Monday, 2019-10-28

xreactx[m]__"/home/vincent/hadk/droid-local-repo/guacamole/droid-hal-guacamole/droid-hal-guacamole-detritus-0.0.6-201910280054.armv7hl.rpm" its built01:36
rinigusThaodan: I presume that you have dtbo and vbmeta which have to be flashed06:50
rinigusTo build dtbo, see, end part of #Build hybris-hal06:51
rinigusTo build vbmeta, see targets used in the last line of the same hadk at Build systemimage vendorimage06:51
rinigusif you have the same system as used in xz2, dtbo is not that critical - it boots with the "wrong" one as well06:52
rinigusto flash: fastboot flash dtbo dtbo.img06:52
rinigusvbmeta, though, is critical and it will not boot without proper flashing of it as its part of verification06:53
rinigusto disable verification (we can later maybe make it properly), flash with06:54
rinigusfastboot --disable-verity --disable-verification flash vbmeta vbmeta.img06:54
rinigusxz2 bootloader has few bugs: according to aosp devs, it does not boot if you are in active slot B. So, I have to always use slot A for booting06:55
rinigusalthough, system_b mounts without issues during sfos boot06:55
rinigusnext, on xz2, I have
rinigusif you keep getting corrupted image message, try to disable all non-needed fancy filesystems in sfos (nfs and friends). that will reduce kernel size and allow it to boot. later we can probably stick them to modules06:57
riniguscurrent working hypothesis is that its just flips corrupted flag when you boot too large kernel06:58
rinigusThaodan: ^06:58
Thaodanrinigus: I don't have the same system but the Xperia 1 with android 10 which has a dtbo partition too and I didn't know if I need to handle that somehow07:39
Thaodanmy device just rebooted my image and didn't show any output on the screen after the sony logo.07:40
ThaodanI was wondering where the issue is and if I had a similar issue like you with the file systems07:41
ThaodanI would kill for dmesg from uart07:41
rinigusThaodan: already getting to sony logo was a success for me with all these bootloader issues :)07:46
rinigusThaodan: use your own advice you gave to me - hybris-recovery :) . I boot that just with fastboot boot hybris-recovery.img .07:47
ThaodanI don't understand which hybris-recovery shows no output on the screen.07:47
Thaodanfastboot boot is broken from the official side no idea why07:48
rinigus... but, as bootloader is buggy, you may have to issue "fastboot reboot bootloader" before fastboot boot07:48
rinigus(assuming that xperia 1 & xz2 share some bugs)07:49
Thaodanquote from jerpelea (sony)  "ok devices that use dtb overlay this functionality can create issues"07:49
rinigusThaodan: these made rndis work for me (thanks to mal):
rinigusI wonder what sony devs mean by issues with dtbo07:51
Thaodanno idea this stuff isn't that good documented imho07:51
ThaodanI already had CONFIG_USB_CONFIGFS_QCRNDIS=y07:53
ThaodanBut I don't had CONFIG_USB_CONFIGFS_RNDIS=y07:53
ThaodanI addded it for testing07:54
rinigusThaodan: if you look at the last version, I had to drop those07:54
rinigusand to make it compile, had to set CONFIG_USB_CONFIGFS_F_GSI=n07:54
ThaodanMine builds fine with them.07:55
ThaodanBut I have 4.14 so no idea.07:55
rinigusalso, xz2 uses b.1 for gadget config, not c.1 as in our hybris scripts. but I am not sure whether its important... and not sure whether I have default c.1 working as well in my last images. but mentioning it to be sure I don't forget it07:56
rinigusThaodan: mine if 4.907:57
ThaodanI know07:57
ThaodanI wish mine would have just pie as it was released on stock07:58
rinigusThaodan: but speaking of dtbo - when you figure out how it works and whether we can mostly ignore it, please share over here. would be good to know09:16
Thaodanmine won't even boot to hybris-recovery09:17
Thaodanreboots after a few sec and the screen is black/off after the sony logo09:20
rinigusThaodan: sounds different to xz2, that I can boot with 'fastboot boot myboot.img'. I guess you will have to flash recovery as boot. did you flash vbmeta as I mentioned above?09:22
Thaodanyes yesterday09:24
Thaodanmade it with make vbmetageimage09:24
rinigusThaodan: and with those disable flags?09:26
Thaodansudo fastboot --disable-verity --disable-verification flash vbmeta ../android-syspart/out/target/product/griffin/vbmeta.img09:26
rinigusThaodan: ok, at least that's not on the way :)09:26
rinigusThaodan: and when you make that dtbo image as in my hadk - you still have issues, right? (aka recovery flipping back to black screen)09:28
Mister_MagisterThaodan: rinigus: u working on griffin or what09:33
rinigusMister_Magister: Theodan is working on it.09:35
Thaodanrinigus works on tama09:35
Mister_Magisteroh sorrey i'm mistaken09:35
rinigustama/akari :)09:35
Mister_Magisterit's first time i see 2 devices sharing same codename09:35
Mister_Magisterthere is sony-griffin and motorola-griffin on which i'm working on09:36
Thaodanmine is kumano/griffin09:36
Thaodanah funny09:36
Mister_Magistergood luck with our griffins :P09:36
Thaodanrinigus: this is how it goes:
Thaodanit stays at black screen for about 40s and the reboots09:37
ThaodanI hope you have more luck than me Mister_Magister09:37
rinigus(I assume that this is with matching dtbo) anything on dmesg in your linux pc?09:38
Mister_MagisterThaodan: :P09:38
Thaodanjust that rinigus09:38
Thaodan[  +0,113955] usb 2-7: new high-speed USB device number 33 using xhci_hcd09:38
Thaodan[  +0,155893] usb 2-7: New USB device found, idVendor=0fce, idProduct=0dde, bcdDevice= 1.0009:38
Thaodan[  +0,000004] usb 2-7: New USB device strings: Mfr=1, Product=2, SerialNumber=309:38
Thaodan[  +0,000003] usb 2-7: Product: Android09:38
Thaodan[  +0,000002] usb 2-7: Manufacturer: Google09:38
Thaodan[  +0,000002] usb 2-7: SerialNumber: QV711BUJ1T09:38
Thaodan[  +2,804515] usb 2-7: USB disconnect, device number 3309:38
Mister_Magisterthat ain't sfos09:39
rinigusThaodan: I think it reboots as it cannot make rndis device. that's recovery, right?09:39
rinigusThaodan: try to make the kernel config closer to mine. for that also disable options as I did in my last config09:40
rinigus... it was only that combination that worked. and I did try quite a few09:41
Thaodanthat could be.09:42
rinigusThaodan: fingers crossed09:44
Thaodanuart without opening the device would be a blessing09:45
Thaodanbtw rinigus if you want to see my progress in porting kumano look at
Thaodancould be also for other devices. I found that easier than using .hadk.env and running everything by hand.09:47
rinigusThaodan: I will look  into it at some point. right now I am stuck on fastboot not recognizing a device at all. or only partially and refusing to flash09:50
rinigushappened after I rebooted my pc. will have to figure out tonight if its pc with some update, usb cable, or some connection between usb3->usb2 device.09:51
Thaodanthats strange that never happened to me. Only once on my notebook which made me reboot the bootloader after each flash.09:52
Thaodanbtw was was the limitation that the android 9 systemimage didn't build inside the ha sdk? My android 10 image builds fine inside it.09:53
rinigusThaodan: it was just unable to compile it. didn't look further into it09:55
T42Luka %lastname% was added by: Luka %lastname%10:26
T42<Luka %lastname%> hi10:26
T42<Luka %lastname%> I am porting to moto g510:26
T42<Luka %lastname%> but ending with10:27
T42<Luka %lastname%> error: Failed build dependencies:10:27
T42<Luka %lastname%> pkgconfig(android-headers) is needed by droid-config-cedric-1-1.armv7hl10:27
T42<Luka %lastname%> repomd-pattern-builder is needed by droid-config-cedric-1-1.armv7hl10:27
T42<Luka %lastname%> ssu-kickstart-configuration is needed by droid-config-cedric-1-1.armv7hl10:27
T42<Luka %lastname%> now only :10:30
T42<Luka %lastname%> repomd-pattern-builder is needed by droid-config-cedric-1-1.armv7hl10:30
T42<Luka %lastname%>         ssu-kickstart-configuration is needed by droid-config-cedric-1-1.armv7hl10:30
T42<Luka %lastname%> here is full output
T42hacker12455 was added by: hacker1245510:33
ThaodanPlease set a username Luka10:34
Thaodanor else communication from the IRC side is pita10:34
T42<Luka %lastname%> @Thaodan what do you mean i have it on tg10:39
Thaodanyou have no @Username set10:40
T42<hacker12455> Maybe something with the tg bridge10:40
T42<hacker12455> Hmm, is there a Matrix bridge?10:41
T42<Luka %lastname%> @luka set10:41
T42<Luka %lastname%> @Luka %lastname% [@luka set], like that?10:41
T42<Luka %lastname%> or just10:41
T42<NotKit> no, username in Telegram settings10:41
T42<Luka %lastname%> @Username set10:41
T42<Lukapanio> ouch sorry10:42
T42<Lukapanio> now done10:42
T42<Lukapanio> so why do i have that error10:43
T42<Lukapanio> ?10:43
vknecht@Lukapanio: first should check your environment (.hadk.env, .mersdk.profile, etc.), "MerSDK PlatformSDK bash-3.2$" as prompt looks fishy10:47
T42<elros34> what release you use for platform sdk and target? check full log (line 32 in your log)10:47
T42<Lukapanio> @vknecht what is bad with  "MerSDK PlatformSDK bash-3.2$"11:01
vknechtdepends on you .mersdk.profile and .hadk.env files, can you show them ?11:03
T42<Lukapanio> @vknecht
T42<Lukapanio> here you are11:06
vknechtok, PS1="MerSDK $PS1" explains it...11:08
vknechtso which release of platformsdk/targets did you install ? and which HADK pdf version are you following ?11:10
T42kiam001 was added by: kiam00111:11
vknechtand what did you do before (make hybris-hal in habuild etc.) ? and can you show your github repos ?11:11
T42<Lukapanio> @vknecht here are all repos
T42<Lukapanio> and i have done mka hybris-hal11:18
vknecht@Lukapanio: it doesn't look like you created your spec files using the templates like instructed in hadk pdf ?11:23
vknechteg. missing %define community_adaptation 1 in droid-config-cedric.spec11:24
T42<Lukapanio> used that guide11:36
T42<Lukapanio> @vknecht so i added that and i have
T42<Lukapanio> ouch i have * Building of community-adaptation finished successfully11:38
T42<Lukapanio> and still error: Failed build dependencies:11:38
T42<Lukapanio>         community-adaptation is needed by droid-config-cedric-1-1.armv7hl11:38
T42<Lukapanio>         repomd-pattern-builder is needed by droid-config-cedric-1-1.armv7hl11:38
T42<Lukapanio>         ssu-kickstart-configuration is needed by droid-config-cedric-1-1.armv7hl11:38
vknechtwell, unless really special needs, you should be using the pdf linked at
vknechtfrom the start...11:41
T42<Lukapanio> ok11:42
T42<Lukapanio> lets do all from the begining to the end11:42
deathmistjusa: so what about these changes to Android audio HAL configs and whatnot in Pie 9.0 and beyond? I don't have a "audio_policy.conf" anywhere on my phone so my old PA config won't work but found a possible replacement "audio_policy_configuration.xml" file
T42<Lukapanio> @vknecht what base is better android 7/8/9/8(treble)12:01
T42<Lukapanio> ?12:01
maldeathmist: that xml is used in sfos pa droid plugin12:02
vknechtgenerally, use the one that works best ; now, with 7 you should have less surprises as sfos porting go (personnal opinion) ;-)12:02
deathmistmal: oh ok, didn't know that. then what about my config file? audio_policy.conf is definitely gone for me on Pie12:04
vknechtnot sure treble is an option12:04
maldeathmist: audio_policy.conf is not needed is xml is used12:06
T42<Lukapanio> but android 7 is 32bit all other is aarch6412:06
T42<Lukapanio> and android 7 is not very stable12:06
vknechtwell, go with 8 then, with 9 it's still quite hardcore12:08
deathmistmal: so I should just remove the " config=/etc/audio_policy.conf" part entirely from my PA config?12:09
maldeathmist: do you have some custom changes to pa config?12:12
deathmistin /etc/audio_policy.conf I used to remove an entire dsd_compress_passthrough section as it caused PA service to fail on startup12:16
T42<Lukapanio> so i must include all partition in fix mountpoint12:17
T42<Lukapanio> or it will be ok to use same as halium?12:17
deathmist@Lukapanio you definitely need your most important ones at least like /system /vendor /persist /firmware etc there, I just put everything there so I wouldn't have problems12:20
T42<Lukapanio> hmmmm there is no vendor12:21
T42<Lukapanio> as phone was released with android 712:22
deathmistthat's because you probably aren't doing a treble port (Android 8+)12:22
T42<Lukapanio> there is treble tree with dirty trick12:23
deathmistif you don't have many partitions just add all of them, I had 70 so I made a quick script that did it for me12:23
T42<Lukapanio> using oem as vendor12:23
T42<Lukapanio> 54 here12:23
jusadeathmist: you could also report to pulseaudio-modules-droid github project if there are crashes.. but yeah, you can pass xml to config variable as well, or omit the variable and the module will try to open xml first and legacy config after12:29
deathmistok, I'll be doing more testing soon! :)12:29
maldeathmist: but you said you don't have audio_policy.conf but then you said you removed something from it?12:30
deathmistmal: that was back in 15.1 where I also had the file in /system iirc12:31
T42<Lukapanio> @deathmist
T42<Lukapanio> here is fixup for halium12:32
T42<Lukapanio> can i use that12:32
T42<Lukapanio> or better idea is to add all partiton12:32
maldeathmist: does it help if you remove the custom config?12:32
T42<Lukapanio> partitons*12:39
T42<Lukapanio> partitions*12:39
maldeathmist: so what pa config do you have now?12:41
deathmistmal: same as I linked but I now removed "config=" arg entirely, still no audio tho. journal should I also get full pulseaudio logs with increased verbosity?12:46
maldeathmist: are you sure the android side audio subsystem is correctly initialized?12:48
deathmistmal: damn doesn't actually look like it, logcat
maldeathmist: that doesn't show any audio related messages, is that complete log after reboot?12:56
deathmistmal: yeah that seems to be all /usr/libexec/droid-hybris/system/bin/logcat gives me after a reboot13:00
deathmistI do however see "load: module=/vendor/lib/hw/" and "dlopen failed:" right under it, looks like I have some permission issue with cutils-trace because that is present there a few times13:01
maldeathmist: some vendor blob mismatch?13:02
deathmistcould that be caused by the kernel? I kinda doubt it as mostly everything else works flawlessly including display, vibration, rotation, proximity, ... old kernel was forked directly from LOS while the current one I've got booted also has linux-stable merged in
malno idea, could be13:13
deathmistisn't the produced boot.img in out/ just the regular LineageOS one? I could try to boot into Android with that and verify that audio works13:14
jusadeathmist: Invalid argument when trying to open hw module is likely permission/initialization issue13:21
maljusa: check the logcat fails to load module13:22
deathmistaudio works fine on Android, at least using the regular LOS 4.4.153 kernel instead of my 4.4.197. will test that older kernel with SFOS too asap with my new defconfig changes13:36
T42<adampigg> mal: my internal turmoil over ; persists.....13:43
T42<adampigg> poll: ; yay or nay?13:43
T42<adampigg> :D13:44
T42<Lukapanio> sb2 -t $VENDOR-$DEVICE-$PORT_ARCH gcc main.c -o test14:14
T42<Lukapanio> sb2: Error: Tools directory '/srv/mer/toolings/motorola-cedric' does not exist.14:14
deathmist@Lukapanio you probably made your sb2 targets wrong. what does "sdk-assistant list" show? I get "SailfishOS-" & "`-oneplus-cheeseburger-armv7hl"14:17
T42<Lukapanio> PlatformSDK bash-3.2$ sdk-assistant list14:18
T42<Lukapanio> SailfishOS-latest14:18
T42<Lukapanio> motorola-cedric-armv7hl14:18
T42<Lukapanio> PlatformSDK bash-3.2$14:18
T42<Lukapanio> why is it armv7hl here?14:18
T42<Lukapanio> and motorola-cedric there?14:18
T42<elros34> it should be: SailfishOS-3.1.014:23
T42<elros34> ├─$VENDOR-$DEVICE-armv7hl14:23
T42<elros34> first is tooling, second target14:23
deathmist@Lukapanio if you pasted exactly how it output then you made both as toolings instead of "SailfishOS-" as tooling and "$VENDOR-$DEVICE-$PORT_ARCH" as target under that tooling. remove both using sdk-assistant (see "sdk-assistant --help"), do "export RELEASE=" and follow for a bit (my own build guide)14:23
T42<Lukapanio> ok thanks14:25
deathmistkeep in mind you can't just blindly follow my build guide either, a bunch of it it for the OnePlus 5 devices & 15.1 specifically and some commands may not work as I have modified SDK environment profile files etc14:33
T42<Lukapanio> i must add droid_target_aarch64 114:42
T42<Lukapanio> ?14:42
deathmist@Lukapanio if your phone SOC is 64-bit capable (armv8 etc) then yes14:45
T42<Lukapanio> yepit is 6414:45
deathmistwhich base were you doing again? you'll prob need this too in your spec if you have a QCOM device to get proper graphics
T42<Lukapanio> my base is los15.114:49
T42<Lukapanio> thanks added14:49
deathmisthmm ok, have you done yet then?14:52
T42<Lukapanio> ouch only first14:53
deathmistno problem, good to catch these things early to avoid needing to rebuild a new image constantly :)14:54
T42<Lukapanio> thanks will add now14:55
deathmistI think you'd want to do as many steps from HADK PDF as possible before even booting for the first time (e.g. droidmedia etc) so graphics will work when you eventually get to that point14:58
deathmisttho it's also nice to see if it even boots to anything, I think actually don't follow everything yet14:59
T42<Lukapanio> lol it wount work from first time15:00
deathmistI know lol just realized, no need to waste time early as we'll prob be swapping around defconfig options and whatnot15:00
T42<Lukapanio> once i had rebuilt kernel 20 times before got postmarketos booting on galaxy tab 4 7.015:01
T42<Lukapanio> but with cedric libhybris is working, haliumworkes not bad15:02
deathmistthe way my phone boots hybris properly is when enable SELinux options such as SECURITY_SELINUX_BOOTPARAM mentioned by kernel checker and do "selinux=0" on boot KERNEL_CMDLINE in some device/**/BoardConfig*.mk file, haven't tried anything else as this has worked till 16.0 too now15:04
T42<Lukapanio> i alredy added selinux=0 audit=015:05
T42<Lukapanio> error: Failed build dependencies:15:18
T42<Lukapanio>         ofono-devel >= 1.21+git57 is needed by ofono-ril-binder-plugin-1.0.10-1.armv7hl15:18
T42<Lukapanio>         pkgconfig(libgbinder-radio) >= 1.0.8 is needed by ofono-ril-binder-plugin-1.0.10-1.armv7hl15:18
T42<Lukapanio>         pkgconfig(libgrilio) >= 1.0.35 is needed by ofono-ril-binder-plugin-1.0.10-1.armv7hl15:18
deathmistcd hybris/mw/ofono-ril-binder-plugin && git reset --hard 1.0.6; cd -15:21
deathmist you can use this for example to track down the more common issues15:21
T42<Lukapanio> thanks15:22
deathmistprob too late now but nfcd-binder-plugin will fail too, that one you'll downgrade to 1.0.415:24
T42<Lukapanio> damn15:24
T42<Lukapanio> i will have problems with unneeded garbage15:24
T42<Lukapanio> as i have no nfc at all15:25
deathmistit will build it as that's a common component in most newer devices, you could fork the submodule and remove the entire build part in build_packages but that's effort to maintain. it won't actually install it on the device unless you specify so in patterns15:26
vknechtpiggz, those « ; » can die, there should be a space before them anyway ;-)15:57
deathmistmal: I may have found a clue about what qti / dataqti is and does: "#Create QTI dir for logs" is some init files and it creates a "/data/vendor/dataqti" dir at least on Pie, imo it's safe to disable16:00
deathmistespecially when it's just being started and dying every 10s on SFOS :p16:01
vknechtwhat's the stance on instanciating a qmlRegister'ed Type and calling view->rootContext()->setContextProperty() on them in main cpp file ? makes them available everywhere without needing to declare Them{} in qml, right ?16:02
vknechtif so, anything against doing this for FSOperations ?16:02
T42<adampigg> vknecht: does that solve your signal problem? i cant understand why? i think hthe current way is preferred than creating a contextProperty for a qobject16:04
vknechtwell, I've ditched signaling16:05
vknechtlemme see where I'm at, and I'll push16:06
Thaodanvknecht: creating the object inside C++ vs. just registering the type has the benifit that you can pass additional arguments from the C++ side to the object while creating it. For example if you have a model for search data you can pass a reference to an object cache.16:08
vknechtpiggz, "static StorageModel *m_model;" to the rescue ;-)
vknechtif FSOperation had only one instance, guess could ditch this trick, or at least initialize a non-static one in constructor16:10
vknechtThaodan, ok16:12
vknechtalready did it for the storage model, no point having more than one, and should then be accessible equally from cameraui, settings overlay (everywhere we need writablelocation, in fact)16:19
vknechtat least now, the list gets refreshed ; still have to iron some detail so that correct entry gets selected and setting saved16:23
vknechtnot sure if it's my usb key, or device' usb low power mode, but the key tends to disappear few seconds after mounting its partitions and running pov16:25
vknechtso it complicates testing a bit16:25
T42<Lukapanio> Error <creator>[10/28 16:19:38] : Unable to find pattern: Jolla Configuration cedric16:28
Thaodanvknecht:  you could look at the usb sysfs power interface: /sys/bus/usb/devices/$id/$id/power/control16:30
Thaodan@Lukapanio: ./hybris/droid-configs/droid-configs-device/helpers/process_patterns.sh16:31
Thaodanmal: thanks I've seen the commit16:36
vknechtThaodan, thanks, I see it's on "auto", will have to dig kernel/doc for possible value16:44
ThaodanSee here16:45
Thaodanyou change it to on  to disable suspend16:46
T42<elros34> could somebody run: "udevadm monitor -k -s cpu", wait for a while or lock/unlock screen so kernel event will be generated. My device generates /devices/system/cpu/cpu1/cpufreq but I suspect it's not common and should be /devices/system/cpu/cpu116:50
vknechtthanks, still having that so must be the key or the connector...16:53
vknechtcrap, have to update my X to 3.1 now there's keepalive in pov16:56
vknechttempted to use graphical updater and watch it crash XD16:58
T42<elros34> thx17:05
deathmist@elros34 I got same results but with cpus 1-7 instead ofc :p17:05
vknechtwell, it's an 8 cores, but maybe there a problem17:06
T42<elros34> ,pu0 is usually always enable, thanks so it's my strange device17:06
T42<elros34> cpu0*17:07
vknechtin aida64 also I see only those 2 suspending17:08
vknechtlatest submodules ok/recommended for 3.1 ?17:21
T42<Lukapanio> Error <creator>[10/28 16:45:26] : URLGrabber error: - e.errno: 1417:24
vknechtseem to be missing environment, like RELEASE env var17:25
T42<Lukapanio> ouch17:26
T42<Lukapanio> yep17:26
T42<Lukapanio> repo problem: nothing provides pulseaudio-module-keepalive >= 1.0.0 needed by pulseaudio-modules-droid-12.2.81-1.armv7hl,17:29
T42<Lukapanio> may be bad release
T42<Lukapanio> ?17:29
T42<Lukapanio> so what version is lastest?17:33
T42<elros34> correct  is
T42<Lukapanio> same with 1217:39
T42<Lukapanio> but it is there17:42
T42<Lukapanio> PlatformSDK bash-3.2$ cat /var/tmp/mic/cache/adaptation-community-common-cedric- | grep keepalive17:42
T42<Lukapanio>       <rpm:entry name="pulseaudio-module-keepalive" flags="GE" epoch="0" ver="1.0.0"/>17:42
T42<Lukapanio> PlatformSDK bash-3.2$17:42
T42<Lukapanio> damn17:42
T42<Lukapanio> why cant it fond that17:43
T42<Lukapanio> @elros34 [correct  is], thanks but not helped17:44
T42<elros34> keepalive was splited recently so try to downgrade pulseaudio-modules-droid or build
T42<Lukapanio> ouch17:47
T42<elros34> hmm, it's already built in common repo so check you repositories in ks file17:50
T42<Bob_Da_Builder> Hello, getting this error when trying to build for a3xelte:18:01
T42<Bob_Da_Builder> ninja: error: '/home/quantumdynamo/ANDROID_ROOT/out/host/linux-x86/obj/STATIC_LIBRARIES/libfdt_intermediates/export_includes', needed by '/home/quantumdynamo/ANDROID_ROOT/out/host/linux-x86/obj/EXECUTABLES/dtbhtoolExynos_intermediates/import_includes', missing and no known rule to make it18:01
T42<Bob_Da_Builder> 18:00:01 ninja failed with: exit status 118:01
*** electro is now known as Guest2475618:04
rinigusevening! I simplified my config with dropping few things in sparse. xz2 (akari) still doesn't boot to gui. corresponding logs18:05
rinigusdmesg with df:
riniguslogcat with complains regarding absent opengl implementation:
rinigusls -lR of /dev:
riniguson user side - it boots to sony logo, lightens up white led , shuts it down after some time. on power press - led goes red and its shutting down nicely18:06
T42<elros34> rinigus: "SELinux:  Starting in permissive mode" is that expected because of some hybris16 specific patches or you just forgot to disable selinux?18:08
riniguselros34: I kept selinux, as it is for xperia 10. is that the place to disable?18:09
rinigus(met "something to disable")18:09
T42<elros34> Usually it must be: SELinux: disabled at boot or something similar but I remember mal mentions some changes which requires selinux partialy enabled18:11
malrinigus: not sure if selinux yet works in current release and even if it does then it would need some configs and correct patch to system/core also18:12
riniguselros34: as far as I understood, I cannot just specify selinux=0 on boot cmd line. if I wish to do so, I will have to drop corresponding patches in hybris18:12
T42<Bob_Da_Builder> (Document)
deathmistmal: couldn't "external/libhybris" be added as an entry in hybris >=15.1 branch manifests? what were the issues people are experiencing as I've not seen anything wrong since I've added it to my local manifests and it has worked always18:14
T42<Lukapanio> Error <creator>[10/28 18:14:09] : Unable to find pattern: Jolla Configuration cedric18:14
T42<Lukapanio> again18:14
T42<Lukapanio> hybris/droid-configs/droid-configs-device/helpers/process_patterns.sh18:14
T42<Lukapanio> not helping18:14
deathmist@Lukapanio: run build_packages -c (ofc replace with full path to scripts)18:14
rinigusmal: my config is at ; its the same as for xperia10, as far as I undestand. or would you expect/suspect to have it different for xz2?18:15
rinigusmal: would you recommend to disable selinux, for now at least (with going through hybris patches to do that properly)?18:16
vknechtis this trick to always start sensors still needed ?
malvknecht: usually no afaik, it sensor service is not running then maybe18:17
T42<Lukapanio> build_packages -c and process_patterns.sh18:19
T42<Lukapanio> not helpes18:19
deathmist@Lukapanio regenerate your kickstart file by following hadk section 8.3 and attempt mic build again18:24
T42<elros34> rinigus: "Failed to find firmware cdsp.mdt" maybe you how some partition not mounted18:29
riniguselros34: /odm is absent and whole /config does not have usb gadget initialized. dev-mtp.mount fails18:31
riniguselros34: mount: /dev/mtp: unknown filesystem type 'functionfs'. I suspect /config is an issue with it18:32
deathmistfor /odm you'll probably need a odm.mount unit like I have done with my /vendor partition18:32
T42<elros34> I don't thinks mtp via functionfs has anything to it18:33
rinigusI had to disable one setting in kernel config to get rndis working - maybe should try with it18:33
rinigusdeathmist: let me check, I think I have one18:38
deathmistby absent do you mean the entire directory is not there or?18:39
rinigusdeathmist: "Dependency failed for Droid mount for /odm." in journal. will have to see which dependency is that18:39
T42<Lukapanio> still same error18:41
T42<elros34> You are missing some package. Follow hadk 8.5, there is instruction what to do: Substitute the line @Jolla Configuration @DEVICE@ with @jolla-hw-adaptation-@DEVICE@18:43
T42<elros34> in your .ks18:43
T42<Lukapanio> yep thanks finaly repo problem: nothing provides pulseaudio-module-keepalive >= 1.0.0 needed by pulseaudio-modules-droid-12.2.81-1.armv7hl is back18:47
T42<elros34> show your ks file18:48
malthat should be in common repo18:48
malrinigus: did fixup fail for odm?18:49
T42<Lukapanio> @elros34 [show your ks file],
rinigusdeathmist: it doesn't mount it => should be no tree under /odm. let me see18:49
malrinigus: on x10 /odm is from oem partition18:50
T42<elros34> @Lukapanio you are missing common repo18:50
rinigusmal: booting to recovery, to check what's odm mount is. yes, I know its oem - same for xz218:51
malrinigus: what was the mount error?18:51
malrinigus: can you show the mount service file18:51
rinigusjust a sec18:52
T42<elros34> @Lukapanio do you have adaptation-community-common-$DEVICE-@RELEASE@ repo in droid-config-$DEVICEssu-kickstarts-1-1.armv7hl?18:53
rinigusmal and elros34: mount service file:
T42<elros34> as you can see fixup fails  What=/dev/block/bootdevice/by-name/oem18:55
riniguselros34: ok, so I have to add block/bootdevice/by-name/oem somehow to fixup?18:56
riniguscurrent fixup is at
riniguslooks like I have to add just oem (without _a or _b). do I understand it correctly?18:58
T42<elros34> yes you need to choose active slot18:59
riniguselros34: I can see that this was done for xperia 10. well, sounds easy to do it :) . thank you and thanks to mal!18:59
T42<elros34> so /dev/block/bootdevice/by-name/oem will be replaced with /dev/sda40 or sda5018:59
T42<elros34> you can do changes in device first19:00
riniguselros34: yes, that's true. device first then...19:02
malrinigus: interesting that you have oem_a and oem_b but fstab has oem19:04
rinigusmal: I think its an expected one as active slot is set for os, isn't it?19:04
T42<Lukapanio> @elros34 [@Lukapanio do you have adaptation-community-co …], sorry for stupid question, but where is it located, can find  only its rpm19:07
T42<elros34> just unpack rpm19:09
T42<Lukapanio> i have repo --name=adaptation-community-common-cedric-@RELEASE@ --baseurl=
malrinigus: maybe19:14
T42<elros34> @Lukapanio so you probably made mistake in 8.3 because it should be also in your generated ks19:14
T42<elros34> don't mixup with adaptation-community-cedric19:16
mal@elros34 does this now work correctly?19:16
rinigusmal and elros34: no gui, but way more interesting logs. journal and logcat coming up in few minutes19:18
T42<elros34> mal: my device generates /devices/system/cpu/cpu1/cpufreq instead /devices/system/cpu/cpu1  but based on logs from vknecht it's not normal behavior. Now it it only adds TEST so shouldn't break anything just suppress spam if path doesn't exist19:19
mal@elros34 ok, I'll need to test on some devices19:22
T42<Lukapanio> thanks @elros34  i passed that19:23
rinigusjournal: ; logcat: ; df -h:
rinigusmal & elros34: pasted logs, haven't had a chance to look through them yet19:33
rinigusat least opengl errors disappeared19:34
malrinigus: related to usb usb_moded[3272]: /config/usb_gadget/g1/functions/rndis_bam.rndis: mkdir failed: No such file or directory19:38
rinigusmal: I think that message was even on stock - or something similar. I think its coming from upstream/upstream android and on sony implementation they use gsi.rndis version. at least that was guessing from reading init scripts in aosp19:40
malrinigus: does test_hwcomposer fail19:40
rinigusmal: how do I run it?19:41
rinigus... I don't have ssh into device19:41
malhmm, good point19:41
malrinigus: you know you can configure usb-moded to use correct paths19:41
rinigusmal: you mean for config of rndis on sailfish boot proper?19:42
rinigus ... sorry, have to go now for 10-20 mins19:43
malrinigus: yes, usb-moded has possibility set those paths etc like replacing rndis_bam.rndis with something else19:43
malrinigus: what rndis options do you have enabled in kernel19:46
malrinigus: and a bit later the rndis.usb0 stuff, could those be something you need?19:46
malrinigus: where is that df -h from?19:47
rinigusmal: I will look into it. I have a set of options for kernel that works with hybris-recovery for rndis. so, something like that should work then for usb-moded, I guess19:54
rinigusmal: logs and df -h are from a small logs script which I run using extra service on boot and shutdown. later I get those by booting into recovery19:54
malrinigus: do you have any logs from recovery which would show what paths it has19:55
rinigusmal: yeah, I am logged in into it. its using usb_gadget/g1/functions/rndis.usb019:57
malrinigus: so add the config mentioned in my link19:57
malthe filename is one line above where my link points to19:58
deathmistmal: I tried the audio fix piggz also used from 4 months ago: "ln -s /system/lib/vndk-28/ /usr/libexec/droid-hybris/system/lib" as I have the same issue, I definitely get different results now like lipstick takes forever to load and logs look more interesting too: logcat journal
malrinigus: also check whether there is c.1 or b.1 folder somewhere under  usb_gadget/20:00
deathmiststill no audio tho sadly just to clarify20:01
rinigusmal: just ini with `gadget_conf_directory = configs/c.1` ? anything else?20:01
rinigus+ [configfs] as a header20:01
rinigusmal: full path on recovery is /config/usb_gadget/g1/configs/c.1/ where rndis.usb0 is added as symlink20:02
malrinigus: ok, config file needs to have 3 lines20:02
malgadget_conf_directory = configs/c.120:02
malfunction_rndis        = rndis.usb020:03
rinigusmal: thank you!20:03
malhopefully that works20:03
malI think deathmist uses the same config20:03
malalthough I don't see that in his configs20:05
deathmistI attempted but it didn't work. my device uses b.1 as config dir on all init files etc so I'll attempt again right now probably, will need to edit /init-debug to as that uses c.1 by default, probably relevant defconfig options:
rinigusmal and deathmist: nope, didn't work. as recovery works with c.1, I think that should be ok. I have to go now for some time, will probably pick up logs later tonight and continue tomorrow.20:10
maltry b.120:14
malrecovery could do things differently20:14
malanyway that kind of config could help, unless something else is missing20:14
maldeathmist: there are some audio related kernel errors in your logs20:16
deathmistmal: this is on the older 153 revision 4.4 kernel now and ofc all works fine on LineageOS side, what could be causing those then? here's also my disabled_services.rc as of now
maldeathmist: do sensors work?20:30
deathmistmal: als & rotation work fine, how should I test the others again?20:32
deathmisttest_sensors seems to list 51 sensors at least20:33
piggzmal: rinigus: whats that configfs stuff for? i think pro1 has hose folders20:39
deathmistholy shit I got MTP working now :DDD20:40
deathmistI used usb-moded-configfs.ini with c.1 path and "it just worked"20:40
piggzyes, we have same errors20:42
deathmistoh I wonder if it was some init scripts that were messing everything up, they were importing some more vendor init usb and configfs rc file and I commented them out now & plan to use some overlay / bind mount solution to use them cleanly20:47
maldeathmist: ok, then sensors are fine20:47
malpiggz: some device use different configfs stuff and might need configuration like that b.1 vc c.1 and rndis name20:48
piggzmal: will i need to change anything for this
piggzOct 28 20:53:52 Sailfish usb_moded[6797]: /config/usb_gadget/g1/functions/rndis.usb0/ethaddr: can't open for writing: No such file or directory20:55
piggzOct 28 20:53:52 Sailfish usb_moded[6797]: /config/usb_gadget/g1/functions/rndis.usb0/wceis: can't open for writing: No such file or directory20:55
malpiggz: do you have any config now?21:03
piggzmal: i just made one based on what you told rinigus21:03
malpiggz: did it change anything? what is broken in your device, all of usb or just some part of it?21:04
piggzmal: not sure if anything changed ..... i have problems getting an address on dhcp for the PC21:07
*** electro is now known as Guest7048621:09
malpiggz: you can see the host_addr and dev_addr in the paths21:09
malmaybe try using one of those, maybe host_addr?21:10
malhmm, those don't seem to be configurable21:12
piggzmal: so, first thing, usb-moded tried to open b.1, not the c.1 i have, so some config is needed21:20
piggzOct 28 21:16:58 Sailfish usb_moded[6878]: /config/usb_gadget/g1/configs/b.1: opendir failed: No such file or directory21:20
malpiggz: what kind of ucb changes did you do to kernel defconfig21:21
piggzmal: none im aware of21:21
malhmm, ok21:22
malpiggz: is that for fxtec?21:22
piggzmal: yes21:22
malpiggz: wondering if you should try replacing USB_CONFIGFS_RNDIS with USB_CONFIGFS_QCRNDIS, although maybe your device is meant to use the normal one21:25
piggzmal: do i need # CONFIG_USB_CONFIGFS_RMNET_BAM is not set21:35
ThaodanMine also uses b121:35
rinigusmal: and switching to b.1 made a trick. I can see rndis_host loaded on my pc dmesg21:35
Thaodanat least on stock aosp21:35
Thaodanrinigus: Do you want to know a funny thing?21:36
rinigus(b.1 was populated on boot to the full os)21:36
rinigusThaodan: sure21:36
Thaodanrinigus: my device is so limited on aosp that there's no gui so that loop maybe something that was intended21:37
malrinigus: nice21:38
rinigusThaodan: so its perfect paperweight. and, as wine, will get better21:38
rinigus... with time21:38
mallooks like b.1 is quite common21:38
rinigusmal: any idea what would be ip address used by regular sfos?21:39
malrinigus: b.1 is the default, maybe the other things is just needed, rndis thingy21:39
malrinigus: is the default21:39
Thaodandmesg output from stock aosp21:40
rinigusmal: let's see21:40
Thaodanrinigus: looks like that. I thought that maybe I can ignore that as SailfishOs doesn't use much of the infrastructure of aosp.21:41
rinigusmal: works nicely. only ... ... don't have nemo's password :)21:41
malrinigus: telnet is not working?21:41
ThaodanI bet I have more setup for Sailfish OS than I needed currently.21:42
rinigusconnection refused21:42
malrinigus: hmm, but that was using the official way which doesn't have telnet21:42
rinigusmal: I can just login into recovery and add ssh keys21:43
rinigusshould work21:43
malprobably yes21:43
vknechton regular systems, we could set passwords in .ks file, maybe that could be done here too ?21:44
rinigusmal & vknecht: in the end, booting to recovery, chroot to rootfs and passwd for root and nemo worked the best. ssh authorized failed somehow (probably made a typo somewhere).21:51
rinigusin short, I can log in21:51
rinigusmal: EGL_PLATFORM=hwcomposer test_hwcomposer -> library "" not found21:55
rinigus(no hits on piggz grep yet)21:55
vknechtlooks like proprietary QC file... try linking it to /dev/null ? ;-)21:59
vknechtor maybe it's somewhere on device in some partition22:01
Thaodanis it in vendor or oem?22:01
malrinigus: don't use nemo as password, at least in the past if password was nemo it is reset during boot because it's unsafe22:06
rinigusno - its not nemo :)22:07
malrinigus: I used to be lazy and used nemo as password and assume it was feature that all passwords are reset, it was just that :D22:07
rinigusgetting late over here. I'll continue tomorrow - thanks for help!22:08
Thaodanbb have a good night22:09
rinigusand just before shutting down all: strace from hwcomposer and hwc2
rinigusboth were run as root and EGL_PLATFORM=hwcomposer22:12

Generated by 2.17.1 by Marius Gedminas - find it at!