Skip to contentSkip to conversation

e1000 only works in 2.6.3 with UP kernel ?

DISCUSSION Patrick Richard 3 participants 5 messages Feb 25, 2004 – Feb 26, 2004

lore
Share & tools
lore
#0 lore raw

Hi,

I have been struggling to get the e1000 driver working _at all_ in 2.6.3. It works fine under 2.4.xx (same machine). This is all on an ASUS P4C800-E deluxe, using onboard e1000 and P4 with HT enabled.

I have tried with / without apic, ACPI (boot option and compiled in), SCSI, libata. Depending on what I have configured in the kernel, it will stop at various places, for example, sometimes we will get a solid lockup when running ifconfig (with ACPI and apic enabled), whereas elsewhere, (no ACPI compiled in, i.e. not just disabled on boot args) it will lock not on ifconfig, but will on first attempy to use the card (i.e. ping thyself).

Anyhow, the *only* configuration I have managed to get working is on a UP-built kernel, built with SMP disabled. Despite having an affect on where it would lock up in the SMP kernel, it seemed to not matter if APIC or ACPI or libata etc. was enabled or not in the UP kernel. If I disabled the SMP at boot with bootargs, it still didn't work.

Is anyone else seeing similar problems ? Seems like an interrupt/ACPI type of thing to me based on what is affecting it but what do I know.

Does anyone have this working on an SMP kernel/machine with this motherboard ? Is this a mis-config on my part ? Like I said it works fine in 2.4 and in UP kernel.

-Pat

#1 lore raw
Quoted 5 lines · Patrick Richard

Hi,

I have been struggling to get the e1000 driver working _at all_ in 2.6.3. It works fine under 2.4.xx (same machine). This is all on an ASUS P4C800-E deluxe, using onboard e1000 and P4 with HT enabled.

[snip]

Quoted 6 lines

Is anyone else seeing similar problems ? Seems like an interrupt/ACPI type of thing to me based on what is affecting it but what do I know.

Does anyone have this working on an SMP kernel/machine with this motherboard ? Is this a mis-config on my part ? Like I said it works fine in 2.4 and in UP kernel.

This is a known problem with 2.6.3
Please try the latest 2.6.3-bk snapshot, the patch that introduced this problem has been reverted.

Signature · 1 line
/Martin
#2 lore raw
Quoted 5 lines · Martin Josefsson

This is a known problem with 2.6.3
Please try the latest 2.6.3-bk snapshot, the patch that introduced this problem has been reverted.

Thanks,

I have just tried bk-7 in SMP with APIC and ACPI enabled and it didn't hang on ifconfig... now testing.

-Pat

#3 lore raw
In reply to Patrick Richard
Quoted 3 lines · Patrick Richard

Hi,

I have been struggling to get the e1000 driver working _at all_ in 2.6.3. It works fine under 2.4.xx (same machine). This is all on an ASUS P4C800-E deluxe, using onboard e1000 and P4 with HT enabled.

Does 2.6.3-bk7 change things?

Jeff

#4 lore raw
Quoted 10 lines · Jeff Garzik
Quoted 3 lines · Patrick Richard

Hi,

I have been struggling to get the e1000 driver working _at all_ in 2.6.3. It works fine under 2.4.xx (same machine). This is all on an ASUS P4C800-E deluxe, using onboard e1000 and P4 with HT enabled.

Does 2.6.3-bk7 change things?

Jeff

Yes,

with ACPI turned on etc. now it no longer hangs on ifconfig (or at first use), and I am succesfully linked up and transferring files. I havn't done any huge tests but so far so good. This is with HT, SMP, ACPI, apic, all enabled. Incidentally now /proc/interrupts is showing better:

Quoted 2 lines
18:    5    0  IO-APIC-level  libata, eth0
23:    0    0  IO-APIC-level  libata

(of course, this is just after reboot)

-Pat