Mark NonNull read family and offset_from as inline(always) by zakrad · Pull Request #161004 · rust-lang/rust

zakrad · 2026-08-12T19:57:22Z

These are one line wrappers that just forward to the raw pointer funcs that are all #[inline(always)], as are NonNull's own write and offset, The read and offset_from were left as plain #[inline]

read, read_volatile, read_unaligned, and offset_from are the only NonNull pointer methods still marked inline; the other 17 are inline(always), as are the raw const/mut pointer equivalents. They are one-line delegating wrappers, so make them consistent -- they now inline in debug like their siblings.

rustbot · 2026-08-12T19:57:26Z

r? @clarfonthey

rustbot has assigned @clarfonthey.
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

Why was this reviewer chosen?

The reviewer was selected based on:

Owners of files modified in this PR: libs
libs expanded to 12 candidates
Random selection from JohnTitor, Mark-Simulacrum, clarfonthey, nia-e

clarfonthey · 2026-08-12T20:06:28Z

Do you have any particular evidence this helps? I figure it probably does, but I could be wrong.

zakrad · 2026-08-12T20:44:48Z

Do you have any particular evidence this helps? I figure it probably does, but I could be wrong.

The raw pointer funcs these wrap are #[inline(always)] #160816 NonNull read + offset_from left out, codegen wise the difference shows in debug, at -Copt-level=0 #[inline] stays an out of line call, while #[inline(always)] gets inlined, its not a release perf change but consistency plus tiny unoptimized codegen

clarfonthey · 2026-08-13T12:12:30Z

Right, I know what inline(always) does, but I mostly am asking whether this actually noticeably improves performance without affecting build times, since that's not always clear.

zakrad · 2026-08-13T15:15:52Z

No evidence it does, it was mostly for consistency, we can have perf run tho

clarfonthey · 2026-08-18T20:21:20Z

@bors try

Mark NonNull read family and offset_from as inline(always)

clarfonthey · 2026-08-18T20:21:33Z

@rust-timer queue

rust-bors · 2026-08-18T21:21:04Z

☀️ Try build successful (CI)
Build commit: 0a988f7 (0a988f73b92c408de20490bfee5e5a2c20460e80)
Base parent: e71c0f1 (e71c0f1e3395b10a8c331317be1a5c107bdf7b2e)

rust-timer · 2026-08-18T22:03:30Z

Finished benchmarking commit (0a988f7): 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 0.6%)

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)	0.6%	[0.6%, 0.6%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Cycles

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

Binary size

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

Bootstrap: 456.919s -> 458.765s (0.40%)
Artifact size: 398.99 MiB -> 399.01 MiB (0.01%)

clarfonthey · 2026-08-18T23:12:14Z

@rust-timer build 0a988f7 profiles=debug

rust-timer · 2026-08-18T23:12:17Z

Cannot attach SHAs to try benchmark request on PR 161004 and SHA 0a988f7: failed to attach SHAs to try benchmark request

Caused by:
0: db error
1: ERROR: duplicate key value violates unique constraint "benchmark_request_tag_key"
DETAIL: Key (tag)=(0a988f7) already exists.

clarfonthey · 2026-08-18T23:13:24Z

@bors try
@rust-timer queue profiles=debug

Mark NonNull read family and offset_from as inline(always)

clarfonthey · 2026-08-18T23:14:00Z

@rust-timer build 93f964f profiles=debug

rust-timer · 2026-08-18T23:14:03Z

Missing artifact for sha 93f964f3b6c9a41e2cc599d2699f5e1aab15459f (https://ci-artifacts.rust-lang.org/rustc-builds/93f964f3b6c9a41e2cc599d2699f5e1aab15459f/rustc-nightly-x86_64-unknown-linux-gnu.tar.xz); not built yet, try again later.

rust-bors · 2026-08-19T00:18:33Z

☀️ Try build successful (CI)
Build commit: 5e7f312 (5e7f312343b2fce41af58de929a6042926fb6715)
Base parent: e71c0f1 (e71c0f1e3395b10a8c331317be1a5c107bdf7b2e)

rust-timer · 2026-08-19T01:10:30Z

Finished benchmarking commit (5e7f312): 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 (primary 1.8%, secondary 2.2%)

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

	mean	range	count
Regressions ❌ (primary)	1.8%	[1.3%, 2.2%]	4
Regressions ❌ (secondary)	2.2%	[1.1%, 4.1%]	11
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.8%	[1.3%, 2.2%]	4

Cycles

Results (secondary 11.3%)

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)	11.3%	[11.3%, 11.3%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

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

Bootstrap: 456.919s -> 457.847s (0.20%)
Artifact size: 398.99 MiB -> 399.66 MiB (0.17%)

clarfonthey · 2026-08-19T02:46:58Z

Unless you have any actual benchmarks to share I'm just going to close this tbh

zakrad · 2026-08-19T18:45:10Z

no benchmarks, fine to close

rustbot assigned clarfonthey Aug 12, 2026

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Aug 12, 2026