Re: [Openocd-development] new lpc2xxx cfg files layout

[David B. <da...@pa...>]

On Wednesday 23 September 2009, Petri Pirttinen wrote:
> > If you don't see what's wrong about forcing every
> > single LPC2xxx chip to use "12 MHz" even if that's
> > not the *correct* frequency for their board ... I
> > don't know what more I could try to explain.
> 
> I may have missed something in here, but why trying to specify a crystal 
> frequency? Can't we use 4 MHz in here, IF we are setting the clock 
> source to internal RC oscillator in reset-init script anyway?

I think Freddie pointed out that the frequency is used in two ways:
(a) configuring the flash write timings; (b) deriving the JTAG clock
rate, which seems to need about 1/8 that clock.  Some such frequency
setting is needed on most targets ... (b) can be avoided if RTCK is
in use, but I don't know that these LPC chips support it.

As for clock selection, crystal vs internal RC etc ... that problem
is related to the "reset init" issues.  It appears that the LPC2xxx
chips can't achieve halt-after-reset, so that some board-specific init
code is going to run ... and may set up the clock to whatever that
board needs.

Now, I think Freddie was being justifiably cautious and avoiding
that reset-init issue with this patch.  But this clock setup issue
doesn't seem like it can be avoided.  Either there's a new reset-init
scheme which sets up a known clock (ideally a crystal clock, since
those RC oscillators can be rather inaccurate) -OR- a board-specific
input parameter tells what value the on-chip setup code will have
arranged.


> And, Freddie, I hope you can bear with this conversation. I think that 
> these things discussed in here does relate to each other, although some 
> of them needs to be addressed separately really. In all and all I'm 
> quite thrilled about this as finally I get some light shred to this 
> LPC2xxx mess. Your proposed script layout is A LOT better than any I've 
> seen so far.
> 
>