T42 | <duhjokersinhell> i would like to port sailfish os to pixel 3a am i in the right place | 00:17 |
---|---|---|
T42 | <duhjokersinhell> how do i start the repo | 00:24 |
mal | which android base does it have? | 01:04 |
T42 | <duhjokersinhell> android 9 pie | 02:03 |
T42 | <XAP2P> who knows about xiaomeme ginkgo/willow port? | 03:02 |
T42 | <birdzhang> @XAP2P i'm doing that, something not working yet, like sound and camera | 03:04 |
T42 | <duhjokersinhell> where can i download the HADK? im reading the pdf but cannot find the link just instructions on setting it up | 03:43 |
mighty17[m] | https://sailfishos.org/content/uploads/2021/06/SailfishOS-HardwareAdaptationDevelopmentKit-4.1.0.0.pdf | 04:31 |
rinigus | @duhjokersinhell: you are probably looking for https://sailfishos.org/wiki/Platform_SDK | 05:48 |
rinigus | piggz: encryption repos moved to sailfishos-open. | 05:48 |
rinigus | piggz: looks like we are missing https://github.com/piggz/sailfish-fpd-community-test | 05:48 |
T42 | <duhjokersinhell> ok ty. is there an updated build guide | 05:49 |
rinigus | for aosp9 based? checkout FAQ as well (see this room header for links) | 05:59 |
T42 | <duhjokersinhell> ok i have been reading the pdf and i am lost. | 06:04 |
T42 | <duhjokersinhell> does the platform sdk replace the hadk? | 06:04 |
rinigus | @duhjokersinhell: no, it doesn't. read hadk and work according to it. when getting lost, ask. but check faq just in case before | 06:05 |
rinigus | as written last night - /home decrypt scripts have now hwcrypt support as well. https://github.com/sailfishos-open/sailfish-device-encryption-community | 06:07 |
rinigus | early days still, have to adjust decrypt script to be able to respond to hw-backed and recovery passwords in parallel | 06:08 |
rinigus | that way user can choose which password to enter... | 06:08 |
rinigus | also scrypt bits are missing for password processing. but most critical part - interaction with android hw key handling - has been checked and worked during a boot as well. | 06:09 |
T42 | <duhjokersinhell> oh i think i am starting to understand the hadk is the pdf? | 06:09 |
T42 | <pherjung> yes | 06:11 |
T42 | <duhjokersinhell> ok ty. lol i was confused thinking it was software | 06:12 |
T42 | <elros34> @duhjokersinhell to save you from issues at the beginning from outdated docs read also this: collabedit.com/kr9xx | 06:54 |
T42 | <adampigg> rinigus: i need -test on volla as its not working :/ | 07:04 |
T42 | <adampigg> Ironic that it was the fist phone to supporrt ot | 07:05 |
rinigus | @adampigg: I preinstall sailfish-fpd-community-test as GUI for FP setup. it works better than the official one on tama | 07:06 |
T42 | <adampigg> heh | 07:06 |
rinigus | but maybe we should move that repo also to -open | 07:06 |
T42 | <adampigg> something crashes android side on new aarch64 port | 07:07 |
T42 | <adampigg> sure, will move when i get to laptop....also needs fixing, systemd unit is still in /lib | 07:08 |
rinigus | @adampigg: ouch. for tama it works well... | 07:08 |
rinigus | but sony have their own oss fpd implementation on aosp. | 07:08 |
rinigus | re unit in /lib - we don't have any in fpd-test, right? | 07:09 |
T42 | <adampigg> no, i mean the main pkg | 07:10 |
rinigus | and the one in fpd proper should be in /usr/lib/systemd/system - as it is expected to be for aarch64 as well | 07:10 |
T42 | <adampigg> unless i installed an old version last night...could have | 07:11 |
rinigus | on tama it is in correct spot. must be old version | 07:12 |
piggz | rinigus: mal: seen this before? | 07:40 |
piggz | [D] unknown:0 - bool makePath(const char*, const char*, const char*, QFileDevice::Permissions) /var/lib/sailfish-fpd-community/100000 QFlags(0x1000|0x2000|0x4000) | 07:40 |
piggz | [D] unknown:0 - void AndroidFP::setGroup(uint32_t, const QString&) 0 "/data/vendor_de/100000/fpdata" | 07:40 |
piggz | Aborted (core dumped) | 07:40 |
rinigus | piggz: no, unfortunately | 08:10 |
rinigus | piggz: but I have it under `/data/system/users/100000/fpdata` | 08:18 |
piggz | rinigus: have you altered fpd-community for that path? | 08:22 |
rinigus | piggz: not that I remember. let me check | 08:23 |
piggz | rinigus: pretty sure the one i mentioned is hard coded | 08:23 |
rinigus | piggz: nope, it pulls sources from old repo (piggz) | 08:24 |
rinigus | so, it should be grabbing path from android side | 08:25 |
rinigus | piggz: it is hard coded but has one if statement | 08:26 |
rinigus | https://github.com/sailfishos-open/sailfish-fpd-community/blob/master/src/androidfp.cpp#L65 | 08:27 |
rinigus | piggz: ^ | 08:27 |
piggz | rinigus: yeah, same issue as before, looking at old source! | 08:27 |
rinigus | piggz: but I presume it was working for you before with vendor_de path. and for tama I had to use the other one | 08:28 |
piggz | rinigus: yeah, it worked previously, but that was android 9/armv7 | 08:31 |
piggz | ive now rebased on los17 aarch64 | 08:31 |
rinigus | for me it stayed put, but maybe it did change for you. | 08:32 |
rinigus | flashing los17 and checking where it stores fingerprints? | 08:33 |
rinigus | piggz: ^ | 08:33 |
piggz | rinigus: looks like it dies in setActiveGroup https://paste.mozilla.org/HacsKBAy | 08:37 |
rinigus | piggz: some null pointer in there. no idea what is it doing. is fingerprint vendor blob or some oss implementation? | 08:42 |
piggz | i presume vendor blob | 08:43 |
piggz | just debugging why its choosing that path | 08:43 |
rinigus | for sony we have sources. | 08:43 |
rinigus | piggz: but maybe it would make sense to check where fp are stored in los. just in case it is a simple fix | 08:44 |
piggz | sure | 08:44 |
rinigus | although I have no idea how frequently android moves it all around | 08:44 |
piggz | rinigus: plan b ... ask erfan and notkit if fp works for them on UT :D | 08:47 |
rinigus | piggz: great plan! assuming that they moved to android 10 | 08:47 |
rinigus | piggz: actually, why did you move to android 10? is there something better on volla with it? | 08:48 |
piggz | rinigus: i know they are on halium 10 base | 08:48 |
piggz | rinigus: yeah, its loads more stable, plus, if i was rebasing to aarch64 anyway, made sense | 08:48 |
piggz | for some unknown reason, kernel was leaking on 9, causing OOM every few days | 08:49 |
rinigus | piggz: I see - good reasons. | 08:50 |
T42 | <wkming> hello everyone , how to get root permission on salifish os ? | 09:08 |
rinigus | @wkming: in settings/development set password and in terminal use it when asked by `devel-su` | 09:17 |
T42 | <XAP2P> oh :D (re @birdzhang: @XAP2P i'm doing tha...) | 09:49 |
T42 | <wkming> thank you (re @SailfishFreenodeIRCBridgeBot: <rinigus>@wkming: in...) | 09:59 |
piggz | rinigus: well, learned something new | 10:34 |
piggz | i didnt know logcat now filtered messages based on user ... i was running logcat wondering why i wasnt seeing messages, runs as root and there they are! | 10:34 |
mighty17[m] | `/home/mighty/hadk/hybris/mw/libhybris/libhybris/hybris/tests/test_common.cpp:142: undefined reference to `hwc2_compat_device_on_hotplug'` any ideas why it breaks? | 10:52 |
mighty17[m] | i think im supposed to build libhybris with the `rpm/dhd/helpers/build_packages.sh` as external/libhybris was missing when i cloned it with repo | 10:55 |
mighty17[m] | oh nvm found it in hadk-faq | 10:58 |
mighty17[m] | `error: Failed build dependencies: pkgconfig(droidmedia) is needed by gstreamer1.0-droid-0.20210820.0-1.armv7hl` | 12:53 |
mighty17[m] | how do i fix this? | 12:54 |
mighty17[m] | same thing happens for gmp droid | 13:10 |
piggz | mighty17[m]: build/package droidmedia | 13:14 |
T42 | <edp_17> mighty17: I was told to use these commits: droidmedia: 9450a66a133ffa4df4af5f1d2af5b09ff794606d | 13:18 |
T42 | <edp_17> gst-droid: 364c13e3f6b3d99a87b58beb526bd13f377ce6f4 | 13:18 |
T42 | <edp_17> gmp-droid: b61b740987a9880fccf5401220644c7dcf184be2 | 13:18 |
T42 | <edp_17> For me all worked fine. | 13:18 |
mighty17[m] | building droidmedia failed, I'll give that commit a go | 13:18 |
mighty17[m] | https://paste.debian.net/1209011/ piggz | 13:20 |
piggz | mighty17[m]: you are not following the hadk | 13:21 |
mighty17[m] | im still on hadk 3.4 due to testing out new kernel and stuff before updating to 4.x | 13:21 |
piggz | droidmedia is built in the ubuntu sdk (for android) | 13:22 |
piggz | then, you package all with build_packages.sh -gg | 13:22 |
mighty17[m] | 3.4's hadk doesnt tell me to build droidmedia* | 13:22 |
mighty17[m] | oh derp | 13:23 |
mighty17[m] | `make -j$(nproc --all) hybris-hal droidmedia` was already done before | 13:23 |
mighty17[m] | wtf its rebuilding now?? im sorry something must be rly broken on my machine | 13:29 |
mighty17[m] | `Problem: geoclue-provider-hybris-hal-0.2.31-1.armv7hl conflicts with geoclue-provider-hybris-binder provided by geoclue-provider-hybris-binder-0.2.31-1.armv7hl` ?? | 14:48 |
mighty17[m] | @pig | 15:51 |
mighty17[m] | * piggz: yup built droidmedia acc to hadk still the same issue | 15:51 |
mighty17[m] | https://irclogs.sailfishos.org/logs/%23sailfishos-porters/%23sailfishos-porters.2021-03-27.log.html i had to follow steps from here for gst-droid | 15:51 |
mighty17[m] | but no clues about gmp-droid | 15:51 |
mighty17[m] | https://paste.debian.net/1209032/ | 15:55 |
T42 | <elros34> if you are building for old sfos release then usually you can't just use latest versions for everything (gmp-droid, gst-droid, droidmedia, droid-hal/config submodules and maybe others) | 15:59 |
mighty17[m] | welp can i go back to the commit? | 16:00 |
mighty17[m] | as the older base i had was cm13 so i cannot copy over the working folders from there to los15.1 | 16:00 |
T42 | <elros34> you can either manually checkout to older version then build package with offline option or apply this https://github.com/mer-hybris/droid-hal-device/pull/288 and then you can specify commit or tag you want to build | 16:02 |
mighty17[m] | oh thats pretty neat! | 16:05 |
mighty17[m] | is there a way i can know commit is of what sfos version? like with branches or do i need to check all commits | 16:06 |
T42 | <elros34> I do not think you can use branches, maybe only for some packages so better either use your cm13 base droid-local-repo as reference of 3.4.0 changelog from sfos forum. | 16:09 |
mighty17[m] | https://github.com/sailfishos/gmp-droid/tree/upgrade-3.4.0 seems to have worked :D | 16:15 |
T42 | <elros34> 3.4.0 changelog version is few commits older, that is why I do not recommended branches but it's up to you | 16:19 |
mighty17[m] | `Unable to find pattern: Jolla Configuration espresso3g`uh oh | 16:27 |
mighty17[m] | well why does it complain about patterns on .4 | 16:30 |
mighty17[m] | 3.4* | 16:30 |
T42 | <edp_17> @elros34 : Hi. I have a question for you. I've found a newer kernel (3.12) for the i9100 but that kernel only works with Android 4.3 (Jelly Bean - CM10.2). My current port uses the original kernel (3.0) with CM12.1 base. Does it worth to try using the newer kernel? If yes, will it work with that cm12.1 base or I should use cm10.2? | 16:32 |
mighty17[m] | s2 has 3.12 kernel wow | 16:33 |
T42 | <elros34> no idea whether is worth or whether it will work out of the box. If you have any issues with current kernel then sure it's worth to try. How can cm10 have newer kernel than cm12? | 16:35 |
mighty17[m] | seems like 3.12 is heavily modified for cm10 | 16:37 |
T42 | <elros34> nice version, IIRC I have 3.3 for one device:) | 16:37 |
T42 | <elros34> remember that nobody use hybris-10 and probably only one insane hybris-11 | 16:39 |
mighty17[m] | @elros34 Unable to find pattern: Jolla Configuration espresso3g , i feel like to give up on this xD | 16:40 |
T42 | <edp_17> Yeah, the guy who ported kernel 3.12 has modified cm10.2 to make it work with that kernel. | 16:40 |
T42 | <edp_17> I've installed that kernel for los14.1 but didn't start. It booted but stuck at the screen after the boot logo. However, when I installed a cm10.2 rom, it works well with that kernel. | 16:42 |
T42 | <elros34> @mighty17[m] really? then maybe you should build 4.1.0 and avoid such a issues. Obviously its fixable, just a matter of providing logs and uploading sources | 16:42 |
mighty17[m] | i mean i know for a fact that 3.4 with its patterns works, but well new android new kernel then lets just use new hadk | 16:42 |
T42 | <edp_17> That's why I asked. Perhaps first I would need to make cm12.1 work with that kernel, then it could be used for sfos as well. | 16:43 |
mighty17[m] | "need to make cm12.1 work with that kernel" have fun with android | 16:44 |
mighty17[m] | im still waiting for sailfish + mainline kenrel | 16:44 |
T42 | <edp_17> Haha. That was a rethorical sentence. :-) | 16:45 |
T42 | <b100dian> But isnt hardware lacking proper drivers preventing that? The vebdor blobs with binder apis are like a new api surface for crippled drivers. I know i fought to drive a vibrator and its not fun. The drivers in question wouldnt pass my review at work, even less so linux reviews. So we get the hardware but cannot use it without provided crutches | 17:45 |
T42 | <b100dian> *vendor | 17:45 |
mighty17[m] | "hardware lacking proper drivers preventing that" wdym sfos on mainline? | 18:22 |
T42 | <skvorl> You can run SFOS on PinePhone. Is it what you are waiting for?) (re @SailfishFreenodeIRCBridgeBot: <mighty17[m]>im stil...) | 18:39 |
mal | mighty17[m]: sfos can run even on qcom devices using mainline kernel, not really good enough for releases but UI works | 18:41 |
rinigus | from experimenting with the systemd-ask-password - looks like Lipstick has some support for it. although, it doesn't really send anything back and just states that the password is incorrect | 18:56 |
rinigus | I guess, half-baked feature in 4.1 SFOS | 18:56 |
T42 | <edp_17> I am still puzzled how the drivers of a newer kernel can work with a device hardware that designed for older kernel. I expect if this was pissible for an android version, it should be possible for another version too. | 19:52 |
T42 | <XAP2P> do you have test builds? well sooooo interesting to see (re @birdzhang: @XAP2P i'm doing tha...) | 20:00 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!