Re: rtc and smp

[Ruediger Oertel <ro suse de>]

On Wed, 29 Sep 1999, Rich Payne wrote:

> 
> I've been working with 2.2.13pre14 on an SMP machine and came across
> something rather strange. If I compile the kernel with Enhanced Real Time
> clock support (config_rtc) then it 'seems' like nothing ever gets
> scheduled on the 2nd cpu. I say 'seems' because I'm looking at tools like
> top (which never shows >50% cpu util) and gtop (which never shows work on
> the 2nd cpu). Howver both CPUs are detected at boot and /proc/cpuinfo
> reports 
[..]
> So, is this a kernel problem or are the userland tools failing? The only
> other way I could thing of testing this would be to run some sort of
> benchmark, get the time it took to run, then run it twice (both at the
> same time that is) and see if the second time is 2x the original (or greater).
> That should in theory tell me if the 2nd CPU is being used.
Yes, that one bit me, when I started on alpha-smp.
The timer irq is used for the rtc, but only the normal timer-irq handler
updates the data for /proc/stat.
But I needed/wanted proc/rtc and since I never needed the irq-features
of the rtc-driver I've been running the rtc-light-patches from
genie.ucd.ie for more than half a year now without problems

> 
> Anybody have any ideas on this? It's seems a little strange, especaially
> when config_rtc says 
> 
> "If you run Linux on a multiprocessor machine and said Y to
> "Symmetric Multi Processing" above, you should say Y here to read
> and set the RTC clock in an SMP compatible fashion."

-- 
with kind regards (mit freundlichem Grinsen),
                                     Ruediger Oertel (ro@suse.de)
-----------------------------------------------------------------
Linux 2.2.12.SuSE (root@fatou) (gcc version egcs-2.91.66 ) #1 SMP
       Total of 2 processors activated (800.36 BogoMIPS)