Store the type of each GVN value #143271

cjgillot · 2025-07-01T08:21:00Z

MIR is fully typed, so type information is an integral part of what defines a value. GVN currently tries to circumvent storing types, which creates all sorts of complexities.

This PR stores the type along with the enum Value when defining a value index. This allows to simplify a lot of code.

Fixes #128094
Fixes #135128

r? @ghost for perf

cjgillot · 2025-07-01T08:21:16Z

@bors try @rust-timer queue

Store the type of each GVN value MIR is fully typed, so type information is an integral part of what defines a value. GVN currently tries to circumvent storing types, which creates all sorts of complexities. This PR stores the type along with the enum `Value` when defining a value index. This allows to simplify a lot of code. r? `@ghost` for perf

bors · 2025-07-01T08:22:31Z

⌛ Trying commit 2684875 with merge 0debe2b...

bors · 2025-07-01T10:38:03Z

☀️ Try build successful - checks-actions
Build commit: 0debe2b (0debe2b1f819bc0387b12f860b7de22201171ddc)

Kobzol · 2025-07-01T20:54:04Z

Note: this benchmark will be delayed a bit, because we just switched to a new benchmarking machine and I need to backfill stable benchmark results. Sorry for that! It should be done sometime tomorrow, hopefully.

rust-timer · 2025-07-02T10:53:27Z

Finished benchmarking commit (0debe2b): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking this pull request means it may be perf-sensitive – we'll automatically label it not fit for rolling up. You can override this, but we strongly advise not to, due to possible changes in compiler perf.

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

Instruction count

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

Max RSS (memory usage)

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

Cycles

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

Binary size

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

Bootstrap: missing data
Artifact size: 372.23 MiB -> 371.76 MiB (-0.13%)

rustbot · 2025-07-03T09:56:59Z

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

cjgillot · 2025-07-03T09:57:09Z

r? compiler

SparrowLii · 2025-07-07T07:07:36Z

Thanks for the great work!
r? @rust-lang/wg-mir-opt
since they should be more familar with this

rustbot · 2025-07-07T07:07:39Z

Failed to set assignee to eddyb: invalid assignee

Note: Only org members with at least the repository "read" role, users with write permissions, or people who have commented on the PR may be assigned.

SparrowLii · 2025-07-07T07:11:47Z

r? @oli-obk
I think you know about GVN or can figure out who are available to review this work?

oli-obk · 2025-07-16T13:53:32Z

compiler/rustc_mir_transform/src/gvn.rs

+        let ty = self.ty(value);
+        let ty = self.ecx.layout_of(ty).ok()?;
        let op = match *self.get(value) {
+            _ if ty.is_zst() => ImmTy::uninit(ty).into(),


Mostly keeping notes to myself: this is why we get a test change from this PR. It is otherwise a transparent change that just tracks more data

oli-obk · 2025-07-16T13:55:42Z

compiler/rustc_mir_transform/src/gvn.rs

-                    ty.backend_repr,
-                    BackendRepr::Scalar(..) | BackendRepr::ScalarPair(..)
-                ) {
+                let variant = if ty.ty.is_enum() { Some(variant) } else { None };


orthogonal, but this check seems like it could just be removed and the if let Some(variant) =variant below could do the enum check instead.

oli-obk · 2025-07-16T14:05:45Z

@bors r+ rollup

bors · 2025-07-16T14:05:48Z

📌 Commit 4a95b16 has been approved by oli-obk

It is now in the queue for this repository.

Store the type of each GVN value MIR is fully typed, so type information is an integral part of what defines a value. GVN currently tries to circumvent storing types, which creates all sorts of complexities. This PR stores the type along with the enum `Value` when defining a value index. This allows to simplify a lot of code. Fixes rust-lang#128094 Fixes rust-lang#135128 r? ``@ghost`` for perf

Store the type of each GVN value MIR is fully typed, so type information is an integral part of what defines a value. GVN currently tries to circumvent storing types, which creates all sorts of complexities. This PR stores the type along with the enum `Value` when defining a value index. This allows to simplify a lot of code. Fixes rust-lang#128094 Fixes rust-lang#135128 r? ```@ghost``` for perf

Rollup of 15 pull requests Successful merges: - #142300 (Disable `tests/run-make/mte-ffi` because no CI runners have MTE extensions enabled) - #143271 (Store the type of each GVN value) - #143293 (fix `-Zsanitizer=kcfi` on `#[naked]` functions) - #143719 (Emit warning when there is no space between `-o` and arg) - #143833 (Ban projecting into SIMD types [MCP838]) - #143846 (pass --gc-sections if -Zexport-executable-symbols is enabled and improve tests) - #143879 (parse `const trait Trait`) - #143891 (Port `#[coverage]` to the new attribute system) - #143967 (constify `Option` methods) - #143985 (rustc_public: de-StableMIR-ize) - #144008 (Fix false positive double negations with macro invocation) - #144010 (Boostrap: add warning on `optimize = false`) - #144034 (tests: Test line number in debuginfo for diverging function calls) - #144049 (rustc-dev-guide subtree update) - #144056 (Copy GCC sources into the build directory even outside CI) r? `@ghost` `@rustbot` modify labels: rollup

Store the type of each GVN value MIR is fully typed, so type information is an integral part of what defines a value. GVN currently tries to circumvent storing types, which creates all sorts of complexities. This PR stores the type along with the enum `Value` when defining a value index. This allows to simplify a lot of code. Fixes rust-lang#128094 Fixes rust-lang#135128 r? ````@ghost```` for perf

Rollup of 12 pull requests Successful merges: - #142300 (Disable `tests/run-make/mte-ffi` because no CI runners have MTE extensions enabled) - #143271 (Store the type of each GVN value) - #143293 (fix `-Zsanitizer=kcfi` on `#[naked]` functions) - #143719 (Emit warning when there is no space between `-o` and arg) - #143846 (pass --gc-sections if -Zexport-executable-symbols is enabled and improve tests) - #143891 (Port `#[coverage]` to the new attribute system) - #143967 (constify `Option` methods) - #144008 (Fix false positive double negations with macro invocation) - #144010 (Boostrap: add warning on `optimize = false`) - #144034 (tests: Test line number in debuginfo for diverging function calls) - #144049 (rustc-dev-guide subtree update) - #144056 (Copy GCC sources into the build directory even outside CI) r? `@ghost` `@rustbot` modify labels: rollup

Store the type of each GVN value MIR is fully typed, so type information is an integral part of what defines a value. GVN currently tries to circumvent storing types, which creates all sorts of complexities. This PR stores the type along with the enum `Value` when defining a value index. This allows to simplify a lot of code. Fixes rust-lang#128094 Fixes rust-lang#135128 r? `````@ghost````` for perf

Rollup of 12 pull requests Successful merges: - #142300 (Disable `tests/run-make/mte-ffi` because no CI runners have MTE extensions enabled) - #143271 (Store the type of each GVN value) - #143280 (Remove duplicate error about raw underscore lifetime) - #143649 (Add test for `default_field_values` and `const_default`) - #143699 (Make `AsyncDrop` check that it's being implemented on a local ADT) - #143719 (Emit warning when there is no space between `-o` and arg) - #143833 (Ban projecting into SIMD types [MCP838]) - #143891 (Port `#[coverage]` to the new attribute system) - #143908 (`tests/ui`: A New Order [0/28] ) - #143909 (docs(alloc::fmt): Make type optional, instead of matching empty string) - #143925 (Make slice comparisons const) - #143997 (Use $crate in macros for rustc_public (aka stable_mir)) r? `@ghost` `@rustbot` modify labels: rollup

Store the type of each GVN value MIR is fully typed, so type information is an integral part of what defines a value. GVN currently tries to circumvent storing types, which creates all sorts of complexities. This PR stores the type along with the enum `Value` when defining a value index. This allows to simplify a lot of code. Fixes rust-lang#128094 Fixes rust-lang#135128 r? ``````@ghost`````` for perf

Rollup of 11 pull requests Successful merges: - #142300 (Disable `tests/run-make/mte-ffi` because no CI runners have MTE extensions enabled) - #143271 (Store the type of each GVN value) - #143293 (fix `-Zsanitizer=kcfi` on `#[naked]` functions) - #143719 (Emit warning when there is no space between `-o` and arg) - #143846 (pass --gc-sections if -Zexport-executable-symbols is enabled and improve tests) - #143891 (Port `#[coverage]` to the new attribute system) - #143967 (constify `Option` methods) - #144008 (Fix false positive double negations with macro invocation) - #144010 (Boostrap: add warning on `optimize = false`) - #144049 (rustc-dev-guide subtree update) - #144056 (Copy GCC sources into the build directory even outside CI) r? `@ghost` `@rustbot` modify labels: rollup

rustbot added the T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. label Jul 1, 2025

This comment has been minimized.

Sign in to view

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Jul 1, 2025

This comment has been minimized.

Sign in to view

cjgillot added 5 commits July 1, 2025 11:37

Simplify assignments.

fba4177

Introduce Value::RawPtr as it behaves differently from other aggregates.

ac70dc8

Store a full Ty with each Value.

be41333

Remove extraneous types.

4750dff

Avoid computing layouts inside coroutines.

cbc3cf7

cjgillot added the A-mir-opt-GVN Area: MIR opt Global Value Numbering (GVN) label Jul 1, 2025

cjgillot force-pushed the gvn-types branch from 2684875 to cbc3cf7 Compare July 2, 2025 09:31

This comment has been minimized.

Sign in to view

Move crashes tests.

4a95b16

rustbot removed the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Jul 2, 2025

cjgillot marked this pull request as ready for review July 3, 2025 09:56

rustbot added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jul 3, 2025

rustbot assigned SparrowLii Jul 3, 2025

rustbot assigned oli-obk Jul 7, 2025

rustbot unassigned SparrowLii Jul 7, 2025

oli-obk approved these changes Jul 16, 2025

View reviewed changes

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 Jul 16, 2025

fmease mentioned this pull request Jul 17, 2025

Rollup of 16 pull requests #144067

Closed

fmease mentioned this pull request Jul 17, 2025

Rollup of 15 pull requests #144071

Closed

matthiaskrgr mentioned this pull request Jul 17, 2025

Rollup of 12 pull requests #144075

Closed

jhpratt mentioned this pull request Jul 18, 2025

Rollup of 12 pull requests #144101

Closed

matthiaskrgr mentioned this pull request Jul 18, 2025

Rollup of 11 pull requests #144109

Open

Store the type of each GVN value #143271

Are you sure you want to change the base?

Store the type of each GVN value #143271

Conversation

cjgillot commented Jul 1, 2025 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cjgillot commented Jul 1, 2025

Uh oh!

This comment has been minimized.

bors commented Jul 1, 2025

Uh oh!

This comment has been minimized.

bors commented Jul 1, 2025

Uh oh!

This comment has been minimized.

Kobzol commented Jul 1, 2025

Uh oh!

This comment has been minimized.

rust-timer commented Jul 2, 2025

Overall result: no relevant changes - no action needed

Instruction count

Max RSS (memory usage)

Cycles

Binary size

Uh oh!

rustbot commented Jul 3, 2025

Uh oh!

cjgillot commented Jul 3, 2025

Uh oh!

SparrowLii commented Jul 7, 2025

Uh oh!

rustbot commented Jul 7, 2025

Uh oh!

SparrowLii commented Jul 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

oli-obk Jul 16, 2025

Choose a reason for hiding this comment

Uh oh!

oli-obk Jul 16, 2025

Choose a reason for hiding this comment

Uh oh!

oli-obk commented Jul 16, 2025

Uh oh!

bors commented Jul 16, 2025

Uh oh!

Uh oh!

cjgillot commented Jul 1, 2025 •

edited by rustbot

Loading

SparrowLii commented Jul 7, 2025 •

edited

Loading