On 7/28/06, Vojtech Pavlik <vojtech@xxxxxxx> wrote:
> I do not understand this. Any polling (in kernel or in userspace) will > wake the CPU, wasting power. The kernel, however, has all the gory details at hand, and can decide much better about the polling frequency, than the (hopefully) hardware agnostic userspace. Imagine your Zaurus: You don't need to poll very often when you are on the flat part of the LiIon discharge curve, you probably want more detailed info near the end.
Another example: on my ThinkPad, the readouts provided by the EC change only every 2 seconds, regardless of how much you poll it. Hence my proposal for a polling-based interface with kernel-side caching. This way, the hardware query rate is the minimum of two rates: what the application can use and what the kernel thinks is the hardware's change rate. This doesn't solve the timer interrupt (tickless kernel) issue. Sane apps would never ask for battery updates at more than a few Hz, are we likely to see tickless machines where this makes a difference beyondnormal load? Shem - To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
