16:53 <javierm> bamse: any thoughts on https://github.com/andersson/rmtfs/pull/19 ? A rmtfs package and deps landed in Fedora, the only bit missing for Chromebooks is a systemd service that could start rmtfs with -o

16:54 <javierm> I would prefer to not ship a Fedora specific unit file but one that could come from rmtfs upstream

16:57 <bamse> javierm: would it make sense to make it a runtime thing instead?

16:57 <bamse> javierm: something where we look for partitions and fall back to the path?

16:58 <javierm> bamse: I thought about that, but didn't know a) how to properly check and b) what path should be used

16:59 <bamse> javierm: okay, let me take another look when my mind is in the right place...

16:59 <javierm> bamse: because the instructions I got was to use https://gitlab.com/postmarketOS/msm-cros-efs-loader (which I already packaged in Fedora) and that populates a /var/lib/rmtfs/ dir

16:59 <javierm> but didn't know if that's a convention or just the path that tool uses...

17:00 <javierm> bamse: I plan to write a Fedora wiki page explaining that one has to run ms-cros-efs-loader once, so that the FCG data is taken from the ChromeOS eMMC boot partition and then enable the rmtfs-dir.service

17:00 <javierm> for WoA I would mention that only enabling rmtfs.service should be enough

17:01 <javierm> *FSG

17:02 <bamse> javierm: my desire is that we can always launch rmtfs.service and it will just work, even in the case where we don't have a modem...so that we don't need to have device-specific configuration

17:03 <bamse> javierm: right now rmtfs fails if i don't have a modem remoteproc, in which case the entire systemd boot is marked as "failed"

17:04 <javierm> bamse: yeah, I see that in HP X2 Chromebook (without my patch) and systemct status says State: degraded

17:05 <javierm> bamse: if rmtfs can be made more smart and figure out what has to do (instead of the user telling it with -P or -o), that you be even better

17:05 <javierm> then we could just have a systemd preset file and enable the service by default

17:06 <javierm> bamse: but I didn't know how to do that, hence my patch to add another systemd service unit

17:07 <javierm> bamse: and that's the reason why I for now just ship rmtfs v1.0 without any additional patches, even when that's not helfpul for my use case :)

17:12 <bamse> javierm: i'll try to find some time to think this through properly

17:13 <javierm> bamse: awesome, thanks a lot!

17:14 <javierm> bamse: that's exactly why I opened that PR, so that you could chime in and tell me what's the proper approach. But unfortunately I'm not that familiar with QC platforms to figure out myself

20:50 <KieranBingham[m]> I've seen mention of "sc7180-trogdor-mipi-camera.dtsi" on patches posted to the kernel regarding Chromebooks. But I can't find any public reference to this file.

20:50 <KieranBingham[m]> Does anyone know if the definition of how cameras are connected on QCOM chromebooks at all is public anywhere?

20:50 <KieranBingham[m]> I understand the ISP being out at the moment but if we can configure in passthrough, we can look at testing with the soft ISP in libcamera

20:52 <javierm> KieranBingham[m]: https://chromium.googlesource.com/chromiumos/third_party/kernel/+/refs/heads/chromeos-6.6/arch/arm64/boot/dts/qcom/sc7180-trogdor-mipi-camera.dtsi

20:57 <KieranBingham[m]> Thanks. I tried different branches, and Google (search) didn't help

20:59 <KieranBingham[m]> There's not a lot to go on in there is there. They must even handle the cameras in userspace. I'm curious to figure out if they're using c-phy or dphy on this platform

21:02 <javierm> KieranBingham[m]: yeah, I don't know tbh

21:07 <KieranBingham[m]> If I can get the i2c interfaces up I think we can query the cameras... Maybe ... But ugh ... There's a lot of obfuscation going on.

21:07 <javierm> KieranBingham[m]: yeah. That's what eballetbo did for the BeaglePlay, where the ISP seems to be implemented in user-space using some GStreamer elements

21:08 <javierm> but he was able to get camera frames using just the CSI-2 and the i2c camera sensors

21:09 <javierm> KieranBingham[m]: maybe you can also get a BeaglePlay for your softisp testing ?

21:10 <KieranBingham[m]> Bah where's the fun in that :-)

21:12 <javierm> KieranBingham[m] :D