Applying #[link_section = ".mysection"] to a function in an impl is now a warning with potential error in the future? #147411
Description
Code
I tried this code:
#[derive(Default)]
pub struct Abc123 {
value: u32,
}
impl Abc123 {
#[no_mangle]
#[link_section = ".mysection"]
fn inside_abc_123(&self) {
println!("Hello World!: {}", self.value);
}
}
#[no_mangle]
#[link_section = ".mysection"]
fn should_always_link() {
println!("Hello World!");
}
fn main() {
// This should be in the .mysection section
should_always_link();
// This too?
Abc123::default().inside_abc_123();
}I expected to see this happen: Code should compile without warnings and the functions should be in .mysection.
Instead, this happened:
➜ rust-playground git:(master) ✗ cargo build
Compiling rust-playground v0.1.0 (/mnt/c/projects/rust-playground)
warning: `#[link_section]` attribute cannot be used on inherent methods
--> src/main.rs:36:5
|
36 | #[link_section = ".mysection"]
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
= warning: this was previously accepted by the compiler but is being phased out; it will become a hard error in a future release!
= help: `#[link_section]` can be applied to statics and functions
= note: `#[warn(unused_attributes)]` (part of `#[warn(unused)]`) on by default
warning: `rust-playground` (bin "rust-playground") generated 1 warnings (run `cargo fix --bin "rust-playground"` to apply 1 suggestions)
Finished `dev` profile [unoptimized + debuginfo] target(s) in 5.53s
➜ rust-playground git:(master) ✗Note: The warning suggests that what I am doing is not working, but this is not true:
➜ rust-playground git:(master) ✗ objdump -t target/debug/rust-playground | grep inside
0000000000056c80 g F .mysection 0000000000000056 inside_abc_123
➜ rust-playground git:(master) ✗ objdump -t target/debug/rust-playground | grep should_always
0000000000056ce0 g F .mysection 0000000000000026 should_always_link
➜ rust-playground git:(master) ✗I would be surprised if this is intentional. Why should a
impl Abc123 {
fn inside_abc_123(&self) {
println!("Hello World!: {}", self.value);
}
}be treated differently from
fn inside_abc_123(this: &Self) {
println!("Hello World!: {}", this.value);
}? Isn't this mostly syntactic sugar?
Version it worked on
It most recently worked on: Rust 1.90
Version with regression
rustc --version --verbose:
rustc 1.92.0-nightly (839222065 2025-10-05)
binary: rustc
commit-hash: 839222065a44ac21c15df68ed2f2f3c0127b0b8e
commit-date: 2025-10-05
host: x86_64-unknown-linux-gnu
release: 1.92.0-nightly
LLVM version: 21.1.2
@rustbot modify labels: +regression-from-stable-to-nightly -regression-untriaged
I think this was introduced with #145085, but it has been quite a while since then, shouldn't this have moved to stable by now?
By the way, ideally I should be able to apply this to any function? Not just foreign ones? The use-case here is that I am working on a library where some functions must be placed in a special section of the binary to improve loading times for these functions.