Unify id-based thread parking implementations
Multiple platforms currently use thread-id-based parking implementations (NetBSD and SGX[^1]). Even though the strategy does not differ, these are duplicated for each platform, as the id is encoded into an atomic thread variable in different ways for each platform.
Since `park` is only called by one thread, it is possible to move the thread id into a separate field. By ensuring that the field is only written to once, before any other threads access it, these accesses can be unsynchronized, removing any restrictions on the size and niches of the thread id.
This PR also renames the internal `thread_parker` modules to `thread_parking`, as that name now better reflects their contents. I hope this does not add too much reviewing noise.
r? `@m-ou-se`
`@rustbot` label +T-libs
[^1]: SOLID supports this as well, I will switch it over in a follow-up PR.
On Linux, use mmap64, open64, openat64, and sendfile64 in place of their
non-LFS counterparts.
This is relevant to #94173.
With these changes (together with rust-lang/backtrace-rs#501), the
simple binaries I produce with rustc seem to have no non-LFS functions,
so maybe #94173 is fixed. But I can't be sure if I've missed something
and maybe some non-LFS functions could sneak in somehow.
Don't link to `libresolv` in libstd on Darwin
Currently we link `libresolv` into every Rust program on apple targets despite never using it (as of https://github.com/rust-lang/rust/pull/44965). I had thought we needed this for `getaddrinfo` or something, but we do not / cannot safely use it.
I'd like to fix this for `libiconv` too (the other library we pull in. that's harder since it's coming in through `libc`, which is https://github.com/rust-lang/libc/pull/2944)).
---
This may warrant release notes. I'm not sure but I've added the flag regardless -- It's a change to the list of dylibs every Rust program pulls in, so it's worth mentioning.
It's pretty unlikely anybody was relying on this being pulled in, and `std` does not guarantee that it will link (and thus transitively provide access to) any particular system library -- anybody relying on that behavior would already be broken when dynamically linking std. That is, there's an outside chance something will fail to link on macOS and iOS because it was accidentally relying on our unnecessary dependency.
(If that *does* happen, that project could be easily fixed by linking libresolv explicitly on those platforms, probably via `#[link(name = "resolv")] extern {}`,` -Crustc-link-lib=resolv`, `println!("cargo:rustc-link-lib=resolv")`, or one of several places in `.config/cargo.toml`)
---
I'm also going to preemptively add the nomination for discussing this in the libs meeting. Basically: Do we care about programs that assume we will bring libraries in that we do not use. `libresolv` and `libiconv` on macOS/iOS are in this camp (`libresolv` because we used to use it, and `libiconv` because the `libc` crate was unintentionally(?) pulling it in to every Rust program).
I'd like to remove them both, but this may cause link issues programs that are relying on `std` to depend on them transitively. (Relying on std for this does not work in all build configurations, so this seems very fragile, and like a use case we should not support).
More generally, IMO we should not guarantee the specific set of system-provided libraries we use (beyond what is implied by an OS version requirement), which means we'd be free to remove this cruft.
Previously, the signal mask is always reset when a child process is
started. This breaks tools like `nohup` which expect `SIGHUP` to be
blocked.
With this change, the default behavior changes to inherit the signal mask.
This also changes the signal disposition for `SIGPIPE` to only be
changed if the `#[unix_sigpipe]` attribute isn't set.
This makes it possible to instruct libstd to never touch the signal
handler for `SIGPIPE`, which makes programs pipeable by default (e.g.
with `./your-program | head -n 1`) without `ErrorKind::BrokenPipe`
errors.
Use `fcntl(fd, F_GETFD)` to detect if standard streams are open
In the previous implementation, if the standard streams were open,
but the RLIMIT_NOFILE value was below three, the poll would fail
with EINVAL:
> ERRORS: EINVAL The nfds value exceeds the RLIMIT_NOFILE value.
Switch to the existing fcntl based implementation to avoid the issue.
Fixes#96621.
Normally, `Thread::spawn` takes care of setting the thread's name, if
one was provided, but since the main thread wasn't created by calling
`Thread::spawn`, we need to call that function in `std::rt::init`.
This is mainly useful for system tools like debuggers and profilers
which might show the thread name to a user. Prior to these changes, gdb
and WinDbg would show all thread names except the main thread's name to
a user. I've validated that this patch resolves the issue for both
debuggers.
In the previous implementation, if the standard streams were open,
but the RLIMIT_NOFILE value was below three, the poll would fail
with EINVAL:
> ERRORS: EINVAL The nfds value exceeds the RLIMIT_NOFILE value.
Switch to the existing fcntl based implementation to avoid the issue.
Mutex and Condvar are being replaced by more efficient implementations, which need thread parking themselves (see #93740). Therefore use the pthread synchronization primitives directly. Also, avoid allocating because the Parker struct is being placed in an Arc anyways.
Move std::sys::{mutex, condvar, rwlock} to std::sys::locks.
This cleans up the the std::sys modules a bit by putting the locks in a single module called `locks` rather than spread over the three modules `mutex`, `condvar`, and `rwlock`. This makes it easier to organise lock implementations, which helps with https://github.com/rust-lang/rust/issues/93740.
L4Re provides limited POSIX support which includes support for
standard I/O streams, and a limited implementation of the standard file
handling API. However, because as a capability based OS it strives to
only make a local view available to each application, there are
currently no standardized special files like /dev/null that could serve
to sanitize closed standard FDs.
For now, skip any attempts to sanitize standard streams until a more
complete POSIX runtime is available.
Move `os_str_bytes` to `sys::unix`
Followup to #84967, with `OsStrExt` and `OsStringExt` moved out of `sys_common`, there is no reason anymore for `os_str_bytes` to live in `sys_common` and not in sys. This pr moves it to the location `sys::unix::os_str` and reuses the code on other platforms via `#[path]` (as is common in `sys`) instead of importing.
In the docs for intrinsics::abort():
* Strengthen the recommendation by to use process::abort instead.
* Document the fact that it (ab)uses an LLVM debug trap and what the
likely consequences are.
* State that the precise behaviour is unstable.
In the docs for process::abort():
* Promise that we have the same behaviour as C `abort()`.
* Document the likely consequences, including, specifically, the
consequences on Unix.
In the internal comment for unix::abort_internal:
* Refer to the public docs for the public API functions.
* Correct and expand the description of libc::abort. Specifically:
* Do not claim that abort() unregisters signal handlers. It doesn't;
it honours the SIGABRT handler.
* Discuss, extensively, the issue with abort() flushing stdio buffers.
* Describe the glibc behaviour in some detail.
Co-authored-by: Mark Wooding <mdw@distorted.org.uk>
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
More ErrorKinds for common errnos
From the commit message of the main commit here (as revised):
```
There are a number of IO error situations which it would be very
useful for Rust code to be able to recognise without having to resort
to OS-specific code. Taking some Unix examples, `ENOTEMPTY` and
`EXDEV` have obvious recovery strategies. Recently I was surprised to
discover that `ENOSPC` came out as `ErrorKind::Other`.
Since I am familiar with Unix I reviwed the list of errno values in
https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/errno.h.html
Here, I add those that most clearly seem to be needed.
`@CraftSpider` provided information about Windows, and references, which
I have tried to take into account.
This has to be insta-stable because we can't sensibly have a different
set of ErrorKinds depending on a std feature flag.
I have *not* added these to the mapping tables for any operating
systems other than Unix and Windows. I hope that it is OK to add them
now for Unix and Windows now, and maybe add them to other OS's mapping
tables as and when someone on that OS is able to consider the
situation.
I adopted the general principle that it was usually a bad idea to map
two distinct error values to the same Rust error code. I notice that
this principle is already violated in the case of `EACCES` and
`EPERM`, which both map to `PermissionDenied`. I think this was
probably a mistake but it would be quite hard to change now, so I
don't propose to do anything about that.
However, for Windows, there are sometimes different error codes for
identical situations. Eg there are WSA* versions of some error
codes as well as ERROR_* ones. Also Windows seems to have a great
many more erorr codes. I don't know precisely what best practice
would be for Windows.
```
<strike>
```
Errno values I wasn't sure about so *haven't* included:
EMFILE ENFILE ENOBUFS ENOLCK:
These are all fairly Unix-specific resource exhaustion situations.
In practice it seemed not very likely to me that anyone would want
to handle these differently to `Other`.
ENOMEM ERANGE EDOM EOVERFLOW
Normally these don't get exposed to the Rust callers I hope. They
don't tend to come out of filesystem APIs.
EILSEQ
Hopefully Rust libraries open files in binary mode and do the
converstion in Rust. So Rust code ought not to be exposed to
EILSEQ.
EIO
The range of things that could cause this is troublesome. I found
it difficult to describe. I do think it would be useful to add this
at some point, because EIO on a filesystem operation is much more
serious than most other errors.
ENETDOWN
I wasn't sure if this was useful or, indeed, if any modern systems
use it.
ENOEXEC
It is not clear to me how a Rust program could respond to this. It
seems rather niche.
EPROTO ENETRESET ENODATA ENOMSG ENOPROTOOPT ENOSR ENOSTR ETIME
ENOTRECOVERABLE EOWNERDEAD EBADMSG EPROTONOSUPPORT EPROTOTYPE EIDRM
These are network or STREAMS related errors which I have never in
my own Unix programming found the need to do anything with. I think
someone who understands these better should be the one to try to
find good Rust names and descriptions for them.
ENOTTY ENXIO ENODEV EOPNOTSUPP ESRCH EALREADY ECANCELED ECHILD
EINPROGRESS
These are very hard to get unless you're already doing something
very Unix-specific, in which case the raw_os_error interface is
probably more suitable than relying on the Rust ErrorKind mapping.
EFAULT EBADF
These would seem to be the result of application UB.
```
</strike>
<i>(omitted errnos are discussed below, especially in https://github.com/rust-lang/rust/pull/79965#issuecomment-810468334)
Redefine `ErrorKind::Other` and stop using it in std.
This implements the idea I shared yesterday in the libs meeting when we were discussing how to handle adding new `ErrorKind`s to the standard library: This redefines `Other` to be for *user defined errors only*, and changes all uses of `Other` in the standard library to a `#[doc(hidden)]` and permanently `#[unstable]` `ErrorKind` that users can not match on. This ensures that adding `ErrorKind`s at a later point in time is not a breaking change, since the user couldn't match on these errors anyway. This way, we use the `#[non_exhaustive]` property of the enum in a more effective way.
Open questions:
- How do we check this change doesn't cause too much breakage? Will a crate run help and be enough?
- How do we ensure we don't accidentally start using `Other` again in the standard library? We don't have a `pub(not crate)` or `#[deprecated(in this crate only)]`.
cc https://github.com/rust-lang/rust/pull/79965
cc `@rust-lang/libs` `@ijackson`
r? `@dtolnay`
Since android ndk version `r23-beta3`, `libgcc` has been replaced with
`libunwind`. This moves the linking of `libgcc`/`libunwind` into the
`unwind` crate where we check if the system compiler can find
`libunwind` and fall back to `libgcc` if needed.
