Procedure to replace a NIC - RESOLVED?

[Steven Stromer]

Steven Stromer wrote:
>> Robert L Cochran wrote:
>>>> I noticed on the Realtek site that there is no downloadable driver 
>>>> for the 2.6 kernel. 
>>>
>>>
>>> http://www.realtek.com.tw/downloads/downloads1-3.aspx?lineid=1&famid=4&series=2003072&Software=True 
>>>
>>>
>>> Download the source code. Unzip the tarball.
>>>
>>> There is a C program in the src directory that has to be corrected
>>> before it will compile:
>>>
>>> http://gentoo-wiki.com/HARDWARE_RTL8168
>>>
>>>> From the r1000_v1.04 directory:
>>>
>>> su
>>> make
>>> depmod -a
>>>
>>> The module will build and install correctly with no changes to any of
>>> the Makefiles.
>>>
>>> Probably you will find the tulip driver has been erroneously loaded. Get
>>> rid of it:
>>>
>>> rmmod tulip
>>>
>>> change to the correct /lib/modules.../net directory then
>>>
>>> insmod r1000.ko
>>>
>>> This should get the Realtek working.
>>>
>>> One last bit of advice -- it took me about 15 minutes of Googling to
>>> locate the links above. That was with time to sip coffee. I got my
>>> Realtek 8168 gigabit adapter for PCIe running as soon as I did the
>>> insmod, and I'm using it now to send this email. Why not try Google a
>>> little more in the future?
>>>
>>> Bob Cochran
>>>
>>>
>> Steven --
>>
>> The r8169 driver sounds like it might have its roots based on the 2.4
>> kernel. I would remove it (rmmod it) and change modprobe.conf to read
>>
>> alias eth0 r1000
>>
>> and reboot.
>>
>> It would be useful to examine your dmesg output. I don't know how else
>> you have changed your network scripts. If my suggestion does not fix
>> your problem, send me your dmesg output as a text file attachment. You
>> can plug a USB flash drive in your machine and then
>>
>> dmesg > /put the name of your usb drive here/dmesg.txt
>>
>> I have several other things to do today, and may not be able to look
>> over your output till tonight, but I'll review it.
>> Bob
> 
> 
> Bob,
> 
> Thank you for your help with my problem. I had changed modprobe.conf to 
> read 'alias eth0 r1000' when I last reported no success, so that hasn't 
> resolved the problem. I rmmod'ed the r8169 driver, but this seems to 
> remove it only until the system is rebooted. dmesg returns some 
> interesting output that may point us to the problem, but I am not able 
> to understand it fully. The pertinent part is as follows:
> 
> PCI: Found IRQ 11 for device 0000:00:0f.0
> PCI: Sharing IRQ 11 with 0000:00:07.5
> PCI: Sharing IRQ 11 with 0000:00:0d.2
> eth0: Identified chip type is 'RTL8169S/8110S'.
> eth0: r10001.04, the Linux device driver for Realtek Ethernet 
> Controllers at 0xc800, 00:40:f4:ee:2f:ff, IRQ 11
> BUG: soft lockup detected on CPU#0!
>  <c043dad5> softlockup_tick+0x90/0xa1  <c0423d50> 
> update_process_times+0x35/0x57
>  <c0406134> timer_interrupt+0x60/0x98  <c043db72> 
> handle_IRQ_event+0x23/0x4c
>  <c043dc17> __do_IRQ+0x7c/0xd1  <c0405035> do_IRQ+0x63/0x80
>  =======================
>  <c04036f2> common_interrupt+0x1a/0x20  <c040d556> delay_tsc+0xb/0x13
>  <c04d8125> __delay+0x9/0xa  <e88160bd> R1000_WRITE_GMII_REG+0x36/0x3c 
> [r1000]
>  <e8817208> r1000_init_one+0x69f/0x892 [r1000]  <c053d851> 
> __driver_attach+0x0/0x8b
>  <c04dfc3b> pci_device_probe+0x36/0x57  <c053d78d> 
> driver_probe_device+0x45/0x9a
>  <c0419231> __cond_resched+0x17/0x28  <c053d8b4> __driver_attach+0x63/0x8b
>  <c053d21f> bus_for_each_dev+0x37/0x59  <c053d6ee> driver_attach+0x16/0x18
>  <c053d851> __driver_attach+0x0/0x8b  <c053cf25> bus_add_driver+0x6f/0x10d
>  <c04dfddf> __pci_register_driver+0x5b/0x7d  <c0433a69> 
> sys_init_module+0x134f/0x14b2
>  <c0423c25> __mod_timer+0x0/0x94  <c0402cb3> syscall_call+0x7/0xb
> Realtek RTL8169/8110 Family Gigabit Ethernet Network Adapter
> Driver version:1.04
> Released date:2006/07/07
> Link Status:Linked
> Link Speed:1000Mbps
> Duplex mode:Full-Duplex
> I/O Base:0xC800(I/O port)
> IRQ:11
> 
> I am trying to research possible causes of the 'BUG: soft lockup 
> detected on CPU#0!', but the lines here are causing me nightmares! on 
> the other hand, there is much more recognition of the card in the final 
> lines than before the switch to the r1000 driver. If you need the rest 
> of the dmesg output, please let me know. I tried to parsse it down to 
> what I thought to be the essentials.
> 
> Thanks again for your help!
> 
> Steven
> 

An update...

This problem was occurring on a fresh install of FC5, so I decided that 
it would take less time to reinstall than any other solution, as I 
NEEDED this NIC working. When I reinstalled FC5 with the NIC in the 
machine, it configured without a hitch. I know that this was a cop-out 
to resolving the problem, but I needed to get on with my life. It also 
shows that this is not a simple 'plug and play' situation, as a warning 
for others who have limited experience with handling such a situation. 
If you return to the beginning of this thread, I was led to believe that 
having the right driver as a module available to the kernel was the only 
item I would have to check on (which was evidently the case, based on 
the successful reinstall), and that if this was the case, the install 
would go flawlessly; three sleepless nights later, I know better. When I 
get the chance, I will try to figure out what files weren't updating 
properly. Thanks again for all the attempts to help me, with special 
thanks to Bob Cochran!

Steven