Wrap private padding fields with `Padding` newtype #4609

mbyx · 2025-08-02T13:16:45Z

Description

Changes the types of fields with name __pad, _pad, __padding, _padding, __rsvd, __unused to use a new type wrapper around MaybeUninit as per #1453

Only modifies private fields so that the changes can be backported.

Sources

Checklist

Relevant tests in libc-test/semver have been updated
No placeholder or unstable values like *LAST or *MAX are
included (see #3131)
Tested locally (cd libc-test && cargo test --target mytarget);
especially relevant for platforms that may not be checked in CI

tgross35 · 2025-08-09T23:05:22Z

src/types.rs

@@ -0,0 +1,12 @@
+use core::mem::MaybeUninit;
+
+/// A transparent wrapper over `MaybeUninit<T>` to represent uninitialized padding.


Suggested change

/// A transparent wrapper over `MaybeUninit<T>` to represent uninitialized padding.

/// A transparent wrapper over `MaybeUninit<T>` to represent uninitialized padding

/// while providing `Default`.

To clarify why we don't just use Default

tgross35 · 2025-08-09T23:06:03Z

@rustbot blocked
on ctest2

tgross35 · 2025-08-10T06:12:12Z

@mbyx could you put up a separate PR that only adds the Padding type? I'd like to start directing new PRs to use this.

tgross35 · 2025-09-23T02:52:07Z

Would you mind rebasing this? CI will still fail because of the eq/ord traits, but with #4711 the Debug messages should be resolved.

tgross35 · 2025-11-02T21:30:52Z

I think most of the errors should be resolved with 141aea1. If you would be able to rebase again, I think this might be able to merge with a few smaller deltas.

@rustbot author

rustbot · 2025-11-02T21:30:56Z

Reminder, once the PR becomes ready for a review, use @rustbot ready.

mbyx · 2025-11-04T14:25:23Z

Rebased. Seems to error because Padding doesn't implement Hash, Eq, and PartialEq, because the s macro tries to implement it for the struct. Would implementing those for Padding be well defined?

Also it seems that the style check fails at the first formatting error? My editor seems to have a different style for imports than libc so I may have caused breakage in more files than just the one that the style check shows.

tgross35 · 2025-11-04T18:59:06Z

Rebased. Seems to error because Padding doesn't implement Hash, Eq, and PartialEq, because the s macro tries to implement it for the struct. Would implementing those for Padding be well defined?

That is what I was thinking, would you be able to add a commit that adds these implementations if extra_traits is enabled? Hash should do nothing and PartialEq should return true, so they have no effect on the results.

Also it seems that the style check fails at the first formatting error? My editor seems to have a different style for imports than libc so I may have caused breakage in more files than just the one that the style check shows.

That config file changed somewhat recently, if you rebased without re-saving the files in your editor or running locally then it might not have picked that up.

mbyx · 2025-11-05T16:32:08Z

That seems to fix most errors, except now we have a bit of an issue. One of the constants in unix/bsd/netbsdlike/netbsd/mod.rs, PTHREAD_MUTEX_INITIALIZER needs to initialize the padding field, but Default::default is not a const function. We can't make a seperate const function either because MaybeUninit::zeroed was made a const function in 1.75 and MSRV is 1.63.

For now I'll try to just not make that type using padding and see if I can get something mergeable.

EDIT: Seems to be many more constants like this. Even if those are ignored for now a good number should still be wrapped.

rustbot · 2025-11-05T16:56:05Z

Some changes occurred in OpenBSD module

cc @semarie

Some changes occurred in the Android module

cc @maurer

Some changes occurred in solarish module

cc @jclulow, @pfmooney

tgross35

All the switches to Padding LGTM, just a few small things about the traits

tgross35 · 2025-11-05T22:53:16Z

src/types.rs

Could you split the changes to this file to a separate commit? It's distinct from the rest of the mechanical changes in this PR.

tgross35 · 2025-11-05T22:54:36Z

src/types.rs

+#[cfg(feature = "extra_traits")]
+impl<T: Copy> Hash for Padding<T> {
+    fn hash<H: hash::Hasher>(&self, _state: &mut H) {}
+}
+
+#[cfg(feature = "extra_traits")]
+impl<T: Copy> PartialEq for Padding<T> {
+    fn eq(&self, _other: &Self) -> bool {
+        true
+    }
+}


These are user-facing, so add a doc comment to each of these impls saying their behavior (i.e. that they give the results they do to act like the fields don't exist)

tgross35 · 2025-11-05T22:57:31Z

You should rebase as well, musl downloads have been flaky so I just switched us to a mirror

rustbot · 2025-11-06T13:30:50Z

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

rustbot added O-android O-arm O-dragonfly O-gnu O-linux O-linux-like O-macos O-mips O-musl O-newlib O-powerpc O-redox O-riscv O-solarish O-sparc O-unix O-wasi O-x86 labels Aug 2, 2025

tgross35 added the stable-nominated This PR should be considered for cherry-pick to libc's stable release branch label Aug 9, 2025

tgross35 reviewed Aug 9, 2025

View reviewed changes

rustbot added the S-blocked label Aug 9, 2025

mbyx force-pushed the libc-wrap-padding branch 4 times, most recently from b1e6991 to e2a5397 Compare August 14, 2025 08:08

tgross35 changed the title ~~libc: wrap padding fields with Padding newtype.~~ Wrap private padding fields with Padding newtype Sep 23, 2025

mbyx force-pushed the libc-wrap-padding branch from e2a5397 to 271d31b Compare September 25, 2025 17:39

rustbot added S-waiting-on-author and removed S-blocked labels Nov 2, 2025

mbyx force-pushed the libc-wrap-padding branch 4 times, most recently from 77679b6 to 3f1c30f Compare November 4, 2025 14:18

mbyx force-pushed the libc-wrap-padding branch 4 times, most recently from f7e75a4 to 33b5c86 Compare November 5, 2025 16:27

mbyx force-pushed the libc-wrap-padding branch 3 times, most recently from a66f7c2 to 6afc113 Compare November 5, 2025 16:43

mbyx marked this pull request as ready for review November 5, 2025 16:56

rustbot added the S-waiting-on-review label Nov 5, 2025

tgross35 requested changes Nov 5, 2025

View reviewed changes

rustbot removed the S-waiting-on-review label Nov 5, 2025

mbyx force-pushed the libc-wrap-padding branch from 6afc113 to 607e39d Compare November 6, 2025 13:30

mbyx added 2 commits November 6, 2025 18:32

libc: add PartialEq, Eq, and Hash trait implementations for Padding.

cd7f8a6

libc: wrap padding fields with Padding newtype.

38d07d4

mbyx force-pushed the libc-wrap-padding branch from 607e39d to 38d07d4 Compare November 6, 2025 13:32

mbyx requested a review from tgross35 November 6, 2025 13:33

		@@ -0,0 +1,12 @@
		use core::mem::MaybeUninit;

		/// A transparent wrapper over `MaybeUninit<T>` to represent uninitialized padding.

	/// A transparent wrapper over `MaybeUninit<T>` to represent uninitialized padding.
	/// A transparent wrapper over `MaybeUninit<T>` to represent uninitialized padding
	/// while providing `Default`.

Wrap private padding fields with Padding newtype #4609

Are you sure you want to change the base?

Wrap private padding fields with Padding newtype #4609

Uh oh!

Conversation

mbyx commented Aug 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Sources

Checklist

Uh oh!

tgross35 Aug 9, 2025

Choose a reason for hiding this comment

Uh oh!

tgross35 commented Aug 9, 2025

Uh oh!

tgross35 commented Aug 10, 2025

Uh oh!

tgross35 commented Sep 23, 2025

Uh oh!

tgross35 commented Nov 2, 2025

Uh oh!

rustbot commented Nov 2, 2025

Uh oh!

mbyx commented Nov 4, 2025

Uh oh!

tgross35 commented Nov 4, 2025

Uh oh!

mbyx commented Nov 5, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Nov 5, 2025

Uh oh!

tgross35 left a comment

Choose a reason for hiding this comment

Uh oh!

tgross35 Nov 5, 2025

Choose a reason for hiding this comment

Uh oh!

tgross35 Nov 5, 2025

Choose a reason for hiding this comment

Uh oh!

tgross35 commented Nov 5, 2025

Uh oh!

rustbot commented Nov 6, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Wrap private padding fields with `Padding` newtype #4609

Wrap private padding fields with `Padding` newtype #4609

mbyx commented Aug 2, 2025 •

edited

Loading

mbyx commented Nov 5, 2025 •

edited

Loading