Duplicate interfaces on e1000

Jose Celestino japc at co.sapo.pt
Mon Jun 11 16:56:55 CDT 2007


Words by Sam Flory [Mon, Jun 11, 2007 at 02:22:33PM -0700]:
> Jose Celestino wrote:
> > I am having the weirdest behaviour on a Dell PE 1850, with Debian 4.0,
> > kernel 2.6.18-4-686-SMP and Intel(R) PRO/1000 Network Driver - version
> > 7.1.9-k4-NAPI.
> 
> 
> > e1000: eth0: e1000_probe: Intel(R) PRO/1000 Network Connection
> > PCI: Enabling device 0000:03:06.0 (0110 -> 0113)
> 
> > ACPI: PCI Interrupt 0000:03:06.0[A] -> GSI 37 (level, low) -> IRQ 225
> 
> > e1000: 0000:03:06.0: e1000_probe: (PCI-X:120MHz:64-bit) 00:04:23:08:be:22
> > e1000: eth0: e1000_probe: Intel(R) PRO/1000 Network Connection
> > PCI: Enabling device 0000:03:06.1 (0110 -> 0113)
> 
> > ACPI: PCI Interrupt 0000:03:06.1[B] -> GSI 38 (level, low) -> IRQ 233
> 
> > e1000: 0000:03:06.1: e1000_probe: (PCI-X:120MHz:64-bit) 00:04:23:08:be:23
> > e1000: eth0: e1000_probe: Intel(R) PRO/1000 Network Connection
> > PCI: Enabling device 0000:04:0b.0 (0110 -> 0113)
> 
> > ACPI: PCI Interrupt 0000:04:0b.0[A] -> GSI 37 (level, low) -> IRQ 225
> 
> > e1000: 0000:04:0b.0: e1000_probe: (PCI-X:133MHz:64-bit) 00:04:23:cb:a2:66
> > e1000: eth0: e1000_probe: Intel(R) PRO/1000 Network Connection
> > PCI: Enabling device 0000:04:0b.1 (0110 -> 0113)
> 
> > ACPI: PCI Interrupt 0000:04:0b.1[B] -> GSI 38 (level, low) -> IRQ 233
> 
> > e1000: 0000:04:0b.1: e1000_probe: (PCI-X:133MHz:64-bit) 00:04:23:cb:a2:67
> 
> 
> > What am I missing?
> 
>   You are sharing interrupts.  Even if this worked for you it's a bad

Yes. But why shouldn't it work anyway? At least acording to
http://www.intel.com/support/network/adapter/pro100/sb/cs-008491.htm it
should:

"The PCI specification does allow for interrupt sharing between PCI
devices (but not non-PCI devices). The drivers for Intel PRO adapters
are designed to share interrupts with other PCI devices."

Will try to reduce the IRQ sharing for performance reasons anyway.

> idea.  I'd be sure you have the latest bios, and that you've tried the
> card in every slot. Also you might have better luck with 2 dual port
> cards than a quad port.  (Assuming you have the solts.)  As a last
> resort you might try disabling every on board device you aren't using in
> the bios.
> 

Well, think I found out what is causing this. Very simple theory :] udev
is renaming the devices as they appear. On /etc/udev/rules.d/z25_persistent-net.rules:

# PCI device 0x8086:0x1076 (e1000)
SUBSYSTEM=="net", DRIVERS=="?*", ATTRS{address}=="00:13:72:63:90:da", NAME="eth2"

# PCI device 0x8086:0x1076 (e1000)
SUBSYSTEM=="net", DRIVERS=="?*", ATTRS{address}=="00:13:72:63:90:db", NAME="eth3"
....

So it matches the behaviour. Can't test it right now but bet it is this.

-- 
Jose Celestino
----------------------------------------------------------------
http://www.msversus.org/     ; http://techp.org/petition/show/1
http://www.vinc17.org/noswpat.en.html
----------------------------------------------------------------
"And on the trillionth day, Man created Gods." -- Thomas D. Pate



More information about the Linux-PowerEdge mailing list