Saturday, 2025-07-26

T42<elros34> just check same things as for previou device: compare backports .config  with modules you have build00:03
T42<edp_17> Thanks. I'll compare with the previous sfos version (4.4). BT is working on it.00:07
T42<elros34> no, check current backports .config file and all built modules from backports ( better do build manually because make backports will probably move .ko files to out/)00:11
T42<edp_17> This logs shows what modules the backports built: https://paste.ubuntu.com/p/W2h9kjwDgk/00:13
T42<elros34> I don't trust log, use find/ls. does .config really have only these few enabled?00:28
T42<edp_17> Here is bluetooth defconfig: https://paste.ubuntu.com/p/f3kKpxwYt7/00:31
T42<edp_17> Only those flags are there.00:31
T42<elros34> .config != defconfig00:32
T42<edp_17> I know: https://paste.ubuntu.com/p/rxQDWSxRKK/00:32
T42<edp_17> And the full .config (from backports): https://paste.ubuntu.com/p/KtzjhsZpQR/00:33
T42<edp_17> I am sure those drivers on device are from kernel and not from backports.00:33
T42<elros34> what if you clean out .config and "make * bluetooth" does it produce some info?00:36
T42<elros34> "make * defconfig-bluetooth"00:37
T42<edp_17> It failed: https://paste.ubuntu.com/p/KKspGbPWwg/00:39
T42<edp_17> Thanks for your support today, I'll continue later.00:40
T42<edp_17> Have a nice night/day.00:40
T42<elros34> I was too lazy to write all the neede args, by * I mean ARCH= CROSS_COMPIILE and so on. Anyway see ya00:43
T42<edp_17> Oh, I see. That is one of the makes command you gave me last time. It was too late for me to think and remember. I'll continue when have time. Thanks.06:56
rinigusb100dian and mal : morning! making progress with aidl volte - can accept calls now and they work07:58
riniguswould have to figure out why call decline is not working07:58
*** poetaster is now known as Guest2277208:27
rinigusOK, qti_ims_call_hangup is not implemented09:24
rinigusto gbinder experts: how I am expected to write AIDL parcelables? I can see that we have gbinder_writer_append_parcelable to dump data buffer. which would work if my struct is fixed length. what about if it with strings?11:40
rinigusdo we have a way where I can compose parcelable via regular gbinder write functions (int32, string, ...), get data pointer and then add via gbinder_writer_append_parcelable to another writer?11:41
malwhat are you trying to write?11:44
rinigusmal: right now testing whether parcelable would work for answering the call. somehow (magically?), just writing 3 int32 into writer is read by qcom lib correctly. it seems like it should be AnswerRequest parcelable instead.11:50
riniguscode where its written without parcelable: https://github.com/rinigus/ofono-binder-plugin-ext-qti/blob/main/src/qti_radio_ext.c#L112411:50
rinigushower, I am preparing for handling hangup but testing on whats working now11:51
T42<abranson> parcelables are just chains of values, but start by printing the constructor java class name that would be used to reassemble them on the other side. that class will contain the serializing and deserializing code, so the android source is the best place to see what you need to write.11:58
rinigus@abranson: that's right. so, for ims we don't really have sources and I have smali decompiled apk instead. and I can on SFOS side read them nicely (well, not super convenient, but OK)12:01
rinigusbut writing parcelable is more complicated if you have strings in it. suddenly I can't just use gbinder_writer_append_parcelable with the struct, but have to compose elements into buffer myself12:02
riniguswhich is something I guess the writer is doing. so, hence the question - is there a way to compose such dynamic struct via writing its elements and then add as parcelable to "real" writer12:02
malmanually writing parcelable as values is simple, my aidl codes have plenty of example12:03
rinigusmal: do you pre-calculate the size then?12:04
malrinigus: check my codes in various places, ofono-binder-plugin or any other, I overwrite the size after writing everything12:05
rinigusexample struct to hangup int, int, string, CallFailCauseResponse, bool12:05
rinigusmal: that's the way android is doing as well. will look for it, thanks!12:05
malrinigus: https://github.com/mer-hybris/ofono-binder-plugin/blob/master/src/binder_network.c#L133712:06
malthat even has comments where it writes dummy size and where it overwrites the size at the end12:08
malI do in few cases precalculate size if it's just basic types like a couple of integers etc12:10
malbut for anything more complex I usually replace the size12:10
rinigusthank you very much! its exactly what's needed12:10
rinigusmal: do you also have to correct values of many constants when going from hidl to aidl ims? quite a few are shifted by one in my case12:12
malthat is a bit problematic12:13
malthere were some conversion things in the apk12:13
rinigusI may have missed the conversions. which one do you keep in mind?12:14
rinigusin my case, for example, CallType voice was 0 in hidl and 1 in aidl12:16
malthere is is that CallType defined nicely in one java file12:18
rinigusmal: yes, that's where I took AIDL value from12:19
malsame for the others also12:19
rinigusI just wonder how it worked for HIDL before. and as you have working hidl code, I wonder if you had to make similar changes for aidl12:20
malyeah, need to convert things, well depends on how you are doing the changes12:20
malif you do completely new plugin or add to existing one12:21
rinigusmal: right now, dropping hidl and going for aidl only. later will discuss with ofono-binder-plugin-ext-qti devs on how to merge it.12:21
rinigusI do wonder where do they hang out...12:21
rinigusas for now, trying to get it working first12:22
malI'm trying to debug that video playback issue on devices using c2 codecs12:22
malI want that to be working before enabling video recording on fp512:22
riniguswell, video recording is handy even if there is issue with playback. btw, in browser on nagara that recorded video plays just fine12:36
malsame on fp5, very odd12:39
malwell, it plays better but not great12:40
malstill hangs sometimes12:40
rinigusfingers crossed you fix it - we will surely all benefit.13:18
rinigusmal: in hangup call, do you fill connUri or does it work without it just fine? will test without soon - just was curious if you had to specify it13:19
rinigus(cutting corners by asking you)13:19
malwell I haven't been able to test my changes properly yet since I can't get registration to happen13:20
rinigusmal: hmm, weird. for me registration works fine. do you want to test with OSS plugin? aidl branch at https://github.com/rinigus/ofono-binder-plugin-ext-qti should register, support the calling to it, and accepting calls13:23
malyeah, I could try with that13:25
malnot right now but a bit later today, need to take a break13:26
rinigusenjoy the break!13:29
malrinigus: btw, could you rebase that aidl branch, it had some changes missing, the removal of libhybris build requirement15:14
rinigusmal: strange, will check15:46
rinigusmal: indeed - done and force pushed15:49
malrinigus: I haven't had luck getting ims registered yet, even with your code15:54
malmust be some issue in the device15:54
malims:imsradio0: QtiRadioRegInfo state:2 radiotech:16 error_code:0 datasz: 3215:54
rinigusor some difference in ims.15:54
malsome android side issue probably15:55
rinigusmal: does it work on sodp?15:55
malI had trouble even there15:55
malmaybe something is missing in the aidl stuff there15:56
rinigusthen no wonder its failing on sfos15:56
rinigusissue in their repo?15:56
malI reported it directly via other way15:57
riniguslet's hope they will fix it15:58
malyeah15:59
malfp6 lineage is not yet ready so that limits a bit getting that running15:59
malfun that there are no errors in logs about ims16:11
riniguserrors are minimal, indeed. not getting any errors on calling hangup either + nothing happens there too.16:36
malI'll go back to video debugging16:38
rinigusgood luck!16:40
malsomething odd in the QoS things during playback17:55
Mister_Magisterodd is my mce pr not being merged yet17:59

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!