Fix HIR visit order
Fixes#61442
When rustc::middle::region::ScopeTree computes its yield_in_scope
field, it relies on the HIR visitor order to properly compute which
types must be live across yield points. In order for the computed scopes
to agree with the generated MIR, we must ensure that expressions
evaluated before a yield point are visited before the 'yield'
expression.
However, the visitor order for ExprKind::AssignOp
was incorrect. The left-hand side of a compund assignment expression is
evaluated before the right-hand side, but the right-hand expression was
being visited before the left-hand expression. If the left-hand
expression caused a new type to be introduced (e.g. through a
deref-coercion), the new type would be incorrectly seen as occuring
*after* the yield point, instead of before. This leads to a mismatch
between the computed generator types and the MIR, since the MIR will
correctly see the type as being live across the yield point.
To fix this, we correct the visitor order for ExprKind::AssignOp
to reflect the actual evaulation order.
rustc_mir: Hide initial block state when defining transfer functions
This PR addresses [this FIXME](2887008e0c/src/librustc_mir/dataflow/mod.rs (L594-L596)).
This makes `sets.on_entry` inaccessible in `{before_,}{statement,terminator}_effect`. This field was meant to allow implementors of `BitDenotation` to access the initial state for each block (optionally with the effect of all previous statements applied via `accumulates_intrablock_state`) while defining transfer functions. However, the ability to set the initial value for the entry set of each basic block (except for START_BLOCK) no longer exists. As a result, this functionality is mostly useless, and when it *was* used it was used erroneously (see #62007).
Since `on_entry` is now useless, we can also remove `BlockSets`, which held the `gen`, `kill`, and `on_entry` bitvectors and replace it with a `GenKill` struct. Variables of this type are called `trans` since they represent a transfer function. `GenKill`s are stored contiguously in `AllSets`, which reduces the number of bounds checks and may improve cache performance: one is almost never accessed without the other.
Replacing `BlockSets` with `GenKill` allows us to define some new helper functions which streamline dataflow iteration and the dataflow-at-location APIs. Notably, `state_for_location` used a subtle side-effect of the `kill`/`kill_all` setters to apply the transfer function, and could be incorrect if a transfer function depended on effects of previous statements in the block on `gen_set`.
Additionally, this PR merges `BitSetOperator` and `InitialFlow` into one trait. Since the value of `InitialFlow` defines the semantics of the `join` operation, there's no reason to have seperate traits for each. We can add a default impl of `join` which branches based on `BOTTOM_VALUE`. This should get optimized away.
Refactor miri pointer checks
Centralize bounds, alignment and NULL checking for memory accesses in one function: `memory.check_ptr_access`. That function also takes care of converting a `Scalar` to a `Pointer`, should that be needed. Not all accesses need that though: if the access has size 0, `None` is returned. Everyone accessing memory based on a `Scalar` should use this method to get the `Pointer` they need.
All operations on the `Allocation` work on `Pointer` inputs and expect all the checks to have happened (and will ICE if the bounds are violated). The operations on `Memory` work on `Scalar` inputs and do the checks themselves.
The only other public method to check pointers is `memory.ptr_may_be_null`, which is needed in a few places. No need for `check_align` or similar methods. That makes the public API surface much easier to use and harder to mis-use.
This should be largely no-functional-change, except that ZST accesses to a "true" pointer that is dangling or out-of-bounds are now considered UB. This is to be conservative wrt. whatever LLVM might be doing.
While I am at it, this also removes the assumption that the vtable part of a `dyn Trait`-fat-pointer is a `Pointer` (as opposed to a pointer cast to an integer, stored as raw bits).
r? @oli-obk
submodules: Update clippy from 5a11ed7b to 8c80b65f
```bash
% git shortlog --no-merges 5a11ed7b92cc4cf40a4568a8fc1ff54b198c333b..
Daniele D'Orazio (4):
make needless_return work with void functions
update tests and fix lints in clippy
more idiomatic code
cargo fmt
Jeremy Stucki (1):
Remove needless lifetimes
Lzu Tao (1):
Fix fallout cause NodeId pruning
Matthias Krüger (1):
readme: update
Mazdak Farrokhzad (5):
Fix fallout from rust-lang/rust PR 60861.
Account for let_chains in collapsible_if ui test cases.
Fix dogfood test failures.
Pacify rustfmt.
Put 'if let' back into comment.
flip1995 (1):
Remove another unnecessary lifetime
```
r? @Manishearth , cc @oli-obk
Remove `ast::Guard`
With the introduction of `ast::ExprKind::Let` in https://github.com/rust-lang/rust/pull/60861, the `ast::Guard` structure is now redundant in terms of representing [`if let` guards](https://github.com/rust-lang/rust/issues/51114) in AST since it can be represented by `ExprKind::Let` syntactically. Therefore, we remove `ast::Guard` here.
However, we keep `hir::Guard` because the semantic representation is a different matter and this story is more unclear right now (might involve `goto 'arm` in HIR or something...).
r? @petrochenkov
Run rustfmt on some libsyntax files
As part of #62008, run rustfmt on:
- src/libsyntax/ext/tt/macro_rules.rs
- src/libsyntax/ext/tt/quoted.rs
There is no semantic change. To fix potential merge conflicts, simply choose the other side then run rustfmt and fix any tidy check (like line length).
Speed up tidy
master:
Time (mean ± σ): 3.478 s ± 0.033 s [User: 3.298 s, System: 0.178 s]
Range (min … max): 3.425 s … 3.525 s 10 runs
This PR:
Time (mean ± σ): 1.098 s ± 0.006 s [User: 783.7 ms, System: 310.2 ms]
Range (min … max): 1.092 s … 1.113 s 10 runs
Alleviates https://github.com/rust-lang/rust/issues/59884. For the most part each commit stands on its own. Timings are on warm filesystem cache.
r? @eddyb
compiletest: Introduce `// {check,build,run}-pass` pass modes
Pass UI tests now have three modes
```
// check-pass
// build-pass
// run-pass
```
mirroring equivalent well-known `cargo` commands.
`// check-pass` will compile the test skipping codegen (which is expensive and isn't supposed to fail in most cases).
`// build-pass` will compile and link the test without running it.
`// run-pass` will compile, link and run the test.
Tests without a "pass" annotation are still considered "fail" tests.
Most UI tests would probably want to switch to `check-pass`.
Tests validating codegen would probably want to run the generated code as well and use `run-pass`.
`build-pass` should probably be rare (linking tests?).
https://github.com/rust-lang/rust/pull/61755 will provide a way to run the tests with any mode, e.g. bump `check-pass` tests to `run-pass` to satisfy especially suspicious people, and be able to make sure that codegen doesn't breaks in some entirely unexpected way.
Tests marked with any mode are expected to pass with any other mode, if that's not the case for some legitimate reason, then the test should be made a "fail" test rather than a "pass" test.
Perhaps some secondary CI can verify this invariant, but that's not super urgent.
`// compile-pass` still works and is equivalent to `build-pass`.
Why is `// compile-pass` bad - 1) it gives an impression that the test is only compiled, but not linked, 2) it doesn't mirror a cargo command.
It can be removed some time in the future in a separate PR.
cc https://github.com/rust-lang/rust/issues/61712
That function also takes care of converting a Scalar to a Pointer, should that be needed. Not all accesses need that though: if the access has size 0, None is returned.
Everyone accessing memory based on a Scalar should use this method to get the Pointer they need.
All operations on the Allocation work on Pointer inputs and expect all the checks to have happened (and will ICE if the bounds are violated).
The operations on Memory work on Scalar inputs and do the checks themselves.
The only other public method to check pointers is memory.ptr_may_be_null, which is needed in a few places.
With this, we can make all the other methods (tests for a pointer being in-bounds and checking alignment) private helper methods, used to implement the two public methods.
That maks the public API surface much easier to use and harder to mis-use.
While I am at it, this also removes the assumption that the vtable part of a `dyn Trait`-fat-pointer is a `Pointer` (as opposed to a pointer cast to an integer, stored as raw bits).
