kernel panic w/ax25 in FC4

[John Bennett <jabennett@insightbb.com>]

Hey guys,

I routinely read the emails in this list and don't remember seeing
anything like this before. If this issue has been previously discussed,
my apologies.

FC4 built from CDs. Updated everything with yum. Latest libax25,
ax25-tools built from tarballs. Rebuilt kernel using
'kernel-2.6.14-1.1656_FC4.src.rpm' sources. Rebuilt kernel because I
needed mkiss for a TNC attached to a serial port. TNC is an MFJ-1274
using TAPR firmware 1.1.8 or 1.1.7b (can't remember which).

Machine in question was a P3-500 w/256 MB. Previously installed and ran
RH9 with a 2.4 kernel on it for the last year with no problems. This was
an 'upgrade' which is actually a new install (never have found that
upgrade paths work well). Due to time constraints, I had to leave and
finish building and configuring the machine remotely.

This build/hardware configuration was similar to several other builds
completed in the last two months.

When starting ax25 (kissattach), there was a kernel panic every time.
Tried 2.6.11 and 2.6.12, but the TIOCSETD error kept popping up. Went
back to 2.6.14. Finally decided there must be a hardware problem with
this particular box that was not on the others. Grabbed another machine
that was identical to an installation running at another QTH (P3-866
w/512MB). Built and checked on the bench here. All worked as it should.

Today I drove to the remote site, installed the new box, and immediately
got a kernel panic. After some investigation, I discovered that the TNC
had dropped out of the KISS mode. If the TNC was disconnected or the
power was off when doing a kissattach, no problems. The instant the TNC
was switched on or was on when invoking kissattach there was a kernel
panic. Using minicom, I put the errant TNC back into the KISS mode.
Everything now works.

I suspect that when mkiss receives certain characters from the TNC when
in the normal (?) mode, it is causing a segmentation fault of some kind
that causes the panic. I know this has not been an issue in the past
with other kernels (i.e., 2.4). While it doesn't happen very often, TNCs
do drop out of KISS if something screwy happens. Hopefully someone on
this list is the maintainer and can take a look at the code. I scanned
through 'kissattach' and didn't see anything that could cause the
problem. It's fairly simple program.

At least the week-long siege of frustration is finally over...

73

-- 
John Bennett
Evansville, IN
Amateur Radio Web: http://n4xi.home.insightbb.com
Where am I? http://www.findu.com/cgi-bin/find.cgi?N4XI
Bike Racing Web: http://xibike.home.insightbb.com
Live Strong - Lance Armstrong Foundation: http://www.laf.org/