libera/#maemo-leste/ Saturday, 2021-10-09

_inkywe don't have libosso-gnomevfs2-dev00:18
_inkyright?00:18
_inkybut mappero needs it. :/00:18
lelMerlijnWajer assigned an issue: https://github.com/maemo-leste/bugtracker/issues/261 (gtk3 port)00:18
lelMerlijnWajer assigned an issue: https://github.com/maemo-leste/bugtracker/issues/261 (gtk3 port)00:18
Wizzupinky: you need to port from gnomevfs to gio, it's usually simple00:21
Wizzuperr _inky ^00:21
Wizzuplibosso-gnomevfs2-dev is dead, basically, iiuc :)00:22
_inkyi don't even understand what you've said00:22
_inkycan i just use gnome-vfs packages?00:22
_inkyinstead of libosso-gnomevfs00:22
_inkyoh there are nothing like that.00:23
_inkyi am not sure what in mappero needs it.00:23
Wizzup_inky: gnomevfs does not exist anymore00:26
Wizzup_inky: so you can no longer link against the library00:26
Wizzupit's mostly used to read/parse the filesystem and such, to do i/o00:26
Wizzuptrhe replacement is "gio"00:26
Wizzupthere are various examples out there, I can take a look with you if you want00:27
_inkyminute00:27
_inkyhttps://github.com/norayr/mappero/00:27
_inkythis is it, fremantle's latest mappero.00:27
_inkyif you can make it use gio instead of libosso-gnomevfs, would be very thankful.00:28
WizzupI'd rather you take a look and try :p00:28
Wizzuplet me see00:28
_inkywhat is gio, is it in debian or maemo repos?00:28
Wizzupare you able to write C?00:29
Wizzupbtw, mappero as you linked doesn't have a debian dir?00:29
_inkywell, i guess so.00:29
_inkyit has, minute00:30
_inkycheck maemo/beowulf branch00:30
_inkyi have a working ready for jenkins port of lagrange - gemini browser.00:31
_inkybut that's another story. i did many changes but not in c source.00:31
Wizzup_inky: hang on, brb00:38
_inkyyes00:39
Wizzup_inky: it seems to just compile, but there is a linking problem00:49
Wizzupworking on it00:49
Wizzupbut pls study the commit so you can do it too :)00:49
Wizzupah ok now I hit the gnome-vfs stuff00:51
Wizzup_inky: https://wizzup.org/0001-Update-debian-dependencies-remove-gvfs-includes.patch00:54
_inkyheh.00:54
_inkyaha let me see00:54
Wizzupthis should getr you to the point where you can build it and vfs complains00:54
Wizzupit doesn't actually do the gvfs porting yet00:54
Wizzupthe first I see is this:00:55
Wizzupmaps.c: In function ‘download_tile’:00:55
Wizzupmaps.c:461:5: error: unknown type name ‘GnomeVFSResult’; did you mean ‘GTestResult’? GnomeVFSResult vfs_result;00:55
_inkyi was guessing so, that it needs the lib to store downloaded tiles.00:55
Wizzupyes, but gio replaces gnome vfs, you just need to do the actual porting00:56
_inkyok i guess it's very late at your side. i'll see what can i do.00:56
_inkythank you.00:56
Wizzupfor example, gnome_vfs_read_entire_file might become https://docs.gtk.org/gio/method.InputStream.read_all.html in combination with something that opens a url as file handle00:58
Wizzupyeah it's 1am so I can't/won't do it now00:58
Wizzuplmk if you run into trouble, I like gps apps :)00:58
Wizzupalso just googling for "gio read remote url" or "glib read remote url" might help00:58
Wizzupfor example that gives this: https://stackoverflow.com/questions/10622642/fetch-a-file-from-web-using-glib-gio-from-c00:59
Wizzuphttps://docs.gtk.org/gio/type_func.File.new_for_uri.html00:59
Wizzupkeep in mind that gnome/gtk broke *all* their doc links00:59
Wizzupit's insane but true, so when you click a link and it doesn't work, search for the method name again in the title bar00:59
inkyokay!00:59
Wizzupin any case dpkg-buildpackage -b -uc should mostly work now with the patch applied01:00
inkyoh indeed it's later here, about 3 am (:01:00
Wizzup(git am blabla.patch)01:00
Wizzupalso just 'make' will work if you did dpkg-buildpackage -b -uc once01:01
WizzupI made one other change:01:01
Wizzup#ifdef MAEMO_CHANGES /* probably not the best macro to check for here */01:01
Wizzup-#    include <device_symbols.h>01:01
Wizzup+//#    include <device_symbols.h> #endif01:01
Wizzupbecause I don't know where that file is, but it's not in the repo01:01
inkyyeah. i hope i'll tinker tomorrow.01:01
Wizzup:)01:01
inkyoh01:01
inkyi see01:01
inkyi suffered with build system of lagranfe gemini browser a lot01:02
inkyit builds but sometimes crashes01:02
inkyalso01:02
inkyit says on console that01:02
inkylibgl error: failed to load omapdrm01:03
inkyit works nice even without it01:03
Wizzupare you building on the droid?01:05
Wizzupooh, sorry, got confused01:05
Wizzupyeah, that libgl error is 'normal'01:05
WizzupI would recommend generally to try building in the VM if you can01:05
_inkyyep, building on the droid (:01:05
Wizzupwould recommend to do it on a VM if you can01:06
_inkyi did so before01:06
Wizzupok01:06
_inkybut it's not that convenient.01:06
Wizzupit's just much faster/easier01:06
Wizzupok :P01:06
_inky(:01:06
Wizzuplet me know wrt mappero, I might have more time later if you need help01:07
Wizzupthere doesn't seem to be too much, maybe 6 places/uses of vfs01:07
_inkyokay, i'll see, i hope will be able to solve this.01:14
Wizzupit's probably ~1-2h if you know what you're doing01:14
Wizzupthere are also examples of this porting in other repos01:14
_inkyi never used gtk and its libs, but i'll see.01:14
Wizzuphttps://github.com/maemo-leste/osso-pdf-viewer/commit/536da6652e44b9e5dc25a9eab004e252e17b452e01:15
Wizzupthat shows how to replace the string escaping (provided I did it right)01:15
Wizzupthere's more porting here https://github.com/maemo-leste/osso-pdf-viewer/commits/master01:15
_inkythaaanks01:16
_inkythis seems very helpful.01:16
Wizzupsure, lmk01:16
mighty17[m]<MartijnBraam[m]> "the pmos n900 kernel is a bit..." <- We should package openpvrsgx kernel for pmos :D06:12
^-^hiOk, so bootlocks only let one type of kernel they have signed to boot07:21
^-^hiAnd Android runs some sort of Linux07:21
^-^hiAnd Linux's kernel could be upgraded on the fly as I heard.07:22
^-^hiSo... can one boot the Android kernel and then change it to mainline Linux after the boot, thereby salvaging the bootlocked phones?07:23
siceloYes. That's what kexecboot does on the motorola droids, and lk2nd in the qcoms.07:36
uvosWizzup: no09:01
uvosWizzup: but i do test it on amd64 arch linux09:01
freemangordonWizzup: yes, please http://46.249.74.23/leste/n900/09:18
freemangordonopenpvrsgx doesn;t boot either10:02
uvosdid you try vanilla?10:02
freemangordonpulling linux stable ATM10:02
dreamerheh. friend had noticed some TI userspace build activations on 3430 and 4430 trees recently. but was wondering why 3630 wasn't done hehe10:16
dreamerthe sgx handling, right?10:16
* mighty17[m] uploaded an image: (84KiB) < https://libera.ems.host/_matrix/media/r0/download/matrix.org/XeWipCSzJhLiLMOSvkCLfsAG/IMG_20211009_133052.jpg >10:24
mighty17[m]Is the green lines GPU issue?10:24
mighty17[m]Or display10:24
Wizzupfreemangordon: trying now10:35
Wizzupfreemangordon: did you append the dtb, btw?10:37
freemangordonyes10:37
freemangordonwait, I have dmesg log10:37
freemangordonlooking at it ATM10:37
freemangordonacx565akm spi0.2: failed to find video source10:38
freemangordonhttps://pastebin.com/1kcKDB2T10:40
freemangordonany idea?10:40
freemangordontmlind: ^^^10:40
Wizzupis this omap2plus_defconfig ?10:41
freemangordonno10:41
freemangordonlemme try with it10:41
uvoswhy are we on rc2 again10:41
freemangordonthis is old dmesg10:41
freemangordonwith vanilla I have the same error10:41
freemangordonuvos: https://pastebin.com/Tbbf9RgL10:42
freemangordonthe same with openpvrsgx tree10:42
siceloNice! It's always the display :-(10:43
Wizzupbtw, I tried on serial and uh it doesn't boot for me on several tries which is weird, also no output on serial, let me try once more10:43
WizzupI thought extra_bootargs with console= set to serial would work but maybe that appends the arg10:44
freemangordonWizzup: don't waste time I have dmesg log10:45
Wizzupok10:45
uvosjust stabbin in the dark here but maybe omapdss dosent probe? boot with initcall_debug=1 nd ignore_loglevel=110:48
uvosmaybe10:48
uvosor since it boots ssh in and take a look around in sys and proc10:49
freemangordonI can;t ssh10:49
freemangordonalso, I see no dss logs, so maybe you are right10:49
freemangordonbut, lemme trey with omap2plus_defconfig first10:50
uvosok10:50
freemangordonwith omap2plus_defconfig it doesn't write anything in /var/log11:12
freemangordonI guess it dies before mounting root11:13
Wizzupis this plain 5.1511:18
WizzupI can try to build it today11:18
freemangordonwait, I think I know what happens11:18
freemangordonbut yeah, you may try vanilla with omap2plus_defconfig to see why it hangs11:19
Wizzuplatest rc?11:19
freemangordonyes11:19
freemangordonrc-411:19
freemangordoniiuc11:19
Wizzupsicelo: (and others) btw, if you know of a way to produce a few serials at not an insanely high cost I'd like to know11:19
Wizzupfreemangordon: ok, but it will be later today, maybe late afternoon or eve11:20
freemangordonok, no hurry11:20
uvossome one has to design them11:20
uvosi can make some11:20
Wizzupuvos: sre designed them11:20
Wizzuphttps://n900.elektranox.org/serial-adapter.html11:20
uvosthere are no files and he used a laser cutter wich i could do but its not a great idea11:20
Wizzupbut we'd need to order the parts and use a cnc11:20
WizzupI also have a laser cutter at the local hackerspace here11:20
Wizzuparen't there files?11:20
uvosi have various cnc machines11:21
Wizzupneat11:21
sicelono there arent. but i'm sure he could share them if we ask11:21
Wizzupi'm sure he'll share the files11:21
freemangordonI think it would be better to re-desigh to use a simple flex cable (an a terminator) that lives under the battery11:22
uvosi dont like the dsign11:22
uvosreally11:22
freemangordonme neither11:22
uvostentioning the flex cable might be hard11:22
uvosan easy way to do it11:22
uvoswould be to use a flex cable with the same pin pattern as the n90011:23
uvosand then using conductive adheasive11:23
freemangordonadhesive?11:23
uvoslike is used on lcd display cables to bond to the glass11:23
uvosyeah11:23
freemangordonthis is even worse11:23
freemangordonI mean:11:23
freemangordonyou can;t reuse that device anymore11:23
uvoswhy just have the cable wrap around the battery11:23
freemangordonunless I missed the point11:24
uvoson the other end of the flex you have a eadge connector11:24
freemangordonuvos: wait, you need to have pogo-pins contacting the MOBO11:24
uvosfreemangordon: no here 1 sec11:24
uvosso you have a flat flex11:26
uvosand a regular pcb11:26
uvoson the regular pcb you have the voltage level conversion and usb uart chip11:26
uvosand a connector like this https://eu.mouser.com/ProductDetail/GCT/FFC2B22-45-T?qs=TuK3vfAjtkVWLNrSep2AyA%3D%3D11:26
uvosthe flat flex is just long enough to wrap arround the battery half way11:27
freemangordonok11:27
uvoson one end it ends with the connector pins like this https://community.voltera.io/uploads/default/original/1X/c94b8fae9c2c7a44bb8ce84afcd97ec08d77c3fa.jpg11:27
uvoson the other end you have the n900 pad pattern exposed11:27
uvosyou bond the n900 end to the mainboard permanently11:27
uvosand wrap the flex around the battery11:27
uvosso you can still close the door11:27
uvosif you want serial you open the door an plug in the regular fr4 pcb11:28
freemangordonuvos: everything is fine, besides the 'bond' part11:28
freemangordonas soon as you disconnect, you have 4-5 cm wire attached directly to SoC hanging in the air11:28
uvos? to the pcb not the soc11:28
uvosand its just a pice that goes around the  battery a half turn11:29
freemangordonit is PCB, but there are no bufers or sumething11:29
freemangordonso basically it is to the SoC11:29
uvosi dont think its a big problem emi wise11:29
uvosyou would only connect the uart pins ofc11:29
uvosthe rest you bond for stability11:29
freemangordonI would say it is, because, as I said, there is no buffersing11:29
uvosbut dont give them any long tail11:29
uvosdont worry  fine electrically11:30
freemangordonso you attach an antenna to the SoC11:30
uvosyeah but just to the uart pins11:30
freemangordondoesn;t really matter, as long as there are no buffers or filter caps11:31
freemangordonafaik this pin is floating11:31
uvosyes it dose its not a big deal you might get some noise on the pins but nothing else11:31
uvosyes its floating11:31
freemangordonwhat about finding a connector that fits pcb pins and use the battery to keep it attached to the PCB?11:31
freemangordonand use bond to attach flex to the connector11:32
uvoswell then you need a grid of pogo pins11:32
freemangordonok11:32
uvosto get defined force11:32
uvosi doubt you can do that in the space availbe11:32
freemangordonit is not that small11:32
uvosits pretty small11:32
* freemangordon checks11:32
uvospogo pins need travel11:32
freemangordonI know11:33
uvosand a backing of sufficant stiffness11:33
freemangordonpcb is ok11:33
Wizzupwell the pogo pin design of nokia works well no?11:33
uvossure11:33
Wizzupat least for me11:33
uvosbut that replaces the whole battery11:33
Wizzupyup11:33
Wizzupbut sre's is kinda simple11:33
freemangordonanyway, back to kernel11:33
Wizzup:)11:33
uvosmachineing a battery shaped object is no issue11:33
uvosthe premis here was that fmg wants the adapter to work with stock battery11:34
uvosanyhow if someone makes cad files for a battery shaped object and a pcb with pogo pins i can machine or 3d print it11:34
freemangordonhmm, maybe I should compile on my laptop (8x i7) not on my desktop (4x i5 from 2010 or something) :)11:35
sicelo:-)11:36
freemangordonnot to say that laptop is with nvme11:36
Wizzupuvos: we'll also need to order parts to get the proper voltage11:36
Wizzuphttps://www.sparkfun.com/short/874511:36
WizzupI have this at home, sre gave me one at fosdem11:36
uvosi would just design a pcb that integrates the voltage shifter and11:36
uvosuart11:36
Wizzupaha11:36
freemangordonWizzup: this level shifter is simple mosfets and resistors11:37
Wizzupfreemangordon: yeah11:37
freemangordonyes, as uvos said11:37
Wizzupwon't it be too much of a time sink?11:37
Wizzupotherwise, sounds like a great idea imho11:37
Wizzupidk how long it takes to do those things11:37
uvosnot very long11:37
uvosbut i dont have any interest in n900 really11:37
freemangordonsuch a simple pcb, maybe a couple of hours11:37
Wizzupok, so we need cad files with the n900 battery size?11:38
uvosyou might not even need to do any voltage shiftig11:38
WizzupbuZz: do you think you can help?11:38
uvosthere are bound to be some usb uart chips with 1.8v logic level11:38
siceloyou do need voltage shifting. the n900 uart is 2.7v11:38
Wizzupuvos: it might be neat to allow connecting random 3.3v uarts to the pcb as well11:39
uvossicelo: omap logic level is 1.811:39
uvosnot sure where 2.7 would come from11:39
siceloi know. but, N900 uart is 2.7v. sre's design uses 2.5v, which is reasonably close11:41
uvoswell its ttl i should have 0.8 as the high threshold so anything in that range will work11:42
uvosas long as its below the maximum rating the pin11:43
uvoslet me check the datasheet11:43
sicelouvos: maybe this will make you understand, http://wiki.maemo.org/N900_Hardware_Hacking#Debug_ports11:45
uvosso omap3 datasheet says that uart is max 2.1V input with ll high thresh of 1.17V low thesh of 0.63V and output of 1.8V11:46
uvosif its doing something else there must be a chip between the uart and the debug port11:47
siceloof course, there is11:48
uvoswell fmg just said there isent11:48
freemangordonwell, we don;t know what is there11:49
uvoswell not omap directly11:49
uvosso i see the issue of putting a 3cm lead on there even less now11:49
freemangordonits ugly11:49
uvosokaaay11:50
uvoswel its not my problem11:50
freemangordonyeah11:50
Wizzupunrelated but I do care about the n900 if only to show that it's possible to make a modern linux smartphone os that mostly just works fine, of course, don't expect browsers to work well12:09
Wizzupbut I think it's nice to support it12:09
freemangordon5.14.10 boots13:16
freemangordonso it is a regression in 5.15 it seems13:17
uvosshould not be to hard to bisct from there13:17
freemangordonyeah13:17
uvoswe really need some kind of automatic testing13:18
uvosthat at least checks if the devices boot & load all requesit modules13:18
uvosn900 not booting has been a theme..13:18
freemangordongoing to check if 5.14.0 boots and then will bisect13:19
sicelonice. that's great progress :-)13:30
freemangordonyep, it does13:31
siceloomap2plus? or your own config?13:31
freemangordonomap2plus13:57
freemangordonWTF? now 5.15-rc4 boots :(13:58
freemangordonthe difference is that I build it on my laptop13:58
sicelo:-)13:59
sicelohehe, i also experienced that a lot - the same kernel would not boot first time, but boot second or third time. i wonder if something doesn't get reset properly somewhere, but my knowledge in this area is lacking.14:01
freemangordonno, kernel build on leste does not boot14:01
freemangordonI tried maybe 10 different kernels14:02
freemangordonthe ones build on ubuntu 18.04 boot14:02
sicelommm. gcc version issues?14:03
freemangordon8.3 vs 8.4, should not make a difference14:03
freemangordonwell, tmlind's kernel does not boot still14:16
Wizzupfreemangordon: did you figure out what was up?14:30
freemangordonno14:30
Wizzupok14:30
freemangordonvanilla boots14:30
freemangordonbut not tmlind's14:30
freemangordonnow I am trying to find tag on his tree that boots to bisect14:30
Wizzupok14:35
Wizzupand tmlinds is both pvr and other patches?14:35
freemangordondroid4-pending-pvr-omapdrm-v5.1514:36
freemangordonwhatever it is14:36
Wizzupok14:36
freemangordonWizzup: if you have time, please compile and try to get boot log14:37
freemangordonbisecting this may turn to be impossible14:37
_uvos_did you merge in rc4?14:54
_uvos_please take rc4 and merge in the pending-pvr branch14:55
freemangordonwhy shall I do that?14:55
freemangordonvanilla rc2 boots fine14:55
_uvos_ok14:55
_uvos_i thought it was rc414:55
freemangordonrc4 too15:00
Wizzupfreemangordon: ok, pls tell me what config and git repo+brachh16:12
freemangordonomap2plus_defconfig16:53
freemangordonhttps://github.com/tmlind/linux_openpvrsgx.git16:53
freemangordondroid4-pending-pvr-omapdrm-v5.1516:54
freemangordonWizzup: ^^^16:54
Wizzupok17:32
Wizzupwill do in 1-2 hrs17:32
freemangordonwhy the hell the whole kernel gets rebuild every time I do make? without changing anything?18:30
Wizzupfreemangordon: shouldn't18:32
freemangordonhmm, yeah, false alarm it seems18:33
Wizzupjust got home18:33
freemangordongood18:33
freemangordonI have 3 steps to finish the bisect18:34
Wizzupok18:36
Wizzupcloning now18:36
freemangordonhmm, I guess it is some stupid setting on config, each bisect step leads to a full rebuild18:45
freemangordonn ever happened before18:45
freemangordonmaybe .ko symbols or something18:45
freemangordonCONFIG_MODVERSIONS that is18:47
Wizzupfreemangordon: how do you make the uImage ?18:52
freemangordoncat arch/arm/boot/zImage arch/arm/boot/dts/omap3-n900.dtb > zImage18:53
freemangordonmkimage -A arm -O linux -T kernel -C none -a 80008000 -e 80008000 -d zImage uImage18:53
Wizzupfreemangordon: it's booting for me19:05
freemangordoncool19:05
freemangordonwell, no idea then19:06
Wizzupwell it doesn't seem to boot all the way to leste19:06
Wizzupthe screen stays on but black19:06
freemangordonthe same here19:06
freemangordonactually it is stuck19:07
freemangordoneventually it will power-down19:07
Wizzup[   15.690734] omapdss_dss 48050000.dss: supply vdda_video not found, using dummy regulator19:07
Wizzup[   15.722290] st-accel-i2c 3-001d: supply vddio not found, using dummy regulator19:07
freemangordonshouldn't be an issue19:07
Wizzupit's recurring19:07
freemangordon ah19:08
mighty17[m]The supply not found happens with me aswell, not an issue19:08
Wizzup[   20.243591] rx51-audio n900-audio: snd_soc_register_card failed (-517)19:08
freemangordondo you have rootfs mounted?19:08
Wizzupyeh19:08
freemangordonhere is does not mount19:08
Wizzup[   17.445861] lp5523x: probe of 2-0032 failed with error -2219:08
freemangordonat least judging by the missing dmesg log in /var/log19:08
freemangordonhmm 223790cf892de9cee6a97fdf95c267ad36507951 is the first bad commit19:10
freemangordondoes not make sense19:10
Wizzuphttps://wizzup.org/log.txt19:10
freemangordondrm: pvrsgx: 1.17.4948957 remove never implemented MODULE_SUPPORTED_DEVICE like commit 6417f03132a6919:10
freemangordonok, lemme try again19:11
Wizzupjust stops there for me19:12
Wizzupit doesn't reset for me either19:12
freemangordoncould you remove dsme from /etc/init.d to see if it will boot to the shell?19:12
mighty17[m]`mmc1: switch to bus width 8 failed`19:12
freemangordonthat's normal19:13
Wizzupmaybe some depmod is necessary19:13
Wizzupunlikely but maybe19:14
freemangordonthe others boot without depmod19:15
freemangordonWizzup: does it boot to shell with dsme removed?19:18
freemangordonor rather - are you able to see any output on the display?19:22
freemangordonhmm, it boots on b98125a4f1287bc29df93c43abe49cc661030ff419:56
Wizzupfreemangordon: with display?20:15
freemangordonyes20:16
freemangordonI mean - I have no other means to distinguish between boot/does not boot but if it displays anything20:17
freemangordonon 4a75f05f070785db5ffcbe21bd45b1482ef99e37 I have display too20:17
freemangordonnot trying the next one20:17
freemangordon614eb7536d3adad342b04fd1666bb3880e660dab that is20:18
freemangordon*now trying20:18
freemangordonWizzup: so, what happens on youre device if you remove dsme?20:20
freemangordondo you get console?20:20
Wizzupcan try momentarily20:32
WizzupI don't think so, I think it blocks on udev events to settle or something20:33
Wizzupbut will try20:33
freemangordonok20:33
Wizzupfreemangordon: so it's stuck here:20:49
Wizzup[   12.403442] udevd[480]: starting eudev-3.2.920:49
Wizzup. ok20:49
Wizzup[ ok ] Synthesizing the initial hotplug events (subsystems)...done.20:49
Wizzup[....] Synthesizing the initial hotplug events (devices)...20:49
Wizzup(with dsme renamed)20:51
Wizzupso no20:51
freemangordonsame here20:52
freemangordonso, actually it does not boot for you :)20:52
freemangordonI mean "same" like :no console"20:52
freemangordonanyway, I am doing a kind of a manual bisect20:52
Wizzupright20:54
Wizzupalpine doesn't have eudev probably? not sure what tmlind usually tests with20:54
freemangordonno idea20:54
freemangordonwhy the hell on every "git checkout" the whole kernel gets rebuild?!?20:55
Wizzupprobably a lot of files change mod time20:57
freemangordonsingle file only20:58
freemangordonjust disabled module symbols versioning, lets see if it will help21:00
freemangordondidn;t help21:05
freemangordonwtf is going on? it is like I am compiling the kernel for a first time :(21:05
freemangordon33bc438d6d8883d77e37b369fe5144ee9b01fad8 breaks it21:17
freemangordonhmm, I think I know why it breaks it, lemme try21:21
Wizzup:)21:24
Wizzuphttps://news.ycombinator.com/item?id=2880955921:25
Wizzupfreemangordon: ^^21:26
freemangordon:)21:27
freemangordonyeah21:27
freemangordonhmm, no, it's something different, I need tmlind21:35
WizzupI guess init=/bin/sh does work?21:36
Wizzupmust, since it starts openrc21:37
freemangordonI'll just revert that commit, to see if pvr works on n90021:42
freemangordonbecause that was the point of the whole exercise21:42
Wizzupreverting 33bc438d6d8883d77e37b369fe5144ee9b01fad8 solves your hangs?21:43
freemangordonyes21:43
Wizzuplet me try21:43
freemangordonwell, on 33bc438d6d8883d77e37b369fe5144ee9b01fad8 it hangs while on d8d18c28963fd9b9ed4425d79c4d5d5d3b496771 it does not21:44
freemangordonnow building with 33bc reverted to see21:44
Wizzupyeah looks like that makes it boot for me too21:48
Wizzupwaiting to get agetty21:48
Wizzup[....] Synthesizing the initial hotplug events (devices)...21:48
Wizzupdone.21:48
Wizzupyeah now it gets past that21:48
Wizzupso I can confirm reverting that commit makes it boot21:49
freemangordonhow you were able to compile so fast?21:50
Wizzupmodel name: AMD Ryzen 7 PRO 4750U with Radeon Graphics21:50
freemangordondoesn;t it recompile the whole kernel when you revert?21:50
Wizzup16 threads21:50
Wizzupthe revert didn't recompile the whole kernel, no21:50
freemangordonit does here21:50
Wizzupmaybe it's a .config option that is enabled?21:50
freemangordonsomething's wrong here it seems21:50
freemangordonomap2plus_defconfig21:51
freemangordonI'll clone tmlind's tree only21:51
freemangordonand try again21:51
Wizzupyeah that's what I did21:51
Wizzupstill, surprising that that would cause udev events to hang21:51
Wizzupmaybe it causes some operation that doesn't finish on n900 and then omapdrm never probes fully?21:52
freemangordoncould be21:52
Wizzuplooks like it21:55
Wizzupon this kernel:21:55
Wizzup[   21.217407] omapdrm omapdrm.0: [drm] fb0: omapdrm frame buffer device21:55
Wizzup[   21.343597] [drm] Initialized omapdrm 1.0.0 20110917 for omapdrm.0 on minor 021:55
Wizzupright above that: [   20.081420] omapdrm omapdrm.0: DMM not available, disable DMM support21:55
Wizzupand then without the commit reverted, it never gets past this:21:56
Wizzup[   19.809112] omapdrm omapdrm.0: DMM not available, disable DMM support21:56
freemangordonmhm21:56
freemangordonhmm, cannot start wpa_supplicant21:56
freemangordonWizzup: any hint how to start usb networking?21:58
Wizzupyeah21:58
Wizzup/usr/sbin/hildon-usb-gadget-network21:58
Wizzupbut if you have the same kernel as me then the usb gadget does not work21:58
freemangordonmaybe I should modprobe something?22:00
WizzupI see no wlan0 btw22:01
freemangordonme too22:01
freemangordonlemme check if it is enabled22:01
Wizzupkind of expect it to be22:01
freemangordonbtw, did you send your patch upstream?22:02
Wizzupnot yet22:03
WizzupI think it just needs modprobe22:03
Wizzupmodprobe wl125122:03
Wizzupmodprobe wl1251_spi22:03
freemangordonlemme check22:03
Wizzupoh yeah without that patch icd2 fails heh22:03
WizzupJan  1 02:16:03 (none) icd2 0.98[2702]: libicd-network-wpasupplicant: close_wpa_control: 0x51bba022:04
WizzupJan  1 02:16:03 (none) icd2 0.98[2702]: libicd-network-wpasupplicant: try_open_wpa_control: 0x5422c022:04
WizzupJan  1 02:16:03 (none) wpa_supplicant[2711]: wlan0: CTRL-EVENT-SCAN-FAILED ret=-2222:04
Wizzupnot yet22:04
freemangordonyse wext22:04
freemangordon*use22:04
Wizzupright, that will work, but not what we use on n900 these days22:05
freemangordonyeah, but I need only shell anyways22:05
Wizzupyeah22:05
WizzupI will let you mess with this now, lmk if you need help22:05
WizzupI'm quite motivated to see if we can move all our pvr stuff to 1.17 somehow..22:05
freemangordonI am in through wifi22:06
Wizzupok, then I will get back to droid3 kexecboot22:06
Wizzupthat's the last thing remaining for me to start writing the news post :D22:07
freemangordon[  808.276550] PVR_K: UM DDK-(4948957) and KM DDK-(4948957) match. [ OK ]22:07
Wizzupthat's the init?22:07
freemangordonmhm22:08
freemangordonand kmscube made it hang :(22:08
freemangordonthe same regression as with 5.1022:09
Wizzupshit22:09
freemangordonyeah22:09
freemangordonok, tomorrow I will try to bisect22:11
Wizzuprough22:12
WizzupI didn't send the wifi patch in part because I'd have to rebase on latest kernel22:12
Wizzupand test it as well22:12
WizzupI might just try to use just modesetting22:13
freemangordonhmm, why wl1251 is not probed?22:15
Wizzupmaybe that's a depmod thing22:15
freemangordonah, right22:15
Wizzupwill verify now22:15
Wizzup(ran depmod -a, rebooting)22:15
WizzupX doesn't start in modesetting without powervr though for me, so need to figure out what's up there22:15
Wizzupeven unaccelerated X is still helpful in testing wifi patch and such22:16
freemangordonmhm22:16
Wizzuphm, didn't help (depmod -a)22:18
freemangordon:(22:19
Wizzupthings really never work with computers :D22:20
freemangordonyeah22:20
Wizzupremoving /dev/dri/card1 and /dev/dri/render129 makes X start at least22:21
freemangordonbisecting will be a nightmare, there are several things broken in 5.10 IIRC22:24
freemangordonlike mmc renaming and panel reset22:24
freemangordonwhat I am missing?22:24
Wizzupwhen exactly did it break?22:25
Wizzupnot exact commit, but version22:25
freemangordonI am not really sure22:25
Wizzupdid 5.9 work?22:25
freemangordonah, in 5.1022:25
freemangordonyes, 5.9 did work22:25
Wizzuphm22:28
Wizzupthere must be a sensible way to debug it22:28
freemangordonyeah, jtag :)22:29
Wizzupdid anyone ever manage to do that? ;)22:29
freemangordonsure, in TI/Nokia ;)22:29
WizzupI wonder if the device I have here can do it22:29
Wizzupparazyd: the latest vm images will lock the screen upon timeout, and then *all* input devices are disabled22:40
Wizzupand then it's not possible to unlock unless you send the 'power' key to the vm22:40
WizzupI don't have this problem on my dev vm, so IDK what's different22:40
Wizzupbut it's likely something inside of the images, rather than leste-config related?22:40
Wizzupparazyd: ah my vm has 'lock screen automatically' disabled23:35
Wizzupbut the fresh image does not23:35

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