| rinigus | @b100dian: was looking/reading last night regarding stock kernels and trying to figure out how can we build one. will continue tonight | 06:31 |
|---|---|---|
| rinigus | did you managed to reproduce getting to gui? | 06:32 |
| T42 | <b100dian> rinigus: I reproduced Sunday evening but I was unable to swipe to unlock, then I tried unrolling the links in /overlays through a script and then I broke something else which I've yet to determine:) Didn't look into it yesterday | 09:16 |
| rinigus | @b100dian: I didn't manage to swipe for unlock either. only could pull menu from top. may have something to do that it didn't ask for pin during setup. which, in turn, maybe is related to keystore being not found messages... | 09:23 |
| rinigus | but sounds like we are in sync then | 09:23 |
| rinigus | @b100dian: as for overlays and symlinks vs bind mounts - I think we can wait with fixing this solution as it will depend on whether they will fix aosp linking or whether it works already on stock. for now, manual copy should be just fine :) | 09:26 |
| T42 | <b100dian> Yes, but for info, this is what I've tried Sunday evening https://github.com/sailfishos-sony-nagara/droid-config-sony-nagara/pull/1/files | 09:29 |
| rinigus | nice processing of symlinks! just I would have made a separate folder for copies and overlaid then that folder instead. or, maybe bind mounts in such "shadow" folders | 09:33 |
| T42 | <b100dian> Yes, this is just the first try which changes the root fs, not very smart - but yeah, we can make it better when needed | 09:34 |
| rinigus | @b100dian: fyi - there was an attempt to port lineage to pdx223, but seem to have stalled. asked about it, no reply yet. https://github.com/sonybasement/device_sony_pdx223/issues/1 | 09:35 |
| rinigus | haven't tested it yet and looks like they started with the binary copy of stock kernel (one of the repos in that org). | 09:36 |
| T42 | <b100dian> they do have a kernel_sony_sm8450 which has 5.10 kernel (don't know the source) where they merged codelinaro stuff for techpack and drivers/staging/qca-* stuff.. but we have 5.4 | 09:51 |
| *** amccarthy is now known as Guest178 | 10:22 | |
| *** amccarthy_ is now known as amccarthy | 10:22 | |
| rinigus | @b100dian: we have 5.10 on stock :) | 10:46 |
| T42 | <b100dian> Oh, ok. Did you alreadt try an AOSP 5.10 boot.img ? | 10:48 |
| rinigus | @b100dian: no, I didn't. | 11:14 |
| T42 | <adampigg> should defaultuser be a member of system ? /sys/class/leds/vibrator/* are rw system:system, and currently, my user isnt a memeber of that group | 17:22 |
| T42 | <elros34> that should be input group | 17:25 |
| T42 | <elros34> long time ago nemo was member of system but that was changed | 17:26 |
| T42 | <adampigg> ok, ill fix the group | 17:27 |
| T42 | <adampigg> well, permission | 17:27 |
| mal | @adampigg where are the permissions handled in .rc files? | 17:28 |
| mal | patch the needed file | 17:28 |
| T42 | <elros34> https://github.com/mer-hybris/hybris-patches/blob/hybris-20.0/system/core/0035-hybris-Allow-input-group-to-use-vibrator.patch | 17:28 |
| mal | some devices have the permissions in device repo if the path is non-standard | 17:29 |
| mal | like here https://github.com/mlehtima/android_device_fairphone_FP5/commit/6fd799b164da3c0d68c0e6ffcead2239da9eba75 | 17:29 |
| T42 | <adampigg> thx ... i have to do it slightly differently as using hybris base image... | 17:56 |
| T42 | <adampigg> s/halium | 17:56 |
| T42 | <adampigg> perhaps they should make a change there | 17:57 |
| T42 | <adampigg> grrr, same issue as x23 ... editing the .rc file works for _most_ of the permissions ... but some get reset back to system:system ... i think some daemon on android side does it.. | 18:07 |
| mal | or is udev doing that? | 18:09 |
| mal | although udev is running earlier | 18:09 |
| T42 | <adampigg> looks like i already tried to solve it with udev | 18:12 |
| T42 | <adampigg> [root@Phone22 defaultuser]# cat /etc/udev/rules.d/70-vibrator.rules | 18:12 |
| T42 | <adampigg> SUBSYSTEM=="leds", KERNEL=="vibrator", ACTION=="add|change", RUN+="/bin/chgrp -R input /sys/class/leds/vibrator/", RUN+="/bin/chmod -R g+w /sys/class/leds/vibrator/" | 18:12 |
| T42 | <elros34> this will not wokr, udev will always be executed before some .rc files | 18:17 |
| T42 | <adampigg> even with the .rc, i have some correct, but not all | 18:18 |
| T42 | <adampigg> https://paste.mozilla.org/WNvGekSX | 18:18 |
| T42 | <elros34> maybe there are more .rc in /vendor or something? | 18:19 |
| T42 | <elros34> maybe waydroid? unless it has some kind of seperation | 18:20 |
| T42 | <adampigg> not waydroid, thats not installed yet... | 18:21 |
| T42 | <adampigg> ill try disabling the vendor vibrator service and see what perms i have then | 18:21 |
| mal | btw, I have wip aidl vibrator code | 18:24 |
| mal | to use vibrator via android service | 18:24 |
| T42 | <adampigg> might be good to use | 18:25 |
| T42 | <adampigg> mal: @elros34 i added a kernel audit watch to /sys/class/leds/vibrator/activate and rebooted, got this output https://paste.mozilla.org/n7iwCgys | 19:10 |
| T42 | <adampigg> there is a chgrp and a chown executed by busybox ?? | 19:10 |
| T42 | <elros34> nice idea for tracking | 19:13 |
| T42 | <adampigg> yeah, unfortunately it didnt give an immediate answer ... but certainly useful for tracking file acceess | 19:14 |
| T42 | <elros34> I assume these a0,a1.. are arguments, maybe they can be more human readable with ausearch -i | 19:39 |
| T42 | <adampigg> so far, im assuming it is the last one ... | 19:45 |
| T42 | <adampigg> in all cases, ogid=998 which is input | 19:46 |
| T42 | <adampigg> the last call is syscall 54 which is fchownat | 19:47 |
| T42 | <adampigg> type=PROCTITLE msg=audit(19/11/24 19:39:06.580:70) : proctitle=/system/bin/init second_stage | 19:48 |
| T42 | <adampigg> type=PATH msg=audit(19/11/24 19:39:06.580:70) : item=0 name=/sys/class/leds/vibrator/activate inode=37519 dev=00:11 mode=file,664 ouid=root ogid=input rdev=00:00 nametype=NORMAL cap_fp=none cap_fi=none cap_fe=0 cap_fver=0 | 19:48 |
| T42 | <adampigg> type=CWD msg=audit(19/11/24 19:39:06.580:70) : cwd=/ | 19:48 |
| T42 | <adampigg> type=SYSCALL msg=audit(19/11/24 19:39:06.580:70) : arch=aarch64 syscall=fchownat per=PER_LINUX32 success=yes exit=0 a0=AT_FDCWD a1=0xb400007db822d910 a2=0x3e8 a3=0x3e8 items=1 ppid=1618 pid=1619 auid=unset uid=root gid=root euid=root suid=root fsuid=root egid=root sgid=root fsgid=root tty=(none) ses=unset comm=init exe=/system/bin/init subj==unconfined key=mon-vibr | 19:48 |
| T42 | <elros34> so it's some .rc in halium | 20:00 |
| T42 | <adampigg> yeah, thats what im searching for | 20:00 |
| T42 | <adampigg> ill be damned if i can find it though! | 20:00 |
| T42 | <b100dian> rinigus: boot img and vendor_boot + vbmeta from aosp + 5.10 kernel does not seem to work on stock vendor + lineage GSI.. | 22:03 |
| T42 | <b100dian> back into lineage gsi adb root, here's the kernel config for reference https://pastebin.ubuntu.com/p/ZdCdthbyz9/ | 22:05 |
| T42 | <b100dian> stock nagara does use this partition though | 22:07 |
| T42 | <b100dian> /dev/block/dm-9 on /vendor_dlkm type ext4 (ro,seclabel,relatime,discard) | 22:07 |
| T42 | <b100dian> https://pastebin.ubuntu.com/p/v47bGYdJVf/ | 22:08 |
| ecrn | when are the miniaf/minisf services needed? | 23:42 |
| mal | miniaf is only used on older android bases | 23:45 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!