Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[David Ranch]

Hello Dave,

Are you talking about this change below? If so, this has no material 
improvements to the AX25 stack other than to improve cross-platform 
compiling.


--David
KI6ZHD


--
Format: 1.8

Date: Sat, 10 Feb 2018 14:18:45 +
Source: libax25
Binary: libax25 libax25-dev
Architecture: source amd64
Version: 0.0.12-rc4-2
Distribution: unstable
Urgency: medium
Maintainer: Debian Hamradio Maintainers
Changed-By: Dave Hibberd
Description:
 libax25- ax25 library for hamradio applications
 libax25-dev - ax25 library development files
Closes: 880394
Changes:
 libax25 (0.0.12-rc4-2) unstable; urgency=medium
 .
   * Updated to DebHelper 11
 - debian/control updated
 - debian/compat updated to level 11
   * Hardening enabled
   * Fix FTCBFS: Let dh_auto_configure pass --host to ./configure.
 - (Closes: #880394)
 - Thanks to Helmut Grohne
   * Fixed lintian Errors - trailing whitespace:
 - debian/changelog
 - debian/rules

--



On 02/10/2018 09:08 AM, David A Aitcheson wrote:

For those NOT on the debian-h...@lists.debian.org reflector...

The AX.25 maintainers are working on AX.25 and updates to AX.25 have
been uploaded to "Debian FTP Masters" by Dave "Hibby" Hibberd.

73
Dave
KB3EFS


On 02/05/2018 02:16 PM, Curt, WE7U wrote:

On Mon, 5 Feb 2018, Jason KG4WSV wrote:


without this kernel support, AX.25 becomes just another application.

To put this into the simplest terms possible (I'm really good at
simple thinking!):

If AX.25 has kernel networking support we can often treat it like an
ethernet or wireless interface and route packets in/out using
completely normal networking software and techniques.

A program doesn't have to be modified specifically for AX.25 in order
to use it with that interface: I could start up a mail program or a
web server and tell it to use that device for instance.


___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir


___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[David A Aitcheson]

For those NOT on the debian-h...@lists.debian.org reflector...

The AX.25 maintainers are working on AX.25 and updates to AX.25 have
been uploaded to "Debian FTP Masters" by Dave "Hibby" Hibberd.

73
Dave
KB3EFS


On 02/05/2018 02:16 PM, Curt, WE7U wrote:
> On Mon, 5 Feb 2018, Jason KG4WSV wrote:
>
>> without this kernel support, AX.25 becomes just another application.
>
> To put this into the simplest terms possible (I'm really good at
> simple thinking!):
>
> If AX.25 has kernel networking support we can often treat it like an
> ethernet or wireless interface and route packets in/out using
> completely normal networking software and techniques.
>
> A program doesn't have to be modified specifically for AX.25 in order
> to use it with that interface: I could start up a mail program or a
> web server and tell it to use that device for instance.
>

___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[David Brooke G6GZH]

On Tue, Feb 06, 2018 at 02:34:18AM +, Max Harper wrote:
> Just as well as I think aprx is having problems also in its latest version.

I noticed a year or so back (maybe longer) that aprx fell foul of a
kernel change which resulted in shorter packets not being sent and the
kernel logging a failure of some kind of minimum length check.

I assume that a sanity check was added in the network stack without
regard to AX.25 but I didn't investigate very deeply. It only seemed to
affect the method which aprx was using; beacon still worked.

I reverted the kernel upgrade and it worked again. I'm no longer using
that system but the SD card image reveals that the last boot was kernel
3.18.16 so that was presumably the one I stuck with.

David G6GZH

___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Max Harper]

Thanks for letting us know. My Pi just came in today so I was going to test 
your fix out tomorrow, can that idea. Just as well as I think aprx is having 
problems also in its latest version. I'm swapping out the uSD card and going 
back to older versions of everything tomorrow instead.
Max KG4PID  From: Tom Russo <ru...@bogodyn.org>
 To: xas...@xastir.org 
 Sent: Monday, February 5, 2018 7:32 PM
 Subject: Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 
users Re: Error in Syslog
   
Folks:  I have done more testing on this "fix" to AX.25 networking, and it
is broken --- it works for transmit, but not receive.  I will not be merging 
it upstream, and if you were planning on giving it a whirl, don't bother.  
I've deleted the branch on which the fix lives, and instead opened a 
"help-wanted" bug report.

On Sun, Jan 28, 2018 at 02:39:37PM -0700, we recorded a bogon-computron 
collision of the <ru...@bogodyn.org> flavor, containing:
> On Sun, Jan 28, 2018 at 02:31:08PM -0700, we recorded a bogon-computron 
> collision of the <ru...@bogodyn.org> flavor, containing:
> > Changing the subject line to get more people to see this.
> > 
> [...]
> > 
> > It took a while, but I think I found some code that should work (using
> > AF_PACKET and SOCK_DGRAM), and am pushing it to github on my own fork so 
> > you 
> > can try it out.  If the warnings go away for you, and AX.25 networking 
> > still 
> > work for everyone who reports in, then I will push it to the main line code.
> > 
> > You can test it out easily:
> > 
> >    git checkout master      # Likely a no-op, but let's be careful
> >    git checkout -b tvr-fix-ax25
> >    git pull https://github.com/tvrusso/Xastir.git fix-ax25
> > 
> > then rebuild the code and try it out.  You should be able to start an AX.25
> > kernel port in Xastir and NOT see those warnings in your syslog.  If it 
> > works,
> > please let me know and I'll merge it to main line.  It is working for me 
> > Just
> > Fine.  But then again, I never had those warnings in syslog.
> 
> [...]
> 
> > I would appreciate it if *everyone* on the list who uses AX.25 networking
> > routinely would test out this code --- it impacts *everyone* who uses AX.25 
> > networking in Xastir.
> 
> I opened a new pull request for this branch to encourage discussion on it:
> https://github.com/Xastir/Xastir/pull/20
> If you try out the code, commenting on the pull request would be welcome.  If
> you spot some horrid error in it, too, that would be a good thing to comment 
> on.
> 
> > >On Sun, Jan 28, 2018 at 12:14:39PM -0700, we recorded a bogon-computron 
> > >collision of the <ru...@bogodyn.org> flavor, containing:
> > >> 
> > >> I just experimented with fixing the preprocessor symbol so it actually
> > >> detects we're running with a newer GLIBC, and it made Xastir AX.25 
> > >> connections
> > >> broken, so this code that has never actually been enabled isn't in fact
> > >> correct.  Turns out that replacing PF_INET,SOCK_PACKET with 
> > >> PF_INET,SOCK_DGRAM,
> > >> as is done in that block of code, was never right.  The correct thing is
> > >> to use AF_PACKET,SOCK_PACKET instead of PF_INET,SOCK_PACKET.  If I fix
> > >> the symbol *and* fix that usage, Xastir builds and does connect to an
> > >> AX.25 interface correctly.
> > >
> > >Actually, SOCK_PACKET is deprecated, too, and should not be used in new
> > >programs.  This is going to take a little more careful experimentation.
> > 
> > 
> > > > On Sun, Jan 28, 2018 at 05:58:54PM +, we recorded a bogon-computron 
> > > > collision of the <kg4...@yahoo.com> flavor, containing:
> > > > > Looking the the file /var/log/syslog, I noticed this...
> > > > > 'Jan 27 07:13:18 Igate kernel: [26779.439570] xastir uses obsolete 
> > > > > (PF_INET,SOCK_PACKET)'
> > > > > 
> > > > > I don't have a clue what it means or if I should even worry about it. 
> > > > > Anyone else see this?
> > > > > I just did an install a week ago so it's the latest version of xastir.
> > > > > Thanks, Max KG4PID??
> > > > > ___
> > > > > Xastir mailing list
> > > > > Xastir@lists.xastir.org
> > > > > http://xastir.org/mailman/listinfo/xastir
> > > > 
> > > > -- 
> > > > Tom Russo    KM5VY  SAR502  DM64ux          http://www.swcp.com/~russo/

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Tom Russo]

Folks:  I have done more testing on this "fix" to AX.25 networking, and it
is broken --- it works for transmit, but not receive.  I will not be merging 
it upstream, and if you were planning on giving it a whirl, don't bother.  
I've deleted the branch on which the fix lives, and instead opened a 
"help-wanted" bug report.

On Sun, Jan 28, 2018 at 02:39:37PM -0700, we recorded a bogon-computron 
collision of the  flavor, containing:
> On Sun, Jan 28, 2018 at 02:31:08PM -0700, we recorded a bogon-computron 
> collision of the  flavor, containing:
> > Changing the subject line to get more people to see this.
> > 
> [...]
> > 
> > It took a while, but I think I found some code that should work (using
> > AF_PACKET and SOCK_DGRAM), and am pushing it to github on my own fork so 
> > you 
> > can try it out.  If the warnings go away for you, and AX.25 networking 
> > still 
> > work for everyone who reports in, then I will push it to the main line code.
> > 
> > You can test it out easily:
> > 
> >git checkout master  # Likely a no-op, but let's be careful
> >git checkout -b tvr-fix-ax25
> >git pull https://github.com/tvrusso/Xastir.git fix-ax25
> > 
> > then rebuild the code and try it out.  You should be able to start an AX.25
> > kernel port in Xastir and NOT see those warnings in your syslog.  If it 
> > works,
> > please let me know and I'll merge it to main line.  It is working for me 
> > Just
> > Fine.  But then again, I never had those warnings in syslog.
> 
> [...]
> 
> > I would appreciate it if *everyone* on the list who uses AX.25 networking
> > routinely would test out this code --- it impacts *everyone* who uses AX.25 
> > networking in Xastir.
> 
> I opened a new pull request for this branch to encourage discussion on it:
> https://github.com/Xastir/Xastir/pull/20
> If you try out the code, commenting on the pull request would be welcome.  If
> you spot some horrid error in it, too, that would be a good thing to comment 
> on.
> 
> > >On Sun, Jan 28, 2018 at 12:14:39PM -0700, we recorded a bogon-computron 
> > >collision of the  flavor, containing:
> > >> 
> > >> I just experimented with fixing the preprocessor symbol so it actually
> > >> detects we're running with a newer GLIBC, and it made Xastir AX.25 
> > >> connections
> > >> broken, so this code that has never actually been enabled isn't in fact
> > >> correct.  Turns out that replacing PF_INET,SOCK_PACKET with 
> > >> PF_INET,SOCK_DGRAM,
> > >> as is done in that block of code, was never right.  The correct thing is
> > >> to use AF_PACKET,SOCK_PACKET instead of PF_INET,SOCK_PACKET.  If I fix
> > >> the symbol *and* fix that usage, Xastir builds and does connect to an
> > >> AX.25 interface correctly.
> > >
> > >Actually, SOCK_PACKET is deprecated, too, and should not be used in new
> > >programs.  This is going to take a little more careful experimentation.
> > 
> > 
> > > > On Sun, Jan 28, 2018 at 05:58:54PM +, we recorded a bogon-computron 
> > > > collision of the  flavor, containing:
> > > > > Looking the the file /var/log/syslog, I noticed this...
> > > > > 'Jan 27 07:13:18 Igate kernel: [26779.439570] xastir uses obsolete 
> > > > > (PF_INET,SOCK_PACKET)'
> > > > > 
> > > > > I don't have a clue what it means or if I should even worry about it. 
> > > > > Anyone else see this?
> > > > > I just did an install a week ago so it's the latest version of xastir.
> > > > > Thanks, Max KG4PID??
> > > > > ___
> > > > > Xastir mailing list
> > > > > Xastir@lists.xastir.org
> > > > > http://xastir.org/mailman/listinfo/xastir
> > > > 
> > > > -- 
> > > > Tom RussoKM5VY   SAR502   DM64ux  
> > > > http://www.swcp.com/~russo/
> > > > Tijeras, NM  QRPL#1592 K2#398  SOC#236
> > > > http://kevan.org/brain.cgi?DDTNM
> > > >  echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] 
> > > > [n-z][a-m]
> > > > 
> > > >  
> > > > 
> > > > 
> > > > ___
> > > > Xastir mailing list
> > > > Xastir@lists.xastir.org
> > > > http://xastir.org/mailman/listinfo/xastir
> > > 
> > > -- 
> > > Tom RussoKM5VY   SAR502   DM64ux  http://www.swcp.com/~russo/
> > > Tijeras, NM  QRPL#1592 K2#398  SOC#236
> > > http://kevan.org/brain.cgi?DDTNM
> > >  echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] 
> > > [n-z][a-m]
> > > 
> > >  
> > > 
> > > 
> > > ___
> > > Xastir mailing list
> > > Xastir@lists.xastir.org
> > > http://xastir.org/mailman/listinfo/xastir
> > 
> > -- 
> > Tom RussoKM5VY   SAR502   DM64ux  http://www.swcp.com/~russo/
> > Tijeras, NM  QRPL#1592 K2#398  SOC#236
> > http://kevan.org/brain.cgi?DDTNM
> >  echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] 
> > [n-z][a-m]
> > 
> >  
> > 
> > 
> > 

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Curt, WE7U]

On Mon, 5 Feb 2018, Jason KG4WSV wrote:


without this kernel support, AX.25 becomes just another application.


To put this into the simplest terms possible (I'm really good at simple 
thinking!):

If AX.25 has kernel networking support we can often treat it like an ethernet 
or wireless interface and route packets in/out using completely normal 
networking software and techniques.

A program doesn't have to be modified specifically for AX.25 in order to use it 
with that interface: I could start up a mail program or a web server and tell 
it to use that device for instance.

--
Curt, WE7U.http://we7u.wetnet.net
"... you can't have 100% security and also then have 100% privacy and zero 
inconvenience. We are going to have to make some choices as a society." - Barack 
Obama
"They who can give up essential liberty to obtain a little temporary safety, deserve 
neither liberty nor safety." - Benjamin Franklin
Coordinate Converter (Android): http://www.sarguydigital.com
___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Jason KG4WSV]

On Mon, Feb 5, 2018 at 10:31 AM, David Ranch  wrote:

>  If you're doing more complex stuff like NETROM, FLEX, routing
> AX.25 packets to AMPR IP-IP networks, etc, it gives you a LOT of flexible
> and powerful options.


To expand a bit on David's comments:

The kernel implementation allows AX.25 to be used as a general purpose
network, integrated with other network stacks. Once popular model of
network communications breaks a network down into 7 layers:

7: Application
6: Presentation
5: Session
4: Transport
3: Network
2: Data Link
1: Physical


The internet is an IP network, so layers 4 and 3 are TCP/IP (or
UDP/IP).  Most local networks use ethernet for layer 2, and RF or CAT5
for layer 1. With an appropriate kernel implementation, we can create
an AX.25 network link that can move IP data
_with_no_change_to_the_application_ by using AX.25 for layer 2 and RF
for layer 1.

without this kernel support, AX.25 becomes just another application.

-Jason
kg4wsv
___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[David Ranch]

Hello Max,


Why does ax.25 need to be in the kernel? What am I installing when I do apt-get 
install ax25? What about the millions of users that don't use ax.25, doesn't 
that just make the kernel a little larger for them without any benefits? Help 
me understand this as I'd really like to know.


Technically speaking, AX.25 doesn't have to be in the kernel.  AX.25 
could be an external program but it becomes much more universal for 
other high up applications *if* it's centralized like this.  If you 
needs for AX.25 are very simple like APRS (just UI packets), being in 
the kernel isn't necessarily as programs like Xastir does all of it's 
own AX.25 UI processing.  If you're doing more complex stuff like 
NETROM, FLEX, routing AX.25 packets to AMPR IP-IP networks, etc, it 
gives you a LOT of flexible and powerful options.  I would guess that 
the second most flexible AX.25 solution out there is BPQ32 which is a 
self-contained program that includes it's own AX.25 stack and it has 
some of it's own benefits (connectors to ARDOP & PACTOR TNCs, etc) but 
it's a bit of an "island".  Third could be Direwolf which has APRS, 
APRStt, and a complete "cutting edge" AX.25 v2.2 stack but it's 
connected modes are only accessible via AGW enabled applications.


Ultimately, Linux's AX.25 kernel code can be considered a "connector of 
AX.25 applications and islands" so to speak.


--David
KI6ZHD
___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Ken Koster]

On Monday, February 5, 2018 6:39:11 AM PST Max Harper wrote:
> Why does ax.25 need to be in the kernel? 

Because it's part of the network stack.

> What am I installing when I do
> apt-get install ax25?

Probably the ax25-tools.

> What about the millions of users that don't use
> ax.25, doesn't that just make the kernel a little larger for them without
> any benefits? 

The kernel doesn't work that way.   The linux kernel is modular and there are 
thousands of 
modules for special purposes that get loaded when needed.   Modules for 
different hardware,  file 
systems, different protocols and interfaces. 

The ax25 kernel support is what provides a network interface (like ethernet 
drivers)  and a 
protocol stack.  It's a module so it doesn't take up space if it's not needed.  
When you setup and 
configure ax25 the kernel module will be loaded and do it's job.

N7IPB

> Help me understand this as I'd really like to know. Max
> KG4PID  From: Tom Russo <ru...@bogodyn.org>
>  To: Xastir - APRS client software discussion <xas...@xastir.org>
>  Sent: Sunday, February 4, 2018 9:46 PM
>  Subject: Re: [Xastir] AX 25 networking kernel warnings --- heads up all
> AX.25 users Re: Error in Syslog
> On Sun, Feb 04, 2018 at 11:16:41AM -0800, we recorded a bogon-computron 
> collision of the 
<xas...@trinnet.net> flavor, containing:
> > Hey Tom, Everyone,
> > 
> > > I would appreciate it if *everyone* on the list who uses AX.25
> > > networking
> > > routinely would test out this code --- it impacts *everyone* who uses
> > > AX.25
> > > networking in Xastir.
> > 
> > The Ax.25 stack in the Linux kernel is falling apart and there have been
> > various discussions about this on the UROnode list and elsewhere.  The
> > situation is that other (non-AX25) Linux kernel developers have been
> > making changes, updates, deprecations, etc. to keep up with general
> > kernel modernization efforts but without testing any of the changes on
> > the AX.25 side of things.  It seems that if it compiles, it must be
> > working.  Not true!  For various classic packet uses, 3.17.x was the
> > last version before things started getting really broken.
> 
> Hmmm.  It still worked for APRS with my 4.4 kernel, but then APRS is all I
> use AX.25 for.
> 
> > Most of these have been reported to the "AX25 kernel maintainers" but
> > those people say they *aren't* the maintainers on the kernel side.. just
> > the userland side.  Putting it another way, we don't have any Linux
> > AX.25 maintainers anymore!  Unless we can find some folk who have the
> > technical chops to fix some of these issues (I wish I did), I imagine
> > the native AX.25 stack in Linux will become toxic and ultimately
> > removed.  If there is anyone who is willing to give this a try, I can
> > connect you with a few other people who have a consolidated list of
> > issues and some possible fixes.  Some of those fixes were previously
> > rejected from the kernel maintainers for various reasons without
> > offering solutions.
> 
> Sigh.  Wish I had the chops for it, but I don't.  It'll be sad if AX.25 dies
> in the Linux kernel.


-- 
Ken - N7IPB
Email: n7...@wetnet.net
JID: n7...@jabber.wetnet.net
PGP Sig: F42B EF90 3CD3 31C7 3056  122E 993A 7B2E 5138 C42A 
I've calculated my velocity with such exquisite precision that I have no idea 
where I am.


signature.asc
Description: This is a digitally signed message part.
___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Max Harper]

Why does ax.25 need to be in the kernel? What am I installing when I do apt-get 
install ax25? What about the millions of users that don't use ax.25, doesn't 
that just make the kernel a little larger for them without any benefits? Help 
me understand this as I'd really like to know.
Max KG4PID  From: Tom Russo <ru...@bogodyn.org>
 To: Xastir - APRS client software discussion <xas...@xastir.org> 
 Sent: Sunday, February 4, 2018 9:46 PM
 Subject: Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 
users Re: Error in Syslog
   
On Sun, Feb 04, 2018 at 11:16:41AM -0800, we recorded a bogon-computron 
collision of the <xas...@trinnet.net> flavor, containing:
> 
> Hey Tom, Everyone,
> 
> > I would appreciate it if *everyone* on the list who uses AX.25 networking
> > routinely would test out this code --- it impacts *everyone* who uses AX.25
> > networking in Xastir.
> 
> The Ax.25 stack in the Linux kernel is falling apart and there have been 
> various discussions about this on the UROnode list and elsewhere.  The 
> situation is that other (non-AX25) Linux kernel developers have been 
> making changes, updates, deprecations, etc. to keep up with general 
> kernel modernization efforts but without testing any of the changes on 
> the AX.25 side of things.  It seems that if it compiles, it must be 
> working.  Not true!  For various classic packet uses, 3.17.x was the 
> last version before things started getting really broken.

Hmmm.  It still worked for APRS with my 4.4 kernel, but then APRS is all I 
use AX.25 for.

> Most of these have been reported to the "AX25 kernel maintainers" but 
> those people say they *aren't* the maintainers on the kernel side.. just 
> the userland side.  Putting it another way, we don't have any Linux 
> AX.25 maintainers anymore!  Unless we can find some folk who have the 
> technical chops to fix some of these issues (I wish I did), I imagine 
> the native AX.25 stack in Linux will become toxic and ultimately 
> removed.  If there is anyone who is willing to give this a try, I can 
> connect you with a few other people who have a consolidated list of 
> issues and some possible fixes.  Some of those fixes were previously 
> rejected from the kernel maintainers for various reasons without 
> offering solutions.

Sigh.  Wish I had the chops for it, but I don't.  It'll be sad if AX.25 dies
in the Linux kernel.

-- 
Tom Russo    KM5VY
Tijeras, NM  

 echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] [n-z][a-m]

___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir


   
___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Tom Russo]

On Sun, Feb 04, 2018 at 11:16:41AM -0800, we recorded a bogon-computron 
collision of the  flavor, containing:
> 
> Hey Tom, Everyone,
> 
> > I would appreciate it if *everyone* on the list who uses AX.25 networking
> > routinely would test out this code --- it impacts *everyone* who uses AX.25
> > networking in Xastir.
> 
> The Ax.25 stack in the Linux kernel is falling apart and there have been 
> various discussions about this on the UROnode list and elsewhere.  The 
> situation is that other (non-AX25) Linux kernel developers have been 
> making changes, updates, deprecations, etc. to keep up with general 
> kernel modernization efforts but without testing any of the changes on 
> the AX.25 side of things.  It seems that if it compiles, it must be 
> working.  Not true!  For various classic packet uses, 3.17.x was the 
> last version before things started getting really broken.

Hmmm.  It still worked for APRS with my 4.4 kernel, but then APRS is all I 
use AX.25 for.

> Most of these have been reported to the "AX25 kernel maintainers" but 
> those people say they *aren't* the maintainers on the kernel side.. just 
> the userland side.  Putting it another way, we don't have any Linux 
> AX.25 maintainers anymore!  Unless we can find some folk who have the 
> technical chops to fix some of these issues (I wish I did), I imagine 
> the native AX.25 stack in Linux will become toxic and ultimately 
> removed.  If there is anyone who is willing to give this a try, I can 
> connect you with a few other people who have a consolidated list of 
> issues and some possible fixes.  Some of those fixes were previously 
> rejected from the kernel maintainers for various reasons without 
> offering solutions.

Sigh.  Wish I had the chops for it, but I don't.  It'll be sad if AX.25 dies
in the Linux kernel.

-- 
Tom RussoKM5VY
Tijeras, NM  

 echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] [n-z][a-m]

___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[David Ranch]

Hey Tom, Everyone,


I would appreciate it if *everyone* on the list who uses AX.25 networking
routinely would test out this code --- it impacts *everyone* who uses AX.25
networking in Xastir.


The Ax.25 stack in the Linux kernel is falling apart and there have been 
various discussions about this on the UROnode list and elsewhere.  The 
situation is that other (non-AX25) Linux kernel developers have been 
making changes, updates, deprecations, etc. to keep up with general 
kernel modernization efforts but without testing any of the changes on 
the AX.25 side of things.  It seems that if it compiles, it must be 
working.  Not true!  For various classic packet uses, 3.17.x was the 
last version before things started getting really broken.


Most of these have been reported to the "AX25 kernel maintainers" but 
those people say they *aren't* the maintainers on the kernel side.. just 
the userland side.  Putting it another way, we don't have any Linux 
AX.25 maintainers anymore!  Unless we can find some folk who have the 
technical chops to fix some of these issues (I wish I did), I imagine 
the native AX.25 stack in Linux will become toxic and ultimately 
removed.  If there is anyone who is willing to give this a try, I can 
connect you with a few other people who have a consolidated list of 
issues and some possible fixes.  Some of those fixes were previously 
rejected from the kernel maintainers for various reasons without 
offering solutions.


--David
KI6ZHD
___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Tom Russo]

On Tue, Jan 30, 2018 at 09:06:10PM +1100, we recorded a bogon-computron 
collision of the  flavor, containing:
> On Sun, 28 Jan 2018 14:31:08 -0700
> Tom Russo  wrote:
> 
> > I would appreciate it if *everyone* on the list who uses AX.25
> > networking routinely would test out this code --- it impacts
> > *everyone* who uses AX.25 networking in Xastir.
> 
> I appreciate the need for testing, and perhaps this would fix my
> recurrent problem with the AX25 networking, but I don't see myself
> doing this in the short term.

I don't think it'll fix anything except to stop some systems from 
warning about obsolete usage, and to prevent Xastir from becoming 
non-functional some day when that obsolete stuff is removed from the kernel.

I'm mostly concerned with making sure this change doesn't *break* anything.

-- 
Tom RussoKM5VY
Tijeras, NM  

 echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] [n-z][a-m]

___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Liz]

On Sun, 28 Jan 2018 14:31:08 -0700
Tom Russo  wrote:

> I would appreciate it if *everyone* on the list who uses AX.25
> networking routinely would test out this code --- it impacts
> *everyone* who uses AX.25 networking in Xastir.

I appreciate the need for testing, and perhaps this would fix my
recurrent problem with the AX25 networking, but I don't see myself
doing this in the short term.

Liz
VK2XSE
___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

Re: [Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Tom Russo]

On Sun, Jan 28, 2018 at 02:31:08PM -0700, we recorded a bogon-computron 
collision of the  flavor, containing:
> Changing the subject line to get more people to see this.
> 
[...]
> 
> It took a while, but I think I found some code that should work (using
> AF_PACKET and SOCK_DGRAM), and am pushing it to github on my own fork so you 
> can try it out.  If the warnings go away for you, and AX.25 networking still 
> work for everyone who reports in, then I will push it to the main line code.
> 
> You can test it out easily:
> 
>git checkout master  # Likely a no-op, but let's be careful
>git checkout -b tvr-fix-ax25
>git pull https://github.com/tvrusso/Xastir.git fix-ax25
> 
> then rebuild the code and try it out.  You should be able to start an AX.25
> kernel port in Xastir and NOT see those warnings in your syslog.  If it works,
> please let me know and I'll merge it to main line.  It is working for me Just
> Fine.  But then again, I never had those warnings in syslog.

[...]

> I would appreciate it if *everyone* on the list who uses AX.25 networking
> routinely would test out this code --- it impacts *everyone* who uses AX.25 
> networking in Xastir.

I opened a new pull request for this branch to encourage discussion on it:
https://github.com/Xastir/Xastir/pull/20
If you try out the code, commenting on the pull request would be welcome.  If
you spot some horrid error in it, too, that would be a good thing to comment on.

> >On Sun, Jan 28, 2018 at 12:14:39PM -0700, we recorded a bogon-computron 
> >collision of the  flavor, containing:
> >> 
> >> I just experimented with fixing the preprocessor symbol so it actually
> >> detects we're running with a newer GLIBC, and it made Xastir AX.25 
> >> connections
> >> broken, so this code that has never actually been enabled isn't in fact
> >> correct.  Turns out that replacing PF_INET,SOCK_PACKET with 
> >> PF_INET,SOCK_DGRAM,
> >> as is done in that block of code, was never right.  The correct thing is
> >> to use AF_PACKET,SOCK_PACKET instead of PF_INET,SOCK_PACKET.  If I fix
> >> the symbol *and* fix that usage, Xastir builds and does connect to an
> >> AX.25 interface correctly.
> >
> >Actually, SOCK_PACKET is deprecated, too, and should not be used in new
> >programs.  This is going to take a little more careful experimentation.
> 
> 
> > > On Sun, Jan 28, 2018 at 05:58:54PM +, we recorded a bogon-computron 
> > > collision of the  flavor, containing:
> > > > Looking the the file /var/log/syslog, I noticed this...
> > > > 'Jan 27 07:13:18 Igate kernel: [26779.439570] xastir uses obsolete 
> > > > (PF_INET,SOCK_PACKET)'
> > > > 
> > > > I don't have a clue what it means or if I should even worry about it. 
> > > > Anyone else see this?
> > > > I just did an install a week ago so it's the latest version of xastir.
> > > > Thanks, Max KG4PID??
> > > > ___
> > > > Xastir mailing list
> > > > Xastir@lists.xastir.org
> > > > http://xastir.org/mailman/listinfo/xastir
> > > 
> > > -- 
> > > Tom RussoKM5VY   SAR502   DM64ux  http://www.swcp.com/~russo/
> > > Tijeras, NM  QRPL#1592 K2#398  SOC#236
> > > http://kevan.org/brain.cgi?DDTNM
> > >  echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] 
> > > [n-z][a-m]
> > > 
> > >  
> > > 
> > > 
> > > ___
> > > Xastir mailing list
> > > Xastir@lists.xastir.org
> > > http://xastir.org/mailman/listinfo/xastir
> > 
> > -- 
> > Tom RussoKM5VY   SAR502   DM64ux  http://www.swcp.com/~russo/
> > Tijeras, NM  QRPL#1592 K2#398  SOC#236
> > http://kevan.org/brain.cgi?DDTNM
> >  echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] 
> > [n-z][a-m]
> > 
> >  
> > 
> > 
> > ___
> > Xastir mailing list
> > Xastir@lists.xastir.org
> > http://xastir.org/mailman/listinfo/xastir
> 
> -- 
> Tom RussoKM5VY   SAR502   DM64ux  http://www.swcp.com/~russo/
> Tijeras, NM  QRPL#1592 K2#398  SOC#236http://kevan.org/brain.cgi?DDTNM
>  echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] [n-z][a-m]
> 
>  
> 
> 
> ___
> Xastir mailing list
> Xastir@lists.xastir.org
> http://xastir.org/mailman/listinfo/xastir

-- 
Tom RussoKM5VY   SAR502   DM64ux  http://www.swcp.com/~russo/
Tijeras, NM  QRPL#1592 K2#398  SOC#236http://kevan.org/brain.cgi?DDTNM
 echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] [n-z][a-m]

 


___
Xastir mailing list
Xastir@lists.xastir.org
http://xastir.org/mailman/listinfo/xastir

[Xastir] AX 25 networking kernel warnings --- heads up all AX.25 users Re: Error in Syslog

[Tom Russo]

Changing the subject line to get more people to see this.

Max reports that on his linux system the kernel is putting warnings in the
syslog about xastir using deprecated arguments to a socket call.  I looked into
it.  

The issue was known to Chuck Byam in 2003, but he made a minor error in the
code he put in to detect GLIBC versions, and so his alternate code has never
been getting built, and the old code just kept on working because it's just
deprecated, not removed.  And that alternate code, which simply used
PF_INET/SOCK_DGRAM instead of PF_INET/SOCK_PACKET, didn't work for me -- it
would compile and Xastir could connect to the socket, but it wouldn't receive
any data.

It took a while, but I think I found some code that should work (using
AF_PACKET and SOCK_DGRAM), and am pushing it to github on my own fork so you 
can try it out.  If the warnings go away for you, and AX.25 networking still 
work for everyone who reports in, then I will push it to the main line code.

You can test it out easily:

   git checkout master  # Likely a no-op, but let's be careful
   git checkout -b tvr-fix-ax25
   git pull https://github.com/tvrusso/Xastir.git fix-ax25

then rebuild the code and try it out.  You should be able to start an AX.25
kernel port in Xastir and NOT see those warnings in your syslog.  If it works,
please let me know and I'll merge it to main line.  It is working for me Just
Fine.  But then again, I never had those warnings in syslog.

You can and should return to the unmodified code with "git checkout master"
and a rebuild after testing out the branch and reporting the results.

BTW, Max, you reported these warnings in this email from 2013:
http://xastir.org/pipermail/xastir/2013-November/021836.html
but unfortunately the conversation turned to the real source of your problem
(broken wireless settings) at that time, and this obsolete socket settings
issue got overlooked.  So these warnings have been around for a good long
time and weren't causing any problem -- but the code was still wrong to be
using obsolete socket types, and someday this would have caused a problem 
when the "SOCK_PACKET" socket type is finally removed.  If it was already
deprecated in 2003, it's reasonable for them to think that folks would have
had time to adapt their code. 

I should note that the common ax25 tool "axlisten" uses "AF_PACKET/SOCK_PACKET" 
to do its work, so one of these days *that* will have some trouble, too.  Max, 
do you see the same warnings show up if you run "sudo axlisten -a -c"?

I would appreciate it if *everyone* on the list who uses AX.25 networking
routinely would test out this code --- it impacts *everyone* who uses AX.25 
networking in Xastir.

>On Sun, Jan 28, 2018 at 12:14:39PM -0700, we recorded a bogon-computron 
>collision of the  flavor, containing:
>> 
>> I just experimented with fixing the preprocessor symbol so it actually
>> detects we're running with a newer GLIBC, and it made Xastir AX.25 
>> connections
>> broken, so this code that has never actually been enabled isn't in fact
>> correct.  Turns out that replacing PF_INET,SOCK_PACKET with 
>> PF_INET,SOCK_DGRAM,
>> as is done in that block of code, was never right.  The correct thing is
>> to use AF_PACKET,SOCK_PACKET instead of PF_INET,SOCK_PACKET.  If I fix
>> the symbol *and* fix that usage, Xastir builds and does connect to an
>> AX.25 interface correctly.
>
>Actually, SOCK_PACKET is deprecated, too, and should not be used in new
>programs.  This is going to take a little more careful experimentation.


> > On Sun, Jan 28, 2018 at 05:58:54PM +, we recorded a bogon-computron 
> > collision of the  flavor, containing:
> > > Looking the the file /var/log/syslog, I noticed this...
> > > 'Jan 27 07:13:18 Igate kernel: [26779.439570] xastir uses obsolete 
> > > (PF_INET,SOCK_PACKET)'
> > > 
> > > I don't have a clue what it means or if I should even worry about it. 
> > > Anyone else see this?
> > > I just did an install a week ago so it's the latest version of xastir.
> > > Thanks, Max KG4PID??
> > > ___
> > > Xastir mailing list
> > > Xastir@lists.xastir.org
> > > http://xastir.org/mailman/listinfo/xastir
> > 
> > -- 
> > Tom RussoKM5VY   SAR502   DM64ux  http://www.swcp.com/~russo/
> > Tijeras, NM  QRPL#1592 K2#398  SOC#236
> > http://kevan.org/brain.cgi?DDTNM
> >  echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] 
> > [n-z][a-m]
> > 
> >  
> > 
> > 
> > ___
> > Xastir mailing list
> > Xastir@lists.xastir.org
> > http://xastir.org/mailman/listinfo/xastir
> 
> -- 
> Tom RussoKM5VY   SAR502   DM64ux  http://www.swcp.com/~russo/
> Tijeras, NM  QRPL#1592 K2#398  SOC#236http://kevan.org/brain.cgi?DDTNM
>  echo "prpv_a'rfg_cnf_har_cvcr" | sed -e 's/_/ /g' | tr [a-m][n-z] [n-z][a-m]
> 
>  
> 
> 
> ___
>