[ECOS] RE: Re: How to set clock related options?

caocangmou caocangmou@sina.com
Fri Sep 26 03:23:00 GMT 2003


Given the configuration :
        Real-time clock numerator : 1000000000
        Real-time clock denominator : 100
        Real-time clock base prescale : 8192   
        OSC/Clock Frequency :20000000
and the discription below,should the Real-time clock period be calculated like this ?:
        Real-time clock period = 10(ms)/((OSC/Clock Frequency)/base prescale/1000(msecs per sec))
                               =10/(20000000/8192/1000)
                               =4
but the value in configuration is 10!Is that right ?


----- Original Message -----
From:Nick Garnett <nickg@ecoscentric.com>
To:Savin Zlobec <savin@elatec.si>
Subject:Re: [ECOS] RE: Re: How to set clock related options?
Date:Thu, 25 Sep 2003 22:48:42 +0800
 >Savin Zlobec <savin@elatec.si> writes:
 >
 >> Savin Zlobec wrote:
 >> 
 >> > caocangmou wrote:
 >> >
 >> >> But, I got a configuration from template(Akizuki H8/3068) :
 >> >>           Real-time clock numerator : 1000000000
 >> >>           Real-time clock denominator : 100
 >> >>           Real-time clock base prescale : 8192           Real-time
 >> >> clock period :10
 >> >>           OSC/Clock Frequency :20000000
 >> >> It seems not the case .What does period really mean?
 >> >>
 >> > CYGNUM_KERNEL_COUNTERS_RTC_PERIOD which defaults to
 >> > CYGNUM_HAL_RTC_PERIOD is
 >> > used to initialize you system clock hardware. Ecos tick is usually
 >> > implemented with a  hw timer which gives
 >> > periodic interrupts. The  CYGNUM_HAL_RTC_PERIOD is calculated so
 >> > that you get an interrupt every
 >> > 10ms (the default) and it depends on your timer hardware. Changing
 >> > the tick requires setting the
 >> > CYGNUM_HAL_RTC_NUMERATOR, CYGNUM_HAL_RTC_DENOMINATOR and
 >> > CYGNUM_HAL_RTC_PERIOD (if it is not automatically calculated from
 >> > the previous two - look how it is
 >> > calculated in cdl - it can only depend on your frequency - in that
 >> > case it is probably set to 10ms).
 >> 
 >> I just realized that what I said about the hw timer it is not entirely
 >> correct. The timer should only be able to generate
 >> an interrupt after a desired amount of time. The kernel takes care of
 >> the periodic aspect by resetting the timer with
 >> HAL_CLOCK_RESET in ISR.
 >
 >Well, if the timer can be configured to generate periodic interrupts
 >then that is what should be done -- you can then leave the
 >HAL_CLOCK_RESET() macro empty. It is much better to let the hardware
 >keep time than try to do it in software. Unless you are very careful,
 >reloading the timer in HAL_CLOCK_RESET() can lead to clock drift.
 >
 >-- 
 >Nick Garnett                    eCos Kernel Architect
 >http://www.ecoscentric.com      The eCos and RedBoot experts
 >
 >
 >-- 
 >Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
 >and search the list archive: http://sources.redhat.com/ml/ecos-discuss
 >
 >
 >

______________________________________

===================================================================

-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss



More information about the Ecos-discuss mailing list