Bug 107861 – [2.6.12 regression] 3c59x driver broken

• Product: Gentoo Release Media
• Status: VERIFIED UPSTREAM

[Description Thomas 2005-10-02 03:40:31 UTC]

This is in fact a linux kernel bug: since 2.6.12 i am unable to access the 3COM
3c590 vortex in my system.  I use the BNC output of the network card (what is
properly detected by the driver), but since the packets don't seem to reach the
driver i doubt, that this matters.

Since the live CD 2005.1 uses a 2.6.12 kernel, this bug prevents me from
installing Gentoo.

Reproducible: Always
Steps to Reproduce:
1. Boot live CD (network adapter is properly detected)
2. Configure eth0
3. Ping other machine on subnet

Actual Results:  
The tx counter does not increase.


Expected Results:  
send out packets and receive responses

[Comment 1 Chris Gianelloni (RETIRED) 2005-10-03 05:15:54 UTC]

Hrrrmn... the driver works fine for me, but I'm using 3C905 and 3C920 chips.

Adding kernel team to CC to see if this is resolved.  Can you boot with older
media and perhaps determine if this has been resolved in a newer kernel?

[Comment 2 Daniel Drake (RETIRED) 2005-10-03 05:26:34 UTC]

There have been power-management related problems come and go recently. It would
be useful to know which "kernels since 2.6.12" have been tested.

[Comment 3 Thomas 2005-10-03 05:58:00 UTC]

I have checked all kernels since 2.6.12 (vanilla kernels from kernel.org):
all 2.6.12 minor versions, all 2.6.13 minor versions until .2, and 2.6.14-rc3
just yesterday.  There also is an entry in the kernel bugzilla (#4818) which
describes this problem; i have attached lots of output there from the working
2.6.11.12 kernel as well as from the not working 2.6.12 and .13.

Regarding Power management: The ACPI bios of this (rather old) motherboard is
not recognized by the kernel; in all the kernels i compiled myself, i have hence
switched it off.

Maybe this is of relevance: when i boot into my home grown system, all eth0
packet counters remain 0.  If i boot the gentoo image, the tx counter is 2,
maybe due to dhcp?

If there is anything i can do to provide more meaningful information, please let
me know.

[Comment 4 Daniel Drake (RETIRED) 2005-10-03 06:11:51 UTC]

will bounce back after fix

[Comment 5 Daniel Drake (RETIRED) 2005-10-03 06:12:08 UTC]

will track upstream bug

[Comment 6 Chris Gianelloni (RETIRED) 2005-11-26 09:42:56 UTC]

Quick question for the OP... what if you boot the CD with acpi=off?

[Comment 7 Chris Gianelloni (RETIRED) 2006-03-15 06:00:03 UTC]

Anybody know if this has been fixed in newer kernels?  I see that the bug is still open upstream, but it could be possible it was fixed and nobody noticed... ;]

[Comment 8 Thomas 2006-03-21 01:05:13 UTC]

The bug still persists in 2.6.16-rc6 as well as in 2.6.15.6.  I compiled both of these kernel on Sunday and the results didn't change since 2.6.12.

I furthermore noticed, that the problem only seems to occur when the NIC is connected to a cheapernet cable (which seems to be less common nowadays ;-).  As soon as i connect the NIC to an (otherwise unconnected) 10baseT switch and ping something, the counters shown by ifconfig increase.

[Comment 9 Daniel Drake (RETIRED) 2007-04-30 12:17:59 UTC]

reporter is no longer using this hardware