The issue here is *connected* AX.25 session so your Winlink
connections would be impacted. APRS would not be impacted since it's
unconnected traffic and still works fine.
Yes, I know that but your subject line implies that you already know the
problem is in the AX.25 stack and that might/probably not be the case. I
am using native Linux AX.25 stack with Dire Wolf DEVELOPMENT version 1.7
A (Feb 15 2021) on 4 different systems and it is working fine.
The way the Canonical backports fixes into older kernels is
difficult to track and they aren't dating things in the
changelog but I see this
The following is the commit summary from kernel.org for kernel 5.11.4
and ax.25 commits. I looked at each of these commits & none of them
should cause your symptom.
2020-11-20 rose: Fix Null pointer dereference in rose_send_frame() Anmol Karn 1
2020-07-23 AX.25: Prevent integer overflows in connect and sendmsg Dan Carpenter 1
2020-07-22 AX.25: Prevent out-of-bounds read in ax25_sendmsg() Peilin Ye 1
2020-07-22 AX.25: Fix out-of-bounds read in ax25_connect() Peilin Ye 1
2020-07-04 Documentation: networking: ax25: drop doubled word Randy Dunlap 1
2020-05-20 ax25: fix setsockopt(SO_BINDTODEVICE) Eric Dumazet 1
2020-04-28 Docs: networking: convert ax25.txt to ReST Mauro Carvalho Chehab 3
2019-09-24 ax25: enforce CAP_NET_RAW for raw sockets Ori Nimron 1
I suggest swapping some big components in your system including NOT
using the internal D-710 TNC. Once you do that you can test against the
Direwolf user land ax.25 stack & the Linux kernel mode ax.25 stack.
Just did a successful connection test between two ARM machines using
these kernels 5.10.17-v7l+ and 5.4.79-v7l+.
Also there was a massive Linux kernel code merge from the middle of
December 2020 to the first week of Jan 2021 that took the kernel from
version 5.4.83 to 5.10.11 that caused a few problems. Not sure what that
means for Ubuntu.
(posted 2021-02-04)
--
* Focal update: v5.4.55 upstream stable release (LP: #1890343)
- AX.25: Fix out-of-bounds read in ax25_connect()
- AX.25: Prevent out-of-bounds read in ax25_sendmsg()
...
- AX.25: Prevent integer overflows in connect and sendmsg
--
* Focal update: v5.4.44 upstream stable release (LP: #1881927)
- ax25: fix setsockopt(SO_BINDTODEVICE)
--
What TNC device are you using? /Basil N7NIX
On this system, it's a D710 in KISS mode connected to a Lenovo T470
(i7-7600U with 16GB RAM) running 64bit Ubuntu 20.04. I'm 99.9% sure
that if I switched away from a serial attached hardware TNC to a
software-based TNC like Direwolf, I would still see the issue. I've
done more testing with reverting the kernel to older versions but the
ones I've tested so far still fail as well:
5.8.0-44 : BAD
5.8.0-43 : BAD
5.8.0-41 : BAD
..
5.8.0-36 : BAD
..
5.4.0-66 : BAD
..
5.4.0-42 : BAD
It's clear that my mistake is that after Canonical pushes a
new kernel version and I apply it, I should reboot then test
things to KNOW if the AX25 stack has been impacted. I skipped
a few reboot cycles as different kernels were installed so I
really don't know where to start. Even then, I'm thinking this
issue might be more of a libc interface issue or something
else since going way back to 5.4.0-42 dated 2020-07-10 is
seeing the same issue.
root@hampacket3:/var/log/apt# dpkg -l | grep -e libc6 -e linux-libc
ii libc6:amd64 2.31-0ubuntu9.2 amd64 GNU C Library: Shared libraries
ii libc6-dbg:amd64 2.31-0ubuntu9.2 amd64 GNU C Library: detached
debugging symbols ii libc6-dev:amd64 2.31-0ubuntu9.2 amd64 GNU C
Library: Development Libraries and Header Files ii
linux-libc-dev:amd64 5.4.0-66.74 amd64 Linux Kernel Headers for
development
That libc6 was installed on "2021-01-28"
It's frustrating as I have no clue where to start as the machine just locks up and doesn't give any hint of where to start troubleshooting. Could be an 64bit thing. Could be an SMP thing. Dunno,
--David KI6ZHD
Basil Gunn