Re: Request to add: LINKTYPE_OASIS

[Denis Ovsienko via tcpdump-workers <tcpdump-workers () lists tcpdump org>]

> --- Begin Message ---
>
>
> From: Denis Ovsienko <denis () ovsienko info>
>
> Date: Tue, 3 Jun 2025 01:30:07 +0100
>
> On Sun, 1 Jun 2025 14:09:41 -0700
> Howard Harte <hharte () magicandroidapps com> wrote:
>
> > In particular, thank you for pointing out the existence of
> > LINKTYPE_RTAC_SERIAL. I will certainly take a closer look at its
> > structure and consider its potential applicability to capturing OASIS
> > Send/Receive traffic.
>
> You are welcome.  Please note I mentioned LINKTYPE_RTAC_SERIAL to make
> it clear that serial port data capture is a problem space of its own,
> not to suggest it necessarily should be a subset of the protocol
> problem space you are looking to solve.
>
> It would help to decide how OASIS Send/Receive relates with the layer
> that transports it.  The original implementation was designed to run
> over serial connections, but there is nothing in the protocol
> specification that would prevent another implementation from running
> over a network protocol (similarly to Kermit over TCP, it seems).
>
> One way to make this design decision would be to say that the transport
> connection specifics is out of scope, and to include only the protocol
> characters in the link-level encoding.  This would be similar to a log
> file of httpd/ftpd/tftpd, supposedly with better granularity of wire
> events, but without the ability to log additional debugging messages.
> Arguably, if such a .pcap file comes from oasis-utils as a by-product
> of driving a serial port, there would be a lot of overlap between the
> .pcap file and the log file.
>
> Another way would be to say that the main focus is the exchange on the
> transport connection, and what higher-level protocol it carries is a
> secondary detail.  In this case LINKTYPE_RTAC_SERIAL may be not trivial
> to implement correctly, but converting the serial connection to a TCP
> connection would put debugging OASIS Send/Receive into the same
> solution space as SMTP, HTTP etc., which most of the time does not even
> require a purpose-designed link-layer header.
>
> Obviously, the latter approach would depend on hypothetical TCP support
> in oasis-utils.  The native OASIS end of the connection could use a
> serial port as before, and the PC end of the serial port would need to
> use a converter (e.g. ser2net) to make it TCP.
>
> Also, before I forget: in any case the .pcap/log file regardless of the
> format would need to log 8-bit characters to make it possible to see if
> any end of the connection violates the 7-bit space of the protocol.
>
> Hopefully this provides some more useful material for this development.
>
> -- 
>     Denis Ovsienko
>
> --- End Message ---
_______________________________________________
tcpdump-workers mailing list -- tcpdump-workers () lists tcpdump org
To unsubscribe send an email to tcpdump-workers-leave () lists tcpdump org
%(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s