[U-Boot] last step to go - dead PHY

[David Collier]

My apologies, I've posted this is AVR32_u-boot where it probably belongs.

But I'm running out of time to ship some prototypes tomorrow, and this is
my last major issue with them, so could I throw myself on the mercy of
the assembled wisdom please? Maybe someone will say - you stupid person,
you need to change this line to that....

We have built a new design based on the ATNGW100, but we were pushed for
space, so we fitted the Nat Semi DP83849C dual PHY

It works great under Linux, but Linux seems to do a probe for PHYs at
multiple addresses on the MII busses.

Yes, I know the very latest u-boot sources can do similar, but I had to
freeze on 2008-10 because of the saveenv issue. Too late to update now.

Anyway, only my 2nd PHY works under u-boot, and that's not the WAN
interface for Linux ( as painted on the box ), so I can't use the same
cable for both.

Could anyone look at the problem report below and see if they can
instantly tell me what to change, or what to examine

Can I squeeze any more progress info out of u-boot ?

Truly grateful.

David Collier

www.dexdyne.com

============= question ==================

Both Ethernets work under Linux.. so it can't be too serious.

the linux startup says:

eth0: attached PHY driver [Generic PHY] (mii_bus:phy_addr=0:10, irq=-1)
eth1: attached PHY driver [Generic PHY] (mii_bus:phy_addr=1:00, irq=-1)

now we have strapped PHYAD4 high, so the 2 phys should have addresses
binary 10000 and 10001 

I can't see any relationship between 0:10 and 1:00 and 10000 and 10001
!!!

well maybe - the first one looks right - binary 10000 is hex 10
but the other port should be on hex 11, not hex 0!!!!!!

---------------
but under u-boot

in my file derived from ATNGW10.c I have

void board_init_info(void)
{
	gd->bd->bi_phy_id[0] = 0x01; 
	gd->bd->bi_phy_id[1] = 0x03; 
}

I THINK the first line refers to the PHY that doesn't work, and the 2nd
to the one that does, but I'm aware of the possibility of a cross-over 

but changing either/both those values seems to make no difference to what
happens....

I still get MACB0 "no PHY" and MACB1 connects...