Don't `alloca` just to look at a discriminant
Today we're making LLVM do a bunch of extra work when you match on trivial stuff like `Option<bool>` or `ControlFlow<u8>`.
This PR changes that so that simple types like `Option<u32>` or `Result<(), Box<Error>>` can stay as `OperandValue::ScalarPair` and we can still read the discriminant from them, rather than needing to write them into memory to have a `PlaceValue` just to get the discriminant out.
Fixes#137503
Currently (PatKind::Wild` (i.e. `_`) gets turned by
`lower_fn_params_to_names` into an empty identifier, which means it is
printed incorrectly by HIR pretty printing.
And likewise for `lower_fn_params_to_names`, which affects some error
messages.
This commit fixes them. This requires a slight tweak in a couple of
places to continue using parameter numbers in some error messages. And
it improves the output of `tests/ui/typeck/cyclic_type_ice.rs`:
`/* _ */` is a better suggestion than `/* */`.
Fix RISC-V VxWorks LLVM target triples
The targets `riscv32-wrs-vxworks` and `riscv64-wrs-vxworks` uses the plain `$ARCH` LLVM triple, which LLVM normalizes to `$ARCH-unknown-unknown`, we should use `$ARCH-unknown-linux-gnu$ABI` which is consistent with the the other VxWorks targets.
Motivation: To make it easier to verify that [`cc-rs`' conversion from `rustc` to Clang/LLVM triples](https://github.com/rust-lang/cc-rs/issues/1431) is correct.
Alternative: Pass `$ARCH-unknown-none` in the other VxWorks LLVM triples, I don't know anything about VxWorks, so am unsure which is the most correct option.
CC target maintainer `@biabbas.`
r? jieyouxu
minor interpreter cleanups
- remove the `eval_inline_asm` hook that `@saethlin` added; the usage never materialized and he agreed with removing it
- I tried merging `init_alloc_extra` and `adjust_global_allocation` and it didn't work; leave a comment as to why. Also, make the allocation code path a bit more clear by renaming `init_alloc_extra` to `init_local_allocation`.
r? `@oli-obk`
Add an opt-out in pretty printing for RTN rendering
Today, we render RPITIT types like `impl Sized { T::method(..) }` when RTN is enabled. This is very useful for diagnostics, since it's often not clear what the `impl Sized` type means by itself, and it makes it clear that that's an RPITIT that can be bounded using RTN syntax. See #115624.
However, since we don't distinguish types that are rendered for the purposes of printing messages vs suggestions, this representation leaks into suggestions and turns into code that can't be parsed. This PR adds a new `with_types_for_suggestion! {}` and `with_types_for_signature! {}` options to the pretty printing architecture to make it clear that we're rendering a type for code suggestions.
This can be applied later as we find that we need it.
Add a .bss-like scheme for encoded const allocs
This check if all bytes are zero feel like it should be too slow, and instead we should have a flag that we track, but that seems hard. Let's see how this perfs first.
Also we can probably stash the "it's all zero actually" flag inside one of the other struct members that's already not using an entire byte. This optimization doesn't fire all that often, so it's possible that by sticking it in the varint length field, this PR actually makes rmeta size worse.
make precise capturing args in rustdoc Json typed
close#137616
This PR includes below changes.
- Add `rustc_hir::PreciseCapturingArgKind` which allows the query system to return a arg's data.
- Add `rustdoc::clean::types::PreciseCapturingArg` and change to use it.
- Add `rustdoc-json-types::PreciseCapturingArg` and change to use it.
- Update `tests/rustdoc-json/impl-trait-precise-capturing.rs`.
- Bump `rustdoc_json_types::FORMAT_VERSION`.
attempt to support `BinaryFormat::Xcoff` in `naked_asm!`
Fixes https://github.com/rust-lang/rust/issues/137219
So, the inline assembly support for xcoff is extremely limited. The LLVM [XCOFFAsmParser](1b25c0c4da/llvm/lib/MC/MCParser/XCOFFAsmParser.cpp) does not support many of the attributes that LLVM itself emits, and that should exist based on [the assembler docs](https://www.ibm.com/docs/en/ssw_aix_71/assembler/assembler_pdf.pdf). It also does accept some that should not exist based on those docs.
So, I've tried to do the best I can given those limitations. At least it's better than emitting the directives for elf and having that fail somewhere deep in LLVM. Given that inline assembly for this target is incomplete (under `asm_experimental_arch`), I think that's OK (and again I don't see how we can do better given the limitations in LLVM).
r? ```@Noratrieb``` (given that you reviewed https://github.com/rust-lang/rust/pull/136637)
It seems reasonable to ping the [`powerpc64-ibm-aix` target maintainers](https://doc.rust-lang.org/rustc/platform-support/aix.html), hopefully they have thoughts too: ```@daltenty``` ```@gilamn5tr```
naked functions: on windows emit `.endef` without the symbol name
tracking issue: https://github.com/rust-lang/rust/issues/90957
fixes https://github.com/rust-lang/rust/issues/138320
The `.endef` directive does not take the name as an argument. Apparently the LLVM x86_64 parser does accept this, but on i686 it's rejected. In general `i686` does some special name mangling stuff, so it's good to include it in the naked function tests.
r? ````@ChrisDenton```` (because windows)
metadata: Ignore sysroot when doing the manual native lib search in rustc
This is the opposite alternative to https://github.com/rust-lang/rust/pull/138170 and another way to make native library search consistent between rustc and linker.
This way the directory list searched by rustc is still a prefix of the directory list considered by linker, but it's a shorter prefix than in #138170.
We can include the sysroot directories into rustc's search again later if the issues with #138170 are resolved, it will be a backward compatible change.
This may break some code doing weird things on unstable rustc, or tier 2-3 targets, like bundling `libunwind.a` or sanitizers into something.
Note that this doesn't affect shipped `libc.a`, because it lives in `self-contained` directories in sysroot, and `self-contained` sysroot is already not included into the rustc's search. All libunwind and sanitizer libs should be moved to `self-contained` sysroot too eventually.
With the consistent search directory list between rustc and linker we can make rustc own the native library search (at least for static libs) and use linker search only as a fallback (like in #123436). This will allow addressing issues like https://github.com/rust-lang/rust/pull/132394 once and for all on all targets.
r? ``@bjorn3``
The place previously used here was that of the temporary holding the
reference returned by `Deref::deref` or `DerefMut::deref_mut`. However,
since the inner pattern of `deref!(inner)` expects the deref-target type
itself, this would ICE when that type was inspected (e.g. by the EUV
case for slice patterns). This adds a deref projection to fix that.
Since current in-tree consumers of EUV (upvar inference and clippy)
don't care about Rvalues, the place could be simplified to
`self.cat_rvalue(pat.hir_id, self.pat_ty_adjusted(subpat)?)` to save
some cycles. I personally find EUV to be a bit fragile, so I've opted
for pedantic correctness. Maybe a `HACK` comment would suffice though?
This means that things like `<usize as Step>::forward_unchecked` and `<PartialOrd for f32>::le` will inline even if we've already done a bunch of inlining to find the calls to them.
Do not register `Self: AutoTrait` when confirming auto trait (in old solver)
Every built-in auto impl for a trait goal like `Ty: Auto` immediately registers another obligation of `Ty: Auto` as one of its nested obligations, leading to us stressing the cycle detection machinery a lot more than we need to. This is because all traits have a `Self: Trait` predicate.
To fix this, remove the call to `impl_or_trait_obligations` in `vtable_auto_impl`, since auto traits do not have where clauses.
r? lcnr
In LLVM v9, lld has supported @response-file
LLVM v9 was released on 2019-09-19.
And the check was added back to 2018-03-14 (1.26.0) via 04442af18b.
It has been more than five years, and we ship our own lld regardlessly.
This should be happily removed.
See also:
* <bb12396f91>
* <https://reviews.llvm.org/D63024>
Calculate predecessor count directly
Avoid allocating a vector of small vectors merely to determine how many
predecessors each basic block has.
Additionally use u8 and saturating operations. The pass only needs to
distinguish between [0..1] and [2..].
The targets used the plain `$ARCH` triple, which LLVM normalizes to
`$ARCH-unknown-unknown`, which is inconsistent with the the other
VxWorks targets which all use `$ARCH-unknown-linux-gnu$ABI`.
Disentangle `ForwardGenericParamBan` and `ConstParamTy` ribs
In #137617, the `ConstParamTy` rib was adjusted to act kinda like the `ForwardGenericParamBan`. However, this means that it no longer served its purpose banning generics from *parent items*. Although we still are checking for param type validity using the `ConstParamTy_` trait, which means that we weren't accepting code we shouldn't, I think it's a bit strange for us not to be rejecting code like this during *resolution* and instead letting these malformed const generics leak into the type system:
```rust
trait Foo<T> {
fn bar<const N: T>() {}
}
```
This PR does a few things:
1. Introduce a `ForwardGenericParamBanReason` enum, and start using the `ForwardGenericParamBan` rib to ban forward-declared params in const tys when `generic_const_parameter_types` is enabled.
2. Start using the `ConstParamTy` rib to ban *all* generics when `generic_const_parameter_types` is disabled.
3. Improve the diagnostics for both of the cases above, and for forward-declared params in parameter defaults too :3
r? `@BoxyUwU` or reassign
Support rmeta inputs for --crate-type=bin --emit=obj
This already works for --emit=metadata, but is possible anytime we're not linking.
Tests:
- `rmeta_bin` checks we're not changing --emit=link (already passes)
- `rmeta_bin-pass` tests the new behavior for --emit=obj (would fail today) and also --emit=metadata which isn't changing
Convert `ShardedHashMap` to use `hashbrown::HashTable`
The `hash_raw_entry` feature (#56167) has finished fcp-close, so the compiler
should stop using it to allow its removal. Several `Sharded` maps were
using raw entries to avoid re-hashing between shard and map lookup, and
we can do that with `hashbrown::HashTable` instead.
change definitely unproductive cycles to error
builds on top of #136824 by adding a third variant to `PathKind` for paths which may change to be coinductive in the future but must not be so right now. Most notably, impl where-clauses of not yet coinductive traits.
With this, we can change cycles which are definitely unproductive to a proper error. This fixes https://github.com/rust-lang/trait-system-refactor-initiative/issues/114. This does not affect stable as we keep these cycles as ambiguous during coherence.
r? ````````@compiler-errors```````` ````````@nikomatsakis````````
Fix false-positive in `expr_or_init` and in the `invalid_from_utf8` lint
This PR fixes the logic for finding initializer in the `expr_or_init` and `expr_or_init_with_outside_body` functions.
If the binding were to be mutable (`let mut`), the logic wouldn't consider that the initializer expression could have been modified and would return the init expression even-trough multiple subsequent assignments could have been done.
Example:
```rust
let mut a = [99, 108, 130, 105, 112, 112]; // invalid, not UTF-8
loop {
a = *b"clippy"; // valid
break;
}
std::str::from_utf8_mut(&mut a); // currently warns, with this PR it doesn't
```
This PR modifies the logic to excludes mutable let bindings.
Found when using `expr_or_init` in https://github.com/rust-lang/rust/pull/119220.
r? compiler
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.
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
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``
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
