Add an optional `llvm = ` argument to rustc_intrinsic. by veluca93 · Pull Request #164544 · rust-lang/rust

veluca93 · 2026-10-18T23:00:40Z

This is meant for usage in core::arch, to define the platform-specific LLVM intrinsics that correspond to each intrinsic.

LLM usage: I used an LLM to point me at places in the code from which to start the implementation, but I did not generate any code. The main non-trivial input from the LLM was to prevent inlining of callees annotated with rustc_llvm_intrinsic. Code has been since re-designed following @bjorn3's and @RalfJung's proposals so there's no trace of any LLM input left :-)

r? RalfJ

rustbot · 2026-10-18T23:00:43Z

Some changes occurred in compiler/rustc_attr_ir

cc @jdonszelmann, @JonathanBrouwer

Some changes occurred in compiler/rustc_attr_parsing

cc @jdonszelmann, @JonathanBrouwer

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

Some changes occurred in compiler/rustc_passes/src/check_attr.rs

cc @jdonszelmann, @JonathanBrouwer

rustbot · 2026-10-18T23:00:46Z

Error: Failed to set assignee to RalfJ: 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.

Please file an issue on GitHub at triagebot if there's a problem with this bot, or reach out on #triagebot on Zulip.

veluca93 · 2026-10-18T23:01:28Z

r? RalfJung

rustbot · 2026-10-18T23:01:32Z

RalfJung is not on the review rotation at the moment.
They may take a while to respond.

tgross35 · 2026-10-18T23:21:04Z

Cc @folkertdev

bjorn3 · 2026-10-19T08:19:04Z

Can you check that all places in the compiler that check for extern "llvm-intrinsic" have been adjusted as appropriate?

veluca93 · 2026-10-19T22:27:46Z

Can you check that all places in the compiler that check for extern "llvm-intrinsic" have been adjusted as appropriate?

I gave a look, and I think so, but I am definitely not knowledgeable enough of compiler internals to be sure.

rustbot · 2026-10-28T12:15:53Z

rustc_codegen_gcc is developed in its own repository. If possible, consider making this change to rust-lang/rustc_codegen_gcc instead.

cc @antoyo, @GuillaumeGomez

Some changes occurred to constck

cc @fee1-dead

Some changes occurred to the CTFE machinery

cc @RalfJung, @oli-obk, @lcnr

rustc_codegen_cranelift is developed in its own repository. If possible, consider making this change to rust-lang/rustc_codegen_cranelift instead.

cc @bjorn3

rustbot · 2026-10-28T19:45:14Z

clippy is developed in its own repository. If possible, consider making this change to rust-lang/rust-clippy instead.

cc @rust-lang/clippy

veluca93 · 2026-11-06T00:06:59Z

I haven't looked too much at the implementation, just the surface area. This looks pretty mechanical (why doesn't GH show a version of git diff --color-moved) but I'm a bit behind on reviews, so perhaps

r? folkertdev

FYI after applying Ralf's suggestion git/github is much happier with the diff, and shows it as a change with a size much more representative of the actual size :-)

rustbot · 2026-11-06T00:07:02Z

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.

tgross35

I guess you're right, this isn't actually that bad. I'll yoink it back.

r? me

View changes since this review

tgross35 · 2026-11-06T07:36:31Z


-                // Intrinsics are language primitives, not regular calls, so treat them separately.
-                if let Some(intrinsic) = tcx.intrinsic(callee) {
+                // Non-backend intrinsics are language primitives, not regular calls, so treat them separately.


I think my idea was that backend intrinsics would always have a fallback body, but that might not necessarily be the case so... what do you think should happen here?

I think that as soon as this lands, we may as well start switching extern "llvm-intrinsic" blocks over - none of those have fallbacks.

Please include a test making sure we enforce constness matching up too.

rustbot · 2026-11-06T08:05:35Z

Reminder, once the PR becomes ready for a review, use @rustbot ready.

veluca93 · 2026-11-06T10:18:14Z

@rustbot ready

RalfJung · 2026-11-06T11:21:38Z

+#[rustc_intrinsic(llvm = "llvm.fshl.i64")]
+fn fshl(a: i64, b: i64, sh: i64) -> i64 {
+    (((a as i128) << 64) | (b as i128) >> sh) as i64
+}
+
+fn main() {
+    const { fshl(1, 2, 3) };
+    //~^ ERROR: cannot call non-const function `fshl` in constants [E0015]
+}


Not sure if this is worth a test, it's just standard const checking. Would be very surprising if we allowed calling a function that's not marked as const.

More interesting is ensuring that const stability checks work properly. You could extend tests/ui/consts/const-unstable-intrinsic.rs to also check backend intrinsics.

View changes since the review

Removed, and added more tests to the other file.

RalfJung · 2026-11-06T14:15:49Z

Btw the PR title and description seem to still refer to an earlier version of this PR.

rustbot · 2026-11-06T14:34:38Z

miri is developed in its own repository. If the Miri part of this change can be broken out, consider making this change to rust-lang/miri instead. However, if Miri needs adjusting for rustc changes, just ignore this message.

cc @rust-lang/miri

RalfJung · 2026-11-07T09:51:54Z


 impl ValidityRequirement {
-    pub fn from_intrinsic(intrinsic: Symbol) -> Option<Self> {
+    pub fn from_intrinsic(intrinsic: Intrinsic) -> Option<Self> {


This is strictly about Rust intrinsics so I don't think it makes sense to change the type here.

View changes since the review

RalfJung · 2026-11-07T09:56:25Z

I was surprised that the interpreter didn't need any patch here. It turns out that the interpreter has relied on the item_name rather than the name recorded in the intrinsic data, which is fine today as the two are equal, but is wrong after this PR.

So this code

rust/src/tools/miri/src/intrinsics/mod.rs

Line 54 in 101dc61

let intrinsic_name = this.tcx.item_name(instance.def_id());

and this

rust/compiler/rustc_const_eval/src/const_eval/machine.rs

Line 489 in 79bcc9d

let intrinsic_name = ecx.tcx.item_name(instance.def_id());

will need adjustment.

View changes since the review

RalfJung · 2026-11-07T09:58:48Z

There's some logic in this file that seems wrong now

rust/compiler/rustc_codegen_cranelift/src/intrinsics/mod.rs

Line 287 in 79bcc9d

let intrinsic = fx.tcx.item_name(instance.def_id());

View changes since the review

RalfJung · 2026-11-07T09:59:17Z

There's logic in this crate that seems wrong now

rust/compiler/rustc_codegen_gcc/src/intrinsic/mod.rs

Line 151 in 81330f9

let name = tcx.item_name(instance.def_id());

View changes since the review

RalfJung · 2026-11-07T10:01:15Z

+#[unstable(feature = "unstable", issue = "42")]
+#[rustc_intrinsic(llvm = "llvm.fshl.i64")]
+pub const unsafe fn fake_size_of_val_llvm<T>(x: *const T) -> usize {
+    0
+}


Please keep the test consistent, and have no fallback body here.

I'd suggest calling it llvm_fshl or so, that seems less confusing.

View changes since the review

RalfJung · 2026-11-07T10:01:32Z

+pub const unsafe fn fake_align_of_val_llvm<T>(x: *const T) -> usize {
+    0
+}


Same here, please remove the fallback body (and pick a better name)

View changes since the review

RalfJung · 2026-11-07T10:01:55Z

+pub const unsafe fn fake_align_of_val_llvm<T>(x: *const T) -> usize {
+    0
+}


also remove this fallback body (and pick a better name)

View changes since the review

rustbot assigned RalfJung Oct 18, 2026

tgross35 reviewed Oct 18, 2026

View reviewed changes

Comment thread tests/ui/intrinsics/rustc-llvm-intrinsic.rs Resolved

veluca93 force-pushed the rustc_llvm_intrinsic branch from 39047e6 to a0ad71b Compare October 19, 2026 07:09

bjorn3 reviewed Oct 19, 2026

View reviewed changes

Comment thread compiler/rustc_codegen_ssa/src/mir/block.rs Resolved

veluca93 marked this pull request as draft October 23, 2026 15:39

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Oct 23, 2026

veluca93 force-pushed the rustc_llvm_intrinsic branch 2 times, most recently from ebce41f to 90416b5 Compare October 28, 2026 12:15

veluca93 marked this pull request as ready for review October 28, 2026 12:15

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Oct 28, 2026

This comment has been minimized.

Sign in to view

bjorn3 reviewed Oct 28, 2026

View reviewed changes

Comment thread compiler/rustc_public_bridge/src/context/impls.rs Resolved

veluca93 force-pushed the rustc_llvm_intrinsic branch from 90416b5 to c906c98 Compare October 28, 2026 19:45

rustbot added the T-clippy Relevant to the Clippy team. label Oct 28, 2026

tgross35 reviewed Oct 28, 2026

View reviewed changes

Comment thread tests/ui/intrinsics/rustc-llvm-intrinsic.rs Resolved

tgross35 requested changes Nov 6, 2026

View reviewed changes

rustbot assigned tgross35 and unassigned folkertdev Nov 6, 2026

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Nov 6, 2026

veluca93 force-pushed the rustc_llvm_intrinsic branch from e294fce to e18a91c Compare November 6, 2026 10:17

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Nov 6, 2026

This comment has been minimized.

Sign in to view

veluca93 force-pushed the rustc_llvm_intrinsic branch from e18a91c to f28292b Compare November 6, 2026 10:26

This comment has been minimized.

Sign in to view

veluca93 force-pushed the rustc_llvm_intrinsic branch from f28292b to 40c3e33 Compare November 6, 2026 10:35

RalfJung reviewed Nov 6, 2026

View reviewed changes

Comment thread tests/ui/intrinsics/rustc-llvm-intrinsic.rs Resolved

Comment thread tests/ui/intrinsics/rustc-llvm-intrinsic.rs Resolved

veluca93 force-pushed the rustc_llvm_intrinsic branch from 40c3e33 to 62cfa98 Compare November 6, 2026 14:34

veluca93 changed the title ~~Add a rustc_llvm_intrinsic attribute.~~ Add an optional llvm = argument to rustc_intrinsic. Nov 6, 2026

Implement rustc_intrinsic(llvm = "...")

e6d0471

veluca93 force-pushed the rustc_llvm_intrinsic branch from 62cfa98 to e6d0471 Compare November 6, 2026 15:06

RalfJung reviewed Nov 7, 2026

View reviewed changes

Comment thread src/tools/miri/tests/pass/intrinsics/backend.rs Resolved

RalfJung reviewed Nov 7, 2026

View reviewed changes

Conversation

veluca93 commented Oct 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Oct 18, 2026

Uh oh!

rustbot commented Oct 18, 2026

Uh oh!

veluca93 commented Oct 18, 2026

Uh oh!

rustbot commented Oct 18, 2026

Uh oh!

Uh oh!

tgross35 commented Oct 18, 2026

Uh oh!

bjorn3 commented Oct 19, 2026

Uh oh!

Uh oh!

veluca93 commented Oct 19, 2026

Uh oh!

rustbot commented Oct 28, 2026

Uh oh!

This comment has been minimized.

Uh oh!

rustbot commented Oct 28, 2026

Uh oh!

Uh oh!

veluca93 commented Nov 6, 2026

Uh oh!

rustbot commented Nov 6, 2026

Uh oh!

tgross35 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!

Uh oh!

Uh oh!

Uh oh!

tgross35 Nov 6, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

rustbot commented Nov 6, 2026

Uh oh!

veluca93 commented Nov 6, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

RalfJung Nov 6, 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!

veluca93 Nov 6, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

RalfJung commented Nov 6, 2026

Uh oh!

rustbot commented Nov 6, 2026

Uh oh!

Uh oh!

RalfJung Nov 7, 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!

RalfJung Nov 7, 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!

RalfJung Nov 7, 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!

RalfJung Nov 7, 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!

veluca93 commented Oct 18, 2026 •

edited

Loading

tgross35 left a comment •

edited by rustbot

Loading

RalfJung Nov 6, 2026 •

edited by rustbot

Loading

RalfJung Nov 7, 2026 •

edited by rustbot

Loading

RalfJung Nov 7, 2026 •

edited by rustbot

Loading

RalfJung Nov 7, 2026 •

edited by rustbot

Loading

RalfJung Nov 7, 2026 •

edited by rustbot

Loading

RalfJung Nov 7, 2026 •

edited by rustbot

Loading

RalfJung Nov 7, 2026 •

edited by rustbot

Loading

RalfJung Nov 7, 2026 •

edited by rustbot

Loading