Date: Tue, 06 Apr 2010 15:47:37 +0300 From: Andriy Gapon <avg@freebsd.org> To: freebsd-hackers@freebsd.org Cc: =?UTF-8?B?RGFnLUVybGluZyBTbcO4cmdyYXY=?= <des@des.no> Subject: Re: periodically save current time to time-of-day hardware Message-ID: <4BBB2D69.80002@freebsd.org> In-Reply-To: <4BB22A42.7050003@icyb.net.ua> References: <4BACC791.70502@icyb.net.ua> <20100327214634.GI32799@server.vk2pj.dyndns.org> <86fx3k7jqi.fsf@ds4.des.no> <201003291044.28544.jhb@freebsd.org> <4BB22A42.7050003@icyb.net.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
on 30/03/2010 19:43 Andriy Gapon said the following: > > do you think that periodic saving of system clock to hardware and making resettodr > asynchronous are dependent issues? Or are they orthogonal and can be implemented > independently? A new version of the patch: http://people.freebsd.org/~avg/periodic_resettodr.diff Some notes: - this patch doesn't implement any resettodr enhancements proposed earlier: making it non-blocking/asynchronous and saving time closer to seconds boundary; - the patch adds code for saving time on shutdown, adds logic for checking ntp synchronization status and for dynamically stoping/(re-)starting the callout via period sysctl; - the patch consists of two pieces: the first is abstracting ntp status check into a separate function and the other is all the time saving code; - I don't feel that kern_ntptime.c is best place for this code, I only chose it because of ntp status checking; perhaps ntp_is_time_error needs to be made public and the code moved to a better file; suggestion are welcome! I also welcome any comments and suggestions on code style, identifier naming, logic, strategy, etc. Thank you! P.S. it seems that machdep is not the best place for "RTC" sysctls, but I am deferring this issue at the moment. -- Andriy Gapon
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4BBB2D69.80002>