How to prevent embedded ppc reset deadlock? (MPC83xx/85xx)

Leon Woestenberg leon.woestenberg at gmail.com
Tue Sep 30 08:05:01 EST 2008


Hello André,

On Mon, Sep 29, 2008 at 11:50 PM, André Schwarz
<Andre.Schwarz at matrix-vision.de> wrote:
> Leon,
>
> you're right.
> PORESET is just there to prevent the core from running as long as power may
> be unstable and/or PLLs are out of lock.
> HRESET is the signal that should reset everything. I did it on my board and
> it works fine.
>
Understood so far.

> Since you also have to assert HRESET when you assert PORESET
>
But when I assert PORESET, the processor will assert HRESET itself
AFAIK, so why do this?

> you can wire-or them with a low drop schottky diode.
>
Ooh, analog electronics, long time ago. Let me think: arrow of diode
symbol pointing from HRESET# to PORESET#, right, so that PORESET#
going low will pull HRESET# low enough, right?

> Hope this helps.
>
Yes it does, thanks.

Regards, Leon.


> Leon Woestenberg wrote:
>>
>> Hello all,
>>
>> not Linux related per se*, but I wonder how your board designs deal
>> with the reset circuitry for embedded PowerPC processors (MPC8313E in
>> my case).
>> My requirement is that both a processor-external hard reset and
>> processor-internal hard reset must both reset the boot device NOR
>> FlashROM, so that it does not remain in write mode (if it is).
>>
>> Given those processor pins:
>>
>> PORESET# (input pin to the processor, power on reset)
>> HRESET# (bidirectional pin on the processor, asserted by processor on
>> hard reset such as watchdog)
>>
>> I see many designs (even the Freescale reference designs) where the
>> HRESET# resets some of the board, but not the FlashROM, and where
>> PORESET# resets the FlashROM. This can cause a deadlock in the case
>> where the watchdog resets during writing to FlashROM, as the FlashROM
>> is not reset and remains in write mode, not allowing the processor to
>> boot from it.
>>
>> I am thinking of using this approach: PORESET# -> processor <-->
>> HRESET# -> board reset.
>>
>> Would that work? or why not?
>>
>> Regards,
>>
>
>
> MATRIX VISION GmbH, Talstraße 16, DE-71570 Oppenweiler  - Registergericht:
> Amtsgericht Stuttgart, HRB 271090
> Geschäftsführer: Gerhard Thullner, Werner Armingeon, Uwe Furtner
>



-- 
Leon



More information about the Linuxppc-dev mailing list