- Aug 31, 2006
- 8,307
- 176
- 106
I was running Debian Lenny 32 bit and the Intel cards (had an MT and GT in the server) worked fine. Decided to move to 64 bit Debian and performed a fresh install. During the install, it used my onboard LAN (Realtek chipset) to grab latest install files.
After the install any time I tried configuring the Intel NICs statically or with DHCP (either the MT, GT, or a PT I had laying around) all of them would hang the system at reboot with:
e1000: eth0: e1000_clean_tx_irq: Detected Tx Unit Hang
If I boot to CD, mount /, and remove the configuration of the cards, I can reboot successfully. But the moment I reconfigure either of them and restart the networking service the error rears its ugly head. The NICs will take the config but I can't ping anywhere on the network.
From Googling around, I've found this is a problem with 2.6 kernels and Intel NICs when paired with either 4GB RAM or the AMD SB700 chipset.
I'm running 2.6.26-2-AMD64 kernel.
Some have suggested an EEPROM update, manually configuring duplex and speed with ethtool, or going to an even newer kernel but with mixed results.
Anyone had this happen on their Linux systems? I've read it's a common bug with 64 bit 2.6 kernels. Were you able to fix it?
After the install any time I tried configuring the Intel NICs statically or with DHCP (either the MT, GT, or a PT I had laying around) all of them would hang the system at reboot with:
e1000: eth0: e1000_clean_tx_irq: Detected Tx Unit Hang
If I boot to CD, mount /, and remove the configuration of the cards, I can reboot successfully. But the moment I reconfigure either of them and restart the networking service the error rears its ugly head. The NICs will take the config but I can't ping anywhere on the network.
From Googling around, I've found this is a problem with 2.6 kernels and Intel NICs when paired with either 4GB RAM or the AMD SB700 chipset.
I'm running 2.6.26-2-AMD64 kernel.
Some have suggested an EEPROM update, manually configuring duplex and speed with ethtool, or going to an even newer kernel but with mixed results.
Anyone had this happen on their Linux systems? I've read it's a common bug with 64 bit 2.6 kernels. Were you able to fix it?
Facebook
Twitter