centralize `-Zmin-function-alignment` logic #142854

folkertdev · 2025-06-21T22:52:36Z

tracking issue: #82232
discussed in: #142824 (comment)

Apply the -Zmin-function-alignment value to the alignment field of the function attributes when those are created, so that individual backends don't need to consider it.

The one exception right now is cranelift, because it can't yet set the alignment for individual functions, but it can (and does) set the global minimum function alignment.

cc @RalfJung I think this is an improvement regardless, is there anything else that should be done for miri?

rustbot · 2025-06-21T22:52:40Z

r? @BoxyUwU

rustbot has assigned @BoxyUwU.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

rustbot · 2025-06-21T22:52:42Z

Some changes occurred to the CTFE machinery

cc @RalfJung, @oli-obk, @lcnr

Some changes occurred to the CTFE / Miri interpreter

cc @rust-lang/miri

Some changes occurred in compiler/rustc_codegen_ssa

cc @WaffleLapkin

Some changes occurred in compiler/rustc_codegen_ssa/src/codegen_attrs.rs

cc @jdonszelmann

jieyouxu · 2025-06-22T01:01:12Z

r? codegen

workingjubilee · 2025-06-22T18:36:13Z

This looks good to me.

r? workingjubilee
@bors r+

bors · 2025-06-22T18:36:15Z

📌 Commit a123a36 has been approved by workingjubilee

It is now in the queue for this repository.

…-alignment, r=workingjubilee centralize `-Zmin-function-alignment` logic tracking issue: rust-lang#82232 discussed in: rust-lang#142824 (comment) Apply the `-Zmin-function-alignment` value to the alignment field of the function attributes when those are created, so that individual backends don't need to consider it. The one exception right now is cranelift, because it can't yet set the alignment for individual functions, but it can (and does) set the global minimum function alignment. cc `@RalfJung` I think this is an improvement regardless, is there anything else that should be done for miri?

Rollup of 6 pull requests Successful merges: - #140136 (Add an aarch64-msvc build running on ARM64 Windows) - #141597 (Document subdirectories of UI tests with README files) - #142823 (Port `#[no_mangle]` to new attribute parsing infrastructure) - #142828 (1.88.0 release notes) - #142854 (centralize `-Zmin-function-alignment` logic) - #142875 (Check rustdoc-json-types FORMAT_VERSION is correctly updated) r? `@ghost` `@rustbot` modify labels: rollup

Rollup of 5 pull requests Successful merges: - #141597 (Document subdirectories of UI tests with README files) - #142823 (Port `#[no_mangle]` to new attribute parsing infrastructure) - #142828 (1.88.0 release notes) - #142854 (centralize `-Zmin-function-alignment` logic) - #142875 (Check rustdoc-json-types FORMAT_VERSION is correctly updated) r? `@ghost` `@rustbot` modify labels: rollup

Rollup merge of #142854 - folkertdev:centralize-min-function-alignment, r=workingjubilee centralize `-Zmin-function-alignment` logic tracking issue: #82232 discussed in: #142824 (comment) Apply the `-Zmin-function-alignment` value to the alignment field of the function attributes when those are created, so that individual backends don't need to consider it. The one exception right now is cranelift, because it can't yet set the alignment for individual functions, but it can (and does) set the global minimum function alignment. cc ``@RalfJung`` I think this is an improvement regardless, is there anything else that should be done for miri?

…no-attributes, r=workingjubilee fix `-Zmin-function-alignment` on functions without attributes tracking issue: rust-lang#82232 related: rust-lang#142854 The minimum function alignment was skipped on functions without attributes (because the logic was in a loop that only runs if there is at least one attribute). The underlying reason we didn't catch this before is that in our testing we generally apply `#[no_mangle]` to functions that are tested. I've added a test now that deliberately has no attributes. r? `@workingjubilee`

Rollup merge of #142923 - folkertdev:min-function-alignment-no-attributes, r=workingjubilee fix `-Zmin-function-alignment` on functions without attributes tracking issue: #82232 related: #142854 The minimum function alignment was skipped on functions without attributes (because the logic was in a loop that only runs if there is at least one attribute). The underlying reason we didn't catch this before is that in our testing we generally apply `#[no_mangle]` to functions that are tested. I've added a test now that deliberately has no attributes. r? `@workingjubilee`

RalfJung · 2025-06-25T16:30:08Z

compiler/rustc_codegen_ssa/src/codegen_attrs.rs

@@ -126,6 +126,12 @@ fn codegen_fn_attrs(tcx: TyCtxt<'_>, did: LocalDefId) -> CodegenFnAttrs {
            }
        }

+        // Apply the minimum function alignment here, so that individual backends don't have to.


It's actually more important than that, if you consider that -C flags can differ between translation units.

This ensures functions defined in a crate compiled with -Cmin-function-alignment get that alignment independent of the crate they are codegen'd in.

If a function from crate A is generated in another crate B compiled with -Zmin-function-alignment=B, I would expect the alignment to be at least B.

We don't say anything about which function is generated where, so I don't see how such a guarantee could be practically useful?

From a const-eval perspective, doing this would be just unsound as it means the result of "which alignment does this function have" could be different in different parts of the program.

RalfJung · 2025-06-25T16:31:50Z

compiler/rustc_const_eval/src/interpret/memory.rs

-                    // Function alignment can be set globally with the `-Zmin-function-alignment=<n>` flag;
-                    // the alignment from a `#[repr(align(<n>))]` is used if it specifies a higher alignment.
-                    let fn_align = self.tcx.codegen_fn_attrs(instance.def_id()).alignment;
-                    let global_align = self.tcx.sess.opts.unstable_opts.min_function_alignment;


Is there any way we can have a lint against accessing tcx.sess.opts in the interpreter? Or is there some way the query machinery could help us ensure this? This is a soundness bug waiting to happen. :/ And it can't be just the interpreter, right? Other queries also must be cross-crate-stable in the same way?

codegen_fn_attrs: make comment more precise Follow-up to rust-lang#142854. r? `@oli-obk` or `@workingjubilee`

Rollup of 5 pull requests Successful merges: - rust-lang/rust#141597 (Document subdirectories of UI tests with README files) - rust-lang/rust#142823 (Port `#[no_mangle]` to new attribute parsing infrastructure) - rust-lang/rust#142828 (1.88.0 release notes) - rust-lang/rust#142854 (centralize `-Zmin-function-alignment` logic) - rust-lang/rust#142875 (Check rustdoc-json-types FORMAT_VERSION is correctly updated) r? `@ghost` `@rustbot` modify labels: rollup

codegen_fn_attrs: make comment more precise Follow-up to rust-lang#142854. r? ``@oli-obk`` or ``@workingjubilee``

codegen_fn_attrs: make comment more precise Follow-up to rust-lang#142854. r? ```@oli-obk``` or ```@workingjubilee```

codegen_fn_attrs: make comment more precise Follow-up to rust-lang#142854. r? ````@oli-obk```` or ````@workingjubilee````

centralize -Zmin-function-alignment logic

a123a36

rustbot assigned BoxyUwU Jun 21, 2025

rustbot assigned WaffleLapkin and unassigned BoxyUwU Jun 22, 2025

ojeda mentioned this pull request Jun 22, 2025

Rust unstable features needed for the kernel Rust-for-Linux/linux#2

Open

98 tasks

rustbot assigned workingjubilee and unassigned WaffleLapkin Jun 22, 2025

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 Jun 22, 2025

jhpratt mentioned this pull request Jun 22, 2025

Rollup of 6 pull requests #142895

Closed

matthiaskrgr mentioned this pull request Jun 23, 2025

Rollup of 5 pull requests #142901

Merged

bors merged commit 57abad8 into rust-lang:master Jun 23, 2025
10 checks passed

rustbot added this to the 1.89.0 milestone Jun 23, 2025

folkertdev mentioned this pull request Jun 23, 2025

fix -Zmin-function-alignment on functions without attributes #142923

Merged

RalfJung reviewed Jun 25, 2025

View reviewed changes

This was referenced Jun 25, 2025

codegen_fn_attrs: make comment more precise #143020

Open

Stabilize -Cmin-function-alignment #142824

Open

samueltardieu added a commit to samueltardieu/rust that referenced this pull request Jun 26, 2025

Rollup merge of rust-lang#143020 - RalfJung:codegen_fn_attrs, r=oli-obk

38f0a16

codegen_fn_attrs: make comment more precise Follow-up to rust-lang#142854. r? `@oli-obk` or `@workingjubilee`

matthiaskrgr added a commit to matthiaskrgr/rust that referenced this pull request Jun 26, 2025

Rollup merge of rust-lang#143020 - RalfJung:codegen_fn_attrs, r=oli-obk

c2b2c66

codegen_fn_attrs: make comment more precise Follow-up to rust-lang#142854. r? ```@oli-obk``` or ```@workingjubilee```

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

centralize `-Zmin-function-alignment` logic #142854

centralize `-Zmin-function-alignment` logic #142854

Uh oh!

folkertdev commented Jun 21, 2025

Uh oh!

rustbot commented Jun 21, 2025

Uh oh!

rustbot commented Jun 21, 2025

Uh oh!

jieyouxu commented Jun 22, 2025

Uh oh!

workingjubilee commented Jun 22, 2025

Uh oh!

bors commented Jun 22, 2025

Uh oh!

Uh oh!

RalfJung Jun 25, 2025

Uh oh!

tmiasko Jun 25, 2025

Uh oh!

RalfJung Jun 25, 2025

Uh oh!

RalfJung Jun 25, 2025

Uh oh!

Uh oh!

centralize -Zmin-function-alignment logic #142854

centralize -Zmin-function-alignment logic #142854

Uh oh!

Conversation

folkertdev commented Jun 21, 2025

Uh oh!

rustbot commented Jun 21, 2025

Uh oh!

rustbot commented Jun 21, 2025

Uh oh!

jieyouxu commented Jun 22, 2025

Uh oh!

workingjubilee commented Jun 22, 2025

Uh oh!

bors commented Jun 22, 2025

Uh oh!

Uh oh!

RalfJung Jun 25, 2025

Choose a reason for hiding this comment

Uh oh!

tmiasko Jun 25, 2025

Choose a reason for hiding this comment

Uh oh!

RalfJung Jun 25, 2025

Choose a reason for hiding this comment

Uh oh!

RalfJung Jun 25, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

centralize `-Zmin-function-alignment` logic #142854

centralize `-Zmin-function-alignment` logic #142854