UMB/MBIM Support on Web Interface

4pack

@LuKePicci I used the FreeBSD 16.0-CURRENT latest VMDK image and brought in the master pfSense source repo. Then I just injected the patch you made and ran make on the umb module.

I DM'd you a link to the compiled module. I can also wrap up my entire development VM if you want that.

LuKePicci

@4pack thank you, ah ok so it's from vanilla Freebsd CURRENT, I fear there might be kernel flags/patch mismatch with the pfsense rebuild of the reference (older) 16-CURRENT. Let me test.

4pack

@LuKePicci Eh there could be. I figured as long as I wasn't replacing the whole kernel and just replacing the UMB module, there wouldn't be too many issues. I haven't panic'd my system yet.

I also quickly ported over your 2.7.2 changes to 2.9.0. Haven't tested them yet though so no guarantees that it works or doesn't break anything.

https://github.com/4pack/pfsense/tree/umb290

stephenw10

If the module loads it should be OK. Usually that sort of mismatch just rejects the module entirely. If there is a mismatch.

4pack

@stephenw10 Is there somewhere else where the 2.9.0 source is posted? I just tried comparing my repo changes on the pfsense master branch to what's installed on my 2.9.0 pfSense instance, and there's quite a few mismatches that would definitely break things.

stephenw10

Not other than github.

4pack

I have opened up a redmine issue for both of the repos.

https://redmine.pfsense.org/issues/17128

4pack

@LuKePicci said in UMB/MBIM Support on Web Interface:

https://github.com/pfsense/pfsense/pull/4760

One thing I did need to add to the pfSense master version was to add umb to the interface ignore list to support removable MBIM devices. Ran into trouble with that one!

LuKePicci

@4pack interesting, we generally had issues with interfaces not being present during boot. This was the same with Dip mode which spawn the module as ue0.

I managed to load the umb module you shared into a 2.9.0 VM.

My plan is to apply umb patches to the files into the 2.9.0 device files directly. Before that, I would generate a diff file between the current master branch and the files into 2.9.0. So in the end I can provide a clean diff including umb only patches for 2.9.0.

4pack

@LuKePicci That's definitely the better idea than what I did!

LuKePicci

Quick update, reconstructed baseline is done (find it in my GH fork), MBIM port to 2.9.0 is done also but I need to test it before publishing that branch as well.