HELP: clock_gettime() problem?

[Joel Sherrill]

Chris Johns wrote:

> Joel Sherrill wrote:
> 
>>
>> I am willing to discuss an optional BSP specific hook to return
>> nanoseconds since the last clock tick if the general consensus
>> is that it is needed.
>>
> 
> I think there is a need. The capture engine is degined to take a pointer 
> to a function that provides more accurate timming (usec). If the RTEMS 
> clock code could be designed to use the same function that would be great.
> 
> Are you thinking about a BSP registering a handler ? This way BSPs who 
> do not need such support do not have to provide it.

That sounds good to me.  I wasn't thinking of anything in particular
except that a need exists. :)

If you can put together a patch and a documentation patch to the
BSP and Device Driver Guide, then I will apply it to the trunk.

I think that at least some of the BSPs in the tree should use
it though.  I would think that psim, erc32, and jmr3904 as
a minimum need to have it early in the game.  With psim going
to the new exception model, that might end up getting it
fixed for more powerpc bsps.

--joel