Add vectored positioned I/O on Unix

[a1phyr]

Add methods for vectored I/O with an offset on File for unix under #![feature(unix_file_vectored_at)].

The new methods are wrappers around preadv and pwritev.

Tracking issue: #89517

[rust-highfive]

r? @Mark-Simulacrum

(rust-highfive has picked a reviewer for you, use r? to override)

[jhwgh1968]

Drive by random comment.

On line 135:

bufs.as_ptr() as *const libc::iovec

Shouldn't that be *mut libc::iovec since it is passed as &mut?

[a1phyr]

Well, I copied read_vectored for this.

Anyway, readv takes a const struct iovec*, so this is the good type. However, according to Rust rules we cannot modify something behind a shared reference, so I think we should get a *mut pointer and cast it to *const.

[Mark-Simulacrum]

Seems reasonable enough. r? @joshtriplett for T-libs-api review as new unstable API

[joshtriplett]

This seems reasonable to me.

I think we may want ReadBuf versions of all of these in the future, but that can be a separate PR.

@bors r+

[bors]

📌 Commit 2265bef has been approved by joshtriplett

[matthiaskrgr]

Build failed in a rollup ( #89763 (comment) )

[RUSTC-TIMING] core test:false 15.649
[RUSTC-TIMING] addr2line test:false 0.427
[RUSTC-TIMING] gimli test:false 4.080
[RUSTC-TIMING] object test:false 4.120
error[E0425]: cannot find function `preadv` in crate `libc`
   --> library/std/src/sys/unix/fd.rs:133:19
133 |             libc::preadv(
    |                   ^^^^^^ help: a function with a similar name exists: `pread`
    |
   ::: /cargo/registry/src/github.com-1ecc6299db9ec823/libc-0.2.103/src/unix/mod.rs:933:5
   ::: /cargo/registry/src/github.com-1ecc6299db9ec823/libc-0.2.103/src/unix/mod.rs:933:5
    |
933 |     pub fn pread(fd: ::c_int, buf: *mut ::c_void, count: ::size_t, offset: off_t) -> ::ssize_t;
    |     ------------------------------------------------------------------------------------------- similarly named function `pread` defined here

error[E0425]: cannot find function `pwritev` in crate `libc`
   --> library/std/src/sys/unix/fd.rs:213:19
    |
213 |             libc::pwritev(
    |                   ^^^^^^^ help: a function with a similar name exists: `pwrite`
   ::: /cargo/registry/src/github.com-1ecc6299db9ec823/libc-0.2.103/src/unix/mod.rs:938:5
    |
    |
938 |     pub fn pwrite(fd: ::c_int, buf: *const ::c_void, count: ::size_t, offset: off_t) -> ::ssize_t;
    |     ---------------------------------------------------------------------------------------------- similarly named function `pwrite` defined here
For more information about this error, try `rustc --explain E0425`.
[RUSTC-TIMING] std test:false 1.946
error: could not compile `std` due to 2 previous errors
Build completed unsuccessfully in 0:15:29

@bors r-

[a1phyr]

I figured out that preadv/pwritev are not POSIX, even if they are defined by most UNIX platforms.

Now preadv/pwritev are used it on all platform for which libc defines them, with a fallback implementation for other platforms.

[bors]

☔ The latest upstream changes (presumably #81156) made this pull request unmergeable. Please resolve the merge conflicts.

[a1phyr]

Rebased

[a1phyr]

Wild guess: preadv was added to bionic in 2015 (cf https://android.googlesource.com/platform/bionic/+/6f4594d). The tests may run with on older version.

To fix this, we could either disable preadv on android or use weak linkage.

[workingjubilee]

Our Android minimum is increasing to Android KitKat. That's lower than Marshmallow, the Oct 2015 release. I need more coffee before I remember the details of how weak symbol linkage works, but what we do here has to work on KitKat and Lollipop, so go with whichever feels right there.

[a1phyr]

I edited the PR to:

• Use weak linkage on Android (with fallback to syscall on 64 bits systems)
• Use &self in the public API
• Add basic tests to make sure everything works as intended enverywhere

@rustbot label -S-waiting-on-author +S-waiting-on-review

[workingjubilee]

Bad news, I looked it up and it seems this was only introduced relatively late in the game by macOS (and iOS). It seems even #104385 won't raise the minimum supported versions high enough to guarantee pwritev on those platforms.

[workingjubilee]

Huzzah, let's see if this goes or if it tells us to do more.

@bors r+

[workingjubilee]

Not again?
@bors r+

[bors]

📌 Commit 92f35b3 has been approved by workingjubilee

It is now in the queue for this repository.

[bors]

⌛ Testing commit 92f35b3 with merge 0fbfc3e...

[bors]

☀️ Test successful - checks-actions
Approved by: workingjubilee
Pushing 0fbfc3e to master...

[rust-timer]

Finished benchmarking commit (0fbfc3e): comparison URL.

Overall result: ❌ regressions - ACTION NEEDED

Next Steps: If you can justify the regressions found in this perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please open an issue or create a new PR that fixes the regressions, add a comment linking to the newly created issue or PR, and then add the perf-regression-triaged label to this PR.

@rustbot label: +perf-regression
cc @rust-lang/wg-compiler-performance

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	0.2%	[0.2%, 0.3%]	4
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.2%	[0.2%, 0.3%]	4

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	1.3%	[1.3%, 1.3%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-5.7%	[-5.7%, -5.7%]	1
Improvements ✅ (secondary)	-2.9%	[-3.0%, -2.7%]	2
All ❌✅ (primary)	-2.2%	[-5.7%, 1.3%]	2

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	0.8%	[0.6%, 1.0%]	8
Regressions ❌ (secondary)	0.8%	[0.8%, 0.8%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.8%	[0.6%, 1.0%]	8

[rylev]

Small regression in Diesel which has started to show some bimodality. I think it's safe to call this triaged.

@rustbot label +perf-regression-triaged