[PATCH 2/4] rtc: sa1100: convert to run-time register mapping

Russell King - ARM Linux [off-list ref] writes:

On Thu, Feb 05, 2015 at 08:18:05PM +0100, Robert Jarzmik wrote:

quoted

As a consequence, I'm pretty much in favor of :
 - making rtc-pxa and rtc-sa1100 exclusive
 - adding to rtc-pxa driver :
   - a new rtc device (ie. it will register 2 rtc devices, one of the sa1100
     kind, one of the pxa specific kind)

This has flaws :
 - a bit of code will be duplicated between rtc-pxa and rtc-sa1100
 - the ordering in rtc-pxa between the 2 rtc devices will be important, as if I
 remember correctly /dev/rtc points to the first /dev/rtc0 registered

But it opens up :
 - DT path to both drivers
 - isolation between sa1100 architecture changes and pxa architecture changes
 - Rob's current patches can remain almost the same

Or you make one export a set of "library" calls which are referenced
by the other driver, and only declare one platform device.  Not
_particularly_ nice but would avoid the code duplication issue (but
has the benefit of not eating up module space, especially when the
strict rw/execute protections are in effect.)

Or even better only export sa1100_rtc_ops, which should be enough, with :
 - Kconfig will have a "select RTC_SRV_SA1100" in the "config RTC_DRV_PXA".
 - struct sa1100_rtc_ops won't be static anymore, it should be exported
 - amend rtc-pxa :
   - the probe/release
   - the interrupt handler should take into account one more bit

That will imply that rtc-pxa is one platform device, which registers 2 rtc
devices. rtc-sa1100 is another platform device. And both will be exclusive.

Yet I don't see the "not eating up module space", unless all sa1100
functions are exported into a "rtc-sa1100-lib.c", which will always be built
builtin into the kernel when CONFIG_RTC is selected.

If there are still 2 modules, rtc-pxa and rtc-sa1100, they'll both eat module
space, unless I'm missing something.

Cheers.

-- 
Robert