libera/#maemo-leste/ Sunday, 2022-04-03

Wizzupuvos: I see yourt new patches for hp detect using new apis, mabye we should go for new kernel in -devel ?09:23
Wizzupbtw, if any anyone has any idea what is failing in parazyd's arm-sdk or image-builder, I'd appreciate it: https://phoenix.maemo.org/view/Images/job/leste-image-bionic/lastFailedBuild/console09:59
Wizzupall image builds seem to fail like this:09:59
WizzupI: Base system installed successfully.09:59
Wizzupblend_bootstrap_setup:3: number expected09:59
Wizzupseems to maybe be this?     [[ -n "$armsdk_version" ]] && req +=(device_name)10:02
Wizzupmaybe it's related to the key expiry somehow10:03
Wizzuptrying now10:08
Wizzuplooks like that was it10:15
Wizzupgood10:21
freemangordon:)10:26
Wizzupfreemangordon: which pkgs do I install to check out the rtcom work?10:28
Wizzuprtcom-accounts-ui ?10:29
Wizzupand librtcom-accounts-ui-client0 ?10:29
Wizzupah, the my information was from abook10:34
freemangordonumm, yes10:35
freemangordonotherwise installing rtcom-accounts-ui  will pull everything needed (in theory)10:35
Wizzupwhat should I see once I install that10:37
Wizzupno cpa plugin, right?10:37
freemangordona new entry in cpl10:37
freemangordonyes, there is10:37
freemangordon"voip and im accounts"10:37
Wizzupok10:38
WizzupI see it now10:38
WizzupI guess I was so used to it on my n90010:38
Wizzup:)10:38
Wizzupit shows an empty wizard atm10:38
Wizzupas expected I guess10:38
freemangordonyeah, I know10:38
freemangordonmhm10:38
Wizzup:)10:38
Wizzupman, so much changed for the news post, going to take some hours to finish this10:39
WizzupI'll make a pdf out of it when I am done and invite some comment10:40
Wizzup(I won't push it to a hidden url like last time, since people picked it up with rss)10:40
freemangordon:)10:40
WizzupErr:1 https://maedevu.maemo.org/leste beowulf InRelease Temporary failure resolving 'maedevu.maemo.org'10:49
Wizzup*sigh*10:49
bencohDNS seems fine from here (?)10:54
Wizzupbencoh: yeah it is the temporary failure in the image build (which also doesn't cause a fatal error) that's annoying11:01
Wizzupthere are some things in arm-sdk that just don't cause fatal errors and it's really annoying11:01
WizzupI don't have the zsh knowhow to fix it11:01
bencohis there a reason why it would temporarily fail to resolve though?11:08
Wizzupno idea11:08
bencohah11:08
bencohyou could add set -e at the top of the script btw11:09
bencohyou'll get a lot of false positives at first, but at least it should catch everything11:09
Wizzupdoes that work with zsh?11:09
bencohhmm11:10
bencoh(damn, bash: zsh: command not found :D lemme check)11:10
bencohlooks like it works, yeah11:11
uvosWizzup: the newer code is only different in implementation, functionally its the same12:04
uvosWizzup: but yes we should update the kernel as allways12:04
Wizzupmhm12:06
lelMerlijnWajer created a repository: https://github.com/maemo-leste-extras/cssufeatures12:24
uvoswhy do we have this terrible hack in af-services that sets the dbus session bus to be the one thats owned by the user user even in other users?12:39
uvosthis breaks quite a few things12:39
Wizzupprobably because without it, everything breaks :)12:41
uvosright but what is everything12:41
uvosbecause this is really broken12:41
Wizzupwhat does it break for you?12:42
uvosit breaks pulse if not in system mode12:42
uvosand it breaks gsettings12:42
uvosgesettings particually is impossible to use with this hack12:42
uvosin a fairly sublte way:12:42
uvosglib reads dconf keys directly, only writing goes thugh the deamon via dbus.12:43
uvosso with this hack12:43
uvosany application using gesettings writes values to the store of the user user12:43
uvosbut reads its settings from whatever user its running as12:43
Wizzupwhat other user do you run the apps as, if not 'user'?12:44
uvoswell lots of stuff runs as root, plenty of stuff also drops privilages to nobody or whatever12:45
uvosexperiamenting: it also breaks all kde applications12:46
uvosthey read settings from ~ ini files12:47
uvosand inform eatch other of settings changes via dbus12:47
Wizzupbencoh: looks like the dns problem might be a real problem in the image generation12:47
Wizzupuvos: and they don't get the dbus address somehow?12:48
uvosthey get get the bus from the evvars12:48
uvoswhich af services sets to user12:48
Wizzupso how does that break them?12:49
WizzupI don't understand12:49
uvosthey read from ~/.config12:49
uvosbut then they inform eatch other via dbus12:49
Wizzupwhat insane behaviour12:49
uvosgesettings works the same12:49
uvosits not really insane12:49
Wizzupprobably to work around slowness in dbus?12:49
uvosno because in kde there is no central settings repo12:50
uvosso eg kwrite owns its ini file12:50
Wizzupit looks like the dns problem shouldn't be too serious at least12:50
uvosand if some other app wants to know when the settings changes12:50
uvosit registers a dbus listener12:50
uvosfor gsettings yeah i think its for paralellisum12:51
uvosanyhow it also breaks pulse12:51
uvosbtw osso-af-startup is extream mess in other ways too12:53
uvosit dose lots of stuff thats not applicable and some stuff thats damageing12:53
sicelok-apps :-p13:53
Wizzup+ ssh amprolla@maedevu.maemo.org -- mkdir -p images/droid4/2022040314:17
WizzupHost key verification failed.14:17
Wizzup+ exit 114:17
Wizzupoh well14:17
Wizzupwill fix that in a little bit14:17
Wizzupsicelo and others, I pushed some wip news post here https://github.com/maemo-leste/maemo-leste.github.io/blob/source/content/maemo-leste-update-april-2022.rst14:31
uvoswhat happend with parazyd anyhow?14:50
uvosnot to pry14:50
uvosidk about stability wrt omap drivers :P14:51
mighty17[m]freemangordon: http://muru.com/linux/d4/0001-egl-dri2-Workaround-for-EGL_EXT_image_dma_buf_import.patch going thru logs, what is this ?14:51
freemangordonmighty17[m]: ummm.... "Signed-off-by: Tony Lindgren <tony@atomide.com>"14:59
mighty17[m]freemangordon: oh my bad, also why did you revert https://github.com/maemo-leste-upstream-forks/mesa/commit/a673ae59b0edd281a20060080554f96331978aa314:59
freemangordonbecause of "if (dri2_dpy->image->base.version >= 8"15:00
freemangordonwell, because at that time it was working without that patch15:01
freemangordonbecause tehre are more checks nad this patch alone is just a hack15:02
freemangordonbut later it got broken by upstream mesa15:02
freemangordonso now it insists on having that extension which was not like that back then15:03
mighty17[m]<mighty17[m]> "freemangordon: http://muru.com/..." <- wouldnt this work here15:04
mighty17[m]` if (dri2_dpy->image->base.version >= 15 &&`15:04
freemangordonsorry, I don;t understand what you are asking15:05
mighty17[m]freemangordon: now it can be applied (as a hack) and it should fix it....?15:05
freemangordonmaybe15:05
freemangordonbut rather not15:05
mighty17[m]freemangordon: tmlind's patch.. instead of reverting why not apply that15:05
freemangordonbecasue it is a hack too15:05
freemangordonthe correct thing is to implement callback in PVR driver to return the supported buffer formats15:06
mighty17[m]oh :(15:06
freemangordonotherwise the applications will just get an emopty list15:06
freemangordon*empty15:06
freemangordonor incorrect15:06
mighty17[m]or we just tell applications we dont support it?15:06
freemangordonwe cannot tell that as it is mandated by mesa15:06
freemangordonwe must support it15:07
mighty17[m]but our blobs dont do it right...15:07
freemangordonblobs? why blobs?15:07
mighty17[m]powervr...?15:07
freemangordonit is not about the blobs15:08
freemangordonit is about mesa pvr driver15:08
freemangordonit must implement that missing piece, even by returning hardcoded data15:08
mighty17[m]oh this is pretty complicated for me :o15:08
freemangordonsec15:09
mighty17[m]freemangordon: so basically, what pvr mesa we use (from chromium) is so old it doesnt support that extension15:09
freemangordonyes15:09
freemangordonsee https://github.com/maemo-leste/sgx-ddk-um/blob/master/dbm/dbm.c#L47615:10
freemangordonthis is REed pvr blobs libdbm15:10
mighty17[m]but https://github.com/maemo-leste-upstream-forks/mesa/commit/6719e058d6b8a38cc66accf1609fcabb66571e86 was added in 18.x and chromium mesa was 19.x?15:10
freemangordonso?15:10
freemangordonthe driver itself implements v815:10
freemangordonof base_image15:11
mighty17[m]freemangordon: oh wait when did this happen :o15:11
freemangordonumm, what you mean?15:11
freemangordonsee the commit log15:11
mighty17[m]ie RE of sgx-ddk-um :P15:11
freemangordonsee the commit log15:11
freemangordononly this is lib REed15:12
mighty17[m]right right, so we gotta implement v9 and so on?15:12
freemangordonno, we need to implement queryDmaBufFormats (IIRC)15:12
freemangordonbut it was a while, it could have been something else we must implement15:13
freemangordonit is just one function15:13
mighty17[m]`.queryDmaBufFormats= PVRDRIQueryDmaBufFormats,` we already have it... maybe something else15:14
freemangordonyeah, we must implement eglQueryDmaBufFormats as now it returns NULL formats15:14
freemangordonyeah, maybe something else, I don;t remember15:14
mighty17[m]yes correct15:14
mighty17[m]https://github.com/xc-racer99/mesa-pvr/blob/mesa-20.3.2-pvr-musl-2/src/mesa/drivers/dri/pvr/pvrext.c#L224 xc-racer has it15:15
freemangordonbut it calls in the blobs15:16
freemangordonna dthis is not supported in our blobs15:16
mighty17[m]riight i am getting it slowly15:16
freemangordon*and this15:16
freemangordonso we return EGL_SUCCESS and empty list15:16
freemangordonor somesuch15:16
mighty17[m]thats a hack :P15:16
mighty17[m]freemangordon: our blobs ie the one you RE'd?15:16
freemangordonno15:17
freemangordonthe real blobc15:17
freemangordonpvr_dri_support.so15:17
freemangordon(IIRC)15:17
mighty17[m]so his mesa is broken?15:17
freemangordonit doesn't have the needed function exported15:17
freemangordonno idea15:17
mighty17[m]im confused, if blobs dont support it why'd it be in mesa15:17
freemangordonbecasue this is crhomioumos mesa15:18
freemangordonfor different driver15:18
freemangordonnot for ours15:18
freemangordonthats why I REed pvr_dri.so that came with our blobs15:19
mighty17[m]ohhh damn15:19
mighty17[m]freemangordon: so atleast we know what func we have :D15:19
freemangordonso what we have in *our* mesa is exactly what comes with DDK 1.17 blobs15:19
mighty17[m]so basically we need to implement what chromiumos mesa had but with funcs in our blobs15:19
freemangordonexactly15:19
freemangordonor just hardcode15:20
Wizzupuvos: @ what happened, your guess is a good as mine15:20
uvosWizzup: ok15:20
* freemangordon is afk15:20
uvosWizzup: newspost looks correct otherwise15:20
mighty17[m]freemangordon: oooh thanks for the guide!15:21
Wizzupin another month or so I'll remove his ssh keys15:21
Wizzupuvos: ok, if you have more things you think should be added lmk15:21
uvossalutem maybe?15:21
uvosmaybe distobute the newspost via salutem :P15:22
Wizzupwas thinking about that15:27
Wizzupyeah15:27
siceloI can somewhat confirm parazyd is fine where he is. I dropped him a message on Facebook a couple of weeks ago15:29
bencohsicelo: oh, that's a good start, if you received some kind of answer15:40
siceloWhen charging, wouldn't it be a good idea to use a different LED color while charging and when full? Anything that prevents this? (Referring to Droid 4)18:45
Wizzupdroid4 has the led override still18:48
Wizzupso it always shows the green one18:49
uvoscharge led is forced by hardware19:15
uvostheres a register bit thats called disable_charge_led in moto kernel sources19:16
uvosbut changing it stops charging entirely for some resaon19:16
uvosso maybe its misslabled19:16
uvosclearly charging with the led is disabled is possible, as android dose so, so really you would just have to diff the registers between android and mainline chargeing19:17
uvosshould not be hard to figure out, but its not a priority19:17
mighty17[m]freemangordon: my try on adding support https://paste.debian.net/1236679/ (old code in comments to refer) basically we do not have `bool *pbHasFormat` and `int iNumFormats;` anymore after your RE, using intel as reference as well https://github.com/xc-racer99/mesa-pvr/blob/mesa-20.3.2-pvr-musl-2/src/mesa/drivers/dri/i965/intel_screen.c#L1348 which basically sets `int *formats` and `int *count` according to the supported extensions20:22
mighty17[m](intel_image_formats/g_asFormats for us)20:22
freemangordonmighty17[m]: sorry, cannot help further without spending some time I don;t have now20:23
mighty17[m]oh alrighty, feel free to have a look once you're free and ping me, its my first time doing it so 😅20:26
freemangordonwell, better do not wait for me, I am not sure I will have time for that soon20:27
mighty17[m]i am unsure if what im doing is correct anyways :P20:33

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