Thursday, 2019-12-05

rinigusspiiroin: you are right, I did start with power press when screen was off. As if we are 'dead'. With sorry press to switch on screen and later press to reboot, it all worked - phone rebooted into full system.05:52
rinigusSounds like my misunderstanding is rare enough to ignore it and focus on other issues. Thank you very much for your help!05:52
spiiroinrinigus: well, now that I spent a bit thinking about this... that long press change ddid make this particular bit a bit more unintuitive05:54
spiiroinbackground: most devices handle power key press while suspended by reporting immediately both key down and key up events -> display power up starts immediately; then came e.g. xperia x that reports key up when user lets go off the button05:57
spiiroin-> it was both a bit non-deterministic and was causing further delay to already slow display wakeup on xperia x05:58
spiiroin-> if there are no long press actions defined, power key down is handled already on key down -> the delay is avoided05:58
spiiroinact dead is a bit special case, but it got affected too: before the change display did not wake up until you lifted finger from button, now it does -> user gets feedback -> maybe expects that it will shutdown/reboot soon?06:00
rinigusspiiroin : my line of thought was that I don't know anything about phone being on. I switched it off and put on charger. So, long press was the same as when phone is really off to see the device.06:03
spiiroinrinigus: there might be room for some usability tweaking here ...06:06
rinigusSure, there is always some :)06:08
deathmistmorning, I assume I can test version OTA going from 3.1 => 3.2? package upgrades already function perfectly with OBS KS :)  and about KS, I probably use a locally generated one instead and add community-adaptation repo in SFOS 3.1 manually before doing the OTA or? all packages on devel are for 3.2 (I didn't have testing setup for 3.1 yet either)07:29
r0kk3rzyour KS has your OBS repo?07:51
deathmistyep, at least my current 3.2 one and I've tested package upgrades already on this version (droid-config + other MW) via zypper ref && zypper dup, the system wasn't left in an unbootable state anymore like before :p07:54
r0kk3rzthen check you get a new kernel flashed up upgrade, otherwise the actual version OTA should be fine07:55
deathmistah nice! alright will do in a few hours, will be nice to see if modules also get upgraded so Wi-Fi works after kernel updates etc :)07:57
deathmistlast thing missing from my the port now is cellular data... then as far as I'm concerne it's ready for a full XDA release07:58
deathmistofc I should still also setup :testing for proper version upgrades and put packages on specific commit hashes and scrub all webhooks08:03
deathmistpiggz: btw has an expired cert since a few days ago ;)08:05
r0kk3rzare you on the porters_ci?08:05
deathmistr0kk3rz: nope, not yet. my gitlab name is same as here and device was oneplus/cheeseburger (and dumpling for 5T)08:06
r0kk3rzdo you want to be?08:06
deathmistfor sure ^^ I've been uploading my local builds on MEGA and GitHub releases page lol08:07
r0kk3rzok, ive added you08:13
r0kk3rzyou can create whatever projects you need08:13
deathmistcool, thanks! I'll look into it after I get :testing going on OBS08:19
T42<adampigg> (Photo, 2560x1440)
T42<adampigg> Pine dev sample and watch devkit :)09:17
vknechtit's christmas already, for you ? :-)09:23
mal@adampigg nice :)09:24
vknechtfor ramdisk command line, where to find values to use for --kernel_offset and --second_offset ? is it ok to just reuse (or ignore) the ones used for ganges and tama ?09:46
T42<adampigg> Yup10:10
deathmistr0kk3rz: kernel updates definitely work too :)  subversion changed, Wi-Fi (module) still works and other misc changes I had all made it in! gonna paccopy stuff to :testing too now and see how that all works11:13
deathmistr0kk3rz: I assume I could make work on the ci docker image? it just needs to be git cloned to be included in built images12:10
Mister_Magistermal: how would you change resolution in phone12:47
Mister_Magisterbasically i need to offset whole display at the top by 53px12:48
malprobably needs some patching in lipstick at least, not sure what else could be needed12:50
Mister_Magisteri was trying to patch some qml's on ApplicationWindow level but that was meh result12:50
T42<adampigg> (Photo, 2560x1440)
T42<adampigg> Mal ^^12:52
Mister_Magistermal: what patching are you talking about?12:52
mal@adampigg cool12:53
malMister_Magister: no idea about details12:53
Mister_Magisterbut like github repo or any hint?12:53
T42<adampigg> Mister_magister yes12:53
malit could be lipstick, libhybris or hwcomposer plugin12:53
Mister_Magisterhwc plugin most likely i think12:53
Mister_Magisterlower the lever the better12:54
Mister_Magisterkrnlyng: maybe you have idea how can i change resolution in hwc plugin?13:00
Mister_Magisterwhat if i just like forced height do be different lmao :P13:05
Mister_Magisterwait isn't that string below an error?13:06
Mister_Magisterhmm but i don't see any code to offset it…13:07
T42<elros34> you could try in createWindow function. I swaped width/height there in native landscape display:
T42<elros34> there are also layer-> which I guess could be used to offset13:14
T42<NotKit> you could possibly hardcode layer size and offset it13:15
T42<NotKit> (though lipstick QML patches seem like a better solution to me)13:15
Mister_Magister@NotKit just a quick fix i wanna have sane display till i get it working then i will try  making some fancy patches to qml's and also some people like notch hidden + sdl apps13:16
Mister_Magister@elros34 @NotKit i thought of using this 2 lines
T42<NotKit> then try to reduce layer height, add size of notch to position y, then lie about screen resolution to the13:18
T42<NotKit> layers above13:18
Mister_Magister@NotKit i don't think i follow13:19
T42<NotKit> hwc2_compat_layer_set_display_frame(layer, 0, 53, width, height);13:25
T42<NotKit> hwc2_compat_layer_set_source_crop(layer, 0.0f, 0.0f, width, height - 53);13:25
T42<NotKit> hwc2_compat_layer_set_visible_region(layer, 0, 0, width, height - 53);13:25
T42<NotKit> maybe like this13:25
T42<NotKit> but you need to make sure QPA reports height - 53 as display height to outside13:26
Mister_Magisteryeah but i don't have to add y to visible region too?13:26
Mister_Magisteri was thinking about that13:26
Mister_Magister@NotKit here?
T42<NotKit> @Mister_Magister [yeah but i don't have to add y to visible regi …], visible region is layer space13:26
Mister_Magisterso it has to fil13:27
Mister_Magisterand the display frame offsets it?13:27
Mister_Magisterwhy not -53 in display frame as height?13:27
Mister_Magisterso many questions13:27
malMister_Magister: so just try and see what works13:27
Mister_Magistermal: i will do!13:28
malMister_Magister: you never learn if you don't just dive in the deep end13:28
Mister_Magistermal: yep!13:28
T42<NotKit> @Mister_Magister [and the display frame offsets it?], display frame is defined not as offset, but portion of screen that a layer covers13:28
malI had to patch hwcomposer for my first port, it was not fun13:28
Mister_Magisterpoor man :P13:29
T42<NotKit> make sure to add some debug prints to check which values do you actually set13:29
malit was actually a buggy android side hwcomposer code, not hwcomposer plugin13:29
T42<NotKit> I think that faking display height will cause createWindow to get height - 53 as argument13:29
Mister_Magister@NotKit thanks i think i will just play like a kid setting random values :P13:30
Mister_Magisterfor fun :313:30
deathmistmal: I can't seem to make packages build on :testing I must be missing something
deathmistlooks like cibot hasn't done anything, I just did a paccopy from :devel and removed all webhooks from _service files13:55
vknechtseems you miss <arch>armv8el</arch> in Meta tab (or repo setting)13:56
vknechtah, no13:57
deathmistin meta tab project path "sailfishos:" has repo "latest" while onyx and my devel seems to have "latest_armv7hl" instead, that must be it14:00
maldeathmist: fixed it14:00
malsome copy paste mistake from devel14:01
rinigusHopefully someone can reply: how to test nfc on sfos, taking into account that there are no aps using it. Or do I miss something?14:09
T42<jgibbon> rinigus: I don't have a supported device, so I can just reiterate what I've read somewhere and for some reason remembered: There seems to be command line tools ndef-read and ndef-write for that, also seems to have a dbus interface14:15
malrinigus: if you have a text/url tag then just use it, there is also one app already for helsinki bus card reading14:15
Mister_Magisterrinigus: you would have to buy NFC url tags they should work ootb14:15
rinigusjgibbon, mal, Mister_Magister : thanks! I'll try to read it with ndef tools. when its approached by other phone with nfc share enabled, I do get vibration and something in logs. now I should be able to see in details.14:17
malrinigus: use nfc logger app14:18
malif you want to what is seen14:18
mal*want to know14:18
rinigusmal: where is that available from?14:19
rinigusmal: thanks!14:21
Mister_Magistermonich: rules!14:22
malthis is the first actual nfc app also available in store14:26
malbtw, does anyone have nfc in their older devices? I mean with older than android 8 based build14:27
deathmistmal: it seems libgbinder (1.0.32), libgbinder-radio (1.0.8) and audiosystem-passthrough (1.0.0) are missing completely from nemo:testing:hw:common/sailfishos_3.2.0.12 (found in nemo:devel:hw:common) which causes my OBS packages to fail building on :testing14:33
Mister_Magistermal: i do14:34
Mister_Magisteron hybris 1314:34
malMister_Magister: which chip does it have?14:36
Mister_Magistersnap 80114:36
malMister_Magister: I meant nfc chip14:36
malshould be visible in dmesg or kernel config or somewhere14:36
maldeathmist: those have never been in testing14:36
vknechtguess I have it too, on 14.1, bcm2079x14:37
Mister_Magistermal: no idea i can check later14:37
deathmisthmm, guess I'll build them manually on testing then14:38
maldeathmist: maybe I should add those there14:38
vknechtgrep'ing for nfc through journal and logcat:
malvknecht: ok, that would then need a new plugin14:44
deathmistmal: thanks for adding the packages :) any reason I wouldn't have "70_sdk-domain" section in my OBS KS file? just comparing it to other ones on porters-ci15:05
vknechtrinigus, any insight about bootloop after ~15s on sony logo, without led ever on16:20
vknechtselinux disabled in sparse, tried with and without selinux=0 in droid-hal-img-boot, I have --os_version and --os_patch_level16:23
vknechtno sign of life from usb16:24
rinigusvknecht: not really... But try to get logs. See my sparse for services and a script in sparse/root. Just have to get in with recovery...16:24
rinigusDoes recovery work?16:24
vknechtdoesn't seem so16:25
rinigusIf lucky, could be selinux related. I may have had similar in 3.2. Disabled selinux in sparse/etc/selinux (can't check, away from pc)16:27
rinigusvknecht: do you have lvm at your userdata partition? I'd suggest to make recovery working first.16:29
vknechtyes, disabled in sparse/etc/selinux/config and have lvm on userdata16:30
rinigusFor that, I installed aosp and twrp. On boot made it log on userdata partition and checked it with twrp after recovery boot16:31
rinigusWith recovery working, we'll get access to logs on lvm16:31
ilpianistawhich are the "phone requirements" to run SailfishOS? Is 1GB of RAM enough?16:53
ilpianistaalso, how do I know if I should build QCOM with BSP or not? And what's "RIL version"?16:55
malilpianista: I have run sailfish on a device with 512 MB of RAM, but at least 1 GB is recommended though17:00
malilpianista: which device are you porting?17:00
malwhich android base?17:01
ilpianistasony xperia m2 (eagle)17:01
ilpianistathe base android just finished the build!17:01
vknechtrinigus, to get logs from recovery, just have to add lines like "dmesg > /data/dmesg.log" in droid-hal-img-boot-sony-loire-pie/initrd/recovery-init ?17:08
vknechtdoesn't seem to work, either it crashes/reboot before init, or /data is wrong17:09
deathmistr0kk3rz mal: "Unable to find pattern: Jolla Configuration $DEVICE" can I run on before mic cmdline perhaps? I think that was the fix, but I've only used it locally for now17:23
maldeathmist: you don't have patterns enabled in testing project17:32
deathmistmal: I see, I may just use meta-packages instead tho (at least attempt first)17:33
rinigusvknecht: (was trying to remember and look through my init) I took classical hybris recovery, without lvm, first. that can write on data partition if its in ext fs. just had to set DATA_PARTITION in init script17:57
rinigusto make it easier, I used aboot to disassemble and reconstruct boot image with my hacks. that saved some time17:58
rinigusbut don't you have any recovery from earlier versions?17:58
rinigusits xperia x, right? we may as well use OFFICIAL one17:58
rinigusit should read lvm just fine :)17:58
rinigusdeathmist: for meta packages, you could use and then add to get corresponding KS18:17
rinigusyou would also need _patterns replacement as in (see also included file in / of that repo)18:18
piggzdeathmist: ill fix my certs now....thatll be why my calendar stopped syncing!18:35
riniguswhen approaching with other phone, nfc logger does push out some logs, ndef-write/read do complain that the 'object is gone'. looking back in this channel logs, seems like only simple tags are supported. as I don't have any, will leave it here, assuming it works.18:39
rinigusquestion regarding switching off NFC: it seems to be able to switch it off in settings and nfc doesn't work after that. however, after reboot, nfc is on again18:40
rinigusis it only my device or is it sfos feature?18:41
deathmistrinigus: damn it automatically enables for me too, I really don't use or care for it and would like it to stay in it's last state too, not just auto-enable and possibly drain battery for no reason19:02
deathmistwhat's also weird is that NFC seems to be the only thing that gets automatically enabling after boot (i.e. cellular, WLAN and BT all load last states)19:06
rinigusdeathmist: exactly. Nfc doesn't seem to be controlled by rfkill, so I wonder how it's enabled/disabled19:13
deathmistwelp, now I'm using meta-packages on droid-configs... let's see if I broke everything :D19:20
deathmistoop, wrong link lol
deathmistaand now the build is passing and cheeseburger CI @ is online & usable! tested produced image: it boots, OTA functions and it's ready for XDA minus cellular data20:52
rinigusdeathmist: congratulations!21:04
deathmistthanks :) I really want to get cellular working now so I can use this as my main OS again, @nitin03 already got cellular on his Poco F1 with mostly same droid-config sources, soo something is wrong on my android side that's pretty much confirmed21:20
r0kk3rzbtw, are you guys hand crafting your meta packages? or using my pattern migration script?22:34
deathmistI just did it all by hand lol, wonder if I can even use my droid-configs for local builds anymore after basically deprecating patterns on my repo22:50
deathmistlocal builds aren't too necessary for me at this point since OTAs for packages and stuff just works, but still would be nice to have around22:51
r0kk3rzpatterns and metapkgs can coexist22:52
r0kk3rzbut yes, once you have OBS then just use that22:52

Generated by 2.17.1 by Marius Gedminas - find it at!