Weird thing - pciconf reading differs in two systems

[Milan Obuch]

Hi,
I was trying to test miniPCI atheros-based card in wrap, using 7-CURRENT. Some 
time ago, it attached as ath0, I did some configuration and it looked OK. 

Yesterday, I tested it again with system built from sources cvsupped at May 
12. No attachment, pciconf -lv said (relevant device)

none0 at pci0:13:0: class=0x020000 card=0x00000000 chip=0xff16168c rev=0x01 
hdr=0x00
    vendor   = 'Atheros Communications Inc.'
    class    = network
    subclass = ethernet

From some earlier conversation I got chips with ID beginning 0xFF are 
pre-production and should not be used.

Just out of curiosity I put this card in my tablet TC1000, with system built 
from sources cvsupped at May 15 and got

ath0 at pci0:10:0: class=0x020000 card=0x000d17f9 chip=0x001b168c rev=0x01 
hdr=0x00

This makes me wonder - how could the same card be identified as two different 
devices? I tested the same card, then another one and the result is the same. 
Working in my tablet, not working in WRAP. There is one difference, however - 
kernel in WRAP has ath built-in, in my tablet I am kldloading it.

If it were the other way round, I would be happier - I can use those cards for 
some radio links, but it would not be less surprising.

Any ideas?

Regards,
Milan

-- 
No need to mail me directly. Just reply to mailing list, please.