Disabling system clock to use external clock driver card

[Michael Westfall]

Thanks.

It looks like the BSP has clockOn() and clockOff() functions to control the
BSP's  default clock.

I think this is what we need.

On Fri, Mar 4, 2016 at 5:56 PM, Gedare Bloom <gedare at rtems.org> wrote:

> On Fri, Mar 4, 2016 at 9:46 AM, Michael Westfall <mwestfal at gemini.edu>
> wrote:
> > Hello,
> >
> > We have a vxWorks driver for a VME IRIG-B time provider card (Bancomm
> > bc635/bc637) which we want to port to  RTEMS.
> >
> > The ISR of the current vxWorks driver disables the system clock via
> > sysClkDisable() in order to take over providing ticks to the kernel via
> > tickAnnounce().
> >
> > rtems_clock_tick() seems to be a direct replacement for the
> tickAnnounce()
> > function, but what is the strategy for disabling/reenabling the system
> > clock?
> >
> > Surely someone else has implemented something similar.
> >
> You should be able to provide a custom clock source for the clock
> driver at the BSP level. Most (all?) BSPs use the framework in
> c/src/lib/libbsp/shared/clockdrv_shell.h to drive the system clock
> ticks by implementing the necessary "hook" functions. You can
> replace/modify the BSP's ckinit.c to use the bc635 card instead. See
> for example c/src/lib/libbsp/i386/pc386/clock/ckinit.c for how the x86
> currently has two options for the clock timer.
>
> This would relieve you from having to track quite so much "state".
> However, I suppose if you wanted to just port your code below, you
> would replace the Clock_isr with this code, and undo the
> Clock_driver_support_initialize_hardware() that your BSP has done.
>
> >
> > Here is our vxWorks ISR:
> >
> > void isr_bc635
> > (
> >  void *parg
> >  )
> > {
> >    register short intnum;
> >    register unsigned short tmask, bcistatus, bcimask;
> >
> >    bcistatus = pbc635->intstat; /* Save copy of status register */
> >    bcimask   = pbc635->mask; /* Save copy of interrupt mask */
> >    /* First check for periodic interrupt */
> >    if ( ((bcistatus & 0x02) & bcimask) != 0)
> >    {
> >       if (bcUseper) /* Using BC periodics for system clock? */
> >       {
> >          if (!HadPerint) /* First Periodic Interrupt? */
> >          {
> >             HadPerint = TRUE;
> >             sysClkDisable(); /* Turn off system clock interrupts */
> >             sysClkRateSet(tickFrequency); /* set clock rate */
> >          }
> >          bcIntCounter++; /* Increment interrupt count */
> >          if (bcUsrClock != NULL) (*bcUsrClock)(bcIntCounter); /* Call
> user
> > routine */
> >          if (bcIntCounter%bcIntPerTick == 0) /* If right number of ticks
> */
> >             tickAnnounce(); /* Announce system clock tick */
> >       }
> >       pbc635->intstat = pbc635->intstat | 0x02; /* Clear interrupt status
> > bit */
> >    }
> >    /* Other interrupts - not periodic pulse */
> >    for (intnum = 1; intnum < 5; intnum++)
> >    {
> >       if (intnum != 2)
> >       {
> >          tmask = 0x01 << (intnum - 1); /* Test value for interrupt */
> >          /* Check for interrupt bit set - request I/O scan, but only if
> > ioscanpvt is valid */
> >          if ( ((bcistatus & tmask) & bcimask) != 0)
> >          {
> >             if (ioscanpvt[intnum-1] != NULL)
> >                scanIoRequest(ioscanpvt[intnum-1]);
> >             pbc635->intstat = pbc635->intstat | tmask;   /* Clear
> interrupt
> > status bit */
> >          }
> >       }
> >    }
> > }
> >
> >
> > --
> > Mike Westfall
> > Gemini Observatory
> >
> > _______________________________________________
> > users mailing list
> > users at rtems.org
> > http://lists.rtems.org/mailman/listinfo/users
>



-- 
Mike Westfall
Gemini Observatory
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20160305/45255fd7/attachment-0002.html>