This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
Re: UART communication & time slices
- To: Andreas dot Karlsson at combitechsystems dot com
- Subject: Re: [ECOS] UART communication & time slices
- From: Jonathan Larmour <jlarmour at redhat dot com>
- Date: Thu, 07 Sep 2000 16:26:58 +0100
- CC: ecos-discuss at sourceware dot cygnus dot com, aeb at arm dot com
- Organization: Red Hat UK Ltd.
- References: <2253171AF143D21185A60000F8FA748B0229A95C@pluto.combitech.se>
Andreas.Karlsson@combitechsystems.com wrote:
>
> I have noticed that a time slice in the config is set to 5.
>
> Is this the same time slice as that one used by the real clock?
I don't understand the question - CYGNUM_KERNEL_SCHED_TIMESLICE_TICKS gives
the number of real time clock ticks to elapse before timeslicing.
> Then one slice is 50 ms which is rather much. Is it just to change the
> values in the configuration tool?
Feel free to reduce CYGNUM_KERNEL_SCHED_TIMESLICE_TICKS if that's what you
want!
> What should the initialization value for AEB be?
I don't understand. There is nothing special about the AEB in this respect
surely? Unless you mean you want to run the real time clock faster itself.
In which case you would change CYGNUM_HAL_RTC_PERIOD in the AEB HAL and
adjust CYGNUM_HAL_RTC_NUMERATOR and/or CYGNUM_HAL_RTC_DENOMINATOR
accordingly.
But if you run the clock faster, you will waste a larger proportion of the
time processing clock interrupts. Similarly, if you reduce the number of
timeslice ticks, the kernel will reschedule more frequently.
Jifl
--
Red Hat, 35 Cambridge Place, Cambridge, UK. CB2 1NS Tel: +44 (1223) 728762
"Plan to be spontaneous tomorrow." || These opinions are all my own fault