Add cfg(target_has_atomic_equal_alignment) and use it for Atomic::from_mut. #76965

m-ou-se · 2020-09-20T10:23:41Z

Fixes some platform-specific problems with #74532 by using the actual alignment of the types instead of hardcoding a few target_archs.

r? @RalfJung

RalfJung · 2020-09-20T10:42:16Z

This is code I never touched and I also don't know what it takes to make sure this new cfg flag is unstable and who would need to approve such an addition. Usually I'd think it requires an RFC, but maybe that is wrong.

But you'll need to find another reviewer, sorry. For now, assigning to the reviewer of the PR this follows up on.
r? @KodrAus

compiler/rustc_feature/src/builtin_attrs.rs

m-ou-se · 2020-09-20T10:57:48Z

~~(If this takes too much time right now, feel free to merge #76967 instead.)~~

Edit: That PR (temporarily reverting Atomic::from_mut) is now merged. I updated this PR to revert that revert.

m-ou-se · 2020-09-20T11:18:16Z

Just like the other target_has_atomic* cfg flags, this flag is internal and only used to communicate from rustc to libcore which functions it should make available. Consumers of libcore should probably use the generic cfg(accessible) (#64797) when that stabilizes.

KodrAus · 2020-09-20T11:31:12Z

Thanks @m-ou-se! It looks like we ended up needing a precise cfg for this after all.

So it looks like we added target_has_atomic_load_store in #65214 without going through a fresh RFC (it was a refactoring of existing functionality though).

This seems like a reasonable addition to the current target_has_atomic feature set to me to gate on platforms where the alignment of an atomic type is the same as its non atomic counterpart.

Having said that, I think @Amanieu is probably the best reviewer for this.

r? @Amanieu

@ghost

…mic-from-mut, r=kodrAus Revert adding Atomic::from_mut. This reverts rust-lang#74532, which made too many assumptions about platforms, breaking some things. Will need to be added later with a better way of gating on proper alignment, without hardcoding cfg(target_arch)s. --- To be merged if fixing from_mut (rust-lang#76965) takes too long. r? @ghost

@ghost

…mic-from-mut, r=kodrAus Revert adding Atomic::from_mut. This reverts rust-lang#74532, which made too many assumptions about platforms, breaking some things. Will need to be added later with a better way of gating on proper alignment, without hardcoding cfg(target_arch)s. --- To be merged if fixing from_mut (rust-lang#76965) takes too long. r? @ghost

joshtriplett · 2020-09-21T14:36:24Z

Adding this new feature-gated cfg for atomic alignment seems perfectly reasonable to me from a language perspective.

Amanieu

Could you add a compile-fail ui test to ensure that from_mut is not available on x86?

library/core/src/sync/atomic.rs

bors · 2020-09-21T17:59:50Z

☔ The latest upstream changes (presumably #77013) made this pull request unmergeable. Please resolve the merge conflicts.

Note that reviewers usually do not review pull requests until merge conflicts are resolved! Once you resolve the conflicts, you should change the labels applied by bors to indicate that your PR is ready for review. Post this as a comment to change the labels:

@rustbot modify labels: +S-waiting-on-review -S-waiting-on-author

This is needed for Atomic::from_mut.

This reverts commit 5ef1db3.

Instead of a few hardcoded cfg(target_arch = ..) like before.

m-ou-se · 2020-09-21T18:59:26Z

Could you add a compile-fail ui test to ensure that from_mut is not available on x86?

Sure. Working on it.

@rustbot modify labels: -S-waiting-on-review +S-waiting-on-author

m-ou-se · 2020-09-21T19:20:23Z

I've added the test. I made it specific for both x86 and linux, because some other x86 platforms do use 64 bit alignment for 64 bit integers:

        target            u64 alignment
i686-unknown-linux-gnu         32
i586-unknown-linux-gnu         32
i686-unknown-linux-musl        32
i586-unknown-linux-musl        32
i686-linux-android             32
i686-unknown-freebsd           32
i686-unknown-openbsd           32
i686-unknown-netbsd            32
i686-unknown-haiku             32
i686-apple-darwin              32
i686-pc-windows-gnu               64
i686-uwp-windows-gnu              64
i686-pc-windows-msvc              64
i686-uwp-windows-msvc             64
i586-pc-windows-msvc              64
i686-unknown-cloudabi          32
i686-unknown-uefi                 64
i686-wrs-vxworks               32

But still, if this test ever happens to run on some x86 Linux target with fully aligned u64s, it'll fail. But that shouldn't affect people cross compiling libcore themselves (-Zbuld-std), as this is a ui test.

@rustbot modify labels: +S-waiting-on-review -S-waiting-on-author

Amanieu · 2020-09-21T22:27:05Z

@bors r+

bors · 2020-09-21T22:27:06Z

📌 Commit bcc1d56 has been approved by Amanieu

@RalfJung

…-from-mut, r=Amanieu Add cfg(target_has_atomic_equal_alignment) and use it for Atomic::from_mut. Fixes some platform-specific problems with rust-lang#74532 by using the actual alignment of the types instead of hardcoding a few `target_arch`s. r? @RalfJung

Rollup of 9 pull requests Successful merges: - rust-lang#76898 (Record `tcx.def_span` instead of `item.span` in crate metadata) - rust-lang#76939 (emit errors during AbstractConst building) - rust-lang#76965 (Add cfg(target_has_atomic_equal_alignment) and use it for Atomic::from_mut.) - rust-lang#76993 (Changing the alloc() to accept &self instead of &mut self) - rust-lang#76994 (fix small typo in docs and comments) - rust-lang#77017 (Add missing examples on Vec iter types) - rust-lang#77042 (Improve documentation for ToSocketAddrs) - rust-lang#77047 (Miri: more informative deallocation error messages) - rust-lang#77055 (Add #[track_caller] to more panicking Cell functions) Failed merges: r? `@ghost`

rust-highfive assigned RalfJung Sep 20, 2020

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Sep 20, 2020

m-ou-se mentioned this pull request Sep 20, 2020

Add Atomic*::from_mut. #74532

Merged

rust-highfive assigned KodrAus and unassigned RalfJung Sep 20, 2020

m-ou-se mentioned this pull request Sep 20, 2020

[AVR] Failure to compile libcore on nightly #76833

Closed

RalfJung reviewed Sep 20, 2020

View reviewed changes

compiler/rustc_feature/src/builtin_attrs.rs Show resolved Hide resolved

m-ou-se mentioned this pull request Sep 20, 2020

Revert adding Atomic::from_mut. #76967

Merged

rust-highfive assigned Amanieu and unassigned KodrAus Sep 20, 2020

Amanieu reviewed Sep 21, 2020

View reviewed changes

library/core/src/sync/atomic.rs Outdated Show resolved Hide resolved

m-ou-se added 6 commits September 21, 2020 20:42

Add cfg(target_has_atomic_equal_alignment).

db74e1f

This is needed for Atomic::from_mut.

Don't unwrap but report a fatal error for TargetDataLayout::parse.

9e3f94d

Add feature gate ui test for cfg(target_has_atomic_load_store).

af56ad7

Add feature gate ui test for cfg(target_has_atomic_equal_alignment).

54fdf54

Revert "Revert adding Atomic::from_mut."

668225d

This reverts commit 5ef1db3.

Gate Atomic::from_mut on cfg(target_has_atomic_equal_alignment).

7a04ff6

Instead of a few hardcoded cfg(target_arch = ..) like before.

m-ou-se force-pushed the fix-atomic-from-mut branch from 950b940 to 7a04ff6 Compare September 21, 2020 18:44

Move use align_of in atomic.rs into the places where it is used.

5d6f1a1

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Sep 21, 2020

Test that AtomicU64::from_mut is not available on x86 linux.

bcc1d56

m-ou-se force-pushed the fix-atomic-from-mut branch from c9c3534 to bcc1d56 Compare September 21, 2020 19:20

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Sep 21, 2020

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Sep 21, 2020

This was referenced Sep 22, 2020

Rollup of 9 pull requests #77061

Closed

Rollup of 9 pull requests #77102

Merged

bors merged commit eaaf5d7 into rust-lang:master Sep 24, 2020

rustbot added this to the 1.48.0 milestone Sep 24, 2020

m-ou-se deleted the fix-atomic-from-mut branch December 30, 2020 21:28

This was referenced Feb 9, 2022

Tracking issue for #[cfg(target_has_atomic = ...)] #32976

Closed

Tracking Issue for cfg_target_has_atomic_equal_alignment #93822

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add cfg(target_has_atomic_equal_alignment) and use it for Atomic::from_mut. #76965

Add cfg(target_has_atomic_equal_alignment) and use it for Atomic::from_mut. #76965

m-ou-se commented Sep 20, 2020

RalfJung commented Sep 20, 2020

m-ou-se commented Sep 20, 2020 •

edited

m-ou-se commented Sep 20, 2020

KodrAus commented Sep 20, 2020

joshtriplett commented Sep 21, 2020

Amanieu left a comment

bors commented Sep 21, 2020

m-ou-se commented Sep 21, 2020

m-ou-se commented Sep 21, 2020

Amanieu commented Sep 21, 2020

bors commented Sep 21, 2020

Add cfg(target_has_atomic_equal_alignment) and use it for Atomic::from_mut. #76965

Add cfg(target_has_atomic_equal_alignment) and use it for Atomic::from_mut. #76965

Conversation

m-ou-se commented Sep 20, 2020

RalfJung commented Sep 20, 2020

m-ou-se commented Sep 20, 2020 • edited

m-ou-se commented Sep 20, 2020

KodrAus commented Sep 20, 2020

joshtriplett commented Sep 21, 2020

Amanieu left a comment

Choose a reason for hiding this comment

bors commented Sep 21, 2020

m-ou-se commented Sep 21, 2020

m-ou-se commented Sep 21, 2020

Amanieu commented Sep 21, 2020

bors commented Sep 21, 2020

m-ou-se commented Sep 20, 2020 •

edited