Bug#426433: xserver-xorg: X server tries to get PCI I/O memory of NIC
reassign 426433 xserver-xorg-core
found 426433 2:1.1.1-21
forcemerge 415543 426433
thank you
Manuel Tobias Schiller wrote:
> After upgrading my beige PowerMac G3 to etch, I noticed that X does not
> come up.
It's actually the fourth report like this. Merging with the others.
> Investigating closer, I found that X apparently tries to
> allocate the PCI I/O memory which belongs to the network adapter in the
> system (see X server log and lspci -v output below). The NIC in question
> is a 3Com Corporation 3c905B 100BaseTX. To make sure that it is not the
> network adapter's fault, I swapped networks cards: Neither the one based
> on the RTL-8193D chipset nor the via-rhine based one could be made
> working. I also tried all the three PCI slots that are available in the
> system. The situation for me is now as follows: Either I don't load a
> driver for the NIC, and X runs fine, or I load a driver and ifup the
> card, and X won't even start. Loading the driver after X has started is
> no good either: My machine will just freeze then.
>
That's a great analysis, thanks a lot. I've been wondering for a while
what was going on on these powermac G3 machine). Now, it's pretty clear.
> 00:0d.0 Ethernet controller: 3Com Corporation 3c905B 100BaseTX [Cyclone] (rev 30)
> I/O ports at fe000400 [size=128]
>
> 00:12.0 VGA compatible controller: ATI Technologies Inc 3D Rage Pro 215GP (rev 5c) (prog-if 00 [VGA])
> I/O ports at fe001000 [size=256]
>
[...]
> (EE) ATI(0): Unable to register the following bus resources:
> [0] 0 0 0xfe000400 - 0xfe0004ff (0x100) IS[B]
Clearly, this range is wrong.
I don't think anybody in the other related bugs already tested with
xserver-xorg-core 1.3 (currently in unstable). I'd like to know whether
the problem occurs there too before forwarding the bug upstream. Could
you try with xserver-xorg-core 1.3?
Thanks again.
Brice
Reply to: