ext/ucred: Support PID in peer creds on macOS
This is a follow-up to https://github.com/rust-lang/rust/pull/75148 (RFC: https://github.com/rust-lang/rust/issues/42839).
The original PR used `getpeereid` on macOS and the BSDs, since they don't (generally) support the `SO_PEERCRED` mechanism that Linux supplies.
This PR splits the macOS/iOS implementation of `peer_cred()` from that of the BSDs, since macOS supplies the `LOCAL_PEERPID` sockopt as a source of the missing PID. It also adds a `cfg`-gated tests that ensures that platforms with support for PIDs in `UCred` have the expected data.
Use is_write_vectored to optimize the write_vectored implementation for BufWriter
In case when the underlying writer does not have an efficient implementation `write_vectored`, the present implementation of
`write_vectored` for `BufWriter` may still forward vectored writes directly to the writer depending on the total length of the data. This misses the advantage of buffering, as the actually written slice may be small.
Provide an alternative code path for the non-vectored case, where the slices passed to `BufWriter` are coalesced in the buffer before being flushed to the underlying writer with plain `write` calls. The buffer is only bypassed if an individual slice's length is at least as large as the buffer.
Remove a FIXME comment referring to #72919 as the issue has been closed with an explanation provided.
Fix SGX CI, take 3
Broken in #79038
r? `@Mark-Simulacrum`
I actually ran `./x.py test --target x86_64-fortanix-unknown-sgx` on the commit before submitting it this time.
Fix incorrect io::Take's limit resulting from io::copy specialization
The specialization introduced in #75272 fails to update `io::Take` wrappers after performing the copy syscalls which bypass those wrappers. The buffer flushing before the copy does update them correctly, but the bytes copied after the initial flush weren't subtracted.
The fix is to subtract the bytes copied from each `Take` in the chain of wrappers, even when an error occurs during the syscall loop. To do so the `CopyResult` enum now has to carry the bytes copied so far in the error case.
Provide IntoInnerError::into_parts
Hi. This is an updated version of the IntoInnerError bits of my previous portmanteau MR #78689. Thanks to `@jyn514` and `@m-ou-se` for helpful comments there.
I have made this insta-stable since it seems like it will probably be uncontroversial, but that is definitely something that someone from the libs API team should be aware of and explicitly consider.
I included a tangentially-related commit providing documentation of the buffer full behaviiour of `&mut [u8] as Write`; the behaviour I am documenting is relied on by the doctest for `into_parts`.
In particular, IntoIneerError only currently provides .error() which
returns a reference, not an owned value. This is not helpful and
means that a caller of BufWriter::into_inner cannot acquire an owned
io::Error which seems quite wrong.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
disable a ptr equality test on Miri
This test relies on deduplication of constants. I do not think that this is a *guarantee* that Rust currently makes, and indeed Miri does not deduplicate constants the same way that rustc does, leading to different behavior in this test.
For now, I propose we simply disable this test in Miri.
Use more std:: instead of core:: in docs for consistency
``@rustbot`` label T-doc
Some cleanup work to use `std::` instead of `core::` in docs as much as possible. This helps with terminology and consistency, especially for newcomers from other languages that have often heard of `std` to describe the standard library but not of `core`.
Edit: I also added more intra doc links when I saw the opportunity.
unix: Extend UnixStream and UnixDatagram to send and receive file descriptors
Add the functions `recv_vectored_fds` and `send_vectored_fds` to `UnixDatagram` and `UnixStream`. With this functions `UnixDatagram` and `UnixStream` can send and receive file descriptors, by using `recvmsg` and `sendmsg` system call.
std::io: Use sendfile for UnixStream
`UnixStream` was forgotten in #75272 .
Benchmark yields the following results.
Before:
`running 1 test
test sys::unix::kernel_copy::tests::bench_file_to_uds_copy ... bench: 54,399 ns/iter (+/- 6,817) = 2409 MB/s`
After:
`running 1 test
test sys::unix::kernel_copy::tests::bench_file_to_uds_copy ... bench: 18,627 ns/iter (+/- 6,007) = 7036 MB/s`
Avoid panic_bounds_check in fmt::write.
Writing any fmt::Arguments would trigger the inclusion of usize formatting and padding code in the resulting binary, because indexing used in fmt::write would generate code using panic_bounds_check, which prints the index and length.
These bounds checks are not necessary, as fmt::Arguments never contains any out-of-bounds indexes.
This change replaces them with unsafe get_unchecked, to reduce the amount of generated code, which is especially important for embedded targets.
---
Demonstration of the size of and the symbols in a 'hello world' no_std binary:
<details>
<summary>Source code</summary>
```rust
#![feature(lang_items)]
#![feature(start)]
#![no_std]
use core::fmt;
use core::fmt::Write;
#[link(name = "c")]
extern "C" {
#[allow(improper_ctypes)]
fn write(fd: i32, s: &str) -> isize;
fn exit(code: i32) -> !;
}
struct Stdout;
impl fmt::Write for Stdout {
fn write_str(&mut self, s: &str) -> fmt::Result {
unsafe { write(1, s) };
Ok(())
}
}
#[start]
fn main(_argc: isize, _argv: *const *const u8) -> isize {
let _ = writeln!(Stdout, "Hello World");
0
}
#[lang = "eh_personality"]
fn eh_personality() {}
#[panic_handler]
fn panic(_: &core::panic::PanicInfo) -> ! {
unsafe { exit(1) };
}
```
</details>
Before:
```
text data bss dec hex filename
6059 736 8 6803 1a93 before
```
```
0000000000001e00 T <T as core::any::Any>::type_id
0000000000003dd0 D core::fmt::num::DEC_DIGITS_LUT
0000000000001ce0 T core::fmt::num:👿:<impl core::fmt::Display for u64>::fmt
0000000000001ce0 T core::fmt::num:👿:<impl core::fmt::Display for usize>::fmt
0000000000001370 T core::fmt::write
0000000000001b30 t core::fmt::Formatter::pad_integral::write_prefix
0000000000001660 T core::fmt::Formatter::pad_integral
0000000000001350 T core::ops::function::FnOnce::call_once
0000000000001b80 t core::ptr::drop_in_place
0000000000001120 t core::ptr::drop_in_place
0000000000001c50 t core::iter::adapters::zip::Zip<A,B>::new
0000000000001c90 t core::iter::adapters::zip::Zip<A,B>::new
0000000000001b90 T core::panicking::panic_bounds_check
0000000000001c10 T core::panicking::panic_fmt
0000000000001130 t <&mut W as core::fmt::Write>::write_char
0000000000001200 t <&mut W as core::fmt::Write>::write_fmt
0000000000001250 t <&mut W as core::fmt::Write>::write_str
```
After:
```
text data bss dec hex filename
3068 600 8 3676 e5c after
```
```
0000000000001360 T core::fmt::write
0000000000001340 T core::ops::function::FnOnce::call_once
0000000000001120 t core::ptr::drop_in_place
0000000000001620 t core::iter::adapters::zip::Zip<A,B>::new
0000000000001660 t core::iter::adapters::zip::Zip<A,B>::new
0000000000001130 t <&mut W as core::fmt::Write>::write_char
0000000000001200 t <&mut W as core::fmt::Write>::write_fmt
0000000000001250 t <&mut W as core::fmt::Write>::write_str
```
Update tests to remove old numeric constants
Part of #68490.
Care has been taken to leave the old consts where appropriate, for testing backcompat regressions, module shadowing, etc. The intrinsics docs were accidentally referring to some methods on f64 as std::f64, which I changed due to being contrary with how we normally disambiguate the shadow module from the primitive. In one other place I changed std::u8 to std::ops since it was just testing path handling in macros.
For places which have legitimate uses of the old consts, deprecated attributes have been optimistically inserted. Although currently unnecessary, they exist to emphasize to any future deprecation effort the necessity of these specific symbols and prevent them from being accidentally removed.
