On Sat, Apr 14, 2012 at 12:16:26AM +0800, Ying-Chun Liu (PaulLiu) wrote:
I haven't completely understood this, and will need more study.
Hi Shawn,
About the above ioctl() related blocks. I think it is used by some user space libs that requires accurate time (47BIT TIME). And it have to be sure that the time continuously flowing so the apps wants to be notified that the rtc is changed dramatically by the set command.
I'm guessing that it might be related to the Multimedia binary libs. A/V sync or something like that. But I don't have those source code. Please make sure those libs are not using it.
I also see other Freescale RTC driver implement the same function. For example: http://tinyurl.com/cnp93jv So I guess it might be a common ioctl that Freescale uses somewhere.
So my plan is to split the IOR to a header file. And keep the rest as is.
'think', 'might' and 'something like that' are the wrong words to convince anybody that these ioctls are needed. So unless you find real usecases and also can convince us that it's not possible to reach your goal using either the existing ioctls or an additional common (non fsl specific) ioctl please just drop them.
Sascha