Skip to contentSkip to conversation

rust: device: allow `dead_code` for `Device<>::parent()`

PATCH In review Miguel Ojeda 3 participants 5 messages Apr 29, 2025

Subsystems (2)
lore
Share & tools
lore

Miguel Ojeda

author maintainer
#0 lore raw

PATCH 1/1 rust: device: allow `dead_code` for `Device<>::parent()`

When CONFIG_AUXILIARY_BUS is disabled, parent() is still dead code:

    error: method `parent` is never used
      --> rust/kernel/device.rs:71:19
       |
    64 | impl<Ctx: DeviceContext> Device<Ctx> {
       | ------------------------------------ method in this implementation
    ...
    71 |     pub(crate) fn parent(&self) -> Option<&Self> {
       |                   ^^^^^^
       |
       = note: `-D dead-code` implied by `-D warnings`
       = help: to override `-D warnings` add `#[allow(dead_code)]`

Thus reintroduce the expect, but now as an allow, and do so as dead_code since that is narrower.

Fixes ce735e73dd59 ("rust: auxiliary: add auxiliary device / driver abstractions")

---
I should have noticed this earlier, sorry. Please feel free to rebase to fix it directly there if you prefer.

 rust/kernel/device.rs | 1 +
 1 file changed, 1 insertion(+)
rust/kernel/device.rs +1 -0
index 40c1f549b0ba..e8ab4b9617db 100644
@@ -68,6 +68,7 @@ pub(crate) fn as_raw(&self) -> *mut bindings::device {
     }
 
     /// Returns a reference to the parent device, if any.
+    #[allow(dead_code)]
     pub(crate) fn parent(&self) -> Option<&Self> {
         // SAFETY:
         // - By the type invariant `self.as_raw()` is always valid.

base-commit: ce735e73dd59b169b877cedd0753297c81c2a091 --
2.49.0

maintainer
#1 lore raw
Quoted 14 lines · Miguel Ojeda

When CONFIG_AUXILIARY_BUS is disabled, parent() is still dead code:

    error: method `parent` is never used
      --> rust/kernel/device.rs:71:19
       |
    64 | impl<Ctx: DeviceContext> Device<Ctx> {
       | ------------------------------------ method in this implementation
    ...
    71 |     pub(crate) fn parent(&self) -> Option<&Self> {
       |                   ^^^^^^
       |
       = note: `-D dead-code` implied by `-D warnings`
       = help: to override `-D warnings` add `#[allow(dead_code)]`
Quoted 5 lines

Thus reintroduce the expect, but now as an allow, and do so as dead_code since that is narrower.

Fixes ce735e73dd59 ("rust: auxiliary: add auxiliary device / driver abstractions")

Good catch, thanks!

I wonder if for such or similar cases we want something like

	macro_rules! cfg_expect {
	    (not($cfg:meta) => $lint:ident, $item:item) => {
		#[cfg(not($cfg))]
		#[expect($lint)]
		$item

		#[cfg($cfg)]
		$item
	    };
	}

which could be used like this:

    cfg_expect!(not(CONFIG_FEATURE) => dead_code,
        pub(crate) fn foo(&self) {
            // noop
        }
    );

Maybe there is a much better solution I don't know about.

#2 lore raw
Quoted 2 lines · Danilo Krummrich

Maybe there is a much better solution I don't know about.

There is cfg_attr for that, and we use it in a couple places, e.g.

#[cfg_attr(not(CONFIG_AUXILIARY_BUS), expect(dead_code))]

But unless we are more or less sure the condition will only ever depend on a single cfg or so, I am not sure if it is worth the complexity (and even then reasonable people could disagree).

I wrote about that approach in the "Conditional compilation" in one of the bullets of:

https://docs.kernel.org/rust/coding-guidelines.html#lints

I hope that helps!

Cheers,
Miguel

maintainer
#3 lore raw
Quoted 7 lines · Miguel Ojeda
Quoted 2 lines · Danilo Krummrich

Maybe there is a much better solution I don't know about.

There is cfg_attr for that, and we use it in a couple places, e.g.

#[cfg_attr(not(CONFIG_AUXILIARY_BUS), expect(dead_code))]

Well, that is much better, thanks!

I prefer this then, I really want to catch if some other code starts using this.

- Danilo

#4 lore raw
Quoted 4 lines · Danilo Krummrich

Well, that is much better, thanks!

I prefer this then, I really want to catch if some other code starts using this.

You're welcome! I will send v2 then, in case it is not "fixup"ed.

Cheers,
Miguel