miri: treat non-memory local variables properly for data race detection
Fixes https://github.com/rust-lang/miri/issues/3242
Miri has an optimization where some local variables are not represented in memory until something forces them to be stored in memory (most notably, creating a pointer/reference to the local will do that). However, for a subsystem triggering on memory accesses -- such as the data race detector -- this means that the memory access seems to happen only when the local is moved to memory, instead of at the time that it actually happens. This can lead to UB reports in programs that do not actually have UB.
This PR fixes that by adding machine hooks for reads and writes to such efficiently represented local variables. The data race system tracks those very similar to how it would track reads and writes to addressable memory, and when a local is moved to memory, the clocks get overwritten with the information stored for the local.
Rollup of 5 pull requests
Successful merges:
- #129195 (Stabilize `&mut` (and `*mut`) as well as `&Cell` (and `*const Cell`) in const)
- #130118 (move Option::unwrap_unchecked into const_option feature gate)
- #130295 (Fix target-cpu fpu features on Armv8-R.)
- #130371 (Correctly account for niche-optimized tags in rustc_transmute)
- #130381 (library: Compute Rust exception class from its string repr)
r? `@ghost`
`@rustbot` modify labels: rollup
Correctly account for niche-optimized tags in rustc_transmute
This is a bit hacky, but it fixes the ICE and makes it possible to run the safe transmute check on every `mem::transmute` check we instantiate. I want to write a lint that needs to do that, but this stands well on its own.
cc `@jswrenn` here's the fix I alluded to yesterday :)
Fixes#123693
Fix target-cpu fpu features on Armv8-R.
This is a follow-up to #123159, but applied to Armv8-R.
This required https://github.com/llvm/llvm-project/pull/88287 to work properly. Now that this change exists in rustc's llvm, we can fix Armv8-R's default fpu features. In Armv8-R's case, the default features from LLVM for floating-point are sufficient, because there is no integer-only variant of this architecture.
Stabilize `&mut` (and `*mut`) as well as `&Cell` (and `*const Cell`) in const
This stabilizes `const_mut_refs` and `const_refs_to_cell`. That allows a bunch of new things in const contexts:
- Mentioning `&mut` types
- Creating `&mut` and `*mut` values
- Creating `&T` and `*const T` values where `T` contains interior mutability
- Dereferencing `&mut` and `*mut` values (both for reads and writes)
The same rules as at runtime apply: mutating immutable data is UB. This includes mutation through pointers derived from shared references; the following is diagnosed with a hard error:
```rust
#[allow(invalid_reference_casting)]
const _: () = {
let mut val = 15;
let ptr = &val as *const i32 as *mut i32;
unsafe { *ptr = 16; }
};
```
The main limitation that is enforced is that the final value of a const (or non-`mut` static) may not contain `&mut` values nor interior mutable `&` values. This is necessary because the memory those references point to becomes *read-only* when the constant is done computing, so (interior) mutable references to such memory would be pretty dangerous. We take a multi-layered approach here to ensuring no mutable references escape the initializer expression:
- A static analysis rejects (interior) mutable references when the referee looks like it may outlive the current MIR body.
- To be extra sure, this static check is complemented by a "safety net" of dynamic checks. ("Dynamic" in the sense of "running during/after const-evaluation, e.g. at runtime of this code" -- in contrast to "static" which works entirely by looking at the MIR without evaluating it.)
- After the final value is computed, we do a type-driven traversal of the entire value, and if we find any `&mut` or interior-mutable `&` we error out.
- However, the type-driven traversal cannot traverse `union` or raw pointers, so there is a second dynamic check where if the final value of the const contains any pointer that was not derived from a shared reference, we complain. This is currently a future-compat lint, but will become an ICE in #128543. On the off-chance that it's actually possible to trigger this lint on stable, I'd prefer if we could make it an ICE before stabilizing const_mut_refs, but it's not a hard blocker. This part of the "safety net" is only active for mutable references since with shared references, it has false positives.
Altogether this should prevent people from leaking (interior) mutable references out of the const initializer.
While updating the tests I learned that surprisingly, this code gets rejected:
```rust
const _: Vec<i32> = {
let mut x = Vec::<i32>::new(); //~ ERROR destructor of `Vec<i32>` cannot be evaluated at compile-time
let r = &mut x;
let y = x;
y
};
```
The analysis that rejects destructors in `const` is very conservative when it sees an `&mut` being created to `x`, and then considers `x` to be always live. See [here](https://github.com/rust-lang/rust/issues/65394#issuecomment-541499219) for a longer explanation. `const_precise_live_drops` will solve this, so I consider this problem to be tracked by https://github.com/rust-lang/rust/issues/73255.
Cc `@rust-lang/wg-const-eval` `@rust-lang/lang`
Cc https://github.com/rust-lang/rust/issues/57349
Cc https://github.com/rust-lang/rust/issues/80384
interpret: get_ptr_alloc_mut: lookup allocation only once
I don't think this will make a big perf difference, but it makes this function symmetric with `get_ptr_alloc` -- and it's always nice to successfully solve a borrow checker puzzle like this. ;)
This is a follow-up to #123159, but applied to Armv8-R.
This required https://github.com/llvm/llvm-project/pull/88287 to work
properly. Now that this change exists in rustc's llvm, we can fix
Armv8-R's default fpu features. In Armv8-R's case, the default features
from LLVM for floating-point are sufficient, because there is no
integer-only variant of this architecture.
Make some lint doctests compatible with `--stage=0`
Currently, running `x test compiler --stage=0` (with `rust.parallel-compiler=false` to avoid other problems) results in two failures, because these lint doctests aren't compatible with the current stage0 compiler.
In theory, the more “correct” solution would be to wrap the opening triple-backtick line in `#[cfg_attr(not(bootstrap), doc = "..."]`. However, that causes a few practical problems:
- `tidy` doesn't understand that syntax, and miscounts the number of backticks in the comment block.
- `lint-docs` doesn't understand that syntax, and thinks it's trying to declare the lint name.
- Working around the above problems would cause more work and more confusion for whoever does the next bootstrap beta bump.
So instead this PR adds some bootstrap gates inside the individual doctests, which end up producing the desired behaviour, and are straightforward to remove.
stabilize `const_extern_fn`
closes https://github.com/rust-lang/rust/issues/64926
tracking issue: https://github.com/rust-lang/rust/issues/64926
reference PR: https://github.com/rust-lang/reference/pull/1596
## Stabilizaton Report
### Summary
Using `const extern "Rust"` and `const extern "C"` was already stabilized (since version 1.62.0, see https://github.com/rust-lang/rust/pull/95346). This PR stabilizes the other calling conventions: it is now possible to write `const unsafe extern "calling-convention" fn` and `const extern "calling-convention" fn` for any supported calling convention:
```rust
const extern "C-unwind" fn foo1(val: u8) -> u8 { val + 1}
const extern "stdcall" fn foo2(val: u8) -> u8 { val + 1}
const unsafe extern "C-unwind" fn bar1(val: bool) -> bool { !val }
const unsafe extern "stdcall" fn bar2(val: bool) -> bool { !val }
```
This can be used to const-ify an `extern fn`, or conversely, to make a `const fn` callable from external code.
r? T-lang
cc `@RalfJung`
const-eval interning: accept interior mutable pointers in final value
…but keep rejecting mutable references
This fixes https://github.com/rust-lang/rust/issues/121610 by no longer firing the lint when there is a pointer with interior mutability in the final value of the constant. On stable, such pointers can be created with code like:
```rust
pub enum JsValue {
Undefined,
Object(Cell<bool>),
}
impl Drop for JsValue {
fn drop(&mut self) {}
}
// This does *not* get promoted since `JsValue` has a destructor.
// However, the outer scope rule applies, still giving this 'static lifetime.
const UNDEFINED: &JsValue = &JsValue::Undefined;
```
It's not great to accept such values since people *might* think that it is legal to mutate them with unsafe code. (This is related to how "infectious" `UnsafeCell` is, which is a [wide open question](https://github.com/rust-lang/unsafe-code-guidelines/issues/236).) However, we [explicitly document](https://doc.rust-lang.org/reference/behavior-considered-undefined.html) that things created by `const` are immutable. Furthermore, we also accept the following even more questionable code without any lint today:
```rust
let x: &'static Option<Cell<i32>> = &None;
```
This is even more questionable since it does *not* involve a `const`, and yet still puts the data into immutable memory. We could view this as promotion [potentially introducing UB](https://github.com/rust-lang/unsafe-code-guidelines/issues/493). However, we've accepted this since ~forever and it's [too late to reject this now](https://github.com/rust-lang/rust/pull/122789); the pattern is just too useful.
So basically, if you think that `UnsafeCell` should be tracked fully precisely, then you should want the lint we currently emit to be removed, which this PR does. If you think `UnsafeCell` should "infect" surrounding `enum`s, the big problem is really https://github.com/rust-lang/unsafe-code-guidelines/issues/493 which does not trigger the lint -- the cases the lint triggers on are actually the "harmless" ones as there is an explicit surrounding `const` explaining why things end up being immutable.
What all this goes to show is that the hard error added in https://github.com/rust-lang/rust/pull/118324 (later turned into the future-compat lint that I am now suggesting we remove) was based on some wrong assumptions, at least insofar as it concerns shared references. Furthermore, that lint does not help at all for the most problematic case here where the potential UB is completely implicit. (In fact, the lint is actively in the way of [my preferred long-term strategy](https://github.com/rust-lang/unsafe-code-guidelines/issues/493#issuecomment-2028674105) for dealing with this UB.) So I think we should go back to square one and remove that error/lint for shared references. For mutable references, it does seem to work as intended, so we can keep it. Here it serves as a safety net in case the static checks that try to contain mutable references to the inside of a const initializer are not working as intended; I therefore made the check ICE to encourage users to tell us if that safety net is triggered.
Closes https://github.com/rust-lang/rust/issues/122153 by removing the lint.
Cc `@rust-lang/opsem` `@rust-lang/lang`
Fix `Parser::break_up_float`'s right span
```rs
use std::mem::offset_of;
fn main() {
offset_of!((u8,), 0.0);
}
```
Before:
```
error[E0609]: no field `0` on type `u8`
--> ./main.rs:4:25
|
4 | offset_of!((u8,), 0.0);
| _____--------------------^-
| | |
| | in this macro invocation
5 | | }
... |
|
= note: this error originates in the macro `offset_of` (in Nightly builds, run with -Z macro-backtrace for more info)
error: aborting due to 1 previous error
```
After:
```
error[E0609]: no field `0` on type `u8`
--> ./main.rs:4:25
|
4 | offset_of!((u8,), 0.0);
| ^
error: aborting due to 1 previous error
```
---
`@rustbot` label +A-parser +D-imprecise-spans
simd_shuffle: require index argument to be a vector
Remove some codegen hacks by forcing the SIMD shuffle `index` argument to be a vector, which means (thanks to https://github.com/rust-lang/rust/pull/128537) that it will automatically be passed as an immediate in LLVM. The only special-casing we still have is for the extra sanity-checks we add that ensure that the indices are all in-bounds. (And the GCC backend needs to do a bunch of work since the Rust intrinsic is modeled after what LLVM expects, which seems to be quite different from what GCC expects.)
Fixes https://github.com/rust-lang/rust/issues/128738, see that issue for more context.
Simplify the canonical clone method and the copy-like forms to copy
Fixes#128081.
The optimized clone method ends up as the following MIR:
```
_2 = copy ((*_1).0: i32);
_3 = copy ((*_1).1: u64);
_4 = copy ((*_1).2: [i8; 3]);
_0 = Foo { a: move _2, b: move _3, c: move _4 };
```
We can transform this to:
```
_0 = copy (*_1);
```
r? `@cjgillot`
Fix `SDKROOT` ignore on macOS
`rustc` has code to detect when `SDKROOT` is obviously set for the wrong platform, so that it can choose to ignore it. This is a pretty important feature for Cargo build scripts and proc macros, since you will often have `SDKROOT` set to an iOS platform there.
However, the code was checking for an old SDK version name `"macosx10.15"` that was previously configured by `add_apple_sdk`, but nowadays configured to the correct `"macosx"`. I think this error was introduced in part https://github.com/rust-lang/rust/pull/77202 and in https://github.com/rust-lang/rust/pull/100286.
Fixes part of https://github.com/rust-lang/rust/issues/80817 (linking with `-Clinker=ld` now works), though more work is still needed in this area, see also https://github.com/rust-lang/rust/issues/129432.
``@rustbot`` label O-macos A-cross
(fix) conflicting negative impl marker
## Context
This MR fixes the error message for conflicting negative trait impls by adding the corresponding the polarity marker to the trait name.
## Issues
- closes#70849
r? `@fmease`
Failing to do this results in the lint example output complaining
about the lint not existing instead of the thing the lint is supposed
to be complaining about.
`ProjectionElem` and `UnOp`/`BinOp` dont need to be `PartialOrd`/`Ord`
These types don't really admit a natural ordering and no code seems to rely on it, so let's remove it.
Don't call closure_by_move_body_def_id on FnOnce async closures in MIR validation
Refactors the check in #129847 to not unncessarily call the `closure_by_move_body_def_id` query for async closures that don't *need* a by-move body.
Fixes#130167
Rollup of 6 pull requests
Successful merges:
- #129320 (Fix crash when labeling arguments for call_once and friends)
- #130266 (target: default to the medium code model on LoongArch targets)
- #130297 (Dataflow cleanups)
- #130299 (Add set_dcx to ParseSess)
- #130301 (some fixes for clashing_extern_declarations lint)
- #130305 (Clippy: consider msrv for const context for const_float_bits_conv)
r? `@ghost`
`@rustbot` modify labels: rollup
some fixes for clashing_extern_declarations lint
There were two issues with the clashing_extern_declarations lint:
- It would accept non-`repr(C)` structs as compatible with each other by comparing their fields in declaration order, but the fields could have different memory order (and with `-Zrandomize-layout`, this can really happen).
- It would accept two types as compatible if `compare_layouts` returns `true`, but that function actually just compared the *ABI*, not the fully layout -- and all sized structs with more than 2 fields have the same ABI (`Abi::Aggregate`), so this missed a *lot* of cases.
We don't currently have a clear spec for what we *want* to consider "clashing" and what is fine, so I otherwise kept the original logic. I hope to have a t-lang discussion about this at some point. But meanwhile, these changes seem like clear bugfixes.
Add set_dcx to ParseSess
After [this](https://github.com/rust-lang/rust/pull/126623) PR was merged, it is no longer possible to inject one's own `Emitter` in the way [described in the Compiler Development Guide](https://rustc-dev-guide.rust-lang.org/rustc-driver-getting-diagnostics.html). The reason is that the `dcx` field in `ParseSess` is no longer public, so it is not possible to update the `dcx` field with a `DiagCtxt` that contains one's own `Emitter` in the `psess_created` callback in `rustc_interface::Config`. The only way I have found to insert my own `DiagCtxt` is by creating an entirely new `ParseSess` and replacing the old one. This is not a good solution as the original `ParseSess` contains fields I would like to keep. (In my case the problem is that I lose the `cfg` and `check-cfg` fields of the original.)
The solution proposed in this PR is to add a `set_dcx` method to `ParseSess`. Per my limited understanding of the rustc codebase this should be fine as `set_dcx` requires a mutable reference to `ParseSess`, which is as far as I know only available in the `psess_created` callback (outside of `rustc_interface::run_compiler`).
If this PR is accepted, I will create a new PR to update the aforementioned example in the Compiler Development Guide.
target: default to the medium code model on LoongArch targets
The Rust LoongArch targets have been using the default LLVM code model so far, which is "small" in LLVM-speak and "normal" in LoongArch-speak. As described in the "Code Model" section of LoongArch ELF psABI spec v20231219 [1], one can only make function calls as far as ±128MiB with the "normal" code model; this is insufficient for very large software containing Rust components that needs to be linked into the big text section, such as Chromium.
Because:
* we do not want to ask users to recompile std if they are to build such software,
* objects compiled with larger code models can be linked with those with smaller code models without problems, and
* the "medium" code model is comparable to the "small"/"normal" one performance-wise (same data access pattern; each function call becomes 2-insn long and indirect, but this may be relaxed back into the direct 1-insn form in a future LLVM version), but is able to perform function calls within ±128GiB,
it is better to just switch the targets to the "medium" code model, which is also "medium" in LLVM-speak.
Relands [2]: #120661
[1]: https://github.com/loongson/la-abi-specs/blob/v2.30/laelf.adoc#code-models
[2]: https://github.com/rust-lang/rust/issues/121289#issuecomment-2333687396
Fix crash when labeling arguments for call_once and friends
When calling a method on Fn* traits explicitly, argument diagnostics should point at the called method (eg Fn::call_once), not the underlying callee.
This PR makes 3 main changes:
* It uses TupleArguments to detect if the user called a Fn* method directly (`my_fn.call_once(…)`) or implicitly (`my_fn(…)`). If it was explicit, argument diagnostics should point at the call_once method, not the underlying callable.
* The previous state was causing confusion between the two arguments lists (which could be different lengths), causing an out-of-bounds slice indexing in #128848. I added a length assert to capture the requirement in case this regresses or happens in another case.
* Unfortunately, this assert tripped when the required arguments information was not available (`self.get_hir_params_with_generics` was returning an empty Vec), so I've updated that to return None when that information is not available. (cc `@strottos` if you have any comments, since you added this function in #121595) Sorry this causes a bunch of indentation changes, recommend reviewing [ignoring whitespace](https://github.com/rust-lang/rust/pull/129320/files?w=1).)
This is my first rustc PR, so please call out if you'd like this split into more commits (or PRs), style nits, etc. I will add a few comments/questions inline. Thank you!
Fixes#128848
