Bug 274765 – [2.6.29 regression] r8169 can't transmit after a while

• Status: RESOLVED TEST-REQUEST
• Product: Gentoo Linux

[Description Manfred Knick 2009-06-20 06:56:58 UTC]

The Network gets loaded, ifconfig / netstat -r show the correct values, but every ping even on local network yields "Destination Host unreachable".

On the connected Switch's LED, rmmod / modprobe lights off / on both LED (= GigaBit, correctly); "ping" gives _some_ correlation to them, but after some while of pings the LEDs don't flicker any more at all ...

Moreover: after letting ping running loose (unsuccessfully) for some minutes, it produced error messages that there would be no space for the buffer available. Notabene: This machine runs with a core of 8 GiB main memory !

I've quickly compiled and investigated with hand-crafted kernels.

RESULT:

- Works beginning with 2.6.27
- Works with 2.6.28.*
- Broken throughout 2.6.29 (at least up to 2.6.29.4 today)
- Works with 2.6.30 again


Reproducible: Always

Steps to Reproduce:





Cross-Reference: This I had first reported in:
    https://bugzilla.redhat.com/show_bug.cgi?id=460747#c89


$ uname -a
Linux mammut 2.6.27-gentoo-r10 #1 SMP Wed Jun 10 09:43:07 CEST 2009 x86_64 AMD Phenom(tm) 9950 Quad-Core Processor AuthenticAMD GNU/Linux



$ emerge --info
Portage 2.1.6.13 (default/linux/amd64/2008.0/desktop, gcc-4.3.3, glibc-2.9_p20081201-r2, 2.6.27-gentoo-r10 x86_64)
=================================================================
System uname: Linux-2.6.27-gentoo-r10-x86_64-AMD_Phenom-tm-_9950_Quad-Core_Processor-with-glibc2.2.5
Timestamp of tree: Sat, 20 Jun 2009 05:15:03 +0000
app-shells/bash:     3.2_p39
dev-java/java-config: 2.1.7
dev-lang/python:     2.5.4-r2
dev-util/cmake:      2.6.4
sys-apps/baselayout: 1.12.12
sys-apps/sandbox:    1.6-r2
sys-devel/autoconf:  2.13, 2.63
sys-devel/automake:  1.4_p6, 1.5, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10.2
sys-devel/binutils:  2.18-r3
sys-devel/gcc-config: 1.4.1
sys-devel/libtool:   1.5.26
virtual/os-headers:  2.6.27-r2
ACCEPT_KEYWORDS="amd64"
CBUILD="x86_64-pc-linux-gnu"
CFLAGS="-march=native        -O2 -pipe"
CHOST="x86_64-pc-linux-gnu"
...
CXXFLAGS="-march=native        -O2 -pipe"
...
LDFLAGS="-Wl,-O1"
...

[Comment 1 Wormo (RETIRED) 2009-06-25 06:57:55 UTC]

Thanks for reporting your test results! Looks like there is a backport to 2.6.29.4 of the fix from 2.6.30 (which you tested): http://patchwork.kernel.org/patch/25515/

[Comment 2 Manfred Knick 2009-06-25 07:56:09 UTC]

(In reply to comment #1)
> Thanks for reporting your test results!
It's my pleasure!

> Looks like there is a backport to
> 2.6.29.4 of the fix from 2.6.30 (which you tested):
Please, be careful and note that just exchanging the provided r8160 module in 2.6.29 did *not* solve the problem, as reported in the cited Fedora bug; there must be another problem in the interaction of the patched r8169 module with the rest of the 2.6.29 kernel. At my place, only 2.6.30 (including the new r8169 patches) worked again.
For production machines, I'm really grateful that Adrian Bunk nourishes 2.6.27 as a stable "long-term edition" again.

[Comment 3 Daniel Drake (RETIRED) 2009-06-28 10:10:23 UTC]

Please try that patch anyway, as it does seem like a likely candidate.

[Comment 4 Manfred Knick 2009-06-28 15:55:41 UTC]

Please, note 

https://bugzilla.redhat.com/show_bug.cgi?id=460747#c99   ff.

[Comment 5 Mike Pagano 2009-08-19 22:34:07 UTC]

Fixed in 2.6.30. If you want to test the patch for possible backport to 2.6.29, please feel free to do so.