Re: [PATCH 0/13] Time: Generic Timeofday Subsystem (v B11)

[Ingo Molnar <mingo@elte.hu>]

* john stultz <johnstul@us.ibm.com> wrote:

> All,
> 	The following patchset applies against 2.6.15-rc1-mm2 and 
> provides a generic timekeeping subsystem that is independent of the 
> timer interrupt. [...]

While we are at introducing and moving around code, i've done a big 
cleanup of all code touched/introduced by your patchset. The 
gtod-B11-cleanup.tar.gz file containing the cleanup patch-series can be 
found at:

   http://redhat.com/~mingo/gtod-patches/

(i'll send the patches individually as well, as replies to your mails).

one side-effect of the cleanups is that the core Linux NTP code has now 
become quite readable, for the first time in history ;-)

the cleanups are also included in the 2.6.14-rt18 tree, and i've tested 
them on a UP and on an SMP box. The cleanups are 99% coding style 
related, but while reviewing the code i've also inserted a few TODO 
entries:

+       /* TODO: bogus limit of 4 CPUs? --mingo */
+/* TODO: why a seqlock? It's only write-locked, so should be a spinlock. */
+                       /* TODO: is 2*time_constant correct? --mingo */
+        * TODO: shouldnt we write-lock xtime_lock below, and then
+        * TODO: shouldnt txc->time be filled in here, within ntp_lock and

	Ingo