On Thu, 11 Aug 2011, Marc Zyngier wrote:

sched_clock() is yet another blocker on the road to the single
image. This patch implements an idea by Russell King:

http://www.spinics.net/lists/linux-omap/msg49561.html

Instead of asking the platform to implement both sched_clock()
itself and the rollover callback, simply register a read()
function, and let the ARM code care about sched_clock() itself,
the conversion to ns and the rollover. sched_clock() uses
this read() function as an indirection to the platform code.

This allow some simplifications and possibly some footprint gain
when multiple platforms are compiled in. Among the drawbacks,
the removal of the *_fixed_sched_clock optimization which could
negatively impact some platforms (sa1100, tegra, versatile
and omap).

Tested on 11MPCore, OMAP4 and Tegra.

[...]

+unsigned long long notrace sched_clock(void)
+{
+	if (read_sched_clock) {
+		u32 cyc = read_sched_clock();
+		return cyc_to_sched_clock(&cd, cyc, sched_clock_mask);
+	}
+
+	return (unsigned long long)(jiffies - INITIAL_JIFFIES)
+					* (NSEC_PER_SEC / HZ);
+}

This bothers me that this function has to include a conditional which 
will _always_ be evaluated as true except during early boot.  Why not 
simply initializing read_sched_clock with a jiffy_shed_clock function 
instead, and let it be overriden during boot?  This way the actual 
sched_clock() code will be more straight forward.

Of course there is the race where read_sched_clock might be updated 
before or after the corresponding clock data/mask are updated leading to 
funky results if you are lucky enough to call sched_clock() during that 
exact wrong moment.  But either we don't care (switching from a jiffy to 
a hardware clock base is going to cause quite a glitch already anyway), 
or simply disabling IRQs during initialization should prevent the race 
(presuming no other CPUs are running at that point).


Nicolas