Subj : Re: Tutorial for rookies
To   : N1uro
From : tenser
Date : Tue Oct 12 2021 10:59 am

On 09 Oct 2021 at 09:16a, N1uro pondered and said...
 
 N1> Hello tenser;
 N1> 
 N1> -=> tenser wrote to tenser <=-
 N1> 
 N1>  te> Perhaps N1URO could give us a rundown on what, precisely, he sees as
 N1>  te> problems with the Linux AX.25 stack that are NOT being addressed.  I'
 N1>  te> seen enough existence proofs of it working just fine to suspect that
 N1>  te> that might be overblown.
 N1> 
 N1> There's way to many to mention and it'd be pointless for me to duplicate
 N1> the data that's already out there.

What I see is modification to the stack based on incremental
updates elsewhere in the kernel.  You had said that there were
two or so outstanding issues that were particularly problematic,
though in exactly what way or whether they prevented Linux
kernel AX.25 from working at all was not specified.,

So I see activity, and my setup works.  Hardly proof, but given
that these issues you referred to don't seem to be tracked
anywhere and we have empirical evidence that the Linux AX.25
stack works, and given that your answer to questions about the
state of things are pretty light on details, one wonders whether
these things are still problems -- indeed, if they ever were at
all.

 N1> As you saw, there's already a patch
 N1> for 6-pack and a few for ROSE... but they seem hesitant to fix the most
 N1> important one which is the NetRom transport's virtual circuit socket NOT
 N1> closing when finished being in use. YO2LOJ, who's on the URONode dev
 N1> team, supplied a patch for it but they seem so hesitant to accept it for
 N1> whatever their political reasons are, and most hams don't want to go
 N1> through having to recompile a kernel.

Ah, now we're getting somewhere: enough information to locate
a patch!  I see that Marius Petrescu lists patched copies of
Linux AX.25 for download on his web site, and he describes a
patch in ax25_subr.c, adding an 'else' at the end of
`ax25_disconnect`.  Immediately obvious is that it does not
follow Linux kernel style.  It's also no immediately clear
that the patch is correct, but without knowing precisely what
the problem is in a little more detail, it's hard to tell.
Again, what's the actual bug here?

But I can find no record of this patch on the LKML or on
linux-hams (searching the archives at marc.info) and only
a few messages from Marius in in 2014 and 2018.  Perhaps
I just haven't looked hard enough, or perhaps it was never
sent upstream?

An invariant of contribution to Linux is that someone really
has to shepherd patches through until they get committed.
Sometimes the maintainers may ask for modifications to
patches.  That's just how it works, hardly "political"
reasons.

--- Mystic BBS v1.12 A47 2021/09/29 (Linux/64)
 * Origin: Agency BBS | Dunedin, New Zealand | agency.bbs.nz (21:1/101)

.