Introduce `ConstAllocation`.
Currently some `Allocation`s are interned, some are not, and it's very
hard to tell at a use point which is which.
This commit introduces `ConstAllocation` for the known-interned ones,
which makes the division much clearer. `ConstAllocation::inner()` is
used to get the underlying `Allocation`.
In some places it's natural to use an `Allocation`, in some it's natural
to use a `ConstAllocation`, and in some places there's no clear choice.
I've tried to make things look as nice as possible, while generally
favouring `ConstAllocation`, which is the type that embodies more
information. This does require quite a few calls to `inner()`.
The commit also tweaks how `PartialOrd` works for `Interned`. The
previous code was too clever by half, building on `T: Ord` to make the
code shorter. That caused problems with deriving `PartialOrd` and `Ord`
for `ConstAllocation`, so I changed it to build on `T: PartialOrd`,
which is slightly more verbose but much more standard and avoided the
problems.
r? `@fee1-dead`
Currently some `Allocation`s are interned, some are not, and it's very
hard to tell at a use point which is which.
This commit introduces `ConstAllocation` for the known-interned ones,
which makes the division much clearer. `ConstAllocation::inner()` is
used to get the underlying `Allocation`.
In some places it's natural to use an `Allocation`, in some it's natural
to use a `ConstAllocation`, and in some places there's no clear choice.
I've tried to make things look as nice as possible, while generally
favouring `ConstAllocation`, which is the type that embodies more
information. This does require quite a few calls to `inner()`.
The commit also tweaks how `PartialOrd` works for `Interned`. The
previous code was too clever by half, building on `T: Ord` to make the
code shorter. That caused problems with deriving `PartialOrd` and `Ord`
for `ConstAllocation`, so I changed it to build on `T: PartialOrd`,
which is slightly more verbose but much more standard and avoided the
problems.
Miri/CTFE: properly treat overflow in (signed) division/rem as UB
To my surprise, it looks like LLVM treats overflow of signed div/rem as UB. From what I can tell, MIR `Div`/`Rem` directly lowers to the corresponding LLVM operation, so to make that correct we also have to consider these overflows UB in the CTFE/Miri interpreter engine.
r? `@oli-obk`
Miri fn ptr check: don't use conservative null check
In https://github.com/rust-lang/rust/pull/94270 I used the wrong NULL check for function pointers: `memory.ptr_may_be_null` is conservative even on machines that support ptr-to-int casts, leading to false errors in Miri.
This fixes that problem, and also replaces that foot-fun of a method with `scalar_may_be_null` which is never unnecessarily conservative.
r? `@oli-obk`
rustc_errors: let `DiagnosticBuilder::emit` return a "guarantee of emission".
That is, `DiagnosticBuilder` is now generic over the return type of `.emit()`, so we'll now have:
* `DiagnosticBuilder<ErrorReported>` for error (incl. fatal/bug) diagnostics
* can only be created via a `const L: Level`-generic constructor, that limits allowed variants via a `where` clause, so not even `rustc_errors` can accidentally bypass this limitation
* asserts `diagnostic.is_error()` on emission, just in case the construction restriction was bypassed (e.g. by replacing the whole `Diagnostic` inside `DiagnosticBuilder`)
* `.emit()` returns `ErrorReported`, as a "proof" token that `.emit()` was called
(though note that this isn't a real guarantee until after completing the work on
#69426)
* `DiagnosticBuilder<()>` for everything else (warnings, notes, etc.)
* can also be obtained from other `DiagnosticBuilder`s by calling `.forget_guarantee()`
This PR is a companion to other ongoing work, namely:
* #69426
and it's ongoing implementation:
#93222
the API changes in this PR are needed to get statically-checked "only errors produce `ErrorReported` from `.emit()`", but doesn't itself provide any really strong guarantees without those other `ErrorReported` changes
* #93244
would make the choices of API changes (esp. naming) in this PR fit better overall
In order to be able to let `.emit()` return anything trustable, several changes had to be made:
* `Diagnostic`'s `level` field is now private to `rustc_errors`, to disallow arbitrary "downgrade"s from "some kind of error" to "warning" (or anything else that doesn't cause compilation to fail)
* it's still possible to replace the whole `Diagnostic` inside the `DiagnosticBuilder`, sadly, that's harder to fix, but it's unlikely enough that we can paper over it with asserts on `.emit()`
* `.cancel()` now consumes `DiagnosticBuilder`, preventing `.emit()` calls on a cancelled diagnostic
* it's also now done internally, through `DiagnosticBuilder`-private state, instead of having a `Level::Cancelled` variant that can be read (or worse, written) by the user
* this removes a hazard of calling `.cancel()` on an error then continuing to attach details to it, and even expect to be able to `.emit()` it
* warnings were switched to *only* `can_emit_warnings` on emission (instead of pre-cancelling early)
* `struct_dummy` was removed (as it relied on a pre-`Cancelled` `Diagnostic`)
* since `.emit()` doesn't consume the `DiagnosticBuilder` <sub>(I tried and gave up, it's much more work than this PR)</sub>,
we have to make `.emit()` idempotent wrt the guarantees it returns
* thankfully, `err.emit(); err.emit();` can return `ErrorReported` both times, as the second `.emit()` call has no side-effects *only* because the first one did do the appropriate emission
* `&mut Diagnostic` is now used in a lot of function signatures, which used to take `&mut DiagnosticBuilder` (in the interest of not having to make those functions generic)
* the APIs were already mostly identical, allowing for low-effort porting to this new setup
* only some of the suggestion methods needed some rework, to have the extra `DiagnosticBuilder` functionality on the `Diagnostic` methods themselves (that change is also present in #93259)
* `.emit()`/`.cancel()` aren't available, but IMO calling them from an "error decorator/annotator" function isn't a good practice, and can lead to strange behavior (from the caller's perspective)
* `.downgrade_to_delayed_bug()` was added, letting you convert any `.is_error()` diagnostic into a `delay_span_bug` one (which works because in both cases the guarantees available are the same)
This PR should ideally be reviewed commit-by-commit, since there is a lot of fallout in each.
r? `@estebank` cc `@Manishearth` `@nikomatsakis` `@mark-i-m`
Always format to internal String in FmtPrinter
This avoids monomorphizing for different parameters, decreasing generic code
instantiated downstream from rustc_middle -- locally seeing 7% unoptimized LLVM IR
line wins on rustc_borrowck, for example.
We likely can't/shouldn't get rid of the Result-ness on most functions, though some
further cleanup avoiding fmt::Error where we now know it won't occur may be possible,
though somewhat painful -- fmt::Write is a pretty annoying API to work with in practice
when you're trying to use it infallibly.
Miri: relax fn ptr check
As discussed in https://github.com/rust-lang/unsafe-code-guidelines/issues/72#issuecomment-1025407536, the function pointer check done by Miri is currently overeager: contrary to our usual principle of only checking rather uncontroversial validity invariants, we actually check that the pointer points to a real function.
So, this relaxes the check to what the validity invariant probably will be (and what the reference already says it is): the function pointer must be non-null, and that's it.
The check that CTFE does on the final value of a constant is unchanged -- CTFE recurses through references, so it makes some sense to also recurse through function pointers. We might still want to relax this in the future, but that would be a separate change.
r? `@oli-obk`
rustc_const_eval: adopt let else in more places
Continuation of #89933, #91018, #91481, #93046, #93590, #94011.
I have extended my clippy lint to also recognize tuple passing and match statements. The diff caused by fixing it is way above 1 thousand lines. Thus, I split it up into multiple pull requests to make reviewing easier. This PR handles rustc_const_eval.
Move ty::print methods to Drop-based scope guards
Primary goal is reducing codegen of the TLS access for each closure, which shaves ~3 seconds of bootstrap time over rustc as a whole.
92d20c4aad removed no-variants special
case from try_destructure_const with expectation that this case would be
handled gracefully when read_discriminant returns an error.
Alas in that case read_discriminant succeeds while returning a
non-existing variant, so the special case is still necessary.
Make it possible to pretty print invalid constants by introducing a
fallible variant of `destructure_const` and falling back to debug
formatting when it fails.
Specifically, rename the `Const` struct as `ConstS` and re-introduce `Const` as
this:
```
pub struct Const<'tcx>(&'tcx Interned<ConstS>);
```
This now matches `Ty` and `Predicate` more closely, including using
pointer-based `eq` and `hash`.
Notable changes:
- `mk_const` now takes a `ConstS`.
- `Const` was copy, despite being 48 bytes. Now `ConstS` is not, so need a
we need separate arena for it, because we can't use the `Dropless` one any
more.
- Many `&'tcx Const<'tcx>`/`&Const<'tcx>` to `Const<'tcx>` changes
- Many `ct.ty` to `ct.ty()` and `ct.val` to `ct.val()` changes.
- Lots of tedious sigil fiddling.
Specifically, change `Ty` from this:
```
pub type Ty<'tcx> = &'tcx TyS<'tcx>;
```
to this
```
pub struct Ty<'tcx>(Interned<'tcx, TyS<'tcx>>);
```
There are two benefits to this.
- It's now a first class type, so we can define methods on it. This
means we can move a lot of methods away from `TyS`, leaving `TyS` as a
barely-used type, which is appropriate given that it's not meant to
be used directly.
- The uniqueness requirement is now explicit, via the `Interned` type.
E.g. the pointer-based `Eq` and `Hash` comes from `Interned`, rather
than via `TyS`, which wasn't obvious at all.
Much of this commit is boring churn. The interesting changes are in
these files:
- compiler/rustc_middle/src/arena.rs
- compiler/rustc_middle/src/mir/visit.rs
- compiler/rustc_middle/src/ty/context.rs
- compiler/rustc_middle/src/ty/mod.rs
Specifically:
- Most mentions of `TyS` are removed. It's very much a dumb struct now;
`Ty` has all the smarts.
- `TyS` now has `crate` visibility instead of `pub`.
- `TyS::make_for_test` is removed in favour of the static `BOOL_TY`,
which just works better with the new structure.
- The `Eq`/`Ord`/`Hash` impls are removed from `TyS`. `Interned`s impls
of `Eq`/`Hash` now suffice. `Ord` is now partly on `Interned`
(pointer-based, for the `Equal` case) and partly on `TyS`
(contents-based, for the other cases).
- There are many tedious sigil adjustments, i.e. adding or removing `*`
or `&`. They seem to be unavoidable.
Lazy type-alias-impl-trait
Previously opaque types were processed by
1. replacing all mentions of them with inference variables
2. memorizing these inference variables in a side-table
3. at the end of typeck, resolve the inference variables in the side table and use the resolved type as the hidden type of the opaque type
This worked okayish for `impl Trait` in return position, but required lots of roundabout type inference hacks and processing.
This PR instead stops this process of replacing opaque types with inference variables, and just keeps the opaque types around.
Whenever an opaque type `O` is compared with another type `T`, we make the comparison succeed and record `T` as the hidden type. If `O` is compared to `U` while there is a recorded hidden type for it, we grab the recorded type (`T`) and compare that against `U`. This makes implementing
* https://github.com/rust-lang/rfcs/pull/2515
much simpler (previous attempts on the inference based scheme were very prone to ICEs and general misbehaviour that was not explainable except by random implementation defined oddities).
r? `@nikomatsakis`
fixes#93411fixes#88236
by using an opaque type obligation to bubble up comparisons between opaque types and other types
Also uses proper obligation causes so that the body id works, because out of some reason nll uses body ids for logic instead of just diagnostics.
Add note suggesting that predicate may be satisfied, but is not `const`
Not sure if we should be printing this in addition to, or perhaps _instead_ of the help message:
```
help: the trait `~const Add` is not implemented for `NonConstAdd`
```
Also added `ParamEnv::is_const` and `PolyTraitPredicate::is_const_if_const` and, in a separate commit, used those in other places instead of `== hir::Constness::Const`, etc.
r? ````@fee1-dead````
Add `intrinsics::const_deallocate`
Tracking issue: #79597
Related: #91884
This allows deallocation of a memory allocated by `intrinsics::const_allocate`. At the moment, this can be only used to reduce memory usage, but in the future this may be useful to detect memory leaks (If an allocated memory remains after evaluation, raise an error...?).
Check `const Drop` impls considering `~const` Bounds
This PR adds logic to trait selection to account for `~const` bounds in custom `impl const Drop` for types, elaborates the `const Drop` check in `rustc_const_eval` to check those bounds, and steals some drop linting fixes from #92922, thanks `@DrMeepster.`
r? `@fee1-dead` `@oli-obk` <sup>(edit: guess I can't request review from two people, lol)</sup>
since each of you wrote and reviewed #88558, respectively.
Since the logic here is more complicated than what existed, it's possible that this is a perf regression. But it works correctly with tests, and that makes me happy.
Fixes#92881
- Also rename a trivial_const_drop to match style of other functions in
the util module.
- Also add a test for `const Drop` that doesn't depend on a `~const`
bound.
- Also comment a bit why we remove the const bound during dropck impl
check.
Remove deprecated LLVM-style inline assembly
The `llvm_asm!` was deprecated back in #87590 1.56.0, with intention to remove
it once `asm!` was stabilized, which already happened in #91728 1.59.0. Now it
is time to remove `llvm_asm!` to avoid continued maintenance cost.
Closes#70173.
Closes#92794.
Closes#87612.
Closes#82065.
cc `@rust-lang/wg-inline-asm`
r? `@Amanieu`
Error when selected impl is not const in constck
Catches bad things when checking a `default_method_body_is_const` body, such as:
```rust
self.map(/* .. */).is_sorted();
```
When `Map` does not yet have a `const` `impl` for `Iterator`.
r? ```@oli-obk```
The field is also renamed from `ident` to `name. In most cases,
we don't actually need the `Span`. A new `ident` method is added
to `VariantDef` and `FieldDef`, which constructs the full `Ident`
using `tcx.def_ident_span()`. This method is used in the cases
where we actually need an `Ident`.
This makes incremental compilation properly track changes
to the `Span`, without all of the invalidations caused by storing
a `Span` directly via an `Ident`.
Remove `NullOp::Box`
Follow up of #89030 and MCP rust-lang/compiler-team#460.
~1 month later nothing seems to be broken, apart from a small regression that #89332 (1aac85bb716c09304b313d69d30d74fe7e8e1a8e) shows could be regained by remvoing the diverging path, so it shall be safe to continue and remove `NullOp::Box` completely.
r? `@jonas-schievink`
`@rustbot` label T-compiler
CTFE eval_fn_call: use FnAbi to determine argument skipping and compatibility
This makes use of the `FnAbi` type in CTFE/Miri, which `@eddyb` has been saying for years is what we should do.^^ `FnAbi` is used to
- determine which arguments to skip (rather than the previous heuristic of skipping ZST arguments with the Rust ABI)
- impose further restrictions on whether caller and callee are consistent in how a given argument is passed
I was hoping it would also simplify the code, but that is not the case -- the previous type compatibility checks are still required (AFAIK), only the ZST skipping is gone and that took barely any code. We also need some hacks because `FnAbi` assumes a certain way of implementing `caller_location` (by passing extra arguments), but Miri can just read the caller location from the call stack so it doesn't need those arguments. (The fact that every backend has to separately implement support for these arguments seems suboptimal -- looks like this might have been better implemented on the MIR level.) To avoid having to implement those unnecessary arguments in Miri, we just compute *whether* the argument is present on the caller/callee side, but don't actually pass that argument around.
I have no idea if this looks the way `@eddyb` thinks it should look... but it makes Miri's test suite pass. ;)
One of rustc's tests fails unfortunately (`ui/const-generics/issues/issue-67739.rs`), some const generic code that is evaluated too early -- I think that should raise `TooGeneric` but instead it ICEs. My assumption is this is some FnAbi code that has not been properly tested on polymorphic code, but it might also be me calling that FnAbi code the wrong way.
r? `@oli-obk` `@eddyb`
Fixes https://github.com/rust-lang/rust/issues/56166
Miri PR at https://github.com/rust-lang/miri/pull/1928
The `AggregateKind` enum ends up in the final mir `Body`. Currently,
any changes to `AdtDef` (regardless of how significant they are)
will legitimately cause the overall result of `optimized_mir` to change,
invalidating any codegen re-use involving that mir.
This will get worse once we start hashing the `Span` inside `FieldDef`
(which is itself contained in `AdtDef`).
To try to reduce these kinds of invalidations, this commit changes
`AggregateKind::Adt` to store just the `DefId`, instead of the full
`AdtDef`. This allows the result of `optimized_mir` to be unchanged
if the `AdtDef` changes in a way that doesn't actually affect any
of the MIR we build.
Remove `SymbolStr`
This was originally proposed in https://github.com/rust-lang/rust/pull/74554#discussion_r466203544. As well as removing the icky `SymbolStr` type, it allows the removal of a lot of `&` and `*` occurrences.
Best reviewed one commit at a time.
r? `@oli-obk`
Looser check for overflowing_binary_op
Fix for issue #91636 tight check resulted in ICE, this makes the check a little looser. It seems `eq` allows comparing of `supertype` and `subtype` if `lhs = supertype` and `rhs = subtype` but not vice versa, is this intended behavior ?
