Re: Current amd64, "nve0: device timeout" with nForce4 ethernet

>>How much memory do you have?
>>
>>I'm too having trouble with the nve driver under FreeBSD, but only if memory 
>>is mapped beyond the 4 GB mark.
>>
>>I have a Tyan S2895 mobo with a memory hole at 2.75 GB for PCI cards. In the 
>>BIOS I can decide how to remap this memory: Disabled/Hardware/Software
>>
>>If I choose to set this setting to "Disabled" my nve driver works fine, with 
>>any other setting I too get "device timeout"'s.
>>
> 
> 
> I have a similar timeouts under current with an NForce2 (Abit NF7) with 1GB of memory. My timeouts usually happen under heavy usage however.
> 
> nve0: <NVIDIA nForce MCP2 Networking Adapter> port 0xc000-0xc007 mem 0xe1004000-0xe1004fff irq 10 at device 4.0 on pci0
> nve0: Ethernet address 00:50:8d:fb:18:a2
> miibus0: <MII bus> on nve0
> rlphy0: <RTL8201L 10/100 media interface> on miibus0
> rlphy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
> nve0: Ethernet address: 00:50:8d:fb:18:a2
> nve0: [GIANT-LOCKED]


To clarify myself (after longer testing):

Without the memory clipping at 4 GB the nve driver doesn't work at all. I 
just receive a message
	nve0: device timeout(n)
with n any positive number (5..11 so far).

With
	options 	MAXMEM=4194303
in my kernel config file the nve driver works mostly.

Yesterday I was able to transfer ~500MB without any error. Then suddenly 
network traffic stops after
	nve0: device timeout(-1)
Only solution was a reboot (I may try compiling the nve driver as a module 
and see if unloading/reloading of the module also helps).

Today I'm getting tons of messages
	nve0: device timeout(4)
(68 of these messages for ~260MB traffic) but after 2 seconds the link works 
again.



Daniel