rust/kernel/devres.rs | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
Devres::new() registers a callback with the C devres subsystem via
devres_node_add(). If the Devres is leaked (e.g. via
core::mem::forget(), which is safe), its Drop impl never runs, and the
devres release callback will revoke the inner Revocable on device
unbind, which drops T in place. If T contains non-'static references,
those may be dangling by that point.
Add a 'static bound to prevent storing types with borrowed data in
Devres.
Fixes: 76c01ded724b ("rust: add devres abstraction")
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
---
rust/kernel/devres.rs | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/rust/kernel/devres.rs b/rust/kernel/devres.rs
index 9e5f93aed20c..2e258d31a45c 100644
--- a/rust/kernel/devres.rs
+++ b/rust/kernel/devres.rs
@@ -122,7 +122,7 @@ struct Inner<T> {
/// # Ok(())
/// # }
/// ```
-pub struct Devres<T: Send> {
+pub struct Devres<T: Send + 'static> {
dev: ARef<Device>,
inner: Arc<Inner<T>>,
}
@@ -184,7 +184,7 @@ pub(super) unsafe fn devres_node_remove(
}
}
-impl<T: Send> Devres<T> {
+impl<T: Send + 'static> Devres<T> {
/// Creates a new [`Devres`] instance of the given `data`.
///
/// The `data` encapsulated within the returned `Devres` instance' `data` will be
@@ -349,7 +349,7 @@ unsafe impl<T: Send> Send for Devres<T> {}
// SAFETY: `Devres` can be shared with any task, if `T: Sync`.
unsafe impl<T: Send + Sync> Sync for Devres<T> {}
-impl<T: Send> Drop for Devres<T> {
+impl<T: Send + 'static> Drop for Devres<T> {
fn drop(&mut self) {
// SAFETY: When `drop` runs, it is guaranteed that nobody is accessing the revocable data
// anymore, hence it is safe not to wait for the grace period to finish.
base-commit: 56785dcb2ef6d3cff82ac33f2e34db94377416a3
--
2.54.0
On Tue, 26 May 2026 02:04:41 +0200, Danilo Krummrich wrote:
> [PATCH] rust: devres: add 'static bound to Devres<T>
Applied, thanks!
Branch: driver-core-testing
Tree: git://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core.git
[1/1] rust: devres: add 'static bound to Devres<T>
commit: 016267b521b1
The patch will appear in the next linux-next integration (typically within 24
hours on weekdays).
The patch is in the driver-core-testing branch and will be promoted to
driver-core-next after validation.
On Tue May 26, 2026 at 1:04 AM BST, Danilo Krummrich wrote:
> Devres::new() registers a callback with the C devres subsystem via
> devres_node_add(). If the Devres is leaked (e.g. via
> core::mem::forget(), which is safe), its Drop impl never runs, and the
> devres release callback will revoke the inner Revocable on device
> unbind, which drops T in place. If T contains non-'static references,
> those may be dangling by that point.
>
> Add a 'static bound to prevent storing types with borrowed data in
> Devres.
The bound should be added on `Devres::new` instead.
Best,
Gary
>
> Fixes: 76c01ded724b ("rust: add devres abstraction")
> Signed-off-by: Danilo Krummrich <dakr@kernel.org>
> ---
> rust/kernel/devres.rs | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/rust/kernel/devres.rs b/rust/kernel/devres.rs
> index 9e5f93aed20c..2e258d31a45c 100644
> --- a/rust/kernel/devres.rs
> +++ b/rust/kernel/devres.rs
> @@ -122,7 +122,7 @@ struct Inner<T> {
> /// # Ok(())
> /// # }
> /// ```
> -pub struct Devres<T: Send> {
> +pub struct Devres<T: Send + 'static> {
> dev: ARef<Device>,
> inner: Arc<Inner<T>>,
> }
> @@ -184,7 +184,7 @@ pub(super) unsafe fn devres_node_remove(
> }
> }
>
> -impl<T: Send> Devres<T> {
> +impl<T: Send + 'static> Devres<T> {
> /// Creates a new [`Devres`] instance of the given `data`.
> ///
> /// The `data` encapsulated within the returned `Devres` instance' `data` will be
> @@ -349,7 +349,7 @@ unsafe impl<T: Send> Send for Devres<T> {}
> // SAFETY: `Devres` can be shared with any task, if `T: Sync`.
> unsafe impl<T: Send + Sync> Sync for Devres<T> {}
>
> -impl<T: Send> Drop for Devres<T> {
> +impl<T: Send + 'static> Drop for Devres<T> {
> fn drop(&mut self) {
> // SAFETY: When `drop` runs, it is guaranteed that nobody is accessing the revocable data
> // anymore, hence it is safe not to wait for the grace period to finish.
>
> base-commit: 56785dcb2ef6d3cff82ac33f2e34db94377416a3
On Wed May 27, 2026 at 4:44 PM CEST, Gary Guo wrote: > On Tue May 26, 2026 at 1:04 AM BST, Danilo Krummrich wrote: >> Add a 'static bound to prevent storing types with borrowed data in >> Devres. > > The bound should be added on `Devres::new` instead. I did consider this as it is generally recommended to minimize bounds on structs. However, a Devres<T> with non-'static T is semantically nonsensical, not just unconstructible, and I think type level bound represents that better.
On Wed May 27, 2026 at 7:04 PM BST, Danilo Krummrich wrote: > On Wed May 27, 2026 at 4:44 PM CEST, Gary Guo wrote: >> On Tue May 26, 2026 at 1:04 AM BST, Danilo Krummrich wrote: >>> Add a 'static bound to prevent storing types with borrowed data in >>> Devres. >> >> The bound should be added on `Devres::new` instead. > > I did consider this as it is generally recommended to minimize bounds on > structs. > > However, a Devres<T> with non-'static T is semantically nonsensical, not just > unconstructible, and I think type level bound represents that better. Technically `Devres` can contain references to the registration, which is known to outlive the bound device. Best, Gary
On Wed May 27, 2026 at 8:07 PM CEST, Gary Guo wrote: > On Wed May 27, 2026 at 7:04 PM BST, Danilo Krummrich wrote: >> On Wed May 27, 2026 at 4:44 PM CEST, Gary Guo wrote: >>> On Tue May 26, 2026 at 1:04 AM BST, Danilo Krummrich wrote: >>>> Add a 'static bound to prevent storing types with borrowed data in >>>> Devres. >>> >>> The bound should be added on `Devres::new` instead. >> >> I did consider this as it is generally recommended to minimize bounds on >> structs. >> >> However, a Devres<T> with non-'static T is semantically nonsensical, not just >> unconstructible, and I think type level bound represents that better. > > Technically `Devres` can contain references to the registration, which is known > to outlive the bound device. That's technically true, but how would you express "outlives the device" as a lifetime bound on Devres? Additionally, if the borrowed data outlives the device bound, you don't need device-managed revocation for it in the first place.
On Wed May 27, 2026 at 7:13 PM BST, Danilo Krummrich wrote: > On Wed May 27, 2026 at 8:07 PM CEST, Gary Guo wrote: >> On Wed May 27, 2026 at 7:04 PM BST, Danilo Krummrich wrote: >>> On Wed May 27, 2026 at 4:44 PM CEST, Gary Guo wrote: >>>> On Tue May 26, 2026 at 1:04 AM BST, Danilo Krummrich wrote: >>>>> Add a 'static bound to prevent storing types with borrowed data in >>>>> Devres. >>>> >>>> The bound should be added on `Devres::new` instead. >>> >>> I did consider this as it is generally recommended to minimize bounds on >>> structs. >>> >>> However, a Devres<T> with non-'static T is semantically nonsensical, not just >>> unconstructible, and I think type level bound represents that better. >> >> Technically `Devres` can contain references to the registration, which is known >> to outlive the bound device. > > That's technically true, but how would you express "outlives the device" as a > lifetime bound on Devres? Just having the lifetime would mean it outlives the device (lifetimes that cannot be guaranteed to outlive the device cannot be used in devres). In a world with registration data, I would imagine all lifetime parameters on registration data are also valid for `Devres`. > > Additionally, if the borrowed data outlives the device bound, you don't need > device-managed revocation for it in the first place. Resources that depend on both registration data and device resource perhaps? Best, Gary
On Wed May 27, 2026 at 9:25 PM CEST, Gary Guo wrote: > On Wed May 27, 2026 at 7:13 PM BST, Danilo Krummrich wrote: >> On Wed May 27, 2026 at 8:07 PM CEST, Gary Guo wrote: >>> On Wed May 27, 2026 at 7:04 PM BST, Danilo Krummrich wrote: >>>> On Wed May 27, 2026 at 4:44 PM CEST, Gary Guo wrote: >>>>> On Tue May 26, 2026 at 1:04 AM BST, Danilo Krummrich wrote: >>>>>> Add a 'static bound to prevent storing types with borrowed data in >>>>>> Devres. >>>>> >>>>> The bound should be added on `Devres::new` instead. >>>> >>>> I did consider this as it is generally recommended to minimize bounds on >>>> structs. >>>> >>>> However, a Devres<T> with non-'static T is semantically nonsensical, not just >>>> unconstructible, and I think type level bound represents that better. >>> >>> Technically `Devres` can contain references to the registration, which is known >>> to outlive the bound device. >> >> That's technically true, but how would you express "outlives the device" as a >> lifetime bound on Devres? > > Just having the lifetime would mean it outlives the device (lifetimes that > cannot be guaranteed to outlive the device cannot be used in devres). > > In a world with registration data, I would imagine all lifetime parameters on > registration data are also valid for `Devres`. [...] > Resources that depend on both registration data and device resource perhaps? You'd still need a separate constructor for this, and I'm still not convinced that this makes a lot of sense. That said if you feel strongly about it, I don't mind moving the 'static bound to new() too much.
On Wed May 27, 2026 at 9:49 PM CEST, Danilo Krummrich wrote: > That said if you feel strongly about it, I don't mind moving the 'static bound > to new() too much. Keeping the bound on the struct for now, given that we agree to move it if it creeps to more places off-list.
On Tue May 26, 2026 at 9:04 AM JST, Danilo Krummrich wrote:
> Devres::new() registers a callback with the C devres subsystem via
> devres_node_add(). If the Devres is leaked (e.g. via
> core::mem::forget(), which is safe), its Drop impl never runs, and the
> devres release callback will revoke the inner Revocable on device
> unbind, which drops T in place. If T contains non-'static references,
> those may be dangling by that point.
>
> Add a 'static bound to prevent storing types with borrowed data in
> Devres.
>
> Fixes: 76c01ded724b ("rust: add devres abstraction")
> Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Reviewed-by: Alexandre Courbot <acourbot@nvidia.com>
On Tue May 26, 2026 at 9:04 AM JST, Danilo Krummrich wrote:
> Devres::new() registers a callback with the C devres subsystem via
> devres_node_add(). If the Devres is leaked (e.g. via
> core::mem::forget(), which is safe), its Drop impl never runs, and the
> devres release callback will revoke the inner Revocable on device
> unbind, which drops T in place. If T contains non-'static references,
> those may be dangling by that point.
>
> Add a 'static bound to prevent storing types with borrowed data in
> Devres.
>
> Fixes: 76c01ded724b ("rust: add devres abstraction")
> Signed-off-by: Danilo Krummrich <dakr@kernel.org>
> ---
Reviewed-by: Eliot Courtney <ecourtney@nvidia.com>
On Tue May 26, 2026 at 2:04 AM CEST, Danilo Krummrich wrote:
> Devres::new() registers a callback with the C devres subsystem via
> devres_node_add(). If the Devres is leaked (e.g. via
> core::mem::forget(), which is safe), its Drop impl never runs, and the
> devres release callback will revoke the inner Revocable on device
> unbind, which drops T in place. If T contains non-'static references,
> those may be dangling by that point.
>
> Add a 'static bound to prevent storing types with borrowed data in
> Devres.
>
> Fixes: 76c01ded724b ("rust: add devres abstraction")
> Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Cc: stable@vger.kernel.org
© 2016 - 2026 Red Hat, Inc.