Hi, On Wed, Mar 19, 2014 at 09:49:31PM +0100, Ralf Baechle wrote: > I think the patch (http://patchwork.linux-mips.org/patch/6419/) should be > applied. The argumentation for reserving ports in the platform code are > the same as on x86 - touch the registers and bad things may happen. This > is because a fair number of older MIPS platforms were based on x86 chipsets > or at least are using very similar designs. If you drop the request_region() from the driver, it will try to probe anyway regardless what the platform code codes. So bad things could happen, no? Currently we can prevent i8042 driver from probing I/O space on PCI-less Octeons (for example), because we define empty I/O space so request_region() by driver will fail. So we can actually prevent bad things from happening. I would call this good design, not an accident. Maybe I'm missing something? Anyway, I don't have strong feelings whether this patch is applied or not. My computers will keep on working on either case. A.
