fail, so use `span_delayed_bug` to avoid an inappropriate ICE in
that case.
Add `File` constructors that return files wrapped with a buffer
In addition to the light convenience, these are intended to raise visibility that buffering is something you should consider when opening a file, since unbuffered I/O is a common performance footgun to Rust newcomers.
ACP: https://github.com/rust-lang/libs-team/issues/446
Tracking Issue: #130804
Apply `EarlyOtherwiseBranch` to scalar value
In the future, I'm thinking of hoisting discriminant via GVN so that we only need to write very little code here.
r? `@cjgillot`
Encode `coroutine_by_move_body_def_id` in crate metadata
We synthesize the MIR for a by-move body for the `FnOnce` implementation of async closures. It can be accessed with the `coroutine_by_move_body_def_id` query. We weren't encoding this query in the metadata though, nor were we properly recording that synthetic MIR in `mir_keys`, so the `optimized_mir` wasn't getting encoded either!
Stacked on top is a fix to consider `DefKind::SyntheticCoroutineBody` to return true in several places I missed. Specifically, we should consider the def-kind in `fn DefKind::is_fn_like()`, since that's what we were using to make sure we ensure `query mir_inliner_callees` before the MIR gets stolen for the body. This led to some CI failures that were caught by miri but which I added a test for.
Remove semi-nondeterminism of `DefPathHash` ordering from inliner
Déjà vu or something because I kinda thought I had put this PR up before. I recall a discussion somewhere where I think it was `@saethlin` mentioning that this check was no longer needed since we have "proper" cycle detection. Putting that up as a PR now.
This may slighlty negatively affect inlining, since the cycle breaking here means that we still inlined some cycles when the def path hashes were ordered in certain ways, this leads to really bad nondeterminism that makes minimizing ICEs and putting up inliner bugfixes difficult.
r? `@cjgillot` or `@saethlin` or someone else idk
coverage: Clarify some parts of coverage counter creation
This is a series of semi-related changes that are trying to make the `counters` module easier to read, understand, and modify.
For example, the existing code happens to avoid ever using the count for a `TerminatorKind::Yield` node as the count for its sole out-edge (since doing so would be incorrect), but doesn't do so explicitly, so seemingly-innocent changes can result in baffling test failures.
This PR also takes the opportunity to simplify some debug-logging code that was making its surrounding code disproportionately hard to read.
There should be no changes to the resulting coverage instrumentation/mappings, as demonstrated by the absence of changes to the coverage test suite.
Given that we directly access the graph predecessors/successors in so many
other places, and sometimes must do so to satisfy the borrow checker, there is
little value in having this trivial helper method.
- Look up the node's predecessors only once
- Get rid of some overly verbose logging
- Explain why some nodes need physical counters
- Extract a helper method to create and set a physical node counter
Simplify the canonical clone method and the copy-like forms to copy
Fixes#128081.
The optimized clone method ends up as the following MIR:
```
_2 = copy ((*_1).0: i32);
_3 = copy ((*_1).1: u64);
_4 = copy ((*_1).2: [i8; 3]);
_0 = Foo { a: move _2, b: move _3, c: move _4 };
```
We can transform this to:
```
_0 = copy (*_1);
```
r? `@cjgillot`
Don't call closure_by_move_body_def_id on FnOnce async closures in MIR validation
Refactors the check in #129847 to not unncessarily call the `closure_by_move_body_def_id` query for async closures that don't *need* a by-move body.
Fixes#130167
- Replace non-standard names like 's, 'p, 'rg, 'ck, 'parent, 'this, and
'me with vanilla 'a. These are cases where the original name isn't
really any more informative than 'a.
- Replace names like 'cx, 'mir, and 'body with vanilla 'a when the lifetime
applies to multiple fields and so the original lifetime name isn't
really accurate.
- Put 'tcx last in lifetime lists, and 'a before 'b.
coverage: Simplify creation of sum counters
A small and self-contained improvement, extracted from some larger changes that I'm still working on.
Ultimately I want to avoid creating these sum counter-expressions in some cases (in favour of just adding physical counters directly to the nodes we care about), so a good incremental move towards that is splitting the “gather edge counters” step out from the ”build a sum of those counters” step.
Creating an extra intermediate vector should have negligible cost (and coverage isn't exercised by the benchmark suite anyway). The removed logging is redundant with the `#[instrument(..)]` logging we already have on the underlying method calls.
some const cleanup: remove unnecessary attributes, add const-hack indications
I learned that we use `FIXME(const-hack)` on top of the "const-hack" label. That seems much better since it marks the right place in the code and moves around with the code. So I went through the PRs with that label and added appropriate FIXMEs in the code. IMO this means we can then remove the label -- Cc ``@rust-lang/wg-const-eval.``
I also noticed some const stability attributes that don't do anything useful, and removed them.
r? ``@fee1-dead``
coverage: Clean up terminology in counter creation
Some of the terminology in this module is confusing, or has drifted out of sync with other parts of the coverage code.
This PR therefore renames some variables and methods, and adjusts comments and debug logging statements, to make things clearer and more consistent.
No functional changes, other than some small tweaks to debug logging.
In all cases the struct can own the relevant thing instead of having a
reference to it. This makes the code simpler, and in some cases removes
a struct lifetime.
These are all functions with a single callsite, where having a separate
function does nothing to help with readability. These changes make the
code a little shorter and easier to read.
There are four related dataflow structs: `MaybeInitializedPlaces`,
`MaybeUninitializedPlaces`, and `EverInitializedPlaces`,
`DefinitelyInitializedPlaces`. They all have a `&Body` and a
`&MoveData<'tcx>` field. The first three use different lifetimes for the
two fields, but the last one uses the same lifetime for both.
This commit changes the first three to use the same lifetime, removing
the need for one of the lifetimes. Other structs that also lose a
lifetime as a result of this are `LivenessContext`, `LivenessResults`,
`InitializationData`.
It then does similar things in various other structs.
Currently it constructs two vectors `calls_to_terminated` and
`cleanups_to_remove` in the main loop, and then processes them after the
main loop. But the processing can be done in the main loop, avoiding the
need for the vectors.
Do not call query to compute coroutine layout for synthetic body of async closure
There is code in the MIR validator that attempts to prevent query cycles when inlining a coroutine into itself, and will use the coroutine layout directly from the body when it detects that's the same coroutine as the one that's being validated. After #128506, this logic didn't take into account the fact that the coroutine def id will differ if it's the "by-move body" of an async closure. This PR implements that.
Fixes#129811
coverage: Count await when the Future is immediately ready
Currently `await` is only counted towards coverage if the containing
function is suspended and resumed at least once. This is because it
expands to code which contains a branch on the discriminant of `Poll`.
By treating it like a branching macro (e.g. `assert!`), these
implementation details will be hidden from the coverage results.
I added a test to ensure the fix works in simple cases, but the heuristic of picking only the first await-related covspan might be unreliable. I plan on testing more thoroughly with a real codebase over the next couple of weeks.
closes#98712
Make `Ty::boxed_ty` return an `Option`
Looks like a good place to use Rust's type system.
---
Most of 4ac7bcbaad/compiler/rustc_middle/src/ty/sty.rs (L971-L1963) looks like it could be moved to `TyKind` (then I guess `Ty` should be made to deref to `TyKind`).
Currently `await` is only counted towards coverage if the containing
function is suspended and resumed at least once. This is because it
expands to code which contains a branch on the discriminant of `Poll`.
By treating it like a branching macro (e.g. `assert!`), these
implementation details will be hidden from the coverage results.
Rename dump of coroutine by-move-body to be more consistent, fix ICE in dump_mir
First, we add a missing match for `DefKind::SyntheticCoroutineBody` in `dump_mir`. Fixes#129703. The second commit (directly below) serves as a test.
Second, we reorder the `dump_mir` in `coroutine_by_move_body_def_id` to be *after* we adjust the body source, and change the disambiguator so it reads more like any other MIR body. This also serves as a test for the ICE, since we're dumping the MIR of a body with `DefKind::SyntheticCoroutineBody`.
Third, we change the parenting of the synthetic MIR body to have the *coroutine-closure* (i.e. async closure) as its parent, so we don't have long strings of `{closure#0}-{closure#0}-{closure#0}`.
try-job: test-various
Move `SanityCheck` and `MirPass`
They are currently in `rustc_middle`. This PR moves them to `rustc_mir_transform`, which makes more sense.
r? ``@cjgillot``
Because that's now the only crate that uses it.
Moving stuff out of `rustc_middle` is always welcome.
I chose to use `impl crate::MirPass`/`impl crate::MirLint` (with
explicit `crate::`) everywhere because that's the only mention of
`MirPass`/`MirLint` used in all of these files. (Prior to this change,
`MirPass` was mostly imported via `use rustc_middle::mir::*` items.)
The actual implementation remains in `rustc_mir_dataflow`, but this
commit moves the `MirPass` impl to `rustc_mir_transform` and changes it
to a `MirLint` (fixing a `FIXME` comment).
(I originally tried moving the full implementation from
`rustc_mir_dataflow` but I had some trait problems with `HasMoveData`
and `RustcPeekAt` and `MaybeLiveLocals`. This commit was much smaller
and simpler, but still will allow some follow-up cleanups.)
Remove `#[macro_use] extern crate tracing`, round 4
Because explicit importing of macros via use items is nicer (more standard and readable) than implicit importing via #[macro_use]. Continuing the work from #124511, #124914, and #125434. After this PR no `rustc_*` crates use `#[macro_use] extern crate tracing` except for `rustc_codegen_gcc` which is a special case and I will do separately.
r? ```@jieyouxu```
Remove `Option<!>` return types.
Several compiler functions have `Option<!>` for their return type. That's odd. The only valid return value is `None`, so why is this type used?
Because it lets you write certain patterns slightly more concisely. E.g. if you have these common patterns:
```
let Some(a) = f() else { return };
let Ok(b) = g() else { return };
```
you can shorten them to these:
```
let a = f()?;
let b = g().ok()?;
```
Huh.
An `Option` return type typically designates success/failure. How should I interpret the type signature of a function that always returns (i.e. doesn't panic), does useful work (modifying `&mut` arguments), and yet only ever fails? This idiom subverts the type system for a cute syntactic trick.
Furthermore, returning `Option<!>` from a function F makes things syntactically more convenient within F, but makes things worse at F's callsites. The callsites can themselves use `?` with F but should not, because they will get an unconditional early return, which is almost certainly not desirable. Instead the return value should be ignored. (Note that some of callsites of `process_operand`, `process_immedate`, `process_assign` actually do use `?`, though the early return doesn't matter in these cases because nothing of significance comes after those calls. Ugh.)
When I first saw this pattern I had no idea how to interpret it, and it took me several minutes of close reading to understand everything I've written above. I even started a Zulip thread about it to make sure I understood it properly. "Save a few characters by introducing types so weird that compiler devs have to discuss it on Zulip" feels like a bad trade-off to me. This commit replaces all the `Option<!>` return values and uses `else`/`return` (or something similar) to replace the relevant `?` uses. The result is slightly more verbose but much easier to understand.
r? ``````@cjgillot``````
Several compiler functions have `Option<!>` for their return type.
That's odd. The only valid return value is `None`, so why is this type
used?
Because it lets you write certain patterns slightly more concisely. E.g.
if you have these common patterns:
```
let Some(a) = f() else { return };
let Ok(b) = g() else { return };
```
you can shorten them to these:
```
let a = f()?;
let b = g().ok()?;
```
Huh.
An `Option` return type typically designates success/failure. How should
I interpret the type signature of a function that always returns (i.e.
doesn't panic), does useful work (modifying `&mut` arguments), and yet
only ever fails? This idiom subverts the type system for a cute
syntactic trick.
Furthermore, returning `Option<!>` from a function F makes things
syntactically more convenient within F, but makes things worse at F's
callsites. The callsites can themselves use `?` with F but should not,
because they will get an unconditional early return, which is almost
certainly not desirable. Instead the return value should be ignored.
(Note that some of callsites of `process_operand`, `process_immedate`,
`process_assign` actually do use `?`, though the early return doesn't
matter in these cases because nothing of significance comes after those
calls. Ugh.)
When I first saw this pattern I had no idea how to interpret it, and it
took me several minutes of close reading to understand everything I've
written above. I even started a Zulip thread about it to make sure I
understood it properly. "Save a few characters by introducing types so
weird that compiler devs have to discuss it on Zulip" feels like a bad
trade-off to me. This commit replaces all the `Option<!>` return values
and uses `else`/`return` (or something similar) to replace the relevant
`?` uses. The result is slightly more verbose but much easier to
understand.
By making it own the index maps, instead of holding references to them.
This requires moving the free function `find_candidate` into
`Candidate::reset_and_find`. It lets the `'alloc` lifetime be removed
everywhere that still has it.
