07:14 <macc24> jenneron: veyron-minnie, gru-kevin, kukui-krane and trogdor-lazor

07:15 <macc24> i think i can reach few other people with other veyron devices

18:34 <qzed> bamse: so... regarding wifi: turns out pd-mapper isn't starting (not sure how I missed that before...)

18:34 <qzed> it expects some json files that I don't have

18:34 <qzed> how can I get or create them?

21:38 <bamse> qzed: windows must be hard coding these things in some driver, but you can find them scattered for other releases...i threw them the ones i use together for you here: https://www.toptal.com/developers/hastebin/omezalosep.yaml

21:38 <qzed> thanks!

21:39 <bamse> and yes, unfortunately you must have these jsn files and pd-mapper running, as well as tqftpserv to then provide the wlanmdsp.mbn

21:39 <bamse> on sc8280xp the tqftpserver is gone again, so i've been considering pushing the pd-mapper implementation into the kernel...

21:40 <bamse> but we have a few other fish to fry before that

21:42 <qzed> okay, so I get that the modem (and other remote procs?) communicate with with user-space (and other things?) via the router, and that the ftp server/remotefs is required to give the remote procs access to the host file system

21:42 <qzed> so what's the rest used for?

21:43 <qzed> or rather, what is PD?

21:44 <qzed> kinda seems like some service management/service setup for the router?

21:49 <qzed> bamse: also, do you mind if I push them to a git repo somewhere?

21:50 <steev> could we try to get them pushed to linux-firmware?

21:50 <steev> sdm845 and sm8250 are there

21:50 <steev> i actually tested those on the c630 and they work, as long as they're in the correct spot

21:54 <bamse> qzed: go ahead, some of them are in linux-firmware already and the others i made up based on the ones that are there

21:54 <qzed> neat, thanks again!

21:55 <bamse> qzed: "pd" is "protection domain" or something like that; the software running on the hexagons consists of a small core and then individual, isolated pds which runs the actual application code

21:55 <qzed> ah, okay

21:56 <bamse> qzed: so on your sc8180x you ahve the root_pd and the modem_pd running already, and based on the information in the .jsn files we tell the root on the modem hexagon that it's also supposed to do wifi on this platform...which causes it to request the firmware for wlan_pd and boot that

21:59 <qzed> okay, got it

21:59 <qzed> next question: in what path(s) is the rmtfs client thing looking for files?

22:00 <bamse> /dev/disk/by-partlabel/

22:01 <bamse> so the systemd service should just work for you...

22:01 <bamse> unless someone did something real funny

22:03 <qzed> ahh, it even tells me when it doesn't find something it's looking for, nice!

22:04 <bamse> there's typically one of warning, that you can ignore

22:05 <bamse> but if rmtfs doesn't work, then the modem will not initialize properly and will crash after 40 seconds

22:05 <qzed> it failed earlier for me with `failed to open '/dev/disk/by-partlabel/modemst1' (requested '/boot/modem_fs1'): No such file or directory`

22:06 <qzed> followed by immediate `fatal error without message`

22:06 <qzed> so I guess I need to figure that out now...

22:07 <qzed> I assume those are the couple of partitions that I can see in the efishell but don't show up in linux... so I guess I'm missing some config option?

22:22 <qzed> I guess I'm missing CONFIG_MTD_QCOMSMEM_PARTS

22:32 <qzed> hmm, that doesn't seem to be all that's required...

22:32 <steev> bamse config for the flex5G only has

22:32 <steev> # CONFIG_MTD_QCOMSMEM_PARTS is not set

22:32 <steev> CONFIG_RPMSG_QCOM_GLINK_SMEM=m

22:32 <steev> CONFIG_QCOM_SMEM=y

22:32 <steev> CONFIG_QCOM_SMEM_STATE=y

22:33 <steev> qzed: check that yours isn't called modemst1

22:34 <qzed> hmm, I do have those too, just with CONFIG_RPMSG_QCOM_GLINK_SMEM=y

22:34 <qzed> it's called modemst1 according to the log above...

22:34 <qzed> why?

22:36 <steev> that, i couldn't tell you

22:37 <steev> on my flex5g (and c630), they're labeled modemst1/modemst2

22:37 <qzed> thanks, I guess I'll try looking for other options that I could be missing then

22:39 <bamse> qzed: hmm, not sure where those partitions are...on ufs they are just normal partitions

22:40 <qzed> oh, hmm.... so I should maybe try adding ufs support... according to the schematic it didn't look like the pro x had that

22:41 <steev> what does the prox use?

22:41 <steev> nvme?

22:41 <qzed> yep

22:41 <steev> oh, it might have those still hiding away on ufs, yeah

22:45 <bamse> doubt that they have a ufs device for those, more likely that there's a spinor somewhere

22:45 <bamse> or perhaps some secret partition on the nvme?

22:46 <qzed> any idea how ufs shows up in the efi-shell? the nvme shows up as Pci(...)/.../NVMe(...), the other partitions which I assume are the ones I want are VenHw(<some guid>)

22:47 <bamse> qzed: you could try using -o to point rmtfs to some other directory and just create some empty files (i think they need to be ~2mb in size)

22:48 <bamse> qzed: that might actually be sufficient to get the modem up, assuming that wifi runs off its own settings/calibration data and not what's provided by rmtfs

22:48 <qzed> I'll give that a try if setting up ufs fails and i can't find out more via windows

22:48 <qzed> I guess I could also dump the partition contents to some usb from the efi shell...

22:49 <bamse> if you can find and access them from efi, then you can certainly do that

22:49 <bamse> and just use -o to override the /dev/disk/by-partlabel path

22:49 <qzed> IIRC I had access to at least some, but I think not all

23:41 <qzed> do these partitions show up in windows on the flex 5? or are they hidden? I can't find them on the spx...

23:45 <bamse> on the UFS-based devices you have 5-6 different LUNs and in windows only the first one is visible

23:46 <qzed> I'm just looking through the device manager... there's some soc partition driver but no indication of ufs

23:47 <qzed> I can see the nvme and nvme volumes... but nothing else