libera/#maemo-leste/ Thursday, 2022-10-13

freemangordonbuZz: 144p works fine in chromium, fullscreen as well07:14
freemangordon240p too, but might not fluid at times07:21
tmlindfreemangordon: if you have a chance, maybe comment on the list regarding sony,acx565akm for the old omapfb? imo, we should do minimal fix, then drop the old omapfb stuff unless there's a reason to keep it still?08:45
tmlindand panel-dsi-cm too can go looks lie08:46
tmlindlike08:46
rafael2kthis 6.0 kernel might be worth trying in the PP, especially in PP 1.1 (Braveheart) as it clearly states it is supported09:09
rafael2kpre-compiled kernels are available here  (still in the 5.19, but mostly same patchset: https://xff.cz/kernels/5.19/ )09:11
rafael2khttps://github.com/megous/linux09:11
rafael2kwe could change to this megous branch in case it works better...09:11
buZzfreemangordon: i wonder why it doesnt for me then :(09:46
buZzwhats the CI channel again?11:13
Wizzup#leste-ci11:14
Wizzup##leste-ci11:14
freemangordonbuZz: what i sthe UA you're using?11:24
buZzehw, lets see11:25
freemangordonmake sure it is iphone6 or windows phone11:27
freemangordonso mobyle YT to open11:28
freemangordon*mobile11:28
buZzits Mozilla/5.0 (iPad; CPU OS 11_3 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/11.0 Tablet/15E148 Safari/604.111:29
buZzwhats yours?11:29
freemangordonwhat browser is that?11:30
freemangordonis that chromium?11:31
buZzyes11:33
buZzjust apt install chromium11:33
freemangordonwhat extension do you use to change the UA?11:33
freemangordonuser-agent switcher for chrome is what I use11:34
buZzi think me too? lets see11:34
freemangordonugh. battery went flat :(11:34
buZzhttps://chrome.google.com/webstore/detail/user-agent-switcher-and-m/bhchdcejhohfmigjafbampogmaanbfkg?hl=en11:34
buZzthis one11:34
freemangordonthis https://chrome.google.com/webstore/detail/user-agent-switcher-for-c/djflhoibgkdhkhhcedjiklpkjnoahfmg11:36
freemangordonbuZz: also, make sure it plays @ 144p11:37
freemangordonso, site must be m.youtube.com11:38
buZzyeah i did11:38
buZzi'll try your plugin first11:38
freemangordond4 just cannot handle desktop site11:38
buZzits grabbing the m.youtube already11:45
freemangordonhmm, it does not play here as well11:45
freemangordonwtf?11:45
freemangordonmaybe they have changed something11:46
freemangordonbuZz: oh, I know what happens11:50
buZzoh?11:51
freemangordonsometimes I have to reboot the device, otherwise, for some reason YT playback is choppy11:51
freemangordonfor some reason, sometimes after reboot GPU is not in a mood11:51
buZzhmm, weird11:51
freemangordonyep, rebooting atm, lets see11:51
buZzwonder if we could detect that, beside yt playback11:51
freemangordonno idea11:52
freemangordonmaybe some difference in regs11:52
buZzi wonder if its better on coldboot vs reboot11:52
buZzor the inverse11:52
freemangordonno pattern I am aware of11:52
freemangordonbut, it happens occasionally11:53
buZzsometimes reboot seems to hang for me too, after motorola logo, taking a long time before kexecboot shows11:53
buZzmaybe 10x longer than 'normal'11:53
freemangordonyep, after reboot it plays 240p with no issues11:55
buZzgee! , just a warm reboot?11:55
freemangordonpower down11:55
buZzso , a cold boot?11:56
freemangordonyes11:56
freemangordonbut I am not sure it matters11:56
freemangordonbut again, no known pattern11:56
buZzweird :)11:57
freemangordonthe only difference I am aware of is that it entered charge mode when it was choppy11:58
freemangordonbut I doubt this is the reason11:58
freemangordonalso, the whole display was full with renderning artefacts11:58
buZzthe ants?11:59
freemangordonlegions of them, everywhere :D11:59
buZzwas your battery really low when driver initialized?11:59
freemangordonyes11:59
buZzhmmmmm :)11:59
freemangordonI think it is related11:59
buZzyeah my suspicion too12:00
buZz+spelling12:00
freemangordongit it :)12:00
buZzjust like how the modem seems to behave weird at <20%12:00
freemangordon*got12:00
freemangordonand my battery is really old12:01
buZzsame here\12:02
freemangordonmost-probably related12:02
freemangordonwill see when replacement battery arrives how it will behave12:02
buZzwhich battery did you order?12:02
freemangordonthe one uvos is using12:02
freemangordondon;t remember the exact model12:03
buZzah nice, i want one too12:03
buZzdoes it have the flex lips like d4 uses already?12:03
freemangordonno, I will have to modify12:03
buZzor would it require that flexpcb design he shared12:03
freemangordonwill see if I will manage to :)12:03
buZzi was looking that we could maybe expand that flexpcb to have the connections for a thermistor too12:03
buZzand maybe make it slightly more universal12:04
freemangordonI will try to plant the whole circuitry from the old battery12:04
buZzjust put the thermistor on that flexpcb12:04
buZzah yeah , that works too12:04
buZzi'm still hunting for a new produced HV lipo that fits the case12:05
freemangordontmlind: TBH, I don;t feel experienced enough to comment - neither do I fully understand what the issue dmitry is fixing is about, nor the follow-up discussion12:05
* buZz puts pizza slices on the table for all12:06
Wizzupour raspi4 images are kinda broken12:33
Wizzupthe kernel compile fails12:33
Wizzupheh12:33
buZz:(12:33
buZzsad12:33
Wizzupsomehow it boots12:33
Wizzup:D12:33
Wizzupbut it's llvmpipe and no modules12:33
buZzoh :P haha12:33
buZzstill had a old kernel on it?12:33
Wizzuphttps://phoenix.maemo.org/view/Images/job/leste-image-raspi4-64bit/114/consoleText12:33
WizzupbuZz: not sure how yet12:33
buZz  Temporary failure resolving 'maedevu.maemo.org'12:34
buZz?12:34
buZzdont pi's normally have a seperate partition with the kernel?12:34
norayralso yewtu.be, i guess this was the domain, is an invidious instance, and it has very leightweight ui, even for a computer12:37
norayri often use it to decrease the load of the machine12:38
norayrbut maybe it doesnt have 144p12:38
Wizzupfreemangordon: where did you put iphbd-dkms ?12:38
Wizzupfreemangordon: the raspi can't find it12:39
Wizzupit's not in the droid4 component is it?12:39
Wizzupdoesn't look like it12:40
Wizzupah nvm, it's complaining about linux headers12:40
Wizzupunsurprisingly..12:40
buZznorayr: i'd like to record using actual youtube in the background though12:41
buZzas google now -charges money- for that luxury12:41
buZz(as youtube premium, one of the features is 'background playback')12:41
Wizzupapparnetly linux-headers-arm64 does not provide linux-headers? eh13:03
buZzmight be the issue i saw earlier? where our kernels dont build source packages13:04
buZzimho -headers is still source13:04
Wizzupno, seems very unrelated :D13:04
buZzhehe ok13:04
tmlindfreemangordon: i replied what i think is going on with the panels, maybe take a look and comment if there's any need for the old omapbf drivers in question13:09
tmlindomapfb panel drivers i mean13:09
buZztmlind: do you have any idea if the omap gpu driver could support .icc profiles for color correction?13:10
buZz-could13:10
tmlindbuZz: no idea whatsoever13:13
buZzalright :) i'll try soon13:13
uvos__buZz: the hang at the motorola logo13:15
buZz-after- the logo13:15
uvos__right also after13:15
buZzscreen is black then13:15
uvos__(ie at black)13:15
buZzyeah13:15
uvos__is some bug in the stock kernel caused by the uart module thats loaded by the intercept13:16
uvos__its related to the modem13:16
buZzhmmm13:16
uvos__if the modem is busy and uses the uart the kernel hangs13:16
buZzso maybe that kexecboot kernel needs some 'ignore uarts' flag ?13:16
uvos__no13:16
buZzoh ok13:16
uvos__so the problem is that the kexec module loads a hacky uart module13:16
uvos__s/loads/depends13:17
buZzwhat would that need a uart for?13:17
uvos__this was done by the linageos people because there is no way to get debug output at all13:17
buZzah13:17
uvos__while writeing the kexec module13:17
uvos__but its buggy somehow13:17
buZzbut then still kinda sounds like we dont need uarts there?13:17
uvos__if the modem uses its uarts while the hacky uart module is active13:17
uvos__stock kernel crashes13:18
uvos__i "fixed" this on my device13:18
uvos__by adding a 10 sec delay13:18
uvos__this ensures the modem has setteled more13:19
uvos__and the crash never happens on my device13:19
uvos__but its not really a fix13:19
uvos__fixing this problem is hard beacuse its in the module you need to get any output at all13:19
buZzcant we just remove that lineageos change then?13:20
uvos__no13:20
uvos__its required to kexec13:20
buZzaw13:20
uvos__you could remove all the debug prints in the kexec module13:20
uvos__and then not use the hacky uart module13:20
buZzyeah that was my suggestion13:20
uvos__but then you have no debug output during kexec13:20
buZzremove the dependancy13:20
uvos__so thats not great either13:20
buZzi guess, but when was last time you debugged kexec?13:20
uvos__well right now13:20
uvos__xD13:20
buZz:P13:21
uvos__buZz: as for if omapdrm could do color transformations13:25
uvos__shure you could do it on a pvr shader no problem13:25
uvos__thats how modesetting works13:25
buZzuvos__: no problem, as in no penalty in speed?13:26
uvos__well not zero13:26
uvos__but low13:26
buZzsounds good13:26
norayrpeople, i want to get linux-image-omap kernel sources, added deb-src13:34
norayrcopied our maemo lines in sources.list, but changed deb to deb-src13:35
norayrapt-get update, apt-get source linux-image-omap doesn't get me the source.13:35
norayri would like to get the source, change a couple of things in light meter driver source, my friend believes there is a bug there.13:35
norayrand metered light differs dramatically from other devices.13:35
buZzyeah there's no source package for the kernels i think :(13:36
buZzwish we did13:36
uvos__norayr: just grab it off git13:51
uvos__since we have quite a few devices that are stuck with gprs/edge in most places now14:06
uvos__maybe we should run a compression proxy like https://github.com/barnacs/compy14:07
buZzssh -C works too :P14:11
uvos__on https webpages?14:20
uvos__not really14:21
buZzhehe, no14:21
buZz:P14:21
buZzbut dont almost all webservers do onthefly gzip nowadays?14:21
uvos__no all, also the greates benefit of compy is that it reencodes the images14:22
uvos__to make them smaller14:22
uvos__this is huge saveings14:22
uvos__for little loss on a phone sized screen14:22
WizzupbuZz: there are source packages for kernels14:32
buZzWizzup: see norayr just now?14:44
norayrbut how the kernel source package is named for droid4?14:48
norayrit said Picking 'omap-linux' as source package instead of 'linux-image-omap'14:49
norayrE: Unable to find a source package for omap-linux14:49
norayri'll take this: https://github.com/maemo-leste/droid4-linux.git14:59
uvos__norayr: maemo-5.18.y branch14:59
norayr5.18.19 tag15:00
norayrwill try to not rebuild everything, but only module. not sure how.15:00
freemangordonno way, you need the whole kernel15:00
uvos__why, you can for sure buid just one module15:00
Wizzupgit checkout https://github.com/maemo-leste/droid4-linux.git -b 5.18.19 --depth 115:00
freemangordonor, you can get kernel headers and build as out-of-tree module15:01
freemangordonuvos__: no, because of the module versioning15:01
norayri want to rmmod the module, and then...15:01
norayrthank you, Wizzup!15:01
freemangordonwell, if he copies debian build scripts etc15:01
freemangordonbut then again, it is easier to just build the whole kernel, copy to device and start playing15:02
uvos__sure15:02
freemangordonwell, at least for me is easier :)15:02
norayri don't understand, debian build scripts will help me to build the whole kernel, not only the module.15:02
norayrok let's see.15:02
uvos__but its hardly impossible to build one module for the installed kernel15:03
norayrWizzup: that's not the branch 5.18.19, i guess, that's tag.15:03
freemangordondebian build tree will help you to build the correct kernel symbols version15:03
uvos__-b command takes tags too15:03
uvos__despite the name15:03
uvos__@norayr15:03
freemangordonif you know what you are doing15:03
norayri don't (:15:03
freemangordonme neither, that's why I am building the whole kernel :)15:03
norayrlet's see, i'll try.15:04
uvos__so regarding the als units15:05
uvos__the cange is trival to do15:05
uvos__but idk if the mapphones are off because its a bug in the kernel with what units it presents on sysfs15:05
uvos__or if the mapphones have a variant of the als chip15:05
uvos__thats different15:05
uvos__in case 1 you can just change the scale file in sysfs15:05
uvos__in the other case you would need to add a dts parameter15:06
uvos__to know what it is you would need to have the chip in some other form than just in a mapphone15:06
uvos__so i decided not to deal with this15:06
norayri'll show you the fix by my friend, i'll understand it better when i apply it.15:07
norayrnot sure it'll change the behaviour of als.15:08
norayrbut let's see.15:08
buZzimho, would be easier if we had kernel source packages :P15:11
norayryes. of course.15:11
norayri was sure we have. (:15:11
norayrshould i also take droid4 config from git?15:12
norayrah, i'll take it from /proc/config.gz15:12
freemangordonomap2plus_defconfig, no?15:12
norayri don't know? there are configs for devices on git.15:13
freemangordonuvos__: Wizzup: we are using omap2plus_defconfig, right?15:13
uvos__freemangordon: yes15:14
uvos__but config.gz will work fine too15:14
uvos__ofc15:14
* DocScrutinizer pokes buZz16:21
buZzhiya :)16:23
buZz*sigh* ok, i think i found why my .deb keeps having nothing inside16:24
buZzthe 'make install' on this project does nothing -_-16:24
buZzguess i'll just have to add a pile of cp commands somewhere16:24
WizzupbuZz: does it not use cmake or autotools?16:28
Wizzupmaybe look how many of the other debian/rules files look16:29
buZzit uses cmake, but somehow it doesnt have any installation files16:29
buZzi did find this https://github.com/diasurgical/devilutionX/blob/master/CMakeLists.txt#L51916:29
Wizzupwhat are 'installation files' ?16:30
buZzi mean , 'make install' after running cmake doesnt copy any compiled binary to anywhere16:31
Wizzupfreemangordon: ok, I have a lime2 with 7" ts and 800x480 screen working16:38
WizzupI had to hack a bit on our sunxi image and copy some olimex files, but it's working16:38
buZzcool!16:40
buZzdecent speed?16:40
Wizzupgimme a bit16:43
buZzall the bits16:44
Wizzupneeded to figure out net.ifnames=0 :)16:50
rafael2knet.ifnames=0 ❤16:52
Wizzupyes16:57
freemangordonI guess I will have to buy another uSD card to resurrect my allwinner17:39
Wizzupwhich allwinner is this?17:53
* Wizzup just bought a few microsd cards17:53
buZzso cheap nowadays :)18:06
buZzhttps://www.dataio.nl/samsung-evo-plus-128gb-micro-sd-geheugenkaart-100mbs/18:06
Wizzupfreemangordon: do you remember if we needed any kernel patches for some of the flickering of lima to go away?18:09
buZzfreemangordon: just to confirm, indeed, 144p playback was bad, reboot, and it was smooth18:11
freemangordonWizzup: q8 a33 is all I remember18:14
Wizzupright the plesio one18:15
WizzupI have a lime2 here showing 4g connectivity :)18:15
freemangordonlemme check in android18:15
Wizzupfixing gps atm for the demo18:15
Wizzupthe raspi also works with 3d and touchscreen,wifi etc18:16
freemangordonnice :)18:17
norayri also feel that charging does something bad to droid4.  i think it is already better than a month ago.18:18
freemangordonnorayr: there was a pile of fixes/improvements top xorg driver18:19
norayryes, i was seeing something is happening. (:18:19
freemangordonWizzup: Q8H_HD18:26
freemangordonI remember I had issues with TS18:29
freemangordonhad to copy firmware or something18:29
Wizzupprobably dts needs updates too for newer kernels18:33
freemangordonmhm18:37
Wizzupit seems that mce sometimes locks the screen even when lock is disabled18:41
freemangordonWizzup: BTW, most-probably the flickering in lima was because our clutter was missing buffer_age support19:02
freemangordonWizzup: if I pull image builder, is it supposed to work?19:10
freemangordonor I have to create some special VM?19:10
Wizzupfreemangordon: pull can work, but I loathe to use it in general19:11
Wizzupfreemangordon: I made a sunxi image today19:11
Wizzupbut it does not have kernel or u-boot19:11
Wizzupbut everything else is there19:11
Wizzup:)19:11
Wizzuphttps://maedevu.maemo.org/images/sunxi/19:12
freemangordonwell, I wonder how iseful it would be without kernel/uboot :)19:12
freemangordon*useful19:12
Wizzupwell, you will have your own u-boot and kernel19:12
Wizzupwon't you?19:12
freemangordonyes, I will19:12
Wizzupthen :)19:13
freemangordonbut isn't it better to have at least kernel?19:13
freemangordonlike, we have different DTBs, but kernel should be the same, no?19:13
Wizzupyes, but I am out of time to dedicate to this and the image builder kernel was -broken-19:13
freemangordonah19:13
Wizzupso I just disabled it so I could at least get base image19:13
freemangordonugh, u-boot won;t even build on my ubuntu :(19:17
Wizzupyou might be able to use the old u-boot if you still have the sd card of image19:17
freemangordonI still have the tree :)19:18
freemangordonjust pulled, but stored sha id before the pull ;)19:18
Wizzup:)19:19
freemangordonnow I have to find boot.scr and mkimage parameters19:20
freemangordonnot that I have sd card to test with :)19:20
WizzupI don't have 3d on the rpi yet, but given that it is a rpi4 and the screen is 800x480, llvmpipe seems to be ok :)19:23
freemangordonWizzup: maybe I can pull image builder and see why kernel is broken19:25
freemangordonbut I will have to setup some recent ubuntu VM19:26
freemangordonor maybe I can use leste?19:26
Wizzupall of those can work19:29
freemangordonWizzup: linux-5.11.y19:38
freemangordonisn;t that too old?19:38
Wizzupyes.19:38
Wizzupthat too19:38
freemangordonhmm, how to build for maemo not for devuan?19:40
uvosWizzup: "it seems that mce sometimes locks the screen even when lock is disabled" what makes you say that?19:59
WizzupI observe it20:01
Wizzupit's probably what you discussed with fmg earlier20:01
Wizzupit's particularly obvious on both the lime and the raspi since they don't have a power button20:02
Wizzupso I can't unlock unless I ssh in and restart mce20:02
Wizzupit only happens once after boot, mind you20:02
uvosit seams to work fine here20:03
uvoswhat are you seeing exatcly20:03
Wizzupthe screen turns off and touching it does not make the screen turn on20:03
Wizzupand the option to lock the screen is explicitly disabled20:04
uvosok20:04
freemangordonand to add to that - mce locks screen on power-up even when keyboard is open20:04
uvosdosent happen here so its probubly something in lock-tklock or lack of https://github.com/maemo-leste/osso-applet-display/pull/220:04
uvosfreemangordon: mce cant know if the keyboard is open on power up20:05
uvosunfortionatly20:05
freemangordonhow's that?20:05
uvosin the fremantle kernel20:05
uvosthere was a hack that allowed it to read that key via sysfs20:05
uvosbut we just have evdev20:05
uvosthat only creates an event when the slide state changes20:06
freemangordonoh20:06
uvosso mce has no idea about the slide state untill it moves once20:06
freemangordonthats....20:06
uvosit assumes closed20:06
freemangordonI am speechless :)20:06
uvosbut fear not20:06
uvosthere is some way to read the key state on startup in the mainline kernel20:06
uvosevtest manages somehow :)20:07
freemangordon:)20:07
freemangordonok, then we just need to find it20:07
Wizzupbtw I'm planning to show off a few devices at openfest and have them xmpp chat with each other20:09
Wizzuphard on devices with no hwkeyboard but eh20:09
uvosbring bluetooth keyboards maybe20:09
freemangordonI have one USB, from the tablet :)20:10
Wizzupthe raspi and lime2 have a usb port20:10
Wizzup;)20:10
WizzupIf I ever finished the qt5 im it would not be a problem hehe20:10
uvosyeah its lack of https://github.com/maemo-leste/osso-applet-display/pull/220:18
uvos@bug Wizzup20:18
uvosbut thats still a bug, the gconf key should not have moved in the transition to mce-rtconf, gesttings comapt work, but it probubly changed its name slightly causing this bug20:19
uvoshttps://github.com/maemo-leste/mce/blob/b1954307d70f7013d6da8104c2b629e491e58af1/src/modules/rtconf-gconf.c#L3520:22
uvosprobubly need a special case for this key here20:22
uvosit probubly ends up in /system/osso/dsm/display/ but should go in /system/osso/dsm/locks/20:22
uvosoh wait there is a special case for it allready20:23
uvoshmm20:23
uvosmaybe touchscreen_keypad_autolock_enabled is not the right match, ill check it out some time soon.20:23
uvosin the meantime the de-gconfig-ing of mces interface users would be good to merge anyhow: https://github.com/maemo-leste/osso-applet-display/pull/2, https://github.com/maemo-leste/osso-applet-notificationlight/pull/1, https://github.com/maemo-leste/hildon-desktop/pull/1920:26
Wizzupuvos: ty @ check out20:33
WizzupI don't want to introduce bugs in the next 2 days :D20:36
* uvos merges mce dev branch into master and builds20:41
Wizzup:D20:44

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