Elaborate trait assumption in `receiver_is_dispatchable`
Fixes#138172. See comment on the linked test.
Probably not a fix for the general problem, bc I think this may still be incomplete for other weird `where` clauses on the receiver. But 🤷, supertraits seems like an obvious one to fix.
Add PeekMut::refresh
I'm not sure if this should go through ACP or not. BinaryHeap is not the most critical data structure in the standard library and it would be understandable if maintainer throughput is thus too limited to accept this PR without a proper design phase that ensures the required understanding of consequence over a longer time period.
This aims to improve the useability of heaps for priority-based work queues. In certain scenarios, modifications on the most relevant or critical items are performed until a condition that determines the work items have been sufficiently addressed. For instance the criticality could be a deadline that is relaxed whenever some part of a work item is completed. Such a loop will repeatedly access the most critical item and put it back in a sorted position when it is complete. Crucially, due to the ordering invariant we know that all necessary work was performed when the completed item remains the most critical. Getting this information from the heap position avoids a (potentially more costly) check on the item state itself.
A customized `drop` with boolean result would avoid up to two more comparisons performed in both the last no-op refresh and Drop code but this occurs once in each execution of the above scenario whereas refresh occurs any number of times. Also note that the comparison overhead of Drop is only taken if the element is mutably inspected to determine the end condition, i.e. not when refresh itself is the break condition.
strip `-Wlinker-messages` wrappers from `rust-lld` rmake test
The `tests/run-make/rust-lld` rmake test is failing locally on my M1, due to linker messages being in a different shape than the test expects: it asserts that the LLD version is the first linker message, which is seemingly not always the case on osx I guess.
```console
thread 'main' panicked at /Users/lqd/rust/lqd-rust/tests/run-make/rust-lld/rmake.rs:24:5:
the LLD version string should be present in the output logs:
warning: linker stderr: rust-lld: directory not found for option -L/usr/local/lib
LLD 20.1.0 (https://github.com/rust-lang/llvm-project.git 1c3bb96fdb6db7b8e8f24edb016099c223fdd27e)
Library search paths:
/Users/lqd/rust/lqd-rust/build/aarch64-apple-darwin/test/run-make/rust-lld/rmake_out
/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/lib
Framework search paths:
/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/System/Library/Frameworks
```
This PR normalizes away the `-Wlinker-messages` wrappers around the linker output, to remove the requirement that the linker version is the first linker message / is prefixed with the warning wrapper in the regex.
(also another strange thing to explain the pre-existing regex: it seems the LLD version is sometimes output on stderr sometimes on stdout cool stuff)
We could do this for the other lld rmake tests, but they're only enabled on x64 linux so less likely to have random linker messages appearing without anyone noticing.
It reinterprets uninitialized memory as initialized and does not drop
existing elements of the Vec. Fix that.
Additionally, make it more general by appending, instead of overwriting
existing elements, and rename it to `append_to_enclave_vec`. A caller
can simply call `.clear()` before, for the old behavior.
Allow bounds checks when enumerating `IndexSlice` to be elided
Without this hint, each loop iteration has to separately bounds check the index. See https://godbolt.org/z/zrfPY4Ten for an example.
This is technically a behaviour change, but only in cases where the compiler is going to crash anyways.
The idea is to identify cases of symbols/identifiers that are not
expected to be used. There isn't a perfectly sharp line between "dummy"
and "not dummy", but I think it's useful nonetheless.
This is never hit in the test suite.
At some point the check should be removed entirely. There are a million
places in the compiler where an empty symbol doesn't make sense, so a
check of this nature has almost zero value. But I'll leave it in place
for now just in case it gets hit by fuzzing or in the wild.
Continuing the work from #137350.
Removes the unused methods: `expect_variant`, `expect_field`,
`expect_foreign_item`.
Every method gains a `hir_` prefix.
To make room for the moving of `Map::attrs` to `TyCtxt::hir_attrs` in
the next commit. (It makes sense to rename the query, because it has
many fewer uses than the method.)
Rollup of 10 pull requests
Successful merges:
- #137715 (Allow int literals for pattern types with int base types)
- #138002 (Disable CFI for weakly linked syscalls)
- #138051 (Add support for downloading GCC from CI)
- #138231 (Prevent ICE in autodiff validation by emitting user-friendly errors)
- #138245 (stabilize `ci_rustc_if_unchanged_logic` test for local environments)
- #138256 (Do not feed anon const a type that references generics that it does not have)
- #138284 (Do not write user type annotation for const param value path)
- #138296 (Remove `AdtFlags::IS_ANONYMOUS` and `Copy`/`Clone` condition for anonymous ADT)
- #138352 (miri native_calls: ensure we actually expose *mutable* provenance to the memory FFI can access)
- #138354 (remove redundant `body` arguments)
r? `@ghost`
`@rustbot` modify labels: rollup
We can create the expected error manually, rather than trying to produce
a real one, so the error conversion test can run on all targets. Before,
it was only running on 64-bit and not miri.
In Fedora, we also found that s390x was not getting the expected error,
"successfully" allocating the huge size because it was optimizing the
real `malloc` call away. It's possible to counter that by looking at the
pointer in any way, like a debug print, but it's more robust to just
deal with errors directly, since this test is only about conversion.
miri native_calls: ensure we actually expose *mutable* provenance to the memory FFI can access
In native call mode, the interpreter memory itself is accessed directly by external code via pointers created from integers and passed via libffi, so we have to ensure the provenance in Miri itself (on the meta level) is sufficiently exposed. So far we only exposed the provenance for read-only accesses. This may we enough as that may actually be the same provenance as for mutable accesses, but it's hard to be sure, and anyway there's no reason to do such a gambit -- we have this function, `prepare_for_native_call`, which iterates all memory the call can access. let's just also (re-)expose Miri's own allocations there. We expose the read-only provenance for all of them and the mutable provenance for the mutable allocations.
r? ``@oli-obk``
Remove `AdtFlags::IS_ANONYMOUS` and `Copy`/`Clone` condition for anonymous ADT
cc #131045, which removed anonymous ADTs from the compiler
I forgot more stuff I guess.
Do not write user type annotation for const param value path
As I noted in the code comment, `DefKind::ConstParam` isn't actually *generic* over its own args, we just use the identity args from the body when lowering the value path so we have something to plug into the `EarlyBinder` we get back from `type_of` for the const param. So skip over it in `write_user_type_annotation_from_args`.
Somewhat unrelated, but I left an explanation for a somewhat mysterious quirk in the THIR lowering of user type annotations for patterns having to do with ctors and their `type_of` not actually being the type of the pattern node it's ascribing.
Fixes#138048
r? ``@BoxyUwU``
Do not feed anon const a type that references generics that it does not have
Fixes#137865
See the comment I left in the code. We could alternatively give these anon consts the generics from the parent, but that would be moving in a GCE-esque direction that we may not want. Open to tweaks here.
r? BoxyUwU
Prevent ICE in autodiff validation by emitting user-friendly errors
This PR moves `valid_ret_activity` and `valid_input_activity` checks to the macro expansion phase in compiler/rustc_builtin_macros/src/autodiff.rs, replacing the following internal compiler error (ICE):
```
error: internal compiler error:
compiler/rustc_codegen_ssa/src/codegen_attrs.rs:935:13:
Invalid input activity Dual for Reverse mode
```
with a more user-friendly message.
The issue specifically affected the test file `tests/ui/autodiff/autodiff_illegal.rs`, impacting the functions `f5` and `f6`.
The ICE can be reproduced by following [Enzyme's Rustbook](https://enzymead.github.io/rustbook/installation.html) installation guide.
Additionally, this PR adds tests for invalid return activity in `autodiff_illegal.rs`, which previously triggered an unnoticed ICE before these fixes.
r? ``@oli-obk``
Add support for downloading GCC from CI
This PR adds a new bootstrap config section called `gcc` and implements a single config `download-ci-gcc`. Its behavior is similar to `download-ci-llvm`. Since https://github.com/rust-lang/rust/pull/137667, we distribute a CI component that contains the prebuilt `libgccjit.so` library on x64 Linux. With `download-ci-gcc`, this component is downloaded from CI to avoid building GCC locally.
This is an MVP of this functionality, designed for local usage. This PR does not enable this functionality on the LLVM 18 PR CI job which builds `cg_gcc`, and does not implement more complex detection logic. It simply uses `false` (build locally) or `true` (download from CI if you're on the right target, if CI download fails, then bootstrap fails).
The original LLVM CI download functionality has a lot of features and complexity, which we don't need for GCC (yet). I don't like how the LLVM CI stuff is threaded through multiple parts of bootstrap, so with GCC I would like to take a more centralized approach, where the `build::Gcc` step handles download from CI internally. This means that:
- For the rest of bootstrap, it should be transparent whether GCC was built locally or downloaded from CI.
- GCC is not downloaded eagerly unless you actually requested GCC (either you requested `x build gcc` or you asked to build/test the GCC backend).
This approach will require some modifications once we extend this feature, but so far I like this approach much more than putting this stuff into `Config[::parse]`, which already does a ton of stuff that it arguably shouldn't (but it's super difficult to extract its logic out).
This PR is an alternative to https://github.com/rust-lang/rust/pull/130749, which did a more 1:1 copy of the `download-ci-llvm` logic.
r? ``@onur-ozkan``
Disable CFI for weakly linked syscalls
Currently, when enabling CFI via -Zsanitizer=cfi and executing e.g. std::sys::random::getrandom, we can observe a CFI violation. This is the case for all consumers of the std::sys::pal::weak::syscall macro, as it is defining weak functions which don't show up in LLVM IR metadata. CFI fails for all these functions.
Similar to other such cases in
https://github.com/rust-lang/rust/issues/115199, this change stops emitting the CFI typecheck for consumers of the macro via the `#[no_sanitize(cfi)]` attribute.
r? ``````@rcvalle``````
Allow int literals for pattern types with int base types
r? ``@BoxyUwU``
I also added an error at layout computation time for layouts that contain wrapping ranges (happens at monomorphization time). This is obviously hacky, but at least prevents such types from making it to codegen for now. It made writing the tests for int literals easier as I didn't have to think about that edge case
Basically this PR allows you to stop using transmutes for creating pattern types and instead just use literals:
```rust
let x: pattern_type!(u32 is 5..10) = 7;
```
works, and if the literal is out of range you get a type mismatch because it just stays at the base type and the base type can't be coerced to the pattern type.
cc ``@joshtriplett`` ``@scottmcm``
Add `#[define_opaques]` attribute and require it for all type-alias-impl-trait sites that register a hidden type
Instead of relying on the signature of items to decide whether they are constraining an opaque type, the opaque types that the item constrains must be explicitly listed.
A previous version of this PR used an actual attribute, but had to keep the resolved `DefId`s in a side table.
Now we just lower to fields in the AST that have no surface syntax, instead a builtin attribute macro fills in those fields where applicable.
Note that for convenience referencing opaque types in associated types from associated methods on the same impl will not require an attribute. If that causes problems `#[defines()]` can be used to overwrite the default of searching for opaques in the signature.
One wart of this design is that closures and static items do not have generics. So since I stored the opaques in the generics of functions, consts and methods, I would need to add a custom field to closures and statics to track this information. During a T-types discussion we decided to just not do this for now.
fixes#131298
