Hija,
I just bought a new motherboard "ECS L7VTA" sporting a VIA KT400 chipset
and found an annoying bug which took me quite some time to track down:
As soon as I enable the APIC mode in the BIOS the onboard PHY seems
to ignore any packets which are thrown at it *after* the kernel
initialised itself which is especially nasty since the system is booting
from network effectively stopping its boot when trying to get an IP
using DHCP or mounting a NFS volume in case the IP is fixed. The onboard
NIC is a VIA Rhine II (VT6102).
The startup process looks like:
- POST
- BIOS check
- PXE BIOS initialisation
- PXE boot into etherboot
- Correct detection and initialisation of the NIC in etherboot
- Boot of linux kernel
- Correct initialisation of system including NIC (via-rhine driver from
Donald Becker as in the standard kernels)
- Endless loop like the following:
-------->
eth0: Setting full-duplex based on MII #1 link partner capability of 41e1.
Sending DCHP requests ...... timed out!
<--------
The NIC initialised itself with the correct interrupt according to the
BIOS screen. As soon as I shut down the APIC mode, everything works as
expected.
Ideas?
--
Servus,
Daniel
On Thu, 13 Mar 2003 19:58:21 +0100, Daniel Egger wrote:
> As soon as I enable the APIC mode in the BIOS the onboard PHY seems
> to ignore any packets which are thrown at it *after* the kernel
> initialised itself which is especially nasty since the system is booting
> from network effectively stopping its boot when trying to get an IP
> using DHCP or mounting a NFS volume in case the IP is fixed. The onboard
> NIC is a VIA Rhine II (VT6102).
You may want to try 2.4.x-ac kernels, I believe Alan fixed some VIA APIC
issues.
Roger
Am Don, 2003-03-13 um 21.24 schrieb Roger Luethi:
> > As soon as I enable the APIC mode in the BIOS the onboard PHY seems
> > to ignore any packets which are thrown at it *after* the kernel
> > initialised itself which is especially nasty since the system is booting
> > from network effectively stopping its boot when trying to get an IP
> > using DHCP or mounting a NFS volume in case the IP is fixed. The onboard
> > NIC is a VIA Rhine II (VT6102).
> You may want to try 2.4.x-ac kernels, I believe Alan fixed some VIA APIC
> issues.
According to Alans changelog the 2.5-ac has a forwardport of the VIA
interrupt line patch, after some difficulties applying 2.5.64-ac3 to
a rsynched post 2.5.64 kernel (some parts of the console changes didn't
apply cleanly) I now tried it and it shows exactly the same symptoms.
Are there more VIA patches floating aroung? A short search on google
didn't find anything particularly interesting.
Almost more annoying is that even after removing the fb support I cannot
see enough of the messages to be helpful here, neither scrolllock nor
shift-pgup help, probably also an interrupt issue though a bit seems
to go through:
atkbd.c: Unknown key (set 2, scancode 0xb6, on isa0060/serio0) pressed.
--
Servus,
Daniel