[PATCH] Add sched_clock to AT91 TCB clocksource driver

Remy Bohmer linux at bohmer.net
Mon Aug 1 14:08:19 EDT 2011


Hi,

> You're forgetting that jiffies doesn't jump about.  A late initializing
> sched_clock could jump.

Agreed. Currently the jiffies based clock does not move at all, thus
it surely does not jump... ;-)

>> Anyway, do you have a better suggestion how to fix this? The tcb
>> clocksource is loaded during a arch_initcall(), perhaps we need
>> something before that point.
>> I do not see an easy way to integrate it in MACHINE_START(.timer).
>> Would 'late_time_init' be a better solution?
>
> late_time_init() is not that much better as that still happens after
> sched_init() has been called.  sched_init() initializes various
> structures which involves reading sched_clock().

But late_time_init does it before the sched_clock is declared stable.
Until that point sched_clock will always return zero, as such we can
IMHO safely assume that the scheduler is designed such that it can
handle fixed zero timestamps for a while. The new sched_clock also
nicely starts counting from zero once enabled, and does not make a
jump start.

> Why can't it initialize itself at the standard point during the boot
> sequence, which is time_init() - which in turn is as you say the
> .timer callback?

In that case it has to replace the PIT based clocksource completely.
Since the TCB based clocksource and event device are
optional/configurable for this architecture and it is implemented as a
clocksource driver instead of part of the BSP, it would result in a
lot of ifdefs throughout that code.
The arch_initcall() has to be removed and replaced by something else...
I will try to work out a different solution, and post an update, see
how that looks like...

Kind regards,

Remy



More information about the linux-arm-kernel mailing list