do generic_const_args Const::Ty hack only on projection consts by khyperia · Pull Request #162937 · rust-lang/rust

khyperia · 2026-09-18T06:20:45Z

Under generic_const_args, we currently do a bit of a hack to lower all consts to Const::Ty, because under generic_const_args, projection consts not marked with #[rustc_always_gca] could be impl'd by a const with a gca! rhs, and Const::Unevaluated does not support direct consts.

Instead, only lower a const to Const::Ty if:

it is a direct const (under min_generic_const_args), OR
it is a projection const (under generic_const_args)

This excludes all anon consts, and regular (not direct) free and inherent consts from being lowered as Const::Ty under full GCA. It does not fix the underlying issue that lowering a potentially regular projection const to Const::Ty is incorrect.

Fixes #162923

Introduced in #162760

r? @BoxyUwU

rustbot · 2026-09-18T06:20:50Z

BoxyUwU is currently at their maximum review capacity.
They may take a while to respond.

khyperia · 2026-09-18T06:22:37Z

wait no I haven't had my coffee yet what am I doing, this isn't the right fix, one moment... (I mean, it worked, but, yeah, actual fix pushed now)

BoxyUwU · 2026-09-18T08:41:15Z

-            if tcx.features().generic_const_args()
-                || matches!(def_kind, DefKind::Const | DefKind::AssocConst)
-                    && tcx.is_direct_const(def_id)
+            if matches!(def_kind, DefKind::Const | DefKind::AssocConst)


haha when I was reviewing that PR I thought about how I hate precedence on these things and can never tell where the parens go 😆

View changes since the review

BoxyUwU · 2026-09-18T08:42:53Z

what does the THIR look like for this test, where does a NamedConst come from with a defid of an anon 🤔

khyperia · 2026-09-18T12:22:17Z

what does the THIR look like for this test, where does a NamedConst come from with a defid of an anon 🤔

comes from here

rust/compiler/rustc_mir_build/src/thir/cx/expr.rs

Line 374 in 50d5409

let kind = ExprKind::NamedConst { def_id: did, args, user_ty: None };

for reference this is what the code looks like:

enum T<const N: u8 = { T::<0>::B as u8 }> {
    A = 2,
    B,
}

the anon const for the default in enum T<const N: u8 = { this one here }> is called T::{constant#0}. the anon const for the A = { this one here }, is called T::A::{constant#0}. lowering a reference to T::<0>::B takes the discr_did (T::A::{constant#0}) and adds discr_offset (1) to it to get T::<0>::B. this THIR expr is then used, casted to u8, and that's the body of the anon const of the default generic param.

T::{constant#0} = (T::A::{constant#0} + 1) as u8;

bit-aloo

LGTM 🚀

View changes since this review

bit-aloo · 2026-09-20T09:18:06Z

Should we mention about anon const exception here?

when considering other wordings for this, I found myself unable write a wording to justify not doing the exception for just projection consts, rather than all const items. So I rewrote this to only do the exception for just projection consts.

BoxyUwU · 2026-09-23T10:54:41Z

can you write a different PR title and description :3

@rustbot author

khyperia · 2026-09-24T06:06:23Z

@rustbot ready

BoxyUwU · 2026-09-24T07:48:44Z

@bors r+ rollup

rust-bors · 2026-09-24T07:48:53Z

📌 Commit acc7501 has been approved by BoxyUwU

It is now in the queue for this repository.

@BoxyUwU

…xyUwU do generic_const_args Const::Ty hack only on projection consts Under `generic_const_args`, we currently do a bit of a hack to lower all consts to `Const::Ty`, because under `generic_const_args`, projection consts not marked with `#[rustc_always_gca]` could be `impl`'d by a const with a `gca!` rhs, and `Const::Unevaluated` does not support direct consts. Instead, only lower a const to `Const::Ty` if: - it is a direct const (under `min_generic_const_args`), OR - it is a projection const (under `generic_const_args`) This excludes all anon consts, and regular (not direct) free and inherent consts from being lowered as `Const::Ty` under full GCA. It does not fix the underlying issue that lowering a potentially regular projection const to `Const::Ty` is incorrect. Fixes rust-lang#162923 Introduced in rust-lang#162760 r? @BoxyUwU

…uwer Rollup of 7 pull requests Successful merges: - #163218 (stdarch subtree update) - #159287 (Build a new incr comp session dir from scratch every time) - #163222 (Enable EII tests for cg_gcc) - #163251 (cg_gcc subtree sync 2026-09-24) - #162595 (avoid accessing uninferred closure upvars in diagnostics) - #162937 (do generic_const_args Const::Ty hack only on projection consts) - #163238 (Fix incorrect use of await in parser suggestion)

@BoxyUwU

Rollup merge of #162937 - khyperia:gca-fix-unreachable, r=BoxyUwU do generic_const_args Const::Ty hack only on projection consts Under `generic_const_args`, we currently do a bit of a hack to lower all consts to `Const::Ty`, because under `generic_const_args`, projection consts not marked with `#[rustc_always_gca]` could be `impl`'d by a const with a `gca!` rhs, and `Const::Unevaluated` does not support direct consts. Instead, only lower a const to `Const::Ty` if: - it is a direct const (under `min_generic_const_args`), OR - it is a projection const (under `generic_const_args`) This excludes all anon consts, and regular (not direct) free and inherent consts from being lowered as `Const::Ty` under full GCA. It does not fix the underlying issue that lowering a potentially regular projection const to `Const::Ty` is incorrect. Fixes #162923 Introduced in #162760 r? @BoxyUwU

…uwer Rollup of 7 pull requests Successful merges: - rust-lang/rust#163218 (stdarch subtree update) - rust-lang/rust#159287 (Build a new incr comp session dir from scratch every time) - rust-lang/rust#163222 (Enable EII tests for cg_gcc) - rust-lang/rust#163251 (cg_gcc subtree sync 2026-09-24) - rust-lang/rust#162595 (avoid accessing uninferred closure upvars in diagnostics) - rust-lang/rust#162937 (do generic_const_args Const::Ty hack only on projection consts) - rust-lang/rust#163238 (Fix incorrect use of await in parser suggestion)

rustbot assigned BoxyUwU Sep 18, 2026

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Sep 18, 2026

khyperia mentioned this pull request Sep 18, 2026

yeet alias new_from_def_id #162760

Merged

khyperia force-pushed the gca-fix-unreachable branch from cabcba9 to d32dd9e Compare September 18, 2026 06:25

BoxyUwU reviewed Sep 18, 2026

View reviewed changes

bit-aloo reviewed Sep 20, 2026

View reviewed changes

gca: fix unreachable

acc7501

khyperia force-pushed the gca-fix-unreachable branch from d32dd9e to acc7501 Compare September 21, 2026 05:05

BoxyUwU reviewed Sep 23, 2026

View reviewed changes

Comment thread compiler/rustc_mir_build/src/builder/expr/as_constant.rs

BoxyUwU reviewed Sep 23, 2026

View reviewed changes

Comment thread compiler/rustc_mir_build/src/builder/expr/as_constant.rs

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 23, 2026

khyperia changed the title ~~generic_const_args: fix unreachable~~ do generic_const_args Const::Ty hack only on projection consts Sep 24, 2026

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 24, 2026

rust-bors Bot 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 24, 2026

This was referenced Sep 24, 2026

Rollup of 6 pull requests #163260

Closed

Rollup of 7 pull requests #163265

Merged

rust-bors Bot merged commit a9ddedf into rust-lang:main Sep 24, 2026
13 checks passed

rustbot added this to the 1.100.0 milestone Sep 24, 2026

khyperia mentioned this pull request Sep 27, 2026

[ICE]: shouldn't have created a ValTree for pattern_type!(u32 is 0..=999999999) #162336

Open

khyperia deleted the gca-fix-unreachable branch September 27, 2026 16:11

khyperia mentioned this pull request Sep 28, 2026

[ICE]: erroneous constant missed by mono item collection #162337

Open

Uh oh!

Conversation

khyperia commented Sep 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Sep 18, 2026

Uh oh!

khyperia commented Sep 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

BoxyUwU Sep 18, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

BoxyUwU commented Sep 18, 2026

Uh oh!

khyperia commented Sep 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

bit-aloo left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

bit-aloo Sep 20, 2026

Choose a reason for hiding this comment

Uh oh!

khyperia Sep 21, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

BoxyUwU commented Sep 23, 2026

Uh oh!

khyperia commented Sep 24, 2026

Uh oh!

BoxyUwU commented Sep 24, 2026

Uh oh!

rust-bors Bot commented Sep 24, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

khyperia commented Sep 18, 2026 •

edited

Loading

khyperia commented Sep 18, 2026 •

edited

Loading

BoxyUwU Sep 18, 2026 •

edited by rustbot

Loading

khyperia commented Sep 18, 2026 •

edited

Loading

bit-aloo left a comment •

edited by rustbot

Loading