Saturday, 2026-08-15

« Friday, 2026-08-14 Index Sunday, 2026-08-16 »
Mister_Magistermal: as for DNS issue x2 had CONFIG_ANDROID_PARANOID_NETWORK=y which g2 nor fp2 had lol, weird mer kernel check didn't catch that one. I love it just two oneliners00:14
Mister_Magisterjust two one line fixes, that shows how close i was00:14
Mister_MagisterI was comparing the defconfigs so many times…00:19
Mister_Magisterssh issue and the fix is another one liner…00:43
Mister_Magistermal: on g2 after fresh install the DPI inside sfos browser for some reason is -1 which breaks everything, ever heard similar issue? how does it get generated?00:48
malyou could try if it happens again01:10
malhttps://github.com/sailfishos/docs.sailfishos.org/blob/master/Support/Help_Articles/Web_Browser/README.md?plain=1#L50901:10
malafter removing that file01:10
Mister_Magisterit happens on every clean install01:11
Mister_Magisterbut i'll experiment later sure01:12
malithttps://github.com/sailfishos/sailfish-components-webview/blob/master/lib/webenginesettings.cpp#L16001:12
maloops01:12
malhttps://github.com/sailfishos/sailfish-components-webview/blob/master/lib/webenginesettings.cpp#L16001:12
malI wonder if display size gets detected incorrectly or something01:12
malyou could try forcing the panel dimension in compositor configs01:13
Mister_Magisteroki doki i'll bother you about it more when i get to it, workaround is quite simple so its low priority xd01:13
Mister_Magisterg2 is fully fixed now, x2 i'm trying to fix ssh with claude01:14
Mister_Magisterfinally i'm excited for 5.101:14
malhttps://github.com/mlehtima/droid-config-fp2-sibon/blob/master/sparse/var/lib/environment/compositor/droid-hal-device.conf01:14
Mister_Magisteralso mal both me (on x2) and user of my port (g2) noticed phones being noticably smoother on 5.101:14
malfp2 has hardcoded physical size01:14
Mister_Magisterso while i talk a lot of poop about jolla, gotta give credit where credit is due :)01:14
Mister_Magistercongratulations01:14
malmaybe the libhybris changes and possibly some other changes helped01:15
Mister_Magistermaybe matching the refresh rate dunno01:16
Mister_Magistermaybe i'll bother claude to also fix the droidmedia/gstdroid issue, did you have time to look into it?01:16
malnot yet01:17
Mister_Magisteroki01:17
*** n2klb_ is now known as n2klb01:43
T42<adampigg> Used to happen on pinetab iirc (re @SailfishFreenodeIRCBridgeBot: <Mister_Magister>mal...)11:25
sharks_Hey all - I'm trying to look into sleep on WiFi now, see dmesg output --> https://paste.opensuse.org/pastes/640c54ccbeb8, it appears like the device is getting a ton of wake-on-wlan events which is really wrecking sleep. With WiFi off I get ~80%+ CPU sleep but with WiFi on I'm seeing 0%. Idle battery drain is <0.6%/hr without WiFi or >3.5%/hr with it. Any ideas where to look to chase this one down?12:13
sharks_After having run `iw phy phy0 wowlan disable`, I have dropped wifi idle usage to <2%/hr, which makes me think that's certainly part of the issue. What would cause these wake on wlan requests? I don't have any open SSH connections or anything. What's the phone trying to do in the background? Who's trying to talk to it while it's sleeping?12:59
malsharks_: do you have https://github.com/mlehtima/droid-config-fp5/blob/master/patterns/patterns-sailfish-device-adaptation-fp5.inc#L9513:10
sharks_@mal, yes I do.13:11
sharks_https://github.com/sharks-dev/droid-config-pdx225/blob/818cce2d54c7e5768086c9c2761c01d30791e891/patterns/patterns-sailfish-device-adaptation-xqcc72.inc#L2213:11
*** sskras is now known as [||16:14
*** [|| is now known as sskras16:32
Mister_Magistermal: so apparently sshd not working on x2 is issue with openssh not with my kernel because there's literally no difference between g2, x2, fp2 kernels in that area lol16:56
Mister_Magistermal: https://github.com/sailfishos/openssh/pull/1317:05
malbut why is that not a problem on other old devices?17:08
Mister_MagisterLLM said race condition and better answer is your guess is as good as mine17:10
Mister_Magisterthe getrandom stuff is fine, seccomp stuff is fine, everything's fine17:10
Mister_Magisterlet me just confirm that the fix actually works (yes sorry i forgot to do that)17:11
Mister_Magisteryeap fix works17:13
Mister_Magisterokay with that i think all 3.4 devices are fixed now and i can move onto rest of my devices17:50
Mister_Magistermal: is it expected for pa-modules-droid to fail to build?18:24
Mister_Magisteri literally grabbed _service from fp4 and it fails xd18:24
Mister_Magistermal: i updated my oneplus6 and now lipstick is segfaulting, got any tips for me?19:00
maljb2q version?19:01
Mister_Magistersame as g2 https://build.sailfishos.org/package/show/nemo:testing:hw:oneplus:enchilada/pulseaudio-modules-droid-jb2q19:02
malMister_Magister: did you update libhybris hwc compat module?19:02
Mister_Magisterno19:02
malfor the lipstick issue19:02
Mister_Magisteryou said if i use older pa-modules-droid (which doesn't build) i don't have to19:03
maleither use older hwc qpa plugin or update and rebuild the android side hwc compat module19:03
Mister_Magisteroh19:03
malMister_Magister: what?19:03
Mister_Magisterwait i got confused19:03
Mister_Magisterso "PSA: it's recommended to use pulseaudio-modules-droid version 14.2.106"19:03
Mister_Magisterthis one does not build19:03
Mister_Magisterhttps://build.sailfishos.org/package/show/nemo:testing:hw:oneplus:enchilada/pulseaudio-modules-droid19:03
malthat for new devices19:03
Mister_Magisterhow much newer?19:04
Mister_Magister"if it doesn't build it's not new enough" got it19:04
Mister_Magisterfound the other PSA, fixing19:04
maljb2q as is android jb to q i.e. 4-1019:04
Mister_Magisterg2 has 5.6.2.27 and it works on it so got confused19:05
Mister_Magisterits so old it doesn't matter19:05
mallet's not confuse two topics, pulseaudio and hwc19:05
Mister_Magisteryeah sorry sorry19:06
malMister_Magister: android 8 and newer (if I remember correctly) is where the hwc plugin version matters19:06
Mister_Magisteri've dropped qpa from .27 to .2619:06
maljust mentioning that there might some performance improvement in newer devices with new hwc plugin19:07
malit add some caching changes19:07
Mister_Magisteryou reaaaaaaaly want me to update it? xd ok what do i rebuild?19:07
mallibhwc2_compat_layer in android side after updating the libhybris sources you use in android side build19:09
Mister_Magistergotcha19:09
Mister_Magistermal: oh no… op6 has broken dns resolving :( I guess its kernel is not new enough (4.9)19:12
Mister_Magistergotta do more backporting!19:12
Mister_Magistermal: guy in my chat asked if we're buddies19:19
Mister_Magisterok qpa fixed it19:23
Mister_Magisternow i still need to fix the dns and since that requires kernel rebiuld i'll update compat too19:24
malquite odd why 4.9 would need backporting for dns19:24
malfrom which kernel is the backport?19:24
Mister_Magisterwell i was fixing it in 3.4 so far19:25
Mister_Magisterthe issue is socket(AF_INET, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, IPPROTO_IP) = -1 EACCES (Permission denied)19:25
Mister_Magisterhttps://github.com/VerdandiTeam/android_kernel_motorola_msm8974/commit/f72e2d04f7a70d53029eec5de5f5f5f58e21a7b2 i think it was this one?19:25
Mister_Magisterhmm nope19:27
Mister_MagisterThen I do wonder19:27
Mister_Magistertime to run claude on op6 kernel!19:30
Mister_Magisterwait19:39
Mister_Magisterwait wait wait19:39
Mister_MagisterI think it was Paranoid android issue19:40
Mister_MagisterCONFIG_ANDROID_PARANOID_NETWORK=y YEAP19:40
malheh, same again19:45
Mister_Magisteryeap!19:46
Mister_Magisterweird it wasn't issue on <5.119:46
Mister_Magistermal: should i check out libhybris to master or some specific tag?19:47
malmaster is fine19:48
malit's used on official devices, remember to make sure you get submodule updated19:48
Mister_Magistersure thing boss19:49
Mister_Magistermal: until the ssh fix lands in sfos i'll keep using fork but i hope it lands19:54
malneed to do some testing, I have kernel header update waiting also19:57
malto 5.1019:58
Mister_MagisterI'm gonna keep poking you about the gst-droid fix because it's incredibly important19:58
Mister_Magistermore than you realise maybe :P19:59
Mister_Magisterdns fixed and now on new compat19:59
Mister_Magistermal: hmm i think it's smoother, i'm back on the .5720:17
Mister_Magisterbut 3.4 devices were also smoother20:18
malyeah there are several changes20:24
Mister_Magisteri'm finally excited for 5.1 and not cursing it for ruining everything, God bless people for creating LLMs20:25
Mister_Magistertwo devices to go21:20
Mister_Magistermal: so there was issue with long shutdown on op6, one part was chre process but it was useless to me on sfos so i just disable it, another was pm-service was not being killed properly, and you won't guess the claude fixed it in dhi with, yes, another one line fix23:23
Mister_Magisterwait no, i misread23:26
malwhat does that process do?23:42
Mister_Magisterthere's like sensors co-processor that doesn't wake up the cpu and handles like lift to wake and stuff from what i gathered23:43
Mister_MagisterI don't use those gestures23:43
Mister_Magisterand screen gestures still work23:45
maldid you update droid-config submodule?23:45
malactually it seems you can't get one fix yet23:46
malhttps://github.com/mer-hybris/droid-hal-configs/commit/fe99b05a85e7f001f79d1be7f030e61a2f879db1 that one, there a changes before that which won't work in 5.1 yet23:47
mal*there are23:47
Mister_MagisterI did update dcd, the shutdown issue is not 5.1 related it has been present since always and i didn't know how to debug it so now i thought what if i give claude root and let it wild run what's the worst that could happen23:48
malyeah, that fix is for an issue that has existed always23:49
malonly some devices suffer from processes that fail to stop properly23:49
Mister_Magisteroh, so maybe with my disabling of chre it will be fine23:50
Mister_Magisteryeah that sounds like my issue alright23:50
malso if the process fails to stop the second attempt just make it timeout twice23:50
Mister_Magisterso i'll just implement the disabling of chre and wait for your mentioned fix23:50
Mister_Magisteryeah here it just times out like 4 times or so23:51
mallike jolla phone had one process that caused shutdown from actdead mode be very slow until I disabled it from there23:51
Mister_Magisterthat sounds promising mal, so when that fix will be available?23:52
malI mean that in your case that commit I linked only helps with slow shutdown to some extent but it might still be a bit slower that usual23:53
Mister_Magisterdisabling chre already cut it down in half, if it helps by like 10 seconds it will be good enough23:54
Mister_Magisterfor the record currently shutdown was taking like 70s23:54
malhow fast is boot on your devices?23:54
Mister_Magisteryou just asked about 7 devices i would have to go and test each of them man23:55
malI mean a rough guess23:55
maldoes it seem slow or reasonable23:55
Mister_Magistereven ignoring duplicates its like 5 xd umm its fine, some are faster than others, let me time 3 of them i have on hand23:56
malbtw, check in journal if you see bluebinder failing once at boot on the newer ones which use it23:56
malthat droid-hal-config commit I linked will be usable in sailfish 5.2 once that is released23:57
Mister_Magister~50s on g2, but it's the 2014 oldest supported device so give ol' grandpa some slack23:58
malyeah23:58
malnot too bad23:58
« Friday, 2026-08-14 Index Sunday, 2026-08-16 »

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