Danilo Krummrich
2025-Aug-28 13:59 UTC
[PATCH v7 5/6] rust: pci: use pci::Vendor instead of bindings::PCI_VENDOR_ID_*
On 8/28/25 3:25 PM, Alexandre Courbot wrote:> On Wed Aug 27, 2025 at 8:12 AM JST, John Hubbard wrote: > <snip> >> diff --git a/rust/kernel/pci/id.rs b/rust/kernel/pci/id.rs >> index 4b0ad8d4edc6..fd7a789e3015 100644 >> --- a/rust/kernel/pci/id.rs >> +++ b/rust/kernel/pci/id.rs >> @@ -118,15 +118,14 @@ fn try_from(value: u32) -> Result<Self, Self::Error> { >> /// ``` >> /// # use kernel::{device::Core, pci::{self, Vendor}, prelude::*}; >> /// fn log_device_info(pdev: &pci::Device<Core>) -> Result<()> { >> -/// // Compare raw vendor ID with known vendor constant >> -/// let vendor_id = pdev.vendor_id(); >> -/// if vendor_id == Vendor::NVIDIA.as_raw() { >> -/// dev_info!( >> -/// pdev.as_ref(), >> -/// "Found NVIDIA device: 0x{:x}\n", >> -/// pdev.device_id() >> -/// ); >> -/// } >> +/// // Get the validated PCI vendor ID >> +/// let vendor = pdev.vendor_id(); >> +/// dev_info!( >> +/// pdev.as_ref(), >> +/// "Device: Vendor={}, Device=0x{:x}\n", >> +/// vendor, >> +/// pdev.device_id() >> +/// ); > > Why not use this new example starting from patch 2, which introduced the > previous code that this patch removes?I think that's because in v2 vendor_id() still returns the raw value. I think it makes a little more sense if this patch simply introduces the example as an example for vendor_id() itself. I think struct Vendor does not necessarily need an example by itself.
John Hubbard
2025-Aug-29 21:38 UTC
[PATCH v7 5/6] rust: pci: use pci::Vendor instead of bindings::PCI_VENDOR_ID_*
On 8/28/25 6:59 AM, Danilo Krummrich wrote:> On 8/28/25 3:25 PM, Alexandre Courbot wrote: >> On Wed Aug 27, 2025 at 8:12 AM JST, John Hubbard wrote: >> <snip> >>> diff --git a/rust/kernel/pci/id.rs b/rust/kernel/pci/id.rs >>> index 4b0ad8d4edc6..fd7a789e3015 100644 >>> --- a/rust/kernel/pci/id.rs >>> +++ b/rust/kernel/pci/id.rs >>> @@ -118,15 +118,14 @@ fn try_from(value: u32) -> Result<Self, Self::Error> { >>> /// ``` >>> /// # use kernel::{device::Core, pci::{self, Vendor}, prelude::*}; >>> /// fn log_device_info(pdev: &pci::Device<Core>) -> Result<()> { >>> -/// // Compare raw vendor ID with known vendor constant >>> -/// let vendor_id = pdev.vendor_id(); >>> -/// if vendor_id == Vendor::NVIDIA.as_raw() { >>> -/// dev_info!( >>> -/// pdev.as_ref(), >>> -/// "Found NVIDIA device: 0x{:x}\n", >>> -/// pdev.device_id() >>> -/// ); >>> -/// } >>> +/// // Get the validated PCI vendor ID >>> +/// let vendor = pdev.vendor_id(); >>> +/// dev_info!( >>> +/// pdev.as_ref(), >>> +/// "Device: Vendor={}, Device=0x{:x}\n", >>> +/// vendor, >>> +/// pdev.device_id() >>> +/// ); >> >> Why not use this new example starting from patch 2, which introduced the >> previous code that this patch removes? > > I think that's because in v2 vendor_id() still returns the raw value. I think itThat is correct.> makes a little more sense if this patch simply introduces the example as an > example for vendor_id() itself. > > I think struct Vendor does not necessarily need an example by itself.I'm not quite sure if you are asking for a change to this patch? The example already exercises .vendor_id(), so...? thanks, -- John Hubbard