libera/#maemo-leste/ Wednesday, 2021-06-02

Wizzupjust debian things: installing hostapd means it gets auto started on boot, and if you configure it (via config file) for usage at some point, it will interfere with wifi00:55
Wizzupand I removed it from the runlevel, but every time it's updated, it gets added back to the runlevel00:55
uvosmore like just devuan things01:09
uvossince debian on systemd dosent have this problem01:09
Wizzupuvos: it doesn't autostart services?01:09
uvosit dose but you can ask systemd if the user disabled it or something01:09
uvos(not sure how it works exactly but it only enables the service once)01:09
uvosmight just be that the systemd dpkg scripts are better idk01:10
cockroachyou should be able to do the same thing with normal init scripts if you do it the "proper way" (rather than just removing the symlink). I think.01:10
cockroachi.e. with update-rc.d01:11
Wizzupuvos: I think this is problem with dpkg01:12
WizzupI think it probaly does the same thing01:12
Wizzupwhen I said 'I removed it from the runlevel', I meant 'rc-update del hostapd'01:13
Wizzupmbut maybe update-rc.d does more01:13
cockroachokay, sorry for the noise then01:13
Wizzupvery welcome :)01:13
inkyjk000 if you come and see this - so you see the mouse pointer on the screen?01:20
Wizzupbencoh: btw, I see this in wpa_supplicant: <3>CTRL-EVENT-REGDOM-CHANGE init=BEACON_HINT type=UNKNOWN02:42
Wizzupbencoh: per nl80211_reg_beacon_hint_event in hostap/src/drivers/driver_nl80211_event.c I assume that actually it does change the regulatory domain based on other beacons02:42
Wizzupdidn't check it really hard, but wanted to follow up on that02:42
Wizzuplooks like wpa_supplicant has various sources of regdom info02:42
Wizzupsee reg_init_str in hostap/wpa_supplicant/events.c02:42
Wizzupfreemangordon: ^^02:43
Wizzupso those beacon hints cause EVENT_CHANNEL_LIST_CHANGED which does seem to change frequencies and such02:46
Wizzupso it looks like the "magic" might just work02:47
Wizzupof course, we can do more, based on modem status for example.02:47
ravelo'lo09:15
diejuse Hi, I am trying with ARM instead ARM64. I am trying to chroot Maemo Leste on an old ARM Android Phone. I get many error messages like "failed to get XImage of pixmap". Any clue/help to solve it?09:28
bencohWizzup: oh, interesting10:23
bencohit's really "magic" :)10:23
uvosdiejuse: sounds more like a 3d driver problem than an x problem per say11:00
uvosor bitedness problem even11:00
Wizzupuvos: it could be older android not supporting X as well11:04
diejuseuvos: however if I open the apps through Xephyr it works and I don't get those messages11:04
WizzupXephyr might not be accelerated11:05
diejuseis there any way to remove Hildon's animations to make it faster?11:14
Wizzupdiejuse:: not trivially, no, you won't have a good experience without 3d generally11:14
inkydiejuse, i don't really understand what you are doing, but would like to. so you have a leste chroot on an android phone. okay. that would work in commandline. then did you use some X surface for android?11:17
diejuseI use "XSDL XServer" app like X Window System / X11 server13:06
diejuseinky13:06
diejuseNow I'm testing it on an old mobile, on an Xperia Pro13:09
diejuseI want to see if Maemo Leste is usable through the chroot technique on old mobiles.13:10
diejuseThe script I made for ARM64 does not work at all because it gives me graphical errors when I open the applications. But if I do a workaround using "xephyr" I can get it to work.13:14
Wizzupdiejuse: so it's likely related to 3D (driver) stuff13:26
diejuseWizzup: It seems to be so. I love Xephyr. I also fixed the drop down menu problem with Xephyr.13:29
diejuseThe only problem is that the performance drops a bit.13:30
uvosdiejuse: you are unlikely to have that mutch luck with the pro really13:30
uvosa android device that old, what dose it run linux 2.6.x?13:30
uvosim kinda supprised glibc still has support for the kernel its running...13:31
uvoslooks like current glibc needs at least 3.213:34
uvosour glibc is older so maybe it works with 2.613:34
uvosif so your setup will die with the upgrade to debian/devuan 1113:34
Wizzupthere's always mainlining :)13:35
diejuseuvos: When I first entered this channel, you told me that what I was trying was not going to work. Now I will surprise you again. :P13:36
uvosheh13:36
uvosyeah maybe im overly sceptic13:36
uvosbut still chroot over ever more ainchent android underpinnings its unsustanable13:37
diejuseFirefox and Chromium don't work well on the Xperia Pro but Dillo works quite well.13:37
diejuseI will make another video and put documentation.13:38
uvosso what kernel is the pro running?13:38
diejuseYes but first I had to update to Android 7 (Lineage OS 14.1)13:38
uvosright but lineageos rarely upgrades the kernel13:39
uvosthey mostly just work with the vendor kernel as long a possible13:39
uvosand then drop the device when not13:39
diejuseWith Android 6 (CyanogenMod 13) it didn't work13:39
uvosthey dropped a huge amount of devices after 14 for this reason13:39
uvossince later android needs a newer kernel13:39
diejuseSimilarly, I managed to get Maemo to work on a Blackberry Key 1 (Android 7) but it didn't work for me on a Blackberry Priv (Android 6)13:40
diejuseuvos: I understand13:40
uvosgo to settings -> about  phone -> android version13:41
uvosand tell my "Kernel version" for the devices13:41
uvosim curious13:41
diejuseone moment13:41
diejuseUvos: Kernel 3.4.0-LX+13:42
uvosok your safe for now :)13:42
diejuse>3,213:42
diejuse>3.213:42
uvosbarely made it :P13:42
uvoswhat device was what?13:42
diejusehehe13:43
diejusethe device? Sony Xperia Pro MK1613:43
diejusebeauty device with physical keyboard13:43
diejuseI bought it second-hand for 30 euros and did several experiments on it.13:44
diejuseI have got xperia mini pro sk17 too. Another beauty device.13:45
Wizzupuvos: for the record I did chat some more with fmg about the whole wlancond and such, and it does do some sensible things like change wireless power saving state if a voip call is active, and other stuff like that, not yet sure where we could fit that, so we might ultimately need a daemon to manage some of that13:47
Wizzupthere's no immediate solution required and for now we'll add stuff to libicd-network-wpasupplicant (like power saving etc)13:47
Wizzupwe'll deal with it once we can make calls and voip calls ;)13:48
uvos"whatever manages modem state" still seams like the sane awnser13:49
uvosregarding voip, if we do decide to mangage modem state in mce, it allready knows the call state so a module could do this with almost no extra code13:49
Wizzupyes, but then mce would have to do all wireless power saving setting13:51
uvoswe could make a global decission that "power managment is mces job" or something like that if you like13:51
Wizzupand then it needs IAP knowledge etc13:51
uvosor a seperate modem manager deamon13:51
WizzupIt does get a bit messy13:51
Wizzupright13:51
Wizzupin any case, it's not super relevant now, just wanted to follow up13:51
Wizzupand yes, btw, a modem module for mce would be good, to online the modem and such13:51
uvosno new modules until we unfork our mces :)13:51
Wizzupyeah, yeah ;)13:54
uvosbtw im not sure we have to do anything special for voip13:58
uvossince kernel has dynamic power-saving and will do latency management based on packet behavior13:59
uvosand the old network latancy interface is gohne13:59
uvosso you cant tell the kernel anymore that you need real time14:00
uvos(and its also not needed anymore)_14:00
WizzupI think 'iw set <dev> power_save on' on n900 will make packet latency go up a _lot_ even when active14:01
Wizzupso probably it will be14:01
uvosright but the interface that allowed userspace to detemine what latency is acceptable (what wlancond almost certenly used) is gone afaik14:03
uvosso you can oly turn dps off14:03
bencohiirc there are 3 mods in maemo5, and two of them are definitely bearable latency-wise14:04
bencohthe super-low-power one is a bit harsh (still iirc)14:04
bencohbut that's the kind of thing you'll notice if you ssh (or other low-latency activity) anyway14:05
Wizzupuvos: possible @ power saving granularity14:12
tmlindWizzup: hmm i naturally backported the musb breaking pinephone fix to the droid4-pending-v5.10 and later trees too..18:25
tmlindso that would explain the reports about usb notworking networking18:25
uvosparazyd: ^^^18:27
tmlindi'll go push the fix for the fix to all the droid4-pending-v5.10 and later branches18:28
uvoswe dont need 5.1218:29
uvosleste is on 5.10 and 5.11 and non leste use is 5.1318:29
uvosfor me at least18:29
tmlindok18:32
tmlindwe really should get m-l kernel working with the ddk-1.17 :(18:34
uvosyeah :|18:36
bencohany link to the musb thing?18:36
tmlindbencoh: https://lore.kernel.org/r/20210528140446.278076-1-thomas.petazzoni@bootlin.com18:41
bencohthanks!18:43
uvostmlind: btw your ttys0 patch + disabling sysreq and disabling kernel consol on serial port makes the boot failures with usb pluged in dissapear18:44
uvosso mabye add that patch to the droid4-pending-* branches18:45
tmlinduvos: ok cool18:45
uvosbut the fact that i have to disable kernel console on serial suggests the port is still misbehaveing somehow18:46
tmlindyeah..18:47
uvos(ie hanging or soemthing)18:47
tmlinddid you set sysrq to 0 in .config?18:47
uvosyeah18:47
tmlindhmm18:47
uvosbut it dont know if its needed18:47
tmlindlet's wait on that one a bit then18:47
uvosthat alone was not nessecary18:47
uvosor enough rather18:47
tmlindyeah18:47
uvosi had to also disable the output18:47
uvosok @ wait18:48
tmlindparazyd: pushed out updated droid4-pending-v5.10 and droid4-pending-v5.11 to github with latest stable stuff merged in + usb networking fix applied18:48
tmlindparazyd: only compile tested those branches, i've been using the usb fix for few days locally though. there was also a trivial charger merge conflict so might be worth testing18:49
uvossame here usb fix has been fine on 5.1118:49
tmlindok18:49
tmlinduvos: i'll wait on aplying the usb fix to the droid4-pending-v5.13 branch and will just merge in -rc5 that should have the fix18:50
uvostmlind: ok :)18:50
tmlindno other users for that branch beyong you and i probably right now :)18:50
uvosright :P18:51
tmlinduvos: fyi somehow wl12xx produces always a sdio read error on first boot on mz61718:51
tmlinduvos: and power key does not get detected by kexecboot18:51
tmlinduvos: so i can't select the kexecboot kernel to boot right now..18:52
uvosok ill check it out once Wizzup's package arrives :)18:52
tmlinduvos: ok, i don't really have a much of a kernel patch for it yet, just nothing in it really just the compatible18:53
uvosthats fine18:53
tmlinduvos: will send to you when you get the package, been trying to figure out the lcd panel18:53
tmlindthe lcd seems to be dsi to a dsi-to-lvds bridge18:53
uvosmostly mapphone-common.dtsi (or what the name is) should have it coverd except lcd and some sensor specifics18:54
uvosinteresing18:54
uvoswierd way to do it18:54
tmlindsomething similar is in exynos5250-arndale.dts, hopefully that compatible will work here too18:54
tmlindit's a toshiba tc358765 that hopefully is very similar to tc358764.c that the mainline kernel driver handles18:55
tmlindno idea what the difference between tc358764 and tc358765 is if anybody cares to try to research :)18:56
uvosi++ :P18:56
tmlindheh18:56
tmlindor a typo18:56
tmlindor charge more money from a big custome18:56
tmlindr18:57
tmlindgoogle search produced this answer:19:00
tmlindThe TC358764 supports panels up to 1366x768, with 24 bits per pixel. The TC358765 supports panels up to 1920x1200, with 18 bits per pixel. Applicable to a range of mobile product platforms such as MIDs, netbooks, smartbooks and eBooks.19:01
tmlindi guess i need to somehow fix the kexecboot power key for mz617 first19:03
uvosthe keycode is likey just different19:04
uvoscheck in android19:04
tmlindnoisy dmesg for v3.0.8 kernel on it shows KEY_POWER debug statements for each press and release..19:05
uvoshmm19:05
uvosok thats wierd then19:05
tmlindit should be the same cpcap power key as on others19:05
tmlindhmm maybe it does not work on xt894 either, let me check19:05
uvosit certly works on xt87519:06
uvosso kexecboot binary isent totaly broken19:06
tmlindyou boot to mainline kernel though?19:06
uvosbut different kernel ofcourse19:06
uvosright19:06
uvosbut ist not a fundamental bug in kexecboot im saying19:06
tmlindheh power key won't do anything for xt894 either with kexecboot :) funny nobody noticed19:07
tmlindso the keycode must not be KEY_POWER, or it's not advertised as a proper keyboard19:08
tmlindhmm there's define VERY_LONG_HOLD_REBOOT, maybe it just needs to be pressed for longer..19:11
tmlindnope, maybe kexecboot only supports one keyboard or something19:15
uvosthat would not work on xt87519:17
uvosxt875 has power and volume on different devices19:17
uvosboth work19:17
uvosnot sure why they used the keyboard matrix for 2 keys btw19:20
uvosmaybe its just inherited form/so that its the same as xt862 or something19:20
tmlinduvos: but xt875 uses mainline kernel with kexecboot, not v3.0.8 kernel19:21
uvostmlind: sure but it still has 2 keyboard event devices19:21
uvosand kexeboot picks up both19:21
uvosso its not that19:21
tmlinduvos: ok19:21
uvosbtw android there is getevent19:22
uvos== evtest19:22
uvos(just about)19:22
uvosif you have root19:22
tmlindok still have usable android on mz61719:23
tmlindso is getevent there built-in?19:23
uvosits part of asop19:23
uvosdosent enforce that they ship it19:24
tmlindok19:24
uvosim booting los on d419:24
tmlindok19:24
uvosgoes absolutly crazy with events on d419:25
tmlinddoes it generate KEY_POWER?19:26
uvosi cant red anything19:26
uvosthe accel is a event device19:26
uvosand it totaly swamps everything19:27
tmlindheh19:27
uvospower goes19:29
uvos001 006b19:29
uvoslooks like its KEY_END19:30
tmlindhuh?19:32
uvosyeah its KEY_END19:33
uvosgetevent -lt /dev/input/event219:33
uvosand press power19:33
tmlindweird19:33
uvosyeah19:33
uvosi absolutly hate useing the terminal in los19:34
uvosthere is no ctrl19:34
uvoshow grr19:34
tmlindok event3 on mz617 stock android, but KEY_END yeah19:35
tmlindok i'll patch kexecboot for that19:35
tmlinduserspace tweaks it in android or what?19:35
uvosgreat19:35
uvoscant parse19:35
uvoswhat userspace tweaks19:35
tmlindhmm no userspace gets run with utagboot though19:35
uvosoh the key_end19:36
tmlindi'm not seeing KEY_END in the v3.0.8 kernel sources19:36
uvosyeah no must be the kernel somehow19:36
tmlindfor the cpcap driver19:36
uvosno idea19:36
tmlindweird19:37
tmlinduvos: ok pushed out updated droid4-xyboard-kexecboot-2017.11 to github, will email out the patch with your in cc19:58
tmlindthanks for helping to debug that one :)19:58
uvosyou are ofc welcome :)19:59
tmlindkexecboot patch sent, detaching from keyboard now, ttyl20:01
parazydtmlind, uvos: Thanks for the notification for the patch21:33
parazydI'll get it ready soon21:33

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!