Eliminate `CrateMetadataRef`. by nnethercote · Pull Request #155663 · rust-lang/rust

nnethercote · 2026-04-23T00:30:14Z

There are a number of things I dislike about CrateMetadataRef.

It contains two fields cstore and cdata. The latter points to data within the former. It's like having an Elem type that has a reference to a vec element and also a reference to the vec itself. Weird.
The cdata field gets a lot of use, and the Deref impl just derefs that field. The cstore field is rarely used.
CrateMetadataRef is not a good name.
Variables named cdata sometimes refer to values of this type and sometimes to values of type CrateMetadata, which is confusing.

The good news is that CrateMetadataRef is not necessary and can be replaced with &CrateMetadata. Why? Everywhere that CrateMetadataRef is used, a TyCtxt is also present, and the CStore is accessible from the TyCtxt with CStore::from_tcx.

So this commit removes CrateMetadataRef and replaces all its uses with &CrateMetadata. Notes:

This requires adding only two uses of CStore::from_tcx, which shows how rarely the cstore field was used.
get_crate_data now matches get_crate_data_mut more closely.
A few variables are renamed for consistency, e.g. data/cmeta -> cdata.
An unnecessary local variable (local_cdata) in decode_expn_id is removed.
All the CrateMetadataRef methods become CrateMetadata methods, and their receiver changes from self to &self.
RawDefId::decode_from_cdata is inlined and removed, because it has a single call site.

r? @mejrs

nnethercote · 2026-04-23T00:30:45Z

Locally this was a perf win, mostly on doc builds. Let's try on CI:

@bors try @rust-timer queue

cc @petrochenkov

Eliminate `CrateMetadataRef`.

rust-bors · 2026-04-23T02:41:19Z

☀️ Try build successful (CI)
Build commit: 073731e (073731e7481a7399dd3febf125eb5483b67d4642, parent: 913e4bea83424658d76712fee9c52452a3a9ef0e)

rust-timer · 2026-04-23T03:30:13Z

Finished benchmarking commit (073731e): comparison URL.

Overall result: ✅ improvements - no action needed

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

@bors rollup=never
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.7%	[0.3%, 1.2%]	2
Improvements ✅ (primary)	-0.6%	[-1.3%, -0.3%]	14
Improvements ✅ (secondary)	-0.8%	[-1.3%, -0.2%]	30
All ❌✅ (primary)	-0.6%	[-1.3%, -0.3%]	14

Max RSS (memory usage)

Results (primary 1.8%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	1.8%	[1.8%, 1.8%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.8%	[1.8%, 1.8%]	1

Cycles

Results (secondary -2.4%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	2.2%	[2.2%, 2.2%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.2%	[-4.3%, -2.1%]	6
All ❌✅ (primary)	-	-	0

Binary size

Results (secondary 0.0%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.0%	[0.0%, 0.0%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Bootstrap: 492.161s -> 492.336s (0.04%)
Artifact size: 394.33 MiB -> 394.26 MiB (-0.02%)

nnethercote · 2026-04-23T03:58:53Z

Perf results are good, and match what I saw locally.

mejrs · 2026-04-23T10:31:46Z

This looks really nice, but I think petrochenkov has some thoughts :)

petrochenkov · 2026-04-23T11:12:07Z

Everywhere that CrateMetadataRef is used, a TyCtxt is also present, and the CStore is accessible from the TyCtxt with CStore::from_tcx

This wasn't the case until recently, only Session was passed around, not full tcx, because no other parts of tcx were used during decoding, that's why cstore had to be preserved in CrateMetadataRef.
Also if CStrore::from_tcx was used often, it could be a pessimization compared to passing cstore around, because it's going through a lock, but fortunately CStore::from_tcx is not used often.

petrochenkov · 2026-04-23T11:21:39Z

@bors r=mejrs,petrochenkov

rust-bors · 2026-04-23T11:21:42Z

📌 Commit f1f655d has been approved by mejrs,petrochenkov

It is now in the queue for this repository.

There are a number of things I dislike about `CrateMetadataRef`. - It contains two fields `cstore` and `cdata`. The latter points to data within the former. It's like having an `Elem` type that has a reference to a vec element and also a reference to the vec itself. Weird. - The `cdata` field gets a lot of use, and the `Deref` impl just derefs that field. The `cstore` field is rarely used. - `CrateMetadataRef` is not a good name. - Variables named `cdata` sometimes refer to values of this type and sometimes to values of type `CrateMetadata`, which is confusing. The good news is that `CrateMetadataRef` is not necessary and can be replaced with `&CrateMetadata`. Why? Everywhere that `CrateMetadataRef` is used, a `TyCtxt` is also present, and the `CStore` is accessible from the `TyCtxt` with `CStore::from_tcx`. So this commit removes `CrateMetadataRef` and replaces all its uses with `&CrateMetadata`. Notes: - This requires adding only two uses of `CStore::from_tcx`, which shows how rarely the `cstore` field was used. - `get_crate_data` now matches `get_crate_data_mut` more closely. - A few variables are renamed for consistency, e.g. `data`/`cmeta` -> `cdata`. - An unnecessary local variable (`local_cdata`) in `decode_expn_id` is removed. - All the `CrateMetadataRef` methods become `CrateMetadata` methods, and their receiver changes from `self` to `&self`. - `RawDefId::decode_from_cdata` is inlined and removed, because it has a single call site.

rustbot · 2026-04-24T00:11:39Z

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.

nnethercote · 2026-04-24T00:13:03Z

I think the perf benefits here are mostly/entirely due to eliminating a CStore::from_tcx call that was also eliminated in #155628, which has now merged. So I suspect this will now be perf-neutral. Let's check.

@bors try @rust-timer queue

Eliminate `CrateMetadataRef`.

rust-bors · 2026-04-24T02:20:23Z

☀️ Try build successful (CI)
Build commit: 9237761 (92377611c4498a8049254f7034ab73d596516797, parent: 36ba2c7712052d731a7082d0eba5ed3d9d56c133)

rust-timer · 2026-04-24T03:00:49Z

Finished benchmarking commit (9237761): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking means the PR may be perf-sensitive. Consider adding rollup=never if this change is not fit for rolling up.

@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

This perf run didn't have relevant results for this metric.

Max RSS (memory usage)

Results (secondary -1.1%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	6.0%	[6.0%, 6.0%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-4.6%	[-6.5%, -2.7%]	2
All ❌✅ (primary)	-	-	0

Cycles

Results (secondary -3.4%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.4%	[-4.3%, -2.5%]	2
All ❌✅ (primary)	-	-	0

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 490.512s -> 492.425s (0.39%)
Artifact size: 394.34 MiB -> 394.30 MiB (-0.01%)

nnethercote · 2026-04-24T04:33:11Z

This is now perf neutral and can be rolled up.

@bors r=mejrs,petrochenkov rollup

rust-bors · 2026-04-24T04:33:14Z

📌 Commit 804e419 has been approved by mejrs,petrochenkov

It is now in the queue for this repository.

@mejrs

…aRef, r=mejrs,petrochenkov Eliminate `CrateMetadataRef`. There are a number of things I dislike about `CrateMetadataRef`. - It contains two fields `cstore` and `cdata`. The latter points to data within the former. It's like having an `Elem` type that has a reference to a vec element and also a reference to the vec itself. Weird. - The `cdata` field gets a lot of use, and the `Deref` impl just derefs that field. The `cstore` field is rarely used. - `CrateMetadataRef` is not a good name. - Variables named `cdata` sometimes refer to values of this type and sometimes to values of type `CrateMetadata`, which is confusing. The good news is that `CrateMetadataRef` is not necessary and can be replaced with `&CrateMetadata`. Why? Everywhere that `CrateMetadataRef` is used, a `TyCtxt` is also present, and the `CStore` is accessible from the `TyCtxt` with `CStore::from_tcx`. So this commit removes `CrateMetadataRef` and replaces all its uses with `&CrateMetadata`. Notes: - This requires adding only two uses of `CStore::from_tcx`, which shows how rarely the `cstore` field was used. - `get_crate_data` now matches `get_crate_data_mut` more closely. - A few variables are renamed for consistency, e.g. `data`/`cmeta` -> `cdata`. - An unnecessary local variable (`local_cdata`) in `decode_expn_id` is removed. - All the `CrateMetadataRef` methods become `CrateMetadata` methods, and their receiver changes from `self` to `&self`. - `RawDefId::decode_from_cdata` is inlined and removed, because it has a single call site. r? @mejrs

Rollup of 9 pull requests Successful merges: - #155684 (Generalize IO Traits for `Arc<T>` where `&T: IoTrait`) - #155081 (Move and clean up some ui test) - #155379 (Avoid query cycles in DataflowConstProp) - #155663 (Eliminate `CrateMetadataRef`.) - #155669 (Add `Sender` diagnostic item for `std::sync::mpsc::Sender`) - #155698 (Syntactically reject tuple index shorthands in struct patterns to fix a correctness regression) - #155703 (Remove myself as a maintainer of `wasm32-wasip1-threads`) - #155706 (Remove `AttributeLintKind` variants - part 7) - #155712 (Forbid `*-pass` and `*-fail` directives in tests/crashes)

@mejrs

Rollup merge of #155663 - nnethercote:eliminate-CrateMetadataRef, r=mejrs,petrochenkov Eliminate `CrateMetadataRef`. There are a number of things I dislike about `CrateMetadataRef`. - It contains two fields `cstore` and `cdata`. The latter points to data within the former. It's like having an `Elem` type that has a reference to a vec element and also a reference to the vec itself. Weird. - The `cdata` field gets a lot of use, and the `Deref` impl just derefs that field. The `cstore` field is rarely used. - `CrateMetadataRef` is not a good name. - Variables named `cdata` sometimes refer to values of this type and sometimes to values of type `CrateMetadata`, which is confusing. The good news is that `CrateMetadataRef` is not necessary and can be replaced with `&CrateMetadata`. Why? Everywhere that `CrateMetadataRef` is used, a `TyCtxt` is also present, and the `CStore` is accessible from the `TyCtxt` with `CStore::from_tcx`. So this commit removes `CrateMetadataRef` and replaces all its uses with `&CrateMetadata`. Notes: - This requires adding only two uses of `CStore::from_tcx`, which shows how rarely the `cstore` field was used. - `get_crate_data` now matches `get_crate_data_mut` more closely. - A few variables are renamed for consistency, e.g. `data`/`cmeta` -> `cdata`. - An unnecessary local variable (`local_cdata`) in `decode_expn_id` is removed. - All the `CrateMetadataRef` methods become `CrateMetadata` methods, and their receiver changes from `self` to `&self`. - `RawDefId::decode_from_cdata` is inlined and removed, because it has a single call site. r? @mejrs

Rollup of 9 pull requests Successful merges: - rust-lang/rust#155684 (Generalize IO Traits for `Arc<T>` where `&T: IoTrait`) - rust-lang/rust#155081 (Move and clean up some ui test) - rust-lang/rust#155379 (Avoid query cycles in DataflowConstProp) - rust-lang/rust#155663 (Eliminate `CrateMetadataRef`.) - rust-lang/rust#155669 (Add `Sender` diagnostic item for `std::sync::mpsc::Sender`) - rust-lang/rust#155698 (Syntactically reject tuple index shorthands in struct patterns to fix a correctness regression) - rust-lang/rust#155703 (Remove myself as a maintainer of `wasm32-wasip1-threads`) - rust-lang/rust#155706 (Remove `AttributeLintKind` variants - part 7) - rust-lang/rust#155712 (Forbid `*-pass` and `*-fail` directives in tests/crashes)

rustbot assigned mejrs Apr 23, 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 Apr 23, 2026