Fixes for Rust Device / Driver abstractions
Subsystems (4)
Danilo Krummrich
maintainercover Fixes for Rust Device / Driver abstractions
This patch series contains a few fixes for the preceding series [1] introducing device / driver Rust abstractions.
Except for the CONFIG_PCI_MSI dependency fix, there's no functional change.
[1] https://lore.kernel.org/rust-for-linux/[email protected]/
Danilo Krummrich (3):
rust: pci: do not depend on CONFIG_PCI_MSI
rust: io: move module entry to its correct location
rust: driver: address soundness issue in RegistrationOps
rust/kernel/driver.rs | 25 ++++++++++++++++++++----- rust/kernel/lib.rs | 6 +++--- rust/kernel/pci.rs | 8 +++++--- rust/kernel/platform.rs | 8 +++++--- 4 files changed, 33 insertions(+), 14 deletions(-)
base-commit: 06e843bbbf2107463249ea6f6b1a736f5647e24a
prerequisite-patch-id: 5ce64efa11560206236db2b2f014ceaa3d2e9fa3
prerequisite-patch-id: e86a1e93ce4e9e8440be0bb641ef99def521b75f
Signature · 1 line
2.47.1
Danilo Krummrich
maintainerPATCH 1/3 rust: pci: do not depend on CONFIG_PCI_MSI
The PCI abstractions do not actually depend on CONFIG_PCI_MSI; it also breaks drivers that only depend on CONFIG_PCI, hence drop it.
While at it, move the module entry to its correct location.
--- rust/kernel/lib.rs | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
rust/kernel/lib.rs +2 -2
@@ -58,6 +58,8 @@pub mod net;pub mod of;pub mod page; +#[cfg( CONFIG_PCI )] +pub mod pci;pub mod pid_namespace;pub mod platform;pub mod prelude;
@@ -84,8 +86,6 @@pub use bindings;pub mod io;pub use macros; -#[cfg(all( CONFIG_PCI , CONFIG_PCI_MSI ))] -pub mod pci;pub use uapi;#[doc(hidden)]
Signature · 1 line
2.47.1
Greg KH
maintainerRe: [PATCH 1/3] rust: pci: do not depend on CONFIG_PCI_MSI
Quoted 8 lines · Danilo Krummrich
The PCI abstractions do not actually depend on
CONFIG_PCI_MSI; it also breaks drivers that only depend onCONFIG_PCI, hence drop it.While at it, move the module entry to its correct location.
Reported-by kernel test robot <[email protected]>Fixes 3a9c09193657 ("rust: pci: add basic PCI device / driver abstractions")
Where did that git id come from? It's 1bd8b6b2c5d38d9881d59928b986eacba40f9da8. I'll go edit it by hand...
thanks,
greg k-h
Danilo Krummrich
maintainerRe: [PATCH 1/3] rust: pci: do not depend on CONFIG_PCI_MSI
Quoted 4 lines · Greg KH
Quoted 3 lines · Danilo Krummrich
The PCI abstractions do not actually depend on
CONFIG_PCI_MSI; it also breaks drivers that only depend onCONFIG_PCI, hence drop it.
Quoted 8 lines
Quoted 5 lines
While at it, move the module entry to its correct location.
Reported-by kernel test robot <[email protected]>Fixes 3a9c09193657 ("rust: pci: add basic PCI device / driver abstractions")Where did that git id come from? It's
1bd8b6b2c5d38d9881d59928b986eacba40f9da8. I'll go edit it by hand...
Oops, that's from my tree. I did not rebase to the driver-core tree, sorry.
Quoted 4 lines
thanks,
greg k-h
Danilo Krummrich
maintainerPATCH 2/3 rust: io: move module entry to its correct location
In reply to Danilo KrummrichThe module entry of io falsely ended up in the "use" block instead of the "mod" block, hence move it to its correct location.
--- rust/kernel/lib.rs | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
rust/kernel/lib.rs +1 -1
@@ -48,6 +48,7 @@pub mod firmware;pub mod fs;pub mod init; +pub mod io;pub mod ioctl;pub mod jump_label;#[cfg( CONFIG_KUNIT )]
@@ -84,7 +85,6 @@#[doc(hidden)] pub use bindings; -pub mod io;pub use macros;pub use uapi;
Signature · 1 line
2.47.1
Danilo Krummrich
maintainerPATCH 3/3 rust: driver: address soundness issue in `RegistrationOps`
In reply to Danilo KrummrichThe RegistrationOps trait holds some obligations to the caller and implementers. While being documented, the trait and the corresponding functions haven't been marked as unsafe.
Hence, markt the trait and functions unsafe and add the corresponding safety comments.
This patch does not include any fuctional changes.
--- rust/kernel/driver.rs | 25 ++++++++++++++++++++----- rust/kernel/pci.rs | 8 +++++--- rust/kernel/platform.rs | 8 +++++--- 3 files changed, 30 insertions(+), 11 deletions(-)
rust/kernel/driver.rs +20 -5
@@ -17,23 +17,35 @@/// For instance, the PCI subsystem would set `RegType` to `bindings::pci_driver` and call /// `bindings::__pci_register_driver` from `RegistrationOps::register` and /// `bindings::pci_unregister_driver` from `RegistrationOps::unregister`. -pub trait RegistrationOps { +/// +/// # Safety +/// +/// A call to [`RegistrationOps::unregister`] for a given instance of `RegType` is only valid if a +/// preceding call to [`RegistrationOps::register`] has been successful. +pub unsafe trait RegistrationOps {/// The type that holds information about the registration. This is typically a struct defined /// by the C portion of the kernel. type RegType :Default ;/// Registers a driver. /// +/// # Safety +/// /// On success, `reg` must remain pinned and valid until the matching call to /// [`RegistrationOps::unregister`]. -fn register ( +unsafe fn register ( reg: &Opaque <Self ::RegType >, name: &'staticCStr , module: &'staticThisModule , ) ->Result ;/// Unregisters a driver previously registered with [`RegistrationOps::register`]. -fn unregister (reg: &Opaque <Self ::RegType >); +/// +/// # Safety +/// +/// Must only be called after a preceding successful call to [`RegistrationOps::register`] for +/// the same `reg`. +unsafe fn unregister (reg: &Opaque <Self ::RegType >); }/// A [`Registration`] is a generic type that represents the registration of some driver type (e.g.
@@ -68,7 +80,8 @@ pub fn new(name: &'static CStr, module: &'static ThisModule) -> impl PinInit<Sel// just been initialised above, so it's also valid for read. let drv =unsafe { &*(ptras *const Opaque <T ::RegType >) }; -T ::register (drv, name, module) +// SAFETY: `drv` is guaranteed to be pinned until `T::unregister`. +unsafe {T ::register (drv, name, module) } }), }) }
@@ -77,7 +90,9 @@ pub fn new(name: &'static CStr, module: &'static ThisModule) -> impl PinInit<Sel#[pinned_drop] impl <T :RegistrationOps >PinnedDrop for Registration <T > {fn drop (self:Pin <&mut Self >) { -T ::unregister (&self.reg); +// SAFETY: The existence of `self` guarantees that `self.reg` has previously been +// successfully registered with `T::register` +unsafe {T ::unregister (&self.reg) }; } }
rust/kernel/pci.rs +5 -3
@@ -23,10 +23,12 @@/// An adapter for the registration of PCI drivers. pub struct Adapter <T :Driver >(T ); -impl <T :Driver + 'static> driver::RegistrationOps for Adapter <T > { +// SAFETY: A call to `unregister` for a given instance of `RegType` is guaranteed to be valid if +// a preceding call to `register` has been successful. +unsafe impl <T :Driver + 'static> driver::RegistrationOps for Adapter <T > {type RegType = bindings::pci_driver ; -fn register ( +unsafe fn register ( pdrv: &Opaque <Self ::RegType >, name: &'staticCStr , module: &'static ThisModule,
@@ -45,7 +47,7 @@ fn register( }) } -fn unregister (pdrv: &Opaque <Self ::RegType >) { +unsafe fn unregister (pdrv: &Opaque <Self ::RegType >) {// SAFETY: `pdrv` is guaranteed to be a valid `RegType`. unsafe { bindings::pci_unregister_driver (pdrv.get ()) } }
rust/kernel/platform.rs +5 -3
@@ -19,10 +19,12 @@/// An adapter for the registration of platform drivers. pub struct Adapter <T :Driver >(T ); -impl <T :Driver + 'static> driver::RegistrationOps for Adapter <T > { +// SAFETY: A call to `unregister` for a given instance of `RegType` is guaranteed to be valid if +// a preceding call to `register` has been successful. +unsafe impl <T :Driver + 'static> driver::RegistrationOps for Adapter <T > {type RegType = bindings::platform_driver ; -fn register ( +unsafe fn register ( pdrv: &Opaque <Self ::RegType >, name: &'staticCStr , module: &'static ThisModule,
@@ -44,7 +46,7 @@ fn register(to_result (unsafe { bindings::__platform_driver_register (pdrv.get (), module.0 ) }) } -fn unregister (pdrv: &Opaque <Self ::RegType >) { +unsafe fn unregister (pdrv: &Opaque <Self ::RegType >) {// SAFETY: `pdrv` is guaranteed to be a valid `RegType`. unsafe { bindings::platform_driver_unregister (pdrv.get ()) }; }
Signature · 1 line
2.47.1
Re: [PATCH 3/3] rust: driver: address soundness issue in `RegistrationOps`
Quoted 6 lines · On Fri, 3 Jan 2025 17:46:03 +0100 Danilo Krummrich <[email protected]> wrote:
The
RegistrationOpstrait holds some obligations to the caller and implementers. While being documented, the trait and the corresponding functions haven't been marked as unsafe.Hence, markt the trait and functions unsafe and add the corresponding safety comments.
nit: markt -> mark
Quoted 6 lines
This patch does not include any fuctional changes.
Reported-by Gary Guo <[email protected]>Signed-off-by Danilo Krummrich <[email protected]>
Quoted 5 lines
--- rust/kernel/driver.rs | 25 ++++++++++++++++++++----- rust/kernel/pci.rs | 8 +++++--- rust/kernel/platform.rs | 8 +++++--- 3 files changed, 30 insertions(+), 11 deletions(-)