Network dies after video card install

ivanatora · 06-20-2007, 10:21 AM

All that happened after installing new nvidia video card (my old one was integrated into the MB).
Network dies randomly 5-15 min after boot. No ping, no www, nothing. I tried ifconfig eth0 down, up, removing and inserting the network card module - nothing helps.
Here is what I've got several times in dmesg:

Code:

NETDEV WATCHDOG: eth0: transmit timed out

After reading from some sites, I thought the problem is something ACPI-related. I precompiled the kernel without any kind of ACPI/AMP features and tried again with no result. I tried that in lilo.conf to the current image: append=pci-noacpi with no result.
Here are some stuff I got in dmesg about ACPI:

Code:

PCI->APIC IRQ transform: 0000:00:01.0[A] -> IRQ 16
PCI->APIC IRQ transform: 0000:00:1b.0[A] -> IRQ 16
PCI->APIC IRQ transform: 0000:00:1c.0[A] -> IRQ 16
PCI->APIC IRQ transform: 0000:00:1c.1[B] -> IRQ 17
PCI->APIC IRQ transform: 0000:00:1d.0[A] -> IRQ 23
PCI->APIC IRQ transform: 0000:00:1d.1[B] -> IRQ 19
PCI->APIC IRQ transform: 0000:00:1d.2[C] -> IRQ 18
PCI->APIC IRQ transform: 0000:00:1d.3[D] -> IRQ 16
PCI->APIC IRQ transform: 0000:00:1d.7[A] -> IRQ 23
PCI->APIC IRQ transform: 0000:00:1f.1[A] -> IRQ 18
PCI->APIC IRQ transform: 0000:00:1f.2[B] -> IRQ 19
PCI->APIC IRQ transform: 0000:00:1f.3[B] -> IRQ 19
PCI->APIC IRQ transform: 0000:03:00.0[A] -> IRQ 16
PCI->APIC IRQ transform: 0000:01:00.0[A] -> IRQ 17

My NIC is Realtek 8169 1Gbit, and my current kernel is 2.6.19

Tuttle · 06-21-2007, 09:08 PM

I had this problem, could it possibly be just a coincidence? My ISP upgraded me to a higher bandwidth, the speed was too high for the modem they gave me, causing it to drop out after 20 minutes or so.
The best way to test this is to switch back to your old video card and see if the problem remains.

ivanatora · 06-22-2007, 12:55 PM

Your problem was due to higher speed, probably flooding the device? I didn't understood what dropped - the modem hung or the NIC? Did you have exactly the same message in dmesg?
I'm sure to me is not due to flood, becouse I'm on 512 kbps and not traffic eater is turned on

All my network load is some www, skype logged in, maybe a SSH connection and that is all. I observe the traffic load and it doesn't hit the ceil. I have done it before - downloading with 2Mbps for whole night. And the problem does exist only from whence I've installed that video card.
I disabled all ACPI functions of my new kernel and the problem still remains.
I'm not sure if it something of interrupts?
Currently this is my:

Code:

cat /proc/interrupts 
           CPU0       
  0:      42639   IO-APIC-edge      timer
  1:        189   IO-APIC-edge      i8042
  2:          0    XT-PIC-XT        cascade
  8:          1   IO-APIC-edge      rtc
 12:       5678   IO-APIC-edge      i8042
 14:       5996   IO-APIC-edge      ide0
 16:      12039   IO-APIC-fasteoi   uhci_hcd:usb5, HDA Intel, nvidia
 17:       1747   IO-APIC-fasteoi   eth0
 18:          0   IO-APIC-fasteoi   uhci_hcd:usb4
 19:          0   IO-APIC-fasteoi   libata, uhci_hcd:usb3
 23:          0   IO-APIC-fasteoi   ehci_hcd:usb1, uhci_hcd:usb2
NMI:          0 
LOC:      42599 
ERR:          0
MIS:          0

As you can see 16 is for the video and 17 is for lan.
BTW I don't know why linux can't identify my video card, this is from lspci -v:

Code:

01:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 01)
        Subsystem: ASRock Incorporation Unknown device 8168
        Flags: bus master, fast devsel, latency 0, IRQ 17
        I/O ports at d800 [size=256]
        Memory at fbfff000 (64-bit, non-prefetchable) [size=4K]
        Expansion ROM at fbfc0000 [disabled] [size=128K]
        Capabilities: [40] Power Management version 2
        Capabilities: [48] Vital Product Data
        Capabilities: [50] Message Signalled Interrupts: 64bit+ Queue=0/1 Enable-
        Capabilities: [60] Express Endpoint IRQ 0
        Capabilities: [84] Vendor Specific Information

03:00.0 VGA compatible controller: nVidia Corporation Unknown device 0393 (rev a1) (prog-if 00 [VGA])
        Flags: bus master, fast devsel, latency 0, IRQ 16
        Memory at fd000000 (32-bit, non-prefetchable) [size=16M]
        Memory at d0000000 (64-bit, prefetchable) [size=256M]
        Memory at fc000000 (64-bit, non-prefetchable) [size=16M]
        I/O ports at ec00 [size=128]
        [virtual] Expansion ROM at febe0000 [disabled] [size=128K]
        Capabilities: [60] Power Management version 2
        Capabilities: [68] Message Signalled Interrupts: 64bit+ Queue=0/0 Enable-
        Capabilities: [78] Express Endpoint IRQ 0

Tuttle · 06-22-2007, 08:29 PM

Quote:

Originally Posted by ivanatora

Your problem was due to higher speed, probably flooding the device? I didn't understood what dropped - the modem hung or the NIC?

It was my modem. NIC was fine, the problem was not caused by my hardware or by Linux.

I am not sure where your problem lies, sorry I am of no help :/

Do you have a 10/100 MB/s card lying around? Maybe use it if you don't need the 1 GB/s bandwidth

Otherwise, I think maybe you should try the default 2.6 kernel (2.6.17.13 or 2.6.18) that ships with slackware, it has fixed hardware problems for me, and might show you where your new kernel is going wrong.

ivanatora · 06-23-2007, 02:37 AM

I'm thingking indeed for separate NIC card instead of the integrated one

Default Slackware kernel, you mean the default config? Becouse I use the production 2.6.19 from kernel.org without any patches. Today I'm going to recompile again... to get ACPI back

Now without these advanced power managments when I shut down my PC it wont power off anything

Tuttle · 06-23-2007, 07:46 AM

Yep, use the default config for the 2.6.18 kernel, do a make oldconfig, make, blablabla and see what happens!