[ECOS] Redboot "reset" command on iq80321 eval board

[Mark Salter]

>>>>> Rick Richardson writes:

> I'm wondering if anyone has seen this behavior before, and knows of
> a fix.

> I have RedBoot in flash on an iq80321 eval board.  The board is
> plugged into a PC, and switches S7E1-X are all set to factory.

> After a hard reset, Redboot comes up fine, with the startup messages
> displayed on the serial port, and the LEDs going thru the normal
> sequence to state "A1".  The board is also reachable via the
> ethernet port.

> I then issue the RedBoot "reset" command.  This causes a soft reset by
> setting a couple of bits in the ATU PCSR, i.e. exactly what a device
> driver on the host side of the bus would want to do if it wanted to
> reset a board that had gone out to lunch.

> The board then goes thru the normal sequence of LED states (scrub
> loop SL, etc.) until it reaches state "A1".  But this time, the
> startup message doesn't come out the serial port, and the board
> cannot be reached via ethernet either.

> It looks like "something bad" happens at or after the point where
> vectors.S does the "bl cyg_START".

> Anybody seen this and have a fix?  I need the soft reset to work!

Hi Rick,

I think I know what is happening here, but I'm not sure how to fix
it. When the 80321 is in a PC, it needs to rely on the PC BIOS to
setup some BARs on the iop321 processor. RedBoot can then determine
the address range that can be used for the PCI-X bus that the
ethernet is on. When you reset the iq80321, but not the PC, RedBoot
probably uses incorrect addresses to setup the ethernet. You can
test this theory by disabling the ethernet with S8E1-2 off.

--Mark



-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss