libera/#maemo-leste/ Wednesday, 2022-07-06

Wizzupuvos: the meta pkg should fix that00:08
uvosWizzup: yeah00:09
uvosthe meta package was uninstalled on all of my devices - again00:09
uvosi never do this intentionally00:09
Wizzupit can't be removed with apt upgrade for sure00:10
uvosxt875 and xt894 boot fine with the packaged kernel00:15
uvosxt860 dosent00:15
uvosnever gets anywhere (no serial out beyond stuff from mbm/stock kernel)00:16
Wizzupyou probably lack the ram thing00:45
Wizzup(or not, I have to check)00:47
Wizzupthere is the &memory0 thing00:48
uvoshttps://github.com/IMbackK/droid4-linux/commit/61e0af838435d02fb651671249c9bd4208384f8e00:49
uvoshttps://github.com/IMbackK/droid4-linux/commits/droid3-51600:49
Wizzuphmm00:53
WizzupI will have to try tomorrow00:54
Wizzupuvos: ok, the n900 patch makes the n900 boot and work01:25
Wizzupts, wifi, etc01:25
Wizzupuvos__: I also get the oops on the n900 when pm is enabled and I reboo01:55
Wizzupt01:55
Wizzupuvos: I think the droid3 just boots for my on new kernel03:20
Wizzupit has the usual freezes though03:20
uvos__devel time for the kernel then08:07
uvos__did we complain on linux-omap about the oops allready08:08
uvos__seams to happen on any omap then with the serial ports suspended08:08
tmlinduvos__: please do, oops trace and how to reproduce easily would be great to debug this08:40
tmlindi tried to reproduce reboot with serial ports idled but did not see any oopses in logs, i can see reboot occasionally fail sitting in some stuck state possibly in the bootloader08:42
Wizzupuvos__: I also think I see better battery life10:33
Wizzupuvos__: wait, ok, maybe I didn't properly test 5.18 on droid310:41
uvos__tmlind: ok will do!10:54
tmlinduvos__: thanks!11:22
Wizzuproot@devuan-droid3:~# uname -a12:06
WizzupLinux devuan-droid3 5.18.9 #1 SMP PREEMPT Tue Jul 5 19:41:44 UTC 2022 armv7l GNU/Linux12:06
Wizzupuvos__: works for me, but still loads of freezes/resets12:06
Wizzupand also a trace from omapdrm12:07
uvos__ok12:10
Wizzupalso interesting:12:10
Wizzup[  298.897949] phy-mapphone-mdm6600 usb-phy@1: modem status: 4 awake12:10
Wizzup[  301.089935] phy-mapphone-mdm6600 usb-phy@1: modem status: 5 asleep12:10
Wizzup[  304.528991] phy-mapphone-mdm6600 usb-phy@1: modem status: 4 awake12:10
Wizzup[  306.729827] phy-mapphone-mdm6600 usb-phy@1: modem status: 5 asleep12:10
uvos__interesting12:10
uvos__modem dosent work on d3 no?12:10
uvos__anyhow thats good enough to push to devel or?12:11
uvos__maybe post the trace somewhere12:11
Wizzupyeah I don't think it works12:11
uvos__and yeah my d3 is also super unstable12:11
uvos__on 5.1512:11
uvos__never lives more than a couple of min12:11
Wizzuphttps://paste.villavu.com/raw/Jy5yAiusGk6GR7PEBuGJGI5nj4LgAM4Evv5PbSfe/12:12
Wizzupit lives if you leave the screen off12:12
uvos__(active use, it idles ok )12:12
Wizzupyeah12:12
uvos__it also gets really hot12:12
uvos__like concerningly hot12:12
uvos__there where the sim tray is12:12
Wizzupprobably modem related12:13
Wizzupso there's the pmic voltage warnings12:13
Wizzupbut you said they were harmless in the past12:13
uvos__those are no real problem12:13
uvos__this just means that mbm sets the voltages higher12:14
uvos__than on d412:14
Wizzuphere's this:12:14
Wizzup[   26.306915] asoc-audio-graph-card soundcard: ASoC: DAPM unknown pin Headphones12:14
uvos__that makese sense12:14
uvos__since its a older variant of the chip12:14
Wizzupbut also not a big deal supposedly12:14
uvos__and we can assume mbm sets the voltages ok12:14
Wizzupthis is modem related:12:14
Wizzup[   23.519958] phy-mapphone-mdm6600 usb-phy@1: Timed out powering up12:14
uvos__so we just need to up the acceptable range in the kenrel12:14
Wizzupbut I think the trace is here:12:15
Wizzup[   13.100494] ------------[ cut here ]------------12:15
Wizzup[   13.101226] WARNING: CPU: 1 PID: 60 at drivers/gpu/drm/drm_bridge.c:1207 drm_bridge_hpd_enable+0x74/0x84 [drm]12:15
Wizzup[   13.116180] Hot plug detection already enabled12:15
Wizzupthere are also errors for emif_probe and for omap-aes12:15
WizzupI don't know how relevant those are12:15
Wizzupit looks like the modem status awake/asleep comes from me turning the screen on/off12:16
uvos__we dont use emif12:16
uvos__unless tmlind added it recently12:16
Wizzupprobably the mce mapphone module12:16
uvos__im suprized it tires to probe12:16
Wizzupin any case yeah the modem doesn't work yet afaik12:16
uvos__tmlind: comment? ^^12:16
uvos__afaik adding the emif binigns to dts breaks on mapphones for unkown reasons12:17
uvos__so we just leave it alone to how mbm/stock kernel sets it up12:17
uvos__ah d4 also has the omapdrm trace12:18
Wizzupoh yeah, there's also this:12:18
Wizzup[    9.597930] abe_cm:clk:0060:0: failed to enable12:18
uvos__the drm trace sounds not very serious but we should check if it happens on mainline12:18
Wizzupright12:18
uvos__and report if so12:18
WizzupI'm not going to look very hard at the d3 atm now, but will/can later12:19
uvos__btw12:20
uvos__actually i imagined seeing the ants with the 5.1812:20
uvos__now that im payin12:20
Wizzupyeah, things look a bit different12:23
Wizzupfirst time that a kernel upgrade yields free improvements huh ;)12:23
uvos__*attention it dosent seem to happen12:24
Wizzupnot sure what others think of this, but I think we have two gpio's in the back of most mapphones13:27
WizzupI think on the bionic we could use that for a keyboard, perhaps even use gpio-i2c or something13:28
Wizzupi2c-gpio is what I mean*13:31
mglbg[m]for the n900, what do i do if i want to turn the screen on when slide is open and off when slide closed? dont mind writing some bash, I just cant find where to look, older forum posts refer to unavailable /sys/devices/platform/slide... kind of locations. any ideas?14:17
mglbg[m]i guess i could get some examples from the keyboard backlight, but the search was unfruitful for now14:18
buZzmglbg[m]: yeah i want similar on d4 aswell :)14:45
sicelo_maybe something in mce config? at least my fremantle n900 works like that. i don't think scripts are the correct solution14:48
sicelo_if uvos__ doesn't have solution for you right away, submit a github issue. for script, you could unlock the device via dbus, and your script could be activated via evdev ... i think the sliders are visible there on both n900 and d414:52
mglbg[m]how do i reach the mce config? meh this sounds a bit too far for me now sadly as i do not know what you are talking about .. looks like i need to be signing up for github..14:54
sicelo_ /etc/mce/mce.ini15:00
mglbg[m]thamks, ill check it out15:08
sicelo_ so you're both saying sliding keyboard up doesn't unlock the device in the default config? if so, it could be a regression, or our wiki is wrong and needs updating. it currently states "Additionally, if your device has sliding keyboard (like the Nokia_N900 and Motorola_Droid_4), then opening the keyboard will also unlock the device. Closing the keyboard will lock the device again, but only if you15:09
sicelo_have not interacted with the screen since the last slide up."15:10
sicelo_ i might be wrong, but i think you should get the desired behavior by setting AutolockWhenSlideOpen to 0 in /etc/mce/mce.ini15:23
sicelo_ah ..  yes i don't know mce anymore as it has undergone a lot of changes.15:25
sicelo_buZz: mglbg[m]: set "UnlockOnSlide" = 1 in /etc/mce/mce.ini. see https://github.com/maemo-leste/mce/blob/4ee0792ee09aa7d3638fc5129532cb0b84ddeddf/config/mce.ini#L13115:25
uvos__this still works15:51
mglbg[m]thanks guys, i tried the UnlockOnSlide in the .d/99- file, does not seem to work yet, trying to figure out if this is due to this setting or some other system changes15:52
uvos__mglbg[m]: that setting is int the lockGeneric section or?15:52
uvos__mglbg[m]: that only applies if you use lock generic15:52
uvos__mglbg[m]: so tklock allways unlocks on slide open15:53
uvos__mglbg[m]: if you want the device to allso allways lock on slide closed you can use lock generic for this15:53
uvos__if for some reason it dosent unlock when you slide it open with the default config that is a bug15:53
uvos__but i kinda doubt it since eveytone would have complained about that15:54
uvos__(and it still works here)15:54
mglbg[m]yeah unlock no problem it is about turning off the sreen when sliding in16:04
buZzyeah 'locking' when closing, thats something maemo 5 did on n900 quite well16:06
mglbg[m]ok, ill check out the tklock16:06
buZzbut havent had similar on maemo 7 yet16:06
bencohbuZz: "locking" when closing only happened if no event (keyboard/touchscreen) happened meanwhile16:10
bencoh(i.e slide open, have a quick look at the screen without touching anything, slide close)16:10
buZzindeed, but on maemo7 thats not happening16:10
bencoh(on maemo5 I mean)16:10
buZzmaybe it always thinks something happened?16:10
Wizzupuvos__: yeah I think we can move to -devel today16:14
bencohbuZz: could be yeah16:16
uvoslock on close is an issue with the button layout on d416:44
uvosyou cant have it lock on close if nothing was pressed and have the volume buttons work16:44
uvosas the volume buttons and the lockslider are on the same input device on d416:44
uvosso yes that dosent work and is a known issue16:44
uvosallways locking on closed is something lock generic can do16:44
uvoshttp://uvos.xyz/maserati/90-lock-on-slide.ini16:47
uvoshere you can put that into /etc/mce.ini.d if you want this behavior16:47
sicelo_< uvos> lock on close is an issue with the button layout on d4 <-- does it work at least for n900?16:48
uvosno its disabled everywhere to avoid filtering away the volume buttons on d416:49
uvosi need to rewrite mces intput pipeline to make it work again16:49
buZzahh17:17
rafael2khey, can anyone remind me again how to set a volume key to open the vkb?18:05
rafael2kit is in transitions.ini if I remember18:05
Wizzupdamnit I have a pulse dial phone at home now18:08
Wizzupbut I can't send '#', so I can't dial into work meetings18:08
buZz-pulse- phone :D18:09
buZzlol, in .bg again Wizzup ? :D18:09
Wizzupnope, amsteradm18:10
Wizzupit has a nice sticker on it with feuerwehr and polizei numbers ;)18:11
buZz:D18:15
rafael2kyou can just play the tones18:19
rafael2kwith an outsiđe speaker18:19
rafael2kcoupled with the phone18:19
Wizzupnot sure if the grandstream ata will turn the pulse dial into dtmf in the call too18:19
WizzupI'll have to check18:19
Wizzupanyway offtopic :p18:20
Wizzupbut fun18:20
rafael2k941 Hz + 1477 Hz18:20
rafael2kcan use csound and maemo, and put the speaker close to the phone18:20
rafael2kpulse is pulse, not DTMF18:21
rafael2kjust play the tones from any other device18:21
rafael2k:P18:21
Wizzupoh heh18:22
rafael2kI used this in a "locked" (with key) phone in the past... you can just tap the hook to create de pulses - but not to create DTMF... ehehehhe18:23
Wizzupso you're suggesting to generate the dtmf tones on my end?18:23
rafael2kATDT vs ATDP18:24
rafael2k:P18:24
Wizzupthe numbers as dtmf tones?18:24
rafael2kbetter not to mix I think18:24
rafael2kyou can try18:24
rafael2konce the call is established via pulse, you can certainly switch to DTMF later18:24
rafael2kwhich will be heard by a third party PABX18:25
Wizzupright18:26
Wizzupthat's what I mean18:26
rafael2kright, it will work, indeed18:26
rafael2kyou can get an acoustic coupler and feel like in the early 80's18:27
rafael2k:P18:27
Wizzupheh18:27
rafael2ksorry, this is mid 70's18:28
rafael2kyou will also get for free a 1200 baud (software) modem if you wish to18:30
Wizzuphehe18:43
buZzso is 5.18.latest on hold?19:24
WizzupbuZz: no19:26
Wizzupit works in -experimental19:26
buZzooo19:26
WizzupI will push to -devel today19:26
buZzsweeeet19:26
buZzi'll port my bermbom mods :P19:26
mglbg[m]<uvos> "http://uvos.xyz/maserati/90-lock..." <- thanks! this is one step further, the display goes off, but the led starts growing brighter white and the screen pops on again...19:28
uvosmglbg[m]: heh it actually suffers from the same issue as autolock19:49
uvossince the d4 now accepts the slider key as a valid event19:49
uvosclosing the slider also wakes it again19:49
uvosso yeah19:49
uvosthats also broken for the same reason the other lock on close is broken19:49
rafael2kbtw, a sample took with the gui app qcam (in libcamera-tools): https://www.abradig.org.br/maemo-crazyness/test.jpg20:47
rafael2kdebayering is really fucked up20:48
rafael2ka guy developing for Mobian, Pavel, is also working on the PP (on Mobian)20:49
rafael2kHe also owns a Droid 4 and a N900 (and may be is even here, I dunno)20:50
Wizzupmachek?20:51
rafael2kyes20:53
rafael2kdo you know him?20:53
Wizzupyup20:53
Wizzuphe toyed a lot with maemo on the d4 and n90020:53
Wizzupbut he's all over the place :)20:53
rafael2kright... will not ask nothing about him not to jeopardize the work... ehehhehe20:54
rafael2khe adds some funny signature is his emails...20:54
rafael2kI better not comment20:55
Wizzupsomething about horses?20:56
rafael2kyes20:56
rafael2k(metaphorically yes)20:56
rafael2kbumped with him in libcamera mailing list20:57
sicelo_https://www.youtube.com/watch?v=fH6zuK2OOVU if you want to watch his talk on phone cameras and linux (before libcamera was a thing, iirc)20:59
Wizzupuvos: in -devel20:59
Wizzupsicelo_: maybe also check new kernel on d420:59
sicelo_rafael2k: ^20:59
Wizzupthings feels different and better20:59
Wizzupit's weird21:00
sicelo_Wizzup: i'll see if i can find time for it21:00
rafael2ksicelo_: ow21:01
rafael2kso no, I though you was joking about horses... he really showed a real horse in his presentation21:02
sicelo_hehe, of course. that's his  pasttime. even on his facebook page, his livejournal account, and youtube channel ... horses :p21:02
rafael2kno nooọ...21:03
rafael2kthese days things are more political21:05
rafael2kshit, every time I read patches from mobian patchset I get scared21:25
rafael2kstuff like: "Not sure how this works exactly. More tests are needed."21:25
rafael2k"[PATCH 158/465] media: ov5640: [Experiment] Try to disable denoising/sharpening"21:26
rafael2kI might try vanilla 5.19 to see what is really needed at some point21:26
Wizzup:)21:27
rafael2kecho 1 > /sys/class/leds/white:flash/brightness echo 0 > /sys/class/leds/white:flash/brightness21:36
rafael2kflash is there21:36
rafael2k: )21:36
Wizzupyeah, but it's not necessarily a standard interface21:36
Wizzupwe can also do flash on the d4 and n900, I think, but it's all custom21:36
rafael2khum... this is what is in mainline21:36
uvosrafael2k: thats the wrong interface21:37
WizzupI imagined v4l had a property for it21:37
rafael2kif (N900) elseif (d4) elseif (PP)21:37
rafael2k: )21:37
uvosrafael2k: its supposed to be exposed via v4l21:37
rafael2khum21:37
uvosrafael2k: but maybe implementing it v4l adds this led class device as a compatability feature21:37
uvosrafael2k: the kernel dose that for lots of things21:37
sicelo_uvos: not necessarily. that's just how d4/n900 do it. but the pp implementation is correct as well21:37
uvosthe d4/n900 dont do it at all21:38
uvosand no its not correct21:38
Wizzupsicelo_: I think the point is that probably want something universal21:38
uvosonly v4l has the timeing information for a  real flash21:38
uvosthe led class driver has no idea when to fire21:38
rafael2kthis is indeed in mainline, so I would not think it is wrong21:38
uvos( more important for a flash thats not a led that can blink for a long time but still)21:38
rafael2ktorch21:39
rafael2kwhen power cuts kick in21:39
uvoswe where talking about using the flash as a flash21:40
uvosofc torch usage has not timeing requirements21:40
rafael2kright, it should be exposed in V4L API too, agreed21:41
rafael2kand indeed, the current for flash for picture is higher than in torch mode, afaik21:41
Wizzupthat would provide a universal interface21:41
rafael2khttps://leste.maemo.org/PinePhone#Enabling_Flash_Light_.2F_Torch21:42
rafael2kadding things here before the wine takes them away ^21:42
sicelo_https://elixir.bootlin.com/linux/latest/source/drivers/media/i2c/adp1653.c#L52 ...21:42
rafael2kbut sometimes simple interfaces are good too21:43
rafael2kindeed sicelo21:44
rafael2ksettings are different, as it should be21:44
sicelo_the adp1653 is the driver on the N900 uvos21:44
rafael2kcho 1 > /sys/class/leds/white\:flash/flash_strobe21:46
rafael2kthis is the flash_stobe in the in PP21:46
rafael2kall also available tru V4L21:46
rafael2kthe LED framework overlap with V4L framework21:46
rafael2kas far as I can see21:47
uvosright as i said often a device is exposed automaticly in the other framework of you register it with one21:47
sicelo_rafael2k: really don't feel bad. the PP is doing the right thing, to the best of my knowledge. i have followed a little bit of the development around its flash :-)21:47
uvosif the frameworks overlap in feature set21:47
uvosalso you can implement a v4l device in terms of a led class driver21:47
Wizzuprafael2k: so for this to be a generic interface there needs to be a way to query what the purpose of LEDs is21:47
Wizzupand sysfs doesn't really do that21:48
Wizzupafaik21:48
uvosthe point is that just a led class driver21:48
uvosis not sufficant21:48
Wizzupthis could just as well be the led on some other part21:48
uvosbecause of timeing requirments21:48
rafael2klinux is a mess, lets agree21:48
rafael2k:P21:48
uvosand also what Wizzup saying21:48
rafael2kls /sys/class/leds21:50
rafael2kblue:indicator green:indicator red:indicator white:flash21:50
rafael2knot thaaaaat bad21:50
Wizzupyes it is21:51
Wizzupbut it doesn'21:51
Wizzupt matter21:51
Wizzupit's good that we know how to do it21:51
rafael2kworking... I would not get worried21:54
sicelo_you're doing awesome work21:55
rafael2k: )21:57
sicelo_n900 camera has been popping dmesg errors in recent kernels, and i hope to fix that. at least i found reloading the driver makes the errors go away. maybe you can help with getting pics out of it as you seem to know a lot about this stuff21:57
rafael2kI thought Pavel made it work21:58
rafael2k:P21:58
sicelo_yes he did. even took pics (matrix style - green)21:58
rafael2kthe made a leeeenghty presentation21:58
sicelo_but he doesn't have time for N900 in general anymore21:59
rafael2kHe mentioned in libcamera list it is in his top priority21:59
sicelo_hehe, when? and what?21:59
rafael2ktoday21:59
sicelo_n900 camera?21:59
* Wizzup bbl22:00
rafael2kyeap22:00
rafael2kand he says he uses Mobian on the PP, and Maemo in Droid 4 and N900 btw22:01
rafael2k"I'm actually using Leste on Droid4" ...22:01
sicelo_rafael2k: i was completely confused earlier when we were talking about horses ... but now looking at the list i see what you meant :-D22:02
rafael2kehehehe22:02
rafael2kthat iconic image came to my mind... this is why I answered "yes"22:03
sicelo_one person i know would really like to see n900 camera working (and mapphones) is Laurent22:05
sicelo_*with libcamera, that is22:05
rafael2kcool22:15
rafael2khe understands _a_lot_22:15
rafael2klibcamera is the place to implement all these ISP pipeline setup madness22:18
sicelo_Wizzup: updating droid 4 kernel shortly. how's the stability, i.e. likelihood of unexpected reboots?23:08
sicelo_hmmm, died while updating. nice23:17
sicelo_ok it didn't. just screen went black and wouldn't react to powerkey for a while23:21
sicelo_Wizzup: does this come with your calls hack ootb? or something needed? i have a call to make tomorrow, so i could test with it perhaps23:24
uvosno hacks23:27
uvosvery stable so far23:27
uvosnot a single reboot in 1week23:27
sicelo_alright23:46
Wizzupsicelo_: no calls hacks23:50
Wizzuplike uvos said23:50
sicelo_alright. i'll use earphones then23:52
uvosearphones?23:54

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