Set groundwork for proper const normalization #140549

BoxyUwU · 2025-05-01T12:37:05Z

r? lcnr

Updates a lot of our normalization/alias infrastructure to be setup to handle mgca aliases and normalization once const items are represented more like aliases than bodies. Inherent associated consts are still super busted, I didn't update the assertions that IACs the right arg setup because that winds up being somewhat involved to do before proper support for normalizing const aliases is implemented.

I dont intend for this to have any effect on stable. We continue normalizing via ctfe on stable and the codepaths in project for consts should only be reachable with mgca or ace.

rustbot · 2025-05-01T12:37:12Z

changes to the core type system

cc @compiler-errors, @lcnr

Some changes occurred in match checking

cc @Nadrieril

BoxyUwU · 2025-05-01T12:38:48Z

compiler/rustc_trait_selection/src/traits/project.rs

+    // system consts that we can retrieve with some `query const_arg_of_alias` query. Evaluating the
+    // constant is "close enough" to getting the actual rhs of the const item for now even if it might
+    // lead to some cycles
+    super::evaluate_const(selcx.infcx, alias_ct, param_env)


feature(associated_const_equality) might be a bit more cycle prone now due to evaluating here instead of normalizing to the impl definition and then evaluating? All the tests still pass though so I guess its fine- and both this and the previous way it worked can't stick around long term anyway.

lcnr

need to rebase

also, pls move the normalize.rs move AliasKind branches into subfn into a separate commit if possible. Are there any other changes to that code while moving them?

compiler/rustc_trait_selection/src/traits/normalize.rs

BoxyUwU · 2025-05-01T13:13:04Z

merge conflicts :despair:

BoxyUwU · 2025-05-01T13:14:17Z

also, pls move the normalize.rs move AliasKind branches into subfn into a separate commit if possible. Are there any other changes to that code while moving them?

Kinda, I had to change stuff from being working on AliasTy to AliasTerm so there's some minor changes for getting the types to line up, but behaviourally speaking it should be the same. I originally had it as its own commit but honestly it didn't really seem to matter much when I was looking over the changes because I had to switch from AliasTy to AliasKind at the same time 🤔 though i guess i could have just not done that... okay lemme split this up

BoxyUwU · 2025-05-01T13:40:31Z

Okay first commit should contain no changes to the norm logic other than replacing the two match arms with an if/else. Second commit should be able to be reviewed to see the actual changes to the logic

BoxyUwU · 2025-05-01T14:18:27Z

Oh

rustbot · 2025-05-01T16:12:53Z

This PR changes a file inside tests/crashes. If a crash was fixed, please move into the corresponding ui subdir and add 'Fixes #' to the PR description to autoclose the issue upon merge.

BoxyUwU · 2025-05-01T16:14:08Z

One of the crashes tests is also fixed by #140553 and in an actually principled way so I've just removed it and will let that PR add the test (which ive verified still ICEs under this PR). The other test I've updated to still reproduce since this PR doesn't really fix it afaik

compiler/rustc_trait_selection/src/traits/normalize.rs

lcnr · 2025-05-01T18:07:42Z

compiler/rustc_trait_selection/src/traits/normalize.rs

+        self.obligations.extend(
+            infcx.tcx.predicates_of(free.def_id).instantiate_own(infcx.tcx, free.args).map(
+                |(mut predicate, span)| {
+                    if free.has_escaping_bound_vars() {


why do that if free has escaping bound vars instead of checking the predicate?

seems generally wrong to be instantiating early binders with bound vars, going to open an issue about this

compiler/rustc_trait_selection/src/traits/normalize.rs

lcnr · 2025-05-02T01:55:15Z

@bors r+ rollup=iffy

bors · 2025-05-02T01:55:18Z

📌 Commit 9ec8373 has been approved by lcnr

It is now in the queue for this repository.

Set groundwork for proper const normalization r? lcnr Updates a lot of our normalization/alias infrastructure to be setup to handle mgca aliases and normalization once const items are represented more like aliases than bodies. Inherent associated consts are still super busted, I didn't update the assertions that IACs the right arg setup because that winds up being somewhat involved to do *before* proper support for normalizing const aliases is implemented. I dont *intend* for this to have any effect on stable. We continue normalizing via ctfe on stable and the codepaths in `project` for consts should only be reachable with mgca or ace.

bors · 2025-05-04T03:12:45Z

⌛ Testing commit 9ec8373 with merge 1bea580...

bors · 2025-05-04T06:22:30Z

☀️ Test successful - checks-actions
Approved by: lcnr
Pushing 1bea580 to master...

github-actions · 2025-05-04T06:24:27Z

What is this?

This is an experimental post-merge analysis report that shows differences in test outcomes between the merged PR and its parent PR.

Comparing 3559e0a (parent) -> 1bea580 (this PR)

Test differences

Show 20 test diffs

Stage 1

[crashes] tests/crashes/133066.rs: pass -> [missing] (J1)

Stage 2

[crashes] tests/crashes/133066.rs: pass -> [missing] (J0)

Additionally, 18 doctest diffs were found. These are ignored, as they are noisy.

Job group index

Test dashboard

Run

cargo run --manifest-path src/ci/citool/Cargo.toml -- \
    test-dashboard 1bea580f364c65bd5f7380a1056e150df7b8a1a6 --output-dir test-dashboard

And then open test-dashboard/index.html in your browser to see an overview of all executed tests.

Job duration changes

dist-x86_64-apple: 11442.2s -> 8427.4s (-26.3%)
dist-apple-various: 5964.0s -> 6832.9s (14.6%)
x86_64-apple-1: 7001.7s -> 7779.5s (11.1%)
aarch64-apple: 4040.3s -> 3621.5s (-10.4%)
x86_64-apple-2: 4179.6s -> 4535.9s (8.5%)
x86_64-msvc-1: 8641.9s -> 9364.2s (8.4%)
test-various: 4267.9s -> 3957.0s (-7.3%)
x86_64-gnu-llvm-19-3: 7168.1s -> 6751.2s (-5.8%)
x86_64-gnu-distcheck: 4773.2s -> 5042.0s (5.6%)
dist-x86_64-msvc-alt: 7853.7s -> 7419.8s (-5.5%)

How to interpret the job duration changes?

Job durations can vary a lot, based on the actual runner instance
that executed the job, system noise, invalidated caches, etc. The table above is provided
mostly for t-infra members, for simpler debugging of potential CI slow-downs.

rust-timer · 2025-05-04T08:00:33Z

Finished benchmarking commit (1bea580): comparison URL.

Overall result: ❌ regressions - please read the text below

Our benchmarks found a performance regression caused by this PR.
This might be an actual regression, but it can also be just noise.

Next Steps:

If the regression was expected or you think it can be justified,
please write a comment with sufficient written justification, and add
@rustbot label: +perf-regression-triaged to it, to mark the regression as triaged.
If you think that you know of a way to resolve the regression, try to create
a new PR with a fix for the regression.
If you do not understand the regression or you think that it is just noise,
you can ask the @rust-lang/wg-compiler-performance working group for help (members of this group
were already notified of this PR).

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

Instruction count

This is the most reliable metric that we have; it was used to determine the overall result at the top of this comment. However, even this metric can sometimes exhibit noise.

	mean	range	count
Regressions ❌ (primary)	0.7%	[0.3%, 1.4%]	3
Regressions ❌ (secondary)	0.5%	[0.1%, 0.6%]	7
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.7%	[0.3%, 1.4%]	3

Max RSS (memory usage)

Results (primary 0.4%, secondary 2.1%)

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)	1.0%	[0.4%, 2.9%]	12
Regressions ❌ (secondary)	2.1%	[2.1%, 2.1%]	2
Improvements ✅ (primary)	-0.6%	[-1.0%, -0.4%]	7
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.4%	[-1.0%, 2.9%]	19

Cycles

Results (primary -0.2%)

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%	[0.5%, 1.4%]	3
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.5%	[-0.7%, -0.4%]	11
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.2%	[-0.7%, 1.4%]	14

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 769.569s -> 770.15s (0.08%)
Artifact size: 365.54 MiB -> 365.38 MiB (-0.04%)

Mark-Simulacrum · 2025-05-05T00:51:11Z

Most regressions are small, and a few look like non-noise. Looking at a function-level diff of the syn-1 benchmark... it looks like the changes are mostly small-ish changes around inlining, which doesn't seem worth further investigation.

<  13,419,979  ???:
   36,250,708    <rustc_trait_selection::traits::select::SelectionContext>::confirm_candidate
  -20,035,552    <rustc_trait_selection::traits::select::SelectionContext>::poly_select::{closure#0}
  -11,388,056    <rustc_trait_selection::traits::select::SelectionContext>::match_impl::{closure#0}
    9,881,516    rustc_metadata::rmeta::decoder::cstore_impl::provide_extern::explicit_predicates_of
   -9,273,148    rustc_query_impl::plumbing::__rust_begin_short_backtrace::<rustc_query_impl::query_impl::explicit_predicates_of::dynamic_query::{closure#2}::{closure#0}, rustc_mid>
   -7,724,715    <rustc_trait_selection::traits::select::SelectionContext>::candidate_from_obligation::{closure#0}::{closure#0}
    6,964,830    <rustc_trait_selection::traits::select::SelectionContext>::assemble_candidates
    5,811,720    <rustc_middle::ty::context::TyCtxt>::recursion_limit
   -1,461,675    rustc_trait_selection::traits::project::opt_normalize_projection_term

rustbot assigned lcnr May 1, 2025

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. WG-trait-system-refactor The Rustc Trait System Refactor Initiative (-Znext-solver) labels May 1, 2025

BoxyUwU commented May 1, 2025

View reviewed changes

lcnr reviewed May 1, 2025

View reviewed changes

compiler/rustc_trait_selection/src/traits/normalize.rs Outdated Show resolved Hide resolved

BoxyUwU force-pushed the proper_const_norm branch from 6c4b953 to ed128eb Compare May 1, 2025 13:22

No-op split into sub functions

53e3907

BoxyUwU force-pushed the proper_const_norm branch from ed128eb to 25821fd Compare May 1, 2025 13:39

BoxyUwU force-pushed the proper_const_norm branch from 25821fd to 9c29713 Compare May 1, 2025 13:43

This comment has been minimized.

Sign in to view

lcnr reviewed May 1, 2025

View reviewed changes

compiler/rustc_trait_selection/src/traits/normalize.rs Outdated Show resolved Hide resolved

lcnr reviewed May 1, 2025

View reviewed changes

compiler/rustc_trait_selection/src/traits/normalize.rs Outdated Show resolved Hide resolved

lcnr reviewed May 1, 2025

View reviewed changes

compiler/rustc_trait_selection/src/traits/normalize.rs Outdated Show resolved Hide resolved

BoxyUwU force-pushed the proper_const_norm branch 3 times, most recently from 5ea4762 to 5fb578e Compare May 1, 2025 19:01

BoxyUwU added 2 commits May 1, 2025 20:05

Set groundwork for proper const normalization

238d113

Crashes tests

9ec8373

BoxyUwU force-pushed the proper_const_norm branch from 5fb578e to 9ec8373 Compare May 1, 2025 19:05

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 May 2, 2025

VlaDexa mentioned this pull request May 2, 2025

Rollup of 40 pull requests #140573

Closed

fmease mentioned this pull request May 2, 2025

GCI: At their def site, actually wfcheck the where-clause & always eval free lifetime-generic constants #136429

Open

bors added the merged-by-bors This PR was explicitly merged by bors. label May 4, 2025

bors merged commit 1bea580 into rust-lang:master May 4, 2025
7 checks passed

rustbot added this to the 1.88.0 milestone May 4, 2025

This was referenced May 4, 2025

Ensure constants are WF before calling into CTFE #137972

Draft

wf: emit projection goal for aliases #140558

Open

rustbot added the perf-regression Performance regression. label May 4, 2025

Mark-Simulacrum added the perf-regression-triaged The performance regression has been triaged. label May 5, 2025

moxian mentioned this pull request May 5, 2025

ICE: expected a const, but found a type #140642

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Set groundwork for proper const normalization #140549

Set groundwork for proper const normalization #140549

BoxyUwU commented May 1, 2025 •

edited

Loading

rustbot commented May 1, 2025

BoxyUwU May 1, 2025

lcnr left a comment

BoxyUwU commented May 1, 2025

BoxyUwU commented May 1, 2025 •

edited

Loading

BoxyUwU commented May 1, 2025

This comment has been minimized.

BoxyUwU commented May 1, 2025

rustbot commented May 1, 2025

BoxyUwU commented May 1, 2025 •

edited

Loading

lcnr May 1, 2025

BoxyUwU May 1, 2025

lcnr commented May 2, 2025

bors commented May 2, 2025

bors commented May 4, 2025

bors commented May 4, 2025

github-actions bot commented May 4, 2025

Stage 1

Stage 2

rust-timer commented May 4, 2025

Mark-Simulacrum commented May 5, 2025

Set groundwork for proper const normalization #140549

Set groundwork for proper const normalization #140549

Conversation

BoxyUwU commented May 1, 2025 • edited Loading

rustbot commented May 1, 2025

BoxyUwU May 1, 2025

Choose a reason for hiding this comment

lcnr left a comment

Choose a reason for hiding this comment

BoxyUwU commented May 1, 2025

BoxyUwU commented May 1, 2025 • edited Loading

BoxyUwU commented May 1, 2025

This comment has been minimized.

BoxyUwU commented May 1, 2025

rustbot commented May 1, 2025

BoxyUwU commented May 1, 2025 • edited Loading

lcnr May 1, 2025

Choose a reason for hiding this comment

BoxyUwU May 1, 2025

Choose a reason for hiding this comment

lcnr commented May 2, 2025

bors commented May 2, 2025

bors commented May 4, 2025

bors commented May 4, 2025

github-actions bot commented May 4, 2025

Test differences

Stage 1

Stage 2

Job duration changes

rust-timer commented May 4, 2025

Overall result: ❌ regressions - please read the text below

Instruction count

Max RSS (memory usage)

Cycles

Binary size

Mark-Simulacrum commented May 5, 2025

BoxyUwU commented May 1, 2025 •

edited

Loading

BoxyUwU commented May 1, 2025 •

edited

Loading

BoxyUwU commented May 1, 2025 •

edited

Loading