Strengthen well known check-cfg names and values test
https://github.com/rust-lang/rust/pull/118494 is changing the implementation of how we expect well known check-cfg names and values, but we currently don't have a test that checks every well known only some of them.
This PR therefore strengthen our well known names/values test to include all of the configs to at least avoid unintended regressions and validate new entry.
*this PR also contains some drive-by consolidation of unexpected `target_os`, `target_arch` into a single file*
r? `@nnethercote` (maybe? feel free to re-assign)
Add more SIMD platform-intrinsics
- [x] simd_masked_load
- [x] LLVM codegen - llvm.masked.load
- [x] cranelift codegen - implemented but untested
- [ ] simd_masked_store
- [x] LLVM codegen - llvm.masked.store
- [ ] cranelift codegen
Also added a run-pass test to test both intrinsics, and additional build-fail & check-fail to cover validation for both intrinsics
Make async generators fused by default
I actually changed my mind about this since the implementation PR landed. I think it's beneficial for `async gen` blocks to be "fused" by default -- i.e., for them to repeatedly return `Poll::Ready(None)` -- rather than panic.
We have [`FusedStream`](https://docs.rs/futures/latest/futures/stream/trait.FusedStream.html) in futures-rs to represent streams with this capability already anyways.
r? eholk
cc ```@rust-lang/wg-async,``` would like to know if anyone else has opinions about this.
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.
update target feature following LLVM API change
LLVM commit e817966718 renamed* the `unaligned-scalar-mem` target feature to `fast-unaligned-access`.
(*) technically the commit folded two previous features into one, but there are no references to the other one in rust.
coverage: Use `SpanMarker` to improve coverage spans for `if !` expressions
Coverage instrumentation works by extracting source code spans from MIR. However, some kinds of syntax are effectively erased during MIR building, so their spans don't necessarily exist anywhere in MIR, making them invisible to the coverage instrumentor (unless we resort to various heuristics and hacks to recover them).
This PR introduces `CoverageKind::SpanMarker`, which is a new variant of `StatementKind::Coverage`. Its sole purpose is to represent spans that would otherwise not appear in MIR, so that the coverage instrumentor can extract them.
When coverage is enabled, the MIR builder can insert these dummy statements as needed, to improve the accuracy of spans used by coverage mappings.
Fixes#115468.
---
```@rustbot``` label +A-code-coverage
Add new targets {x86_64,i686}-win7-windows-msvc
This PR adds two new Tier 3 targets, x86_64-win7-windows-msvc and i686-win7-windows-msvc, that aim to support targeting Windows 7 after the `*-pc-windows-msvc` target drops support for it (slated to happen in 1.76.0).
# Tier 3 target policy
> At this tier, the Rust project provides no official support for a target, so we place minimal requirements on the introduction of targets.
>
> A proposed new tier 3 target must be reviewed and approved by a member of the compiler team based on these requirements. The reviewer may choose to gauge broader compiler team consensus via a [Major Change Proposal (MCP)](https://forge.rust-lang.org/compiler/mcp.html).
>
> A proposed target or target-specific patch that substantially changes code shared with other targets (not just target-specific code) must be reviewed and approved by the appropriate team for that shared code before acceptance.
>
> - A tier 3 target must have a designated developer or developers (the "target maintainers") on record to be CCed when issues arise regarding the target. (The mechanism to track and CC such developers may evolve over time.)
This is me, `@roblabla` on github.
> - Targets must use naming consistent with any existing targets; for instance, a target for the same CPU or OS as an existing Rust target should use the same name for that CPU or OS. Targets should normally use the same names and naming conventions as used elsewhere in the broader ecosystem beyond Rust (such as in other toolchains), unless they have a very good reason to diverge. Changing the name of a target can be highly disruptive, especially once the target reaches a higher tier, so getting the name right is important even for a tier 3 target.
I went with naming the target `x86_64-win7-windows-msvc`, inserting the `win7` in the vendor field (usually set to to `pc`). This is done to avoid ecosystem churn, as quite a few crates have `cfg(target_os = "windows")` or `cfg(target_env = "msvc")`, but nearly no `cfg(target_vendor = "pc")`. Since my goal is to be able to seamlessly swap to the `win7` target, I figured it'd be easier this way.
> - Target names should not introduce undue confusion or ambiguity unless absolutely necessary to maintain ecosystem compatibility. For example, if the name of the target makes people extremely likely to form incorrect beliefs about what it targets, the name should be changed or augmented to disambiguate it.
I believe the naming is pretty explicit.
> - If possible, use only letters, numbers, dashes and underscores for the name. Periods (`.`) are known to cause issues in Cargo.
The name comforms to this requirement.
> - Tier 3 targets may have unusual requirements to build or use, but must not create legal issues or impose onerous legal terms for the Rust project or for Rust developers or users.
> - The target must not introduce license incompatibilities.
> - Anything added to the Rust repository must be under the standard Rust license (`MIT OR Apache-2.0`).
> - The target must not cause the Rust tools or libraries built for any other host (even when supporting cross-compilation to the target) to depend on any new dependency less permissive than the Rust licensing policy. This applies whether the dependency is a Rust crate that would require adding new license exceptions (as specified by the `tidy` tool in the rust-lang/rust repository), or whether the dependency is a native library or binary. In other words, the introduction of the target must not cause a user installing or running a version of Rust or the Rust tools to be subject to any new license requirements.
> - Compiling, linking, and emitting functional binaries, libraries, or other code for the target (whether hosted on the target itself or cross-compiling from another target) must not depend on proprietary (non-FOSS) libraries. Host tools built for the target itself may depend on the ordinary runtime libraries supplied by the platform and commonly used by other applications built for the target, but those libraries must not be required for code generation for the target; cross-compilation to the target must not require such libraries at all. For instance, `rustc` built for the target may depend on a common proprietary C runtime library or console output library, but must not depend on a proprietary code generation library or code optimization library. Rust's license permits such combinations, but the Rust project has no interest in maintaining such combinations within the scope of Rust itself, even at tier 3.
> - "onerous" here is an intentionally subjective term. At a minimum, "onerous" legal/licensing terms include but are *not* limited to: non-disclosure requirements, non-compete requirements, contributor license agreements (CLAs) or equivalent, "non-commercial"/"research-only"/etc terms, requirements conditional on the employer or employment of any particular Rust developers, revocable terms, any requirements that create liability for the Rust project or its developers or users, or any requirements that adversely affect the livelihood or prospects of the Rust project or its developers or users.
As far as I understand it, this target has exactly the same legal situation as the existing Tier 1 x86_64-pc-windows-msvc.
> - Neither this policy nor any decisions made regarding targets shall create any binding agreement or estoppel by any party. If any member of an approving Rust team serves as one of the maintainers of a target, or has any legal or employment requirement (explicit or implicit) that might affect their decisions regarding a target, they must recuse themselves from any approval decisions regarding the target's tier status, though they may otherwise participate in discussions.
> - This requirement does not prevent part or all of this policy from being cited in an explicit contract or work agreement (e.g. to implement or maintain support for a target). This requirement exists to ensure that a developer or team responsible for reviewing and approving a target does not face any legal threats or obligations that would prevent them from freely exercising their judgment in such approval, even if such judgment involves subjective matters or goes beyond the letter of these requirements.
Understood.
> - Tier 3 targets should attempt to implement as much of the standard libraries as possible and appropriate (core for most targets, alloc for targets that can support dynamic memory allocation, std for targets with an operating system or equivalent layer of system-provided functionality), but may leave some code unimplemented (either unavailable or stubbed out as appropriate), whether because the target makes it impossible to implement or challenging to implement. The authors of pull requests are not obligated to avoid calling any portions of the standard library on the basis of a tier 3 target not implementing those portions.
This target supports the whole libstd surface, since it's essentially reusing all of the x86_64-pc-windows-msvc target. Understood.
> - The target must provide documentation for the Rust community explaining how to build for the target, using cross-compilation if possible. If the target supports running binaries, or running tests (even if they do not pass), the documentation must explain how to run such binaries or tests for the target, using emulation if possible or dedicated hardware if necessary.
Wrote some documentation on how to build, test and cross-compile the target in the `platform-support` part. Hopefully it's enough to get started.
> - Tier 3 targets must not impose burden on the authors of pull requests, or other developers in the community, to maintain the target. In particular, do not post comments (automated or manual) on a PR that derail or suggest a block on the PR based on a tier 3 target. Do not send automated messages or notifications (via any medium, including via ``@`)` to a PR author or others involved with a PR regarding a tier 3 target, unless they have opted into such messages.
> - Backlinks such as those generated by the issue/PR tracker when linking to an issue or PR are not considered a violation of this policy, within reason. However, such messages (even on a separate repository) must not generate notifications to anyone involved with a PR who has not requested such notifications.
Understood.
> - Patches adding or updating tier 3 targets must not break any existing tier 2 or tier 1 target, and must not knowingly break another tier 3 target without approval of either the compiler team or the maintainers of the other tier 3 target.
> - In particular, this may come up when working on closely related targets, such as variations of the same architecture with different features. Avoid introducing unconditional uses of features that another variation of the target may not have; use conditional compilation or runtime detection, as appropriate, to let each target run code supported by that target.
Understood.
> If a tier 3 target stops meeting these requirements, or the target maintainers no longer have interest or time, or the target shows no signs of activity and has not built for some time, or removing the target would improve the quality of the Rust codebase, we may post a PR to remove it; any such PR will be CCed to the target maintainers (and potentially other people who have previously worked on the target), to check potential interest in improving the situation.
Understood.
Add emulated TLS support
This is a reopen of https://github.com/rust-lang/rust/pull/96317 . many android devices still only use 128 pthread keys, so using emutls can be helpful.
Currently LLVM uses emutls by default for some targets (such as android, openbsd), but rust does not use it, because `has_thread_local` is false.
This commit has some changes to allow users to enable emutls:
1. add `-Zhas-thread-local` flag to specify that std uses `#[thread_local]` instead of pthread key.
2. when using emutls, decorate symbol names to find thread local symbol correctly.
3. change `-Zforce-emulated-tls` to `-Ztls-model=emulated` to explicitly specify whether to generate emutls.
r? `@Amanieu`
Rollup of 6 pull requests
Successful merges:
- #117586 (Uplift the (new solver) canonicalizer into `rustc_next_trait_solver`)
- #118502 (fix: correct the arg for 'suggest to use associated function syntax' diagnostic)
- #118694 (Add instance evaluation and methods to read an allocation in StableMIR)
- #118715 (privacy: visit trait def id of projections)
- #118730 (recurse into refs when comparing tys for diagnostics)
- #118736 (temporarily revert "ice on ambguity in mir typeck")
r? `@ghost`
`@rustbot` modify labels: rollup
This is for post-monomorphization cycles. These are only caught later
(in drop elaboration for the example that I saw), so we need to handle
them here.
This issue wasn't noticed before because exhaustiveness only checked
inhabitedness when `exhaustive_patterns` was on. The preceding commit
now check inhabitedness always, which revealed the problem.
- `ConstructorSet` knows about both empty and nonempty constructors;
- If an empty constructor is present in the column, we output it in
`split().present`.
tidy error: /git/rust/compiler/rustc_ast_pretty/src/pprust/state.rs:1165: unexplained "```ignore" doctest; try one:
* make the test actually pass, by adding necessary imports and declarations, or
* use "```text", if the code is not Rust code, or
* use "```compile_fail,Ennnn", if the code is expected to fail at compile time, or
* use "```should_panic", if the code is expected to fail at run time, or
* use "```no_run", if the code should type-check but not necessary linkable/runnable, or
* explain it like "```ignore (cannot-test-this-because-xxxx)", if the annotation cannot be avoided.
tidy error: /git/rust/compiler/rustc_ast_pretty/src/pprust/state.rs:1176: unexplained "```ignore" doctest; try one:
* make the test actually pass, by adding necessary imports and declarations, or
* use "```text", if the code is not Rust code, or
* use "```compile_fail,Ennnn", if the code is expected to fail at compile time, or
* use "```should_panic", if the code is expected to fail at run time, or
* use "```no_run", if the code should type-check but not necessary linkable/runnable, or
* explain it like "```ignore (cannot-test-this-because-xxxx)", if the annotation cannot be avoided.
In all four of Break, Closure, Ret, Yeet, the needs_par_as_let_scrutinee
is guaranteed to return true because the .precedence().order() of those
expr kinds is <= AssocOp::LAnd.precedence().
The relevant functions in rustc_ast::util::parser are:
fn needs_par_as_let_scrutinee(order: i8) -> bool {
order <= prec_let_scrutinee_needs_par() as i8
}
fn prec_let_scrutinee_needs_par() -> usize {
AssocOp::LAnd.precedence()
}
The .precedence().order() of Closure is PREC_CLOSURE (-40) and of Break,
Ret, Yeet is PREC_JUMP (-30).
The value of AssocOp::LAnd.precedence() is 6.
So this commit causes no change in behavior, only potentially
performance by doing a redundant call to contains_exterior_struct_lit in
those four cases. This is fine because Break, Closure, Ret, Yeet should
be exceedingly rare in the position of a let scrutinee.
- Add `use Mode::*` to avoid all the qualifiers.
- Reorder the variants. The existing order makes no particular sense,
which has bugged me for some time. I've chosen an order that makes
sense to me.
These don't really make sense since C string literals were added. This
commit removes them in favour for `mode: Mode` args. `ascii_check` still
has a `characters_should_be_ascii: bool` arg.
Also, `characters_should_be_ascii` is renamed to be shorter.
temporarily revert "ice on ambguity in mir typeck"
Reverts #116530 as a temporary measure to fix#117577. That issue should be ultimately fixed by checking WF of type annotations prior to normalization, which is implemented in #104098 but this PR is intended to be backported to beta.
r? ``@compiler-errors`` (the reviewer of the reverted PR)
recurse into refs when comparing tys for diagnostics
before:
after:
this diff from the test suite is also quite nice imo:
```diff
`@@` -4,8 +4,8 `@@` error[E0308]: mismatched types
LL | debug_assert_eq!(iter.next(), Some(value));
| ^^^^^^^^^^^ expected `Option<<I as Iterator>::Item>`, found `Option<&<I as Iterator>::Item>`
|
- = note: expected enum `Option<<I as Iterator>::Item>`
- found enum `Option<&<I as Iterator>::Item>`
+ = note: expected enum `Option<_>`
+ found enum `Option<&_>`
```
privacy: visit trait def id of projections
Fixes#117997.
A refactoring in #117076 changed the `DefIdVisitorSkeleton` to avoid calling `visit_projection_ty` for `ty::Projection` aliases, and instead just iterate over the args - this makes sense, as `visit_projection_ty` will indirectly visit all of the same args, but in doing so, will also create a `TraitRef` containing the trait's `DefId`, which also gets visited. The trait's `DefId` isn't visited when we only visit the arguments without separating them into `TraitRef` and own args first.
Eventually this influences the reachability set and whether a function is encoded into the metadata.
Add instance evaluation and methods to read an allocation in StableMIR
The instance evaluation is needed to handle intrinsics such as `type_id` and `type_name`.
Since we now use Allocation to represent all evaluated constants, provide a few methods to help process the data inside an allocation.
I've also started to add a structured way to get information about the compilation target machine. For now, I've only added information needed to process an allocation.
r? ``````@ouz-a``````
Uplift the (new solver) canonicalizer into `rustc_next_trait_solver`
Uplifts the new trait solver's canonicalizer into a new crate called `rustc_next_trait_solver`.
The crate name is literally a bikeshed-avoidance name, so let's not block this PR on that -- renames are welcome later.
There are a host of other changes that were required to make this possible:
* Expose a `ConstTy` trait to get the `Interner::Ty` from a `Interner::Const`.
* Expose some constructor methods to construct `Bound` variants. These are currently methods defined on the interner themselves, but they could be pulled into traits later.
* Expose a `IntoKind` trait to turn a `Ty`/`Const`/`Region` into their corresponding `*Kind`s.
* Some minor tweaks to other APIs in `rustc_type_ir`.
The canonicalizer code itself is best reviewed **with whitespace ignored.**
r? ``@lcnr``
Explicitly implement `DynSync` and `DynSend` for `TyCtxt`
This is an attempt to short circuit trait resolution. It should get a perf run for bootstrap impact.
Introduce support for `async gen` blocks
I'm delighted to demonstrate that `async gen` block are not very difficult to support. They're simply coroutines that yield `Poll<Option<T>>` and return `()`.
**This PR is WIP and in draft mode for now** -- I'm mostly putting it up to show folks that it's possible. This PR needs a lang-team experiment associated with it or possible an RFC, since I don't think it falls under the jurisdiction of the `gen` RFC that was recently authored by oli (https://github.com/rust-lang/rfcs/pull/3513, https://github.com/rust-lang/rust/issues/117078).
### Technical note on the pre-generator-transform yield type:
The reason that the underlying coroutines yield `Poll<Option<T>>` and not `Poll<T>` (which would make more sense, IMO, for the pre-transformed coroutine), is because the `TransformVisitor` that is used to turn coroutines into built-in state machine functions would have to destructure and reconstruct the latter into the former, which requires at least inserting a new basic block (for a `switchInt` terminator, to match on the `Poll` discriminant).
This does mean that the desugaring (at the `rustc_ast_lowering` level) of `async gen` blocks is a bit more involved. However, since we already need to intercept both `.await` and `yield` operators, I don't consider it much of a technical burden.
r? `@ghost`
never_patterns: Parse match arms with no body
Never patterns are meant to signal unreachable cases, and thus don't take bodies:
```rust
let ptr: *const Option<!> = ...;
match *ptr {
None => { foo(); }
Some(!),
}
```
This PR makes rustc accept the above, and enforces that an arm has a body xor is a never pattern. This affects parsing of match arms even with the feature off, so this is delicate. (Plus this is my first non-trivial change to the parser).
~~The last commit is optional; it introduces a bit of churn to allow the new suggestions to be machine-applicable. There may be a better solution? I'm not sure.~~ EDIT: I removed that commit
r? `@compiler-errors`
A refactoring in #117076 changed the `DefIdVisitorSkeleton` to avoid
calling `visit_projection_ty` for `ty::Projection` aliases, and instead
just iterate over the args - this makes sense, as `visit_projection_ty`
will indirectly visit all of the same args, but in doing so, will also
create a `TraitRef` containing the trait's `DefId`, which also gets
visited. The trait's `DefId` isn't visited when we only visit the
arguments without separating them into `TraitRef` and own args first.
Signed-off-by: David Wood <david@davidtw.co>
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.
There are cases where coverage instrumentation wants to show a span for some
syntax element, but there is no MIR node that naturally carries that span, so
the instrumentor can't see it.
MIR building can now use this new kind of coverage statement to deliberately
include those spans in MIR, attached to a dummy statement that has no other
effect.
Resolve associated item bindings by namespace
This is the 3rd commit split off from #118360 with tests reblessed (they no longer contain duplicated diags which were caused by 4c0addc80a) & slightly adapted (removed supertraits from a UI test, cc #118040).
> * Resolve all assoc item bindings (type, const, fn (feature `return_type_notation`)) by namespace instead of trying to resolve a type first (in the non-RTN case) and falling back to consts afterwards. This is consistent with RTN. E.g., for `Tr<K = {…}>` we now always try to look up assoc consts (this extends to supertrait bounds). This gets rid of assoc tys shadowing assoc consts in assoc item bindings which is undesirable & inconsistent (types and consts live in different namespaces after all)
> * Consolidate the resolution of assoc {ty, const} bindings and RTN (dedup, better diags for RTN)
> * Fix assoc consts being labeled as assoc *types* in several diagnostics
> * Make a bunch of diagnostics translatable
Fixes#112560 (error → pass).
As discussed
r? `@compiler-errors`
---
**Addendum**: What I call “associated item bindings” are commonly referred to as “type bindings” for historical reasons. Nowadays, “type bindings” include assoc type bindings, assoc const bindings and RTN (return type notation) which is why I prefer not to use this outdated term.
coverage: Merge refined spans in a separate final pass
Pulling this merge step out of `push_refined_span` and into a separate pass lets us push directly to `refined_spans` instead of calling a helper method.
Because the compiler can now see partial borrows of `refined_spans`, we can remove some extra code that was jumping through hoops to satisfy the borrow checker.
---
``@rustbot`` label +A-code-coverage
Tell MirUsedCollector that the pointer alignment checks calls its panic symbol
Fixes https://github.com/rust-lang/rust/pull/118683 (not an issue, but that PR is a basically a bug report)
When we had `panic_immediate_abort` start adding `#[inline]` to this panic function, builds started breaking because we failed to write up the MIR assert terminator to the correct panic shim. Things happened to work before by pure luck because without this feature enabled, the function we're inserting calls to is `#[inline(never)]` so we always generated code for it.
r? bjorn3
coverage: Avoid unnecessary macros in unit tests
These macros don't provide enough value to justify their complexity, when they can just as easily be functions instead.
---
`@rustbot` label +A-code-coverage
The instance evaluation is needed to handle intrinsics such as
`type_id` and `type_name`.
Since we now use Allocation to represent all evaluated constants,
provide a few methods to help process the data inside an allocation.
`GenKillAnalysis` has five methods that take a transfer function arg:
- `statement_effect`
- `before_statement_effect`
- `terminator_effect`
- `before_terminator_effect`
- `call_return_effect`
All the transfer function args have type `&mut impl GenKill<Self::Idx>`,
except for `terminator_effect`, which takes the simpler `Self::Domain`.
But only the first two need to be `impl GenKill`. The other
three can all be `Self::Domain`, just like `Analysis`. So this commit
changes the last two to take `Self::Domain`, making `GenKillAnalysis`
and `Analysis` more similar.
(Another idea would be to make all these methods `impl GenKill`. But
that doesn't work: `MaybeInitializedPlaces::terminator_effect` requires
the arg be `Self::Domain` so that `self_is_unwind_dead(place, state)`
can be called on it.)
This results in two non-generic types being used: `BorrowckResults` and
`BorrowckFlowState`. It's a net reduction in lines of code, and a little
easier to read.
Rollup of 6 pull requests
Successful merges:
- #116420 (discard invalid spans in external blocks)
- #118686 (Only check principal trait ref for object safety)
- #118688 (Add method to get type of an Rvalue in StableMIR)
- #118707 (Ping GuillaumeGomez for changes in rustc_codegen_gcc)
- #118712 (targets: remove not-added {i386,i486}-unknown-linux-gnu)
- #118719 (CFI: Add char to CFI integer normalization)
Failed merges:
- #117586 (Uplift the (new solver) canonicalizer into `rustc_next_trait_solver`)
r? `@ghost`
`@rustbot` modify labels: rollup
It is used just once. With it removed, the relevant code is a little
boilerplate-y but much easier to read, and is the same length. Overall I
think it's an improvement.
targets: remove not-added {i386,i486}-unknown-linux-gnu
These files were added to the repository but never wired up so they could be used - and that was a few years ago without anyone noticing - so let's remove these, they can be re-added if someone wants them.
cc #80662
r? ```@pnkfelix``` (familiar with the tier policy and Wesley is on vacation)
Add method to get type of an Rvalue in StableMIR
Provide a method to StableMIR users to retrieve the type of an Rvalue operation. There were two possible implementation:
1. Create the logic inside stable_mir to process the type according to the Rvalue semantics, which duplicates the logic of `rustc_middle::mir::Rvalue::ty()`.
2. Implement the Rvalue translation from StableMIR back to internal representation, invoke the `rustc_middle::mir::Rvalue::ty()`, and translate the return value to StableMIR.
I chose the first one for now since the duplication was fairly small, and the option 2 would require way more work to translate everything back to rustc internal representation. If we eventually add those translations, we could easily swap to the option 2.
```@compiler-errors``` / ```@ouz-a``` Please let me know if you have any strong opinion here.
r? ```@compiler-errors```
Only check principal trait ref for object safety
It should make things a bit faster, in case we end up registering a bunch of object safety preds.
r? ```@ghost```
discard invalid spans in external blocks
Fixes#116203
This PR has discarded the invalid `const_span`, thereby making the format more neat.
r? ``@Nilstrieb``
Avoid adding builtin functions to `symbols.o`
We found performance regressions in #113923. The problem seems to be that `--gc-sections` does not remove these symbols. I tested that lld removes these symbols, but ld and gold do not.
I found that `used` adds symbols to `symbols.o` at 3e202ead60/compiler/rustc_codegen_ssa/src/back/linker.rs (L1786-L1791).
The PR removes builtin functions.
Note that under LTO, ld still preserves these symbols. (lld will still remove them.)
The first commit also fixes#118559. But I think the second commit also makes sense.
compile-time evaluation: detect writes through immutable pointers
This has two motivations:
- it unblocks https://github.com/rust-lang/rust/pull/116745 (and therefore takes a big step towards `const_mut_refs` stabilization), because we can now detect if the memory that we find in `const` can be interned as "immutable"
- it would detect the UB that was uncovered in https://github.com/rust-lang/rust/pull/117905, which was caused by accidental stabilization of `copy` functions in `const` that can only be called with UB
When UB is detected, we emit a future-compat warn-by-default lint. This is not a breaking change, so completely in line with [the const-UB RFC](https://rust-lang.github.io/rfcs/3016-const-ub.html), meaning we don't need t-lang FCP here. I made the lint immediately show up for dependencies since it is nearly impossible to even trigger this lint without `const_mut_refs` -- the accidentally stabilized `copy` functions are the only way this can happen, so the crates that popped up in #117905 are the only causes of such UB (in the code that crater covers), and the three cases of UB that we know about have all been fixed in their respective crates already.
The way this is implemented is by making use of the fact that our interpreter is already generic over the notion of provenance. For CTFE we now use the new `CtfeProvenance` type which is conceptually an `AllocId` plus a boolean `immutable` flag (but packed for a more efficient representation). This means we can mark a pointer as immutable when it is created as a shared reference. The flag will be propagated to all pointers derived from this one. We can then check the immutable flag on each write to reject writes through immutable pointers.
I just hope perf works out.
