Implement a lint for implicit autoref of raw pointer dereference - take 2
*[t-lang nomination comment](https://github.com/rust-lang/rust/pull/123239#issuecomment-2727551097)*
This PR aims at implementing a lint for implicit autoref of raw pointer dereference, it is based on #103735 with suggestion and improvements from https://github.com/rust-lang/rust/pull/103735#issuecomment-1370420305.
The goal is to catch cases like this, where the user probably doesn't realise it just created a reference.
```rust
pub struct Test {
data: [u8],
}
pub fn test_len(t: *const Test) -> usize {
unsafe { (*t).data.len() } // this calls <[T]>::len(&self)
}
```
Since #103735 already went 2 times through T-lang, where they T-lang ended-up asking for a more restricted version (which is what this PR does), I would prefer this PR to be reviewed first before re-nominating it for T-lang.
----
Compared to the PR it is as based on, this PR adds 3 restrictions on the outer most expression, which must either be:
1. A deref followed by any non-deref place projection (that intermediate deref will typically be auto-inserted)
2. A method call annotated with `#[rustc_no_implicit_refs]`.
3. A deref followed by a `addr_of!` or `addr_of_mut!`. See bottom of post for details.
There are several points that are not 100% clear to me when implementing the modifications:
- ~~"4. Any number of automatically inserted deref/derefmut calls." I as never able to trigger this. Am I missing something?~~ Fixed
- Are "index" and "field" enough?
----
cc `@JakobDegen` `@WaffleLapkin`
r? `@RalfJung`
try-job: dist-various-1
try-job: dist-various-2
Rollup of 8 pull requests
Successful merges:
- #137653 (Deprecate the unstable `concat_idents!`)
- #138957 (Update the index of Option to make the summary more comprehensive)
- #140006 (ensure compiler existance of tools on the dist step)
- #140143 (Move `sys::pal::os::Env` into `sys::env`)
- #140202 (Make #![feature(let_chains)] bootstrap conditional in compiler/)
- #140236 (norm nested aliases before evaluating the parent goal)
- #140257 (Some drive-by housecleaning in `rustc_borrowck`)
- #140278 (Don't use item name to look up associated item from trait item)
r? `@ghost`
`@rustbot` modify labels: rollup
Move `sys::pal::os::Env` into `sys::env`
Although `Env` (as `Vars`), `Args`, path functions, and OS constants are publicly exposed via `std::env`, their implementations are each self-contained. Keep them separate in `std::sys` and make a new module, `sys::env`, for `Env`.
Also fix `unsafe_op_in_unsafe_fn` for Unix and update the `!DynSend` and `!DynSync` impls which had grown out of sync with the platforms (see #48005 for discussion on that).
r? joboet
Tracked in #117276.
Work around cygwin issue on condvar timeout
This workaround *just works*... Actually I don't quite understand why does it work in such way. With a simple test on Cygwin, it seems that the maximum value of `tv_sec` could be 12899331056917, while the maximum value of `tv_nsec` should be a value floating around 464600000. A larger `timespec` could block the syscall forever.
r? `@joboet`
Although `Env` (as `Vars`), `Args`, path functions, and OS constants are
publicly exposed via `std::env`, their implementations are each
self-contained. Keep them separate in `std::sys` and make a new module,
`sys::env`, for `Env`.
Clarify why SGX code specifies linkage/symbol names for certain statics
Specifying linkage/symbol name is solely to ensure a single instance between the `std` crate and its unit tests.
Also update the symbol names as items have moved around a bit. The actual name isn't that important, it just needs to be unique. But for debugging it can be useful for it to point to the right place.
Hermit: Unify `std::env::args` with Unix
The only differences between these implementations of `std::env::args` are that Unix uses relaxed ordering, but Hermit uses acquire/release, and Unix truncates `argv` at the first null pointer, but Hermit doesn't. Since Hermit aims for Unix compatibility, unify it with Unix.
The atomic orderings were established in https://github.com/rust-lang/rust/pull/74006 (cc `@euclio)` for Unix and https://github.com/rust-lang/rust/pull/100579 (cc `@joboet)` for Hermit and, before those, they used mutexes and non-atomic statics. I think the difference in orderings is simply from them being changed at different times. The commented explanation for using acquire/release for Hermit is “to broadcast writes by the OS”. I'm not experienced enough with atomics to accurately judge, but I think acquire/release is stronger than needed. Either way, they should match.
Truncating at the first null pointer seems desirable, though I don't know whether it is necessary in practice on Hermit.
cc `@mkroening` `@stlankes` for Hermit
Move `pal::env` to `std::sys::env_consts`
Combine the `std::env::consts` platform implementations as a single file. Use the Unix file as the base, since it has 28 entries, and fold the 8 singleton platforms into it. The Unix file was roughly grouped into Linux, Apple, BSD, and everything else, roughly in alphabetical order. Alphabetically order them to make it easier to maintain and discard the Unix-specific groups to generalize it to all platforms.
I'd prefer to have no fallback implementation, as I consider it a bug; however TEEOS, Trusty, and Xous have no definitions here. Since they otherwise have `pal` abstractions, that indicates that there are several platforms without `pal` abstractions which are also missing here. To support unsupported, create a little macro to handle the fallback case and not introduce ordering between the `cfg`s like `cfg_if!`.
I've named the module `std::sys::env_consts`, because they are used in `std::env::consts` and I intend to use the name `std::sys::env` for the combination of `Args` and `Vars`.
cc `@joboet` `@ChrisDenton`
Tracked in #117276.
They were roughly grouped into Linux, Apple, BSD, and everything else,
roughly in alphabetical order. Alphabetically order them to make it
easier to maintain and discard the Unix-specific groups to generalize it
to all platforms.
Fix: Map EOPNOTSUPP to ErrorKind::Unsupported on Unix
This change maps the EOPNOTSUPP errno value (95) to std::io::ErrorKind::Unsupported in the decode_error_kind function for Unix platforms. Previously, it was incorrectly mapped to ErrorKind::Uncategorized.
Fixes#139803
std/thread: Use default stack size from menuconfig for NuttX
* Update comments to clarify the usage of zero as an indication for default stack size configuration
* Adjust conditional compilation to reflect the changes in stack size handling for the NuttX platform
This change improves clarity and consistency in stack size configuration across platforms.
std: sys: process: uefi: Use NULL stdin by default
According to the docs in `Command::output`:
> By default, stdout and stderr are captured (and used to provide the
resulting output). Stdin is not inherited from the parent and any attempt by the child process to read from the stdin stream will result in the stream immediately closing.
This was being violated by UEFI which was inheriting stdin by default.
While the docs don't explicitly state that the default should be NULL, the behaviour seems like reading from NULL.
UEFI however, has a bit of a problem. The `EFI_SIMPLE_TEXT_INPUT_PROTOCOL` only provides support for reading 1 key press. This means that you either get an error, or it is assumed that the keypress was read successfully. So there is no way to have a successful read of length 0. Currently, I am returning UNSUPPORTED error when trying to read from NULL stdin. On linux however, you will get a read of length 0 for Null stdin.
One possible way to get around this is to translate one of the UEFI errors to a read 0 (Maybe unsupported?). It is also possible to have a non-standard error code, but well, not sure if we go that route.
Alternatively, if meaning of Stdio::Null is platform dependent, it should be fine to keep the current behaviour of returning an error.
cc ```@nicholasbishop``` ```@dvdhrm```
This change maps the EOPNOTSUPP errno value (95) to std::io::ErrorKind::Unsupported in the decode_error_kind function for Unix platforms. Previously, it was incorrectly mapped to ErrorKind::Uncategorized.
Fixes#139803
Also update the symbol names as items have moved around a bit. The actual
name isn't that important, it just needs to be unique. But for debugging
it can be useful for it to point to the right place.
The only differences between these implementations are that Unix uses
relaxed ordering, but Hermit uses acquire/release, and Unix truncates
`argv` at the first null pointer, but Hermit doesn't. Since Hermit aims
for Unix compatibility, unify it with Unix.
Stdio::MakePipe is not supported.
For Stdio::Null, return UNSUPPORTED. This is treated as read(0).
Additionally, have infinte loop on the notify function to prevent
wait_for_key from returning.
Signed-off-by: Ayush Singh <ayush@beagleboard.org>
According to the docs in `Command::output`:
> By default, stdout and stderr are captured (and used to provide the
resulting output). Stdin is not inherited from the parent and any attempt
by the child process to read from the stdin stream will result in the
stream immediately closing.
This was being violated by UEFI which was inheriting stdin by default.
While the docs don't explicitly state that the default should be NULL,
the behaviour seems like reading from NULL.
UEFI however, has a bit of a problem. The `EFI_SIMPLE_TEXT_INPUT_PROTOCOL`
only provides support for reading 1 key press. This means that you
either get an error, or it is assumed that the keypress was read
successfully. So there is no way to have a successful read of length 0.
Currently, I am returning UNSUPPORTED error when trying to read from
NULL stdin. On linux however, you will get a read of length 0 for Null
stdin.
One possible way to get around this is to translate one of the UEFI
errors to a read 0 (Maybe unsupported?). It is also possible to have a
non-standard error code, but well, not sure if we go that route.
Alternatively, if meaning of Stdio::Null is platform dependent, it
should be fine to keep the current behaviour of returning an error.
Signed-off-by: Ayush Singh <ayush@beagleboard.org>
* Update comments to clarify the usage of zero as an indication for default stack size configuration
* Adjust conditional compilation to reflect the changes in stack size handling for the NuttX platform
This change improves clarity and consistency in stack size configuration across platforms.
Signed-off-by: Huang Qi <huangqi3@xiaomi.com>
Use `with_native_path` for Windows
Ideally, each platform should use their own native path type internally. This will, for example, allow passing a UTF-16 string directly to `std::fs::File::open` and therefore avoid the need for allocating a new null-terminated wide string. However, doing that for every function and platform all at once makes for a large PR that is way too prone to breaking. So this just does some of the Windows parts.
As with the previous Unix PR (#138832) this is intended to be merely a refactoring so I've avoided anything that may require more substantial changes.
std: Fix build for NuttX targets
Fix std build for all NuttX targets. It is the single largest set of failures on <https://does-it-build.noratrieb.dev/>. Although, ESP-IDF also requires these same gates, there are other issues for those targets.
This can verified be running `x check library/std --target=` for all NuttX targets.
cc ``@no1wudi``
Rollup of 6 pull requests
Successful merges:
- #139107 (std: make `cmath` functions safe)
- #139607 (Add regression test for #127424)
- #139691 (Document that `opt-dist` requires metrics to be enabled)
- #139707 (Fix comment in bootstrap)
- #139708 (Fix name of field in doc comment)
- #139709 (bootstrap: fix typo in doc string)
r? `@ghost`
`@rustbot` modify labels: rollup
std: make `cmath` functions safe
The floating point intrinsics are more difficult, I'll probably wait until #119899 has merged before making them safe as well.
Update windows-bindgen to 0.61.0
This updates the automatically generate Windows API bindings. Not much changed this time:
- There's now `Default` implementations for many types, which is convenient. It does however conflict with one place where we implemented a non-zeroed default (to set the length field). But that's no big problem.
- The `--no-core` flag has been renamed to `--no-deps` to more accurately reflect its meaning (i.e. generate all necessary code without requiring additional dependencies).
- The `--link` flag allows us to set the location of the `link!` macro. Currently we use our workspace's `windows_targets` crate but we could move it into library/std using `--link`. However, this would need to be co-ordinated with the `backtrace` crate (which is a separate crate but included in std using `#[path]`). So I've left that for another time.
Try not to use verbatim paths in `Command::current_dir`
If possible, we should try not to use verbatim paths in `Command::current_dir`. It might work but it might also break code in the subprocess that assume the current directory isn't verbatim (including Windows APIs). cc ``@ehuss``
Side note: we now have a lot of ad-hoc fixes like this spread about the place. It'd be good to make a proper `WindowsPath` type that handles all this in one place. But that's a bigger job for another PR.
If possible, we should try not to use verbatim paths in Command::current_dir. It might work but it might also break code (including some Windows APIs) that assume the current directory isn't verbatim.
