Bug 11298 - RTL8111C onboard NIC, intermittently, does not connect after boot.
Summary: RTL8111C onboard NIC, intermittently, does not connect after boot.
Status: CLOSED PATCH_ALREADY_AVAILABLE
Alias: None
Product: Drivers
Classification: Unclassified
Component: Network (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: Francois Romieu
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-08-10 00:09 UTC by Dave Plater
Modified: 2008-08-12 12:46 UTC (History)
2 users (show)

See Also:
Kernel Version: 2.6.26
Subsystem:
Regression: Yes
Bisected commit-id:


Attachments
nic not working version shown as ff (43.22 KB, application/octet-stream)
2008-08-10 06:04 UTC, Dave Plater
Details
boot.omsg is saved from a successful driver load (44.13 KB, text/plain)
2008-08-10 06:17 UTC, Dave Plater
Details
A good and bad hwinfo --netcard (3.82 KB, text/plain)
2008-08-10 06:18 UTC, Dave Plater
Details
Yast2 hardware info save (363.35 KB, text/plain)
2008-08-10 06:20 UTC, Dave Plater
Details
2.6.26 to 2.6.27-rc2 r8169 patch (9.76 KB, text/plain)
2008-08-10 06:47 UTC, Francois Romieu
Details

Description Dave Plater 2008-08-10 00:09:50 UTC
Latest working kernel version: 2.6.26
Earliest failing kernel version:2.6.24
Distribution:opensuse 11.0
Hardware Environment:MSI-945GCM7 MB w/celeron e1200 dual core
Software Environment:32 bit suse linux 11.0
Problem Description:RTL8111C onboard NIC, intermittently, does not connect after boot. when it doesn't connect version number shown as 0xff instead of 0x02. After googling RTL8111C I found similar problem reported for archlinux and ubuntu but no one has reported the version number change AFAIK. I had to compile the driver into (CONFIG_R8169=y) the kernels prior to 2.6.26 in order for the nic to work after boot 99% of the time. Whatever has happened in  the 2.6.26 version has enabled me to use the stock kernel with no changes but I now have the same behaviour as the compiled in driver in previous kernels. See https://bugzilla.novell.com/show_bug.cgi?id=412823 .

Steps to reproduce:2.6.25 kernel and 945GCM7 w/RTL8111C
Comment 1 Andrew Morton 2008-08-10 00:38:27 UTC
> Latest working kernel version: 2.6.26
> Earliest failing kernel version:2.6.24

I will assume that these are backwards, and shall mark it as a regression.
Comment 2 Dave Plater 2008-08-10 00:56:49 UTC
In case I don't understand the field properly. When I first had the 945GCM7 I had a 2.6.24 kernel installed. With 2.6.24 and 2.6.25 kernels I had to compile the driver into the kernel, as opposed to loading as a module, with the 2.6.26 kernel the problem is 99% solved, maybe I shouldn't have filled in the  Latest working kernel version: field?
Comment 3 Francois Romieu 2008-08-10 05:46:36 UTC
Dave Plater 2008-08-10 00:09:
[...]
> when it doesn't connect version number shown as 0xff instead of
> 0x02.

I guess 'lspci -vvxx' would claim an 'unknown header format' and
report only 0xff in the PCI configuration space of the device.

Similar reports have been fixed with commit (included in 2.6.27-rc1)
77332894c21165404496c56763d7df6c15c4bb09

Can you give 2.6.27-rc1 or 2.6.27-rc2 a try and report if there is
any difference ?

In any case, I would appreciate a complete dmesg from boot including
the XID line emitted by the driver.

-- 
Ueimor
Comment 4 Dave Plater 2008-08-10 06:04:40 UTC
Created attachment 17163 [details]
nic not working version shown as ff
Comment 5 Dave Plater 2008-08-10 06:17:16 UTC
Created attachment 17164 [details]
boot.omsg is saved from a successful driver load

The only difference I could see between when the nic works and when it doesn't is the version number loaded as ff instead of 02. This shows up in hwinfo --netcard too. The only other difference I have found is ethtool shows the nic running at 1000 Mb/s, which is a result of no external connection maybe, and 100Mb/s when it works.
Comment 6 Dave Plater 2008-08-10 06:18:42 UTC
Created attachment 17165 [details]
A good and bad hwinfo --netcard
Comment 7 Dave Plater 2008-08-10 06:20:39 UTC
Created attachment 17166 [details]
Yast2 hardware info save
Comment 8 Dave Plater 2008-08-10 06:23:39 UTC
Although I have an nVidia card, this bug was present with onboard Intel 945 graphics and I saw no change in the behaviour at all.
Comment 9 Dave Plater 2008-08-10 06:35:40 UTC
I will try 2.6.27-rc2 on Monday, I am unable to download large files at the moment.
Comment 10 Francois Romieu 2008-08-10 06:47:53 UTC
Created attachment 17167 [details]
2.6.26 to 2.6.27-rc2 r8169 patch

Dave Plater  2008-08-10 06:35:40 :
[...]
> I will try 2.6.27-rc2 on Monday, I am unable to download large files at the
> moment.

If you are in a hurry I have attached the (~10 ko) patch against 2.6.26.

Thanks for your help.

-- 
Ueimor
Comment 11 Dave Plater 2008-08-11 01:18:11 UTC
2.6.27-rc2 installed, I see an extra message, r8169: eth0: link up, displayed twice. I had a failure first boot this morning with 2.6.26 kernel but could not make it happen again. Do you have any idea about what causes the incorrect version number?
Comment 12 Francois Romieu 2008-08-11 12:14:06 UTC
Dave Plater  2008-08-11 01:18:11 :
[...]
> Do you have any idea about what causes the incorrect version number?

It is probably commit 77332894c21165404496c56763d7df6c15c4bb09.

May I close the bug ?

-- 
Ueimor
Comment 13 Dave Plater 2008-08-11 23:09:03 UTC
Sure, no more problems so far and the first boot seemed to consistently fail with 2.6.26 kernel.

Note You need to log in before you can comment on or make changes to this bug.