Add a `layout_of_val` intrinsic for when you want both `size_of_val`+`align_of_val` by scottmcm · Pull Request #152786 · rust-lang/rust

scottmcm · 2026-02-18T06:58:49Z

EDIT: github stopped picking up pushes here, so moved to #152867

Today we're emitting a bunch more MIR and LLVM-IR than we need for stuff like this.

Closes #152641 because this solves the motivating example from that PR of avoiding the alignment transmutes when getting a layout from a DST, just in a better way.

While I was in the area I also put a more specific alignment !range metadata on loading alignment from a vtable and some extra no-wrap flags when calculating the size of an ADT with an unsized tail.

scottmcm · 2026-02-18T07:06:36Z

@bors try @rust-timer queue

Add a `layout_of_val` intrinsic for when you want both `size_of_val`+`align_of_val`

It was already computing both anyway with `size_of_val::size_and_align_of_dst`, so we just need to return them both instead of ignoring one.

rust-bors · 2026-02-18T09:18:44Z

☀️ Try build successful (CI)
Build commit: f751a56 (f751a568aedd4c46059df62535b974a66346fd29, parent: 8387095803f21a256a9a772ac1f9b41ed4d5aa0a)

rust-timer · 2026-02-18T09:58:56Z

Finished benchmarking commit (f751a56): comparison URL.

Overall result: ✅ improvements - 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

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
Improvements ✅ (primary)	-0.4%	[-0.5%, -0.3%]	2
Improvements ✅ (secondary)	-1.1%	[-3.1%, -0.4%]	14
All ❌✅ (primary)	-0.4%	[-0.5%, -0.3%]	2

Max RSS (memory usage)

Results (primary 0.2%, secondary -2.7%)

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

	mean	range	count
Regressions ❌ (primary)	2.8%	[2.0%, 3.5%]	2
Regressions ❌ (secondary)	1.7%	[1.7%, 1.7%]	1
Improvements ✅ (primary)	-2.4%	[-2.9%, -2.0%]	2
Improvements ✅ (secondary)	-3.3%	[-5.5%, -2.2%]	8
All ❌✅ (primary)	0.2%	[-2.9%, 3.5%]	4

Cycles

Results (primary -2.1%, secondary -3.5%)

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)	-2.1%	[-2.1%, -2.1%]	1
Improvements ✅ (secondary)	-3.5%	[-3.5%, -3.5%]	1
All ❌✅ (primary)	-2.1%	[-2.1%, -2.1%]	1

Binary size

Results (primary -0.4%, secondary -1.6%)

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
Improvements ✅ (primary)	-0.4%	[-0.8%, -0.0%]	25
Improvements ✅ (secondary)	-1.6%	[-3.9%, -0.3%]	13
All ❌✅ (primary)	-0.4%	[-0.8%, -0.0%]	25

Bootstrap: 480.396s -> 480.025s (-0.08%)
Artifact size: 397.88 MiB -> 397.84 MiB (-0.01%)

scottmcm · 2026-02-18T18:01:46Z

@bors try @rust-timer queue

Add a `layout_of_val` intrinsic for when you want both `size_of_val`+`align_of_val`

rust-bors · 2026-02-18T20:13:44Z

☀️ Try build successful (CI)
Build commit: 20e17c0 (20e17c08eb05c07af1938fee20fae4fb3351712d, parent: c043085801b7a884054add21a94882216df5971c)

scottmcm · 2026-02-19T06:15:01Z

I'd tried something else that just didn't work, so I reverted back to what got the previous fully-green perf run (#152786 (comment)).

rustbot · 2026-02-19T06:15:17Z

Some changes occurred to the intrinsics. Make sure the CTFE / Miri interpreter
gets adapted for the changes, if necessary.

cc @rust-lang/miri, @RalfJung, @oli-obk, @lcnr

⚠️ #[miri::intrinsic_fallback_is_spec] must only be used if the function actively checks for all UB cases,
and explores the possible non-determinism of the intrinsic.

cc @rust-lang/miri

rustbot · 2026-02-19T06:15:19Z

r? @JonathanBrouwer

rustbot has assigned @JonathanBrouwer.
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: codegen, compiler
codegen, compiler expanded to 68 candidates
Random selection from 15 candidates

RalfJung · 2026-02-19T07:29:07Z

compiler/rustc_codegen_ssa/src/size_of_val.rs

-            // Alignment is always nonzero.
-            bx.range_metadata(align, WrappingRange { start: 1, end: !0 });
+            // Alignment is always a power of two, thus 1..=0x800…000.
+            let align_bound = size_bound + 1;


In fact the bound is Align::MAX, is it worth emitting that?

Oh, really good point 👍 I was stuck still thinking of it as a load of ptr::Alignment, but it's not.

I don't know how much it'll really matter, since isize::MAX+1 is enough to prove that size+align doesn't unsigned-overflow, but might as well include it because it's easy, cheap, clearly not worse, and makes the test verification check slightly nicer.

RalfJung · 2026-02-19T07:32:37Z

compiler/rustc_codegen_ssa/src/mir/intrinsic.rs

+                    sym::size_of_val => llsize,
+                    sym::align_of_val => llalign,
+                    sym::layout_of_val => {
+                        // Use the builder so we're insulated from the in-memory field order


We're still relying on the in-source field order here. That deserves a comment on both ends (here and in the library). Or can the field lookup here be done by name?

…lignment from a vtable

…udly and clearly ICE and fail library tests.

scottmcm · 2026-02-19T20:57:54Z

Something really odd is happening here where github is noticing the pushes in my fork but not in this pr.

turning it off and on again...

rust-log-analyzer · 2026-02-19T21:04:22Z

The job tidy failed! Check out the build log: (web) (plain enhanced) (plain)

Click to see the possible cause of the failure (guessed by this bot)

finished building tool typos
error: `implemention` should be `implementation`
   ╭▸ library/core/src/alloc/layout.rs:29:16
   │
29 │ // BEWARE! The implemention of the `layout_of_val` intrinsic is coupled to the
   ╰╴               ━━━━━━━━━━━━
rerun tidy with `--extra-checks=spellcheck --bless` to fix typos
tidy [extra_checks]: checks with external tool 'typos' failed
tidy [extra_checks]: FAIL
tidy: The following check failed: extra_checks
Bootstrap failed while executing `test src/tools/tidy tidyselftest --extra-checks=py,cpp,js,spellcheck`
Command `/checkout/obj/build/x86_64-unknown-linux-gnu/stage1-tools-bin/rust-tidy --root-path=/checkout --cargo-path=/checkout/obj/build/x86_64-unknown-linux-gnu/stage0/bin/cargo --output-dir=/checkout/obj/build --concurrency=4 --npm-path=/node/bin/yarn --extra-checks=py,cpp,js,spellcheck` failed with exit code 1
Created at: src/bootstrap/src/core/build_steps/tool.rs:1612:23
Executed at: src/bootstrap/src/core/build_steps/test.rs:1365:29

Command has failed. Rerun with -v to see more details.
Build completed unsuccessfully in 0:02:38
  local time: Thu Feb 19 21:02:30 UTC 2026
  network time: Thu, 19 Feb 2026 21:02:30 GMT
##[error]Process completed with exit code 1.
##[group]Run echo "disk usage:"