Enable support for device ID structures that do not contain
context/data field (usually named `driver_data`), making the trait
usable in a wider range of subsystems and buses.
Several such structures are defined in
include/linux/mod_devicetable.h.
This refactoring is a preparation for enabling the PHY abstractions to
use device_id trait.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@gmail.com>
---
rust/kernel/auxiliary.rs | 6 ++++--
rust/kernel/device_id.rs | 26 +++++++++++++++-----------
rust/kernel/of.rs | 3 ++-
rust/kernel/pci.rs | 3 ++-
4 files changed, 23 insertions(+), 15 deletions(-)
diff --git a/rust/kernel/auxiliary.rs b/rust/kernel/auxiliary.rs
index d2cfe1eeefb6..7b8798599128 100644
--- a/rust/kernel/auxiliary.rs
+++ b/rust/kernel/auxiliary.rs
@@ -147,8 +147,10 @@ pub const fn new(modname: &'static CStr, name: &'static CStr) -> Self {
unsafe impl RawDeviceId for DeviceId {
type RawType = bindings::auxiliary_device_id;
- const DRIVER_DATA_OFFSET: usize =
- core::mem::offset_of!(bindings::auxiliary_device_id, driver_data);
+ const DRIVER_DATA_OFFSET: Option<usize> = Some(core::mem::offset_of!(
+ bindings::auxiliary_device_id,
+ driver_data
+ ));
fn index(&self) -> usize {
self.0.driver_data
diff --git a/rust/kernel/device_id.rs b/rust/kernel/device_id.rs
index 3dc72ca8cfc2..b7d00587a0e2 100644
--- a/rust/kernel/device_id.rs
+++ b/rust/kernel/device_id.rs
@@ -25,8 +25,9 @@
/// transmute; however, const trait functions relies on `const_trait_impl` unstable feature,
/// which is broken/gone in Rust 1.73.
///
-/// - `DRIVER_DATA_OFFSET` is the offset of context/data field of the device ID (usually named
-/// `driver_data`) of the device ID, the field is suitable sized to write a `usize` value.
+/// - If [`RawDeviceId::DRIVER_DATA_OFFSET`] is `Some(offset)`, it's the offset of
+/// context/data field of the device ID (usually named `driver_data`) of the device ID,
+/// the field is suitable sized to write a `usize` value.
///
/// Similar to the previous requirement, the data should ideally be added during `Self` to
/// `RawType` conversion, but there's currently no way to do it when using traits in const.
@@ -37,7 +38,7 @@ pub unsafe trait RawDeviceId {
type RawType: Copy;
/// The offset to the context/data field.
- const DRIVER_DATA_OFFSET: usize;
+ const DRIVER_DATA_OFFSET: Option<usize> = None;
/// The index stored at `DRIVER_DATA_OFFSET` of the implementor of the [`RawDeviceId`] trait.
fn index(&self) -> usize;
@@ -77,14 +78,17 @@ impl<T: RawDeviceId, U, const N: usize> IdArray<T, U, N> {
// SAFETY: by the safety requirement of `RawDeviceId`, we're guaranteed that `T` is
// layout-wise compatible with `RawType`.
raw_ids[i] = unsafe { core::mem::transmute_copy(&ids[i].0) };
- // SAFETY: by the safety requirement of `RawDeviceId`, this would be effectively
- // `raw_ids[i].driver_data = i;`.
- unsafe {
- raw_ids[i]
- .as_mut_ptr()
- .byte_add(T::DRIVER_DATA_OFFSET)
- .cast::<usize>()
- .write(i);
+
+ if let Some(data_offset) = T::DRIVER_DATA_OFFSET {
+ // SAFETY: by the safety requirement of `RawDeviceId`, this would be effectively
+ // `raw_ids[i].driver_data = i;`.
+ unsafe {
+ raw_ids[i]
+ .as_mut_ptr()
+ .byte_add(data_offset)
+ .cast::<usize>()
+ .write(i);
+ }
}
// SAFETY: this is effectively a move: `infos[i] = ids[i].1`. We make a copy here but
diff --git a/rust/kernel/of.rs b/rust/kernel/of.rs
index 40d1bd13682c..0ca1692d61f3 100644
--- a/rust/kernel/of.rs
+++ b/rust/kernel/of.rs
@@ -19,7 +19,8 @@
unsafe impl RawDeviceId for DeviceId {
type RawType = bindings::of_device_id;
- const DRIVER_DATA_OFFSET: usize = core::mem::offset_of!(bindings::of_device_id, data);
+ const DRIVER_DATA_OFFSET: Option<usize> =
+ Some(core::mem::offset_of!(bindings::of_device_id, data));
fn index(&self) -> usize {
self.0.data as usize
diff --git a/rust/kernel/pci.rs b/rust/kernel/pci.rs
index f6b19764ad17..dea49abe7cd3 100644
--- a/rust/kernel/pci.rs
+++ b/rust/kernel/pci.rs
@@ -168,7 +168,8 @@ pub const fn from_class(class: u32, class_mask: u32) -> Self {
unsafe impl RawDeviceId for DeviceId {
type RawType = bindings::pci_device_id;
- const DRIVER_DATA_OFFSET: usize = core::mem::offset_of!(bindings::pci_device_id, driver_data);
+ const DRIVER_DATA_OFFSET: Option<usize> =
+ Some(core::mem::offset_of!(bindings::pci_device_id, driver_data));
fn index(&self) -> usize {
self.0.driver_data
--
2.43.0On 6/23/25 8:09 AM, FUJITA Tomonori wrote: > Enable support for device ID structures that do not contain > context/data field (usually named `driver_data`), making the trait > usable in a wider range of subsystems and buses. > > Several such structures are defined in > include/linux/mod_devicetable.h. > > This refactoring is a preparation for enabling the PHY abstractions to > use device_id trait. > > Signed-off-by: FUJITA Tomonori <fujita.tomonori@gmail.com> Acked-by: Danilo Krummrich <dakr@kernel.org>
On Fri, 4 Jul 2025 00:15:19 +0200 Danilo Krummrich <dakr@kernel.org> wrote: > On 6/23/25 8:09 AM, FUJITA Tomonori wrote: >> Enable support for device ID structures that do not contain >> context/data field (usually named `driver_data`), making the trait >> usable in a wider range of subsystems and buses. >> Several such structures are defined in >> include/linux/mod_devicetable.h. >> This refactoring is a preparation for enabling the PHY abstractions to >> use device_id trait. >> Signed-off-by: FUJITA Tomonori <fujita.tomonori@gmail.com> > > Acked-by: Danilo Krummrich <dakr@kernel.org> Thanks a lot! Miguel suggested that splitting the RawDeviceId trait might lead to a cleaner design, and I also tried that approach [v2]. But just to confirm ― do you prefer the original v1 approach instead? https://lore.kernel.org/lkml/CANiq72k0sdUoBxVYghgh50+ZRV2gbDkgVjuZgJLtj=4s9852xg@mail.gmail.com/ [v2]: https://lore.kernel.org/rust-for-linux/20250701141252.600113-1-fujita.tomonori@gmail.com/ Either way works for me. Sorry Miguel, I forgot to fix the comment typo you pointed out in v1. I'll correct it in v3.
On Fri, Jul 04, 2025 at 08:41:59AM +0900, FUJITA Tomonori wrote: > On Fri, 4 Jul 2025 00:15:19 +0200 > Danilo Krummrich <dakr@kernel.org> wrote: > > > On 6/23/25 8:09 AM, FUJITA Tomonori wrote: > >> Enable support for device ID structures that do not contain > >> context/data field (usually named `driver_data`), making the trait > >> usable in a wider range of subsystems and buses. > >> Several such structures are defined in > >> include/linux/mod_devicetable.h. > >> This refactoring is a preparation for enabling the PHY abstractions to > >> use device_id trait. > >> Signed-off-by: FUJITA Tomonori <fujita.tomonori@gmail.com> > > > > Acked-by: Danilo Krummrich <dakr@kernel.org> > > Thanks a lot! > > Miguel suggested that splitting the RawDeviceId trait might lead to a > cleaner design, and I also tried that approach [v2]. But just to > confirm ― do you prefer the original v1 approach instead? Sorry, I didn't catch v2. From only looking at patch 1, I indeed prefer v1. > https://lore.kernel.org/lkml/CANiq72k0sdUoBxVYghgh50+ZRV2gbDkgVjuZgJLtj=4s9852xg@mail.gmail.com/ Looking at this reply to patch 3, I can see why Miguel suggested it though. Given that, I think splitting the trait is the correct thing to do. I'll reply to v2. > [v2]: https://lore.kernel.org/rust-for-linux/20250701141252.600113-1-fujita.tomonori@gmail.com/ > > Either way works for me. > > Sorry Miguel, I forgot to fix the comment typo you pointed out in v1. I'll correct it in v3.
© 2016 - 2025 Red Hat, Inc.