On 2006-08-08, Steve Underwood <ste...@coppice.org> wrote:
>>I have similar symptoms. It seems to be caused by the fact I'm
>>powering the target via the JTAG interface, and msp430-gdbproxy
>>isn't waiting long enough for the 32KHz oscillator to startup.
>>It only waits 250ms before it gives up on the target and shuts
>>off the power (the 32KHz oscillator takes about 700ms to
>>start).
>>
>>Once it works, it keeps working because msp430-gdbproxy doesn't
>>shut off the power when it exits. So the next time it runs,
>>the power has been on for >700ms and everything is fine.
>>
>>Once msp430-gdbproxy gets unhappy, it shuts off the power.
>>Thereafter it won't leave the power on long enough for the
>>target oscillator to start. If I power the target externally
>>for a second before starting msp430-gdbproxy, then it works.
>
> You can disconnect the 32kHz crystal, and gdbproxy will still work.
One would think it ought to. I'm stumped as to why gdbproxy
can't talk to my target board.
> If it didn't, designs which have no 32kHz oscillator could not
> be debugged. In fact, since debug operation does not set the
> load capacitors for the crystal (it has no why to know what
> the appropriate setting might be), in a large number of
> designs with a 32kHz crystal, the oscillator will probably not
> start.
>
> JTAG and Spy-Bi-Wire operation is completely independent of the 32kHz
> oscillator.
I'm glad to hear that -- I didn't think that JTAG ought to
depend on the CPU clock, but I couldn't see any other
difference between the times when gdbproxy can/can't find the
target.
--
Grant Edwards grante Yow! YOW!! I am having
at FUN!!
visi.com
