tcpdump-htdocs icon indicating copy to clipboard operation
tcpdump-htdocs copied to clipboard

Add ALDL link-layer type

Open jakem72360 opened this issue 2 years ago • 8 comments

PR to match #1143.

Added link-layer type 297 for GM ALDL as described here: Holden - ALDL (8192 Baud)

jakem72360 avatar Dec 12 '22 03:12 jakem72360

You should probably link to https://holden.asger.xyz/wiki/Holden:ALDL_(8192_Baud) in the page for LINKTYPE_ALDL, to describe at least some of the ALDL payload.

guyharris avatar Dec 12 '22 06:12 guyharris

And if it's different in the 160 baud link and the 8192 baud link, should this be LINKTYPE_ALDL_8192_BAUD?

guyharris avatar Dec 12 '22 06:12 guyharris

And, as the Wikipedia page for ALDL indicates, how model-dependent is the format?

guyharris avatar Dec 12 '22 06:12 guyharris

And, as the Wikipedia page for ALDL indicates, how model-dependent is the format?

Documentation is scarce, but the majority of ALDL devices that operate at 8192 baud has a consistent frame format of <id> <len> <data+n> <checksum>.

I considered creating LINKTYPE_ALDL8192 instead, but since it's the most common, I figured LINKTYPE_ALDL with a possibly separate LINKTYPE_ALDL160 would be more appropriate in future.

jakem72360 avatar Dec 12 '22 07:12 jakem72360

I can change the definition to LINKTYPE_ALDL_8192_BAUD if that would be more clear.

As for the site link, the site itself is a WIP, and the domain isn't finalized. Hence why I was hesitant to include it in long-term documentation at this stage.

jakem72360 avatar Dec 12 '22 07:12 jakem72360

Documentation is scarce, but the majority of ALDL devices that operate at 8192 baud has a consistent frame format of <id> <len> <data+n> <checksum>.

...but is <data> model-dependent (so that dissecting anything more than <id>, <len>, and <checksum> would require some other mechanism to specify the model)?

guyharris avatar Dec 12 '22 09:12 guyharris

Documentation is scarce, but the majority of ALDL devices that operate at 8192 baud has a consistent frame format of <id> <len> <data+n> <checksum>.

...but is <data> model-dependent (so that dissecting anything more than <id>, <len>, and <checksum> would require some other mechanism to specify the model)?

It is, but I'd argue that transcends the link layer. Detecting the vehicle model can be done with relatively straightforward heuristics. The purpose of my pull request, is to allow captured ALDL frames to be passed between programs. Anything outside of the message ID, length and checksum are beyond the link-layer, and is the responsibility of a program/user to interpret the meaning of the data section.

jakem72360 avatar Dec 12 '22 11:12 jakem72360

This needs to be updated, as 297, 298, and 299 have been assigned to other link-layer types.

guyharris avatar Aug 06 '23 06:08 guyharris