Off-button reaction time
lars at metafoo.de
Thu Aug 18 08:29:43 EDT 2011
On 08/17/2011 07:02 PM, Bas Wijnen wrote:
> Andrea Bolognani schreef op wo 17-08-2011 om 12:17 [+0200]:
>> Hold on a second: either I’m utterly mistaken, or the kernel is still far
>> from being kicked in when the power button is pressed.
> That is correct. But the delay is made by the rtc, which can be set up
> when the kernel is running. It will retain its value as long as the rtc
> has power (that is, when you remove the battery, you lose both the
> current time and this setting).
>> I always assumed
>> the circuitry needed that delay, and it was not avoidable.
> It is avoidable, and I don't see a reason to use it.
>> Are you telling me that I’ve been wrong the whole time, and that it is
>> possible to achieve instant—boot making changes to the software only?
> Yes. If the kernel sets the rtc register, it will instant-boot from then
> on, until the power on the rtc goes down (which should never happen).
> I propose to write the rtc register on every boot, so that even if the
> battery was removed and the setting was lost, it still only results in
> the delay once.
> I'm talking about register HWFCR, page 131 of the programmer's manual.
> Writing 0 to this register (0x10003024) will remove the delay.
mth has already a working patch for this.
We've decided to use a common tree for nanonote and opendingux kernel
development, to avoid having patches only in one of the trees.
More information about the discussion