[PATCH v2] parport_pc: Also enable driver for PCI systems

Arnd Bergmann arnd at arndb.de
Wed Feb 16 09:17:44 PST 2022


On Wed, Feb 16, 2022 at 5:35 PM Maciej W. Rozycki <macro at orcam.me.uk> wrote:
>
> On Wed, 16 Feb 2022, Christoph Hellwig wrote:
>
> > > > Instead of adding generic-y just ad a mandatory-y in
> > > > include/asm-generic/Kbuild.
> > >
> > >  I'm inconvinced.  Not all archs want it, 5 don't.
> >
> > Which is exactly what mandatory-y is for.  Provide the asm-generic
> > version by default, but let architectures override it.
>
>  I don't think so.  Those 5 architectures don't want it at all; 7 other
> ones have their own versions.
>
>  Otherwise we could blanket-list all asm-generic headers as mandatory-y.

I think ideally the PCI driver should be a separate file from the rest, or
possibly it could get split up even further.

parport_pc_probe_port()/parport_pc_unregister_port() are already exported
by the driver and used by some of the front-ends. The parport_pc_pci_driver
looks like it could easily go into one file using module_pci_driver(), while
the platform driver stays in the existing file and the legacy detection logic
goes into a third one. The powerpc and sparc versions could technically
also be separate drivers, but I wouldn't take the rework that far.

       Arnd



More information about the linux-riscv mailing list