Add fallback for `intrinsics::fabs` by N1ark · Pull Request #159605 · rust-lang/rust

N1ark · 2026-07-20T12:57:47Z

Add a fallback for the fabs intrinsic. Because it uses const traits and const ops it required adding rustc_const_unstable -- i'm not certain it's right but it seems to make sense. I also added #[miri::intrinsic_fallback_is_spec], though im not sure if I need to modify the const eval code to remove fabs handling or to remove it from any of the backends

r? folkertdev

rustbot · 2026-07-20T12:57:50Z

Any special-casing of Miri in the standard library requires review.

cc @rust-lang/miri

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

folkertdev

I'm still skeptical that this is an improvement over the naive "just duplicate it 4 times" approach.

View changes since this review

folkertdev · 2026-07-20T13:42:05Z

+#[miri::intrinsic_fallback_is_spec]
+pub const fn fabs<T: bounds::FloatPrimitive>(x: T) -> T {
+    // SAFETY: the implementation of `FloatPrimitive` guarantees that `T` is a
+    // floating-point type, and that `T::Unsigned` is the corresponding unsigned
+    // integer type with the same size as `T`. Both are valid for all bit patterns.
+    let bits: T::Unsigned = unsafe { transmute_unchecked(x) };
+    let abs = bits & !T::SIGN_MASK;
+    // SAFETY: same as above
+    unsafe { transmute_unchecked(abs) }
+}


@RalfJung does this spark joy? I'd much rather have the one-liner duplicated 4 times than the (trivial, but still) unsafety and const trait stuff that is required.

The question is whether it sparks more or less joy than copy-paste. The status quo is definitely not joyful. ;)

But yeah I agree needing unsafe is pretty bad.

does it spark joy now ? ^-^

pub const fn fabs<T: const bounds::FloatPrimitive>(x: T) -> T { T::from_bits(x.to_bits() & !T::SIGN_MASK) }

Yes that seems exactly as readable as the per-type version, just less duplicated :)

N1ark · 2026-07-20T16:35:10Z

+#[rustc_const_unstable(feature = "core_intrinsics", issue = "none")]
+const unsafe impl FloatPrimitive for f16 {
+    type Unsigned = u16;
+    const SIGN_MASK: Self::Unsigned = f16::SIGN_MASK;
+    fn to_bits(self) -> Self::Unsigned {
+        f16::to_bits(self)
+    }
+    fn from_bits(bits: Self::Unsigned) -> Self {
+        f16::from_bits(bits)
+    }
+}


this can be macro'd if you'd rather i have no strong opinions on that

View changes since the review

Let's hold off on using macros, this seems manageable now and a macro won't work for fallbacks that need a per-type implementation (e.g. sin or fma)

folkertdev

So how will this work for the heterogeneous fallbacks (like fma)? Just add more methods to the trait?

View changes since this review

folkertdev · 2026-07-20T16:45:47Z

+#[rustc_const_unstable(feature = "core_intrinsics", issue = "none")]
+const unsafe impl FloatPrimitive for f16 {
+    type Unsigned = u16;
+    const SIGN_MASK: Self::Unsigned = f16::SIGN_MASK;
+    fn to_bits(self) -> Self::Unsigned {
+        f16::to_bits(self)
+    }
+    fn from_bits(bits: Self::Unsigned) -> Self {
+        f16::from_bits(bits)
+    }
+}


Let's hold off on using macros, this seems manageable now and a macro won't work for fallbacks that need a per-type implementation (e.g. sin or fma)

RalfJung · 2026-07-20T16:54:54Z

So how will this work for the heterogeneous fallbacks (like fma)? Just add more methods to the trait?

Either that, more a per-operation trait defined locally by a macro.

N1ark · 2026-07-20T17:05:35Z

edit: Ralf beat me to it

So how will this work for the heterogeneous fallbacks (like fma)? Just add more methods to the trait?

for those i think what's easiest (which @RalfJung ) commented on Zulip. i dont think i would format it like that, but something along the lines of

intrinsic_dispatch_on_type! {
    pub fn fma<T: bounds::FloatPrimitive>(x: T, y: T, z: T) -> T;
    f16 => {
      fma(x as f64, y as f64, z as f64) as f16
    }
    f32 => { 
      libm::fmaf(x, y, z)
    }
    // etc
}

which then would create a trait for that specific intrinsic that does the thing, adds it to the trait bound and what not? or add it to the FloatPrimitive trait directly, i dont feel strongly abt this

yes it duplicates the impl but it still allows a pretty good cleanup in the rest of the compiler which personally i think is worth the code here (also avoids duplicating comments which can be quite long and technical:) )

folkertdev · 2026-07-22T18:20:41Z

Well let's see how this works out in practice. One thing to note is that the GCC and Cranelift backends don't yet make use of some of the newly-added default (e.g. rust-lang/rustc_codegen_cranelift#1675 can hopefully land tomorrow when nightly include the portable_simd sync) so there may be merge conflicts if you make more progress here in the coming days.

@bors r+

rust-bors · 2026-07-22T18:20:44Z

📌 Commit 74fb1f9 has been approved by folkertdev

It is now in the queue for this repository.

Add fallback for `intrinsics::fabs` Add a fallback for the `fabs` intrinsic. Because it uses const traits and const ops it required adding `rustc_const_unstable` -- i'm not certain it's right but it seems to make sense. I also added `#[miri::intrinsic_fallback_is_spec]`, though im not sure if I need to modify the const eval code to remove `fabs` handling or to remove it from any of the backends r? folkertdev

Rollup of 8 pull requests Successful merges: - #159504 (Abort const-eval queries early when there are generics in the type) - #159523 (std: fix Xous UDP recv length over-report and OOB panic) - #159605 (Add fallback for `intrinsics::fabs`) - #159699 (bump std libc to 0.2.189) - #159306 (Add new variant to iterating-updating-mutref borrowck test) - #159346 (fs::hard_link: use linkat on Android) - #159513 (Consider `()` as suspicious only when expecting `!` for runtime symbols) - #159734 (Document the link_section attribute)

Rollup merge of #159605 - N1ark:fabs-fallback, r=folkertdev Add fallback for `intrinsics::fabs` Add a fallback for the `fabs` intrinsic. Because it uses const traits and const ops it required adding `rustc_const_unstable` -- i'm not certain it's right but it seems to make sense. I also added `#[miri::intrinsic_fallback_is_spec]`, though im not sure if I need to modify the const eval code to remove `fabs` handling or to remove it from any of the backends r? folkertdev

Rollup of 8 pull requests Successful merges: - rust-lang/rust#159504 (Abort const-eval queries early when there are generics in the type) - rust-lang/rust#159523 (std: fix Xous UDP recv length over-report and OOB panic) - rust-lang/rust#159605 (Add fallback for `intrinsics::fabs`) - rust-lang/rust#159699 (bump std libc to 0.2.189) - rust-lang/rust#159306 (Add new variant to iterating-updating-mutref borrowck test) - rust-lang/rust#159346 (fs::hard_link: use linkat on Android) - rust-lang/rust#159513 (Consider `()` as suspicious only when expecting `!` for runtime symbols) - rust-lang/rust#159734 (Document the link_section attribute)

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 Jul 20, 2026

rustbot assigned folkertdev Jul 20, 2026

folkertdev reviewed Jul 20, 2026

View reviewed changes

N1ark force-pushed the fabs-fallback branch from fd18f83 to baaa071 Compare July 20, 2026 16:32

N1ark commented Jul 20, 2026

View reviewed changes

RalfJung reviewed Jul 20, 2026

View reviewed changes

Comment thread library/core/src/intrinsics/bounds.rs Outdated

folkertdev reviewed Jul 20, 2026

View reviewed changes

Add fallback for fabs

74fb1f9

N1ark force-pushed the fabs-fallback branch from baaa071 to 74fb1f9 Compare July 20, 2026 17:07

rust-bors Bot 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 22, 2026

JonathanBrouwer mentioned this pull request Jul 22, 2026

Rollup of 7 pull requests #159741

Closed

JonathanBrouwer mentioned this pull request Jul 22, 2026

Rollup of 7 pull requests #159742

Closed

This was referenced Jul 22, 2026

Rollup of 9 pull requests #159747

Closed

Rollup of 8 pull requests #159748

Merged

rust-bors Bot merged commit c5f9a51 into rust-lang:main Jul 23, 2026
13 checks passed

rustbot added this to the 1.99.0 milestone Jul 23, 2026

Uh oh!

Uh oh!

Conversation

N1ark commented Jul 20, 2026

Uh oh!

rustbot commented Jul 20, 2026

Uh oh!

folkertdev left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

folkertdev Jul 20, 2026

Choose a reason for hiding this comment

Uh oh!

RalfJung Jul 20, 2026

Choose a reason for hiding this comment

Uh oh!

N1ark Jul 20, 2026

Choose a reason for hiding this comment

Uh oh!

RalfJung Jul 20, 2026

Choose a reason for hiding this comment

Uh oh!

N1ark Jul 20, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

folkertdev Jul 20, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

folkertdev left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

folkertdev Jul 20, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

RalfJung commented Jul 20, 2026

Uh oh!

N1ark commented Jul 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

edit: Ralf beat me to it

Uh oh!

folkertdev commented Jul 22, 2026

Uh oh!

rust-bors Bot commented Jul 22, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

folkertdev left a comment •

edited by rustbot

Loading

N1ark Jul 20, 2026 •

edited by rustbot

Loading

folkertdev left a comment •

edited by rustbot

Loading

N1ark commented Jul 20, 2026 •

edited

Loading