[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: iptables: Protocol wrong type for socket.

Camaleón wrote:
On Tue, 07 Aug 2012 11:30:36 -0500, hvw59601 wrote:

Camaleón wrote:


Obviously I have not configured the kernel correctly because with the
Debian kernels there are no errors.

But what kernel configuration parameter might be missing/wrong?
Can't tell for the exact parameter/option that controls this but what I
use to do when have to compile a new kernel is using the current
".config" file which comes along with Debian stock kernel to avoid

But the difference in kernel sizes between a kernel with the Debian
.config and one with my own .config for this particular box is huge:
32MB vs. 4MB.

32 MiB for a kernel is not that huge O:-)

Okay, point taken.

But I figured it out: I compiled 3.4.7 from kernel.org with a .config
from Debian's kernel in experimental, 3.5.0. I booted that and compared
the lsmod from that with the one of 3.4.7 with my own .config and it
turned out that there were 2 modules present in the former that were not
in mine: nf_connttrack_ipv4 and nf_defrag_ipv4. I reconfigured and now
it works like a charm.

Ha, well done.

Which reminds me another trick: the "localmodconfig" parameter for the "make" that suits perfect for these situations.

Right. That would have been easier. I just looked at streamline_config.pl, which is what that executes. Boot the kernel with the Debian .config and execute that perl script: it would have found those 2 modules and set them in the config that it generates.

This is all for figuring out
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=683768 and hibernate
times for the various kernels that changed dramatically beginning with
3.2.21-1, from 10s. to 35s.

Strange enough hibernate for 3.4.7 with Debian .config takes 24s. and
with my own config 10s. Who knows what is going on.

You say it: who knows what's happening. Hibernating and suspension are two things I never use (neither on windows systems not netbooks, laptops, desktops...), mainly because I find them pretty useless but specially in linux also because they can break very easily -a simple kernel change, a problematic driver... and you're hosed- and doing a hard reset is something a would like to avoid as much as I can.

True, but having everything where you left off when you boot again is very tempting and powering off is enforced in this locality. But I average about 12 hibernates with this desktop before a hickup and I have to do a hard reset, without a clue of went wrong of course...

Thanks Camaleón.

You're welcome and good luck with your debugging :-)


Reply to: