| Wizzup | freemangordon: ugh libosso uses __malloc_hook which is now deprecated/gone | 00:30 |
|---|---|---|
| sicelo | freemangordon: re bq2415x: it was the positioning :-) | 04:41 |
| sicelo | originally i had https://paste.debian.net/1331047/ ... i.e. i was calling `power_supply_changed()` after the call to `...update_reported_mode()`. now i call it just before. | 04:42 |
| sicelo | reported_mode actually processes data from isp170x, not bq2415x. so what was happening is that isp fires a notification, and at that time bq2415x starts the 2s internal timer, so at this time, it reports online 0. | 04:44 |
| sicelo | meanwhile, reported mode has already set its state to a charging mode. a couple of seconds later, isp170x fires another event with unchanged values. i absolutely don't know where this 2nd event comes from, but you can see it in https://paste.debian.net/1331046/ | 04:50 |
| sicelo | when that 2nd event arrives, reported_mode simply does nothing since it already updated the mode internally during the 1st isp170x event. calling `power_supply_changed()` before reported_mode causes the bq24 properties to be rechecked every time isp170x reports an event, and thus problem solved :-) | 04:54 |
| sicelo | on that note, uvos thanks for merging the change. would appreciate a build so we can have testing | 06:09 |
| tmlind | Wizzup, uvos: i may have a kexecboot patch to possibly fix that oops issue, needs testing though, not sure why/if i ever pushed it out.. | 07:37 |
| tmlind | well looking at the description, it seems like a workaround, but might be acceptable if it helps: http://muru.com/linux/0001-kexecboot-Only-umount-the-read-only-kernel-partition.patch | 07:40 |
| uvos | tmlind: reading the commit i dont think it helps | 10:31 |
| uvos | tmlind: since thats exactly what happened, the device failed to kexec because the dtb was missing and then the stock kernel crashed | 10:31 |
| Wizzup | uvos: do you need me to kick a build for droid4-linux ? | 10:32 |
| uvos | so your patch dosent help here because its no change | 10:32 |
| uvos | Wizzup: have not done the changelog | 10:32 |
| Wizzup | uvos: I think he's saying it might prevent the oops that followed | 10:32 |
| uvos | Wizzup: no look at the patch | 10:32 |
| uvos | Wizzup: it moves the unmount behind the kexec so it dosent happen at all in the happy path | 10:32 |
| Wizzup | ok | 10:33 |
| uvos | in unhappy path it still happens and stil crashes | 10:33 |
| sicelo | mmm, my leste vm just broke from apt upgrade. https://paste.debian.net/1331486/ | 10:39 |
| Wizzup | I had this too recently | 10:39 |
| Wizzup | sec | 10:40 |
| * Wizzup starts mz617 | 10:40 | |
| Wizzup | freemangordon: it doesn't look like this mem saw stuff is used a lot https://github.com/search?q=org%3Amaemo-leste%20osso_mem_saw_enable&type=code | 10:40 |
| uvos | Wizzup: the tiny mz617 image has the wrong leste-config | 10:41 |
| Wizzup | uvos: huh really? | 10:41 |
| uvos | i have /lib/udev/rules.d/85-input-devices.rules which is breaking my ts | 10:41 |
| uvos | yeah | 10:41 |
| Wizzup | which one is installed | 10:42 |
| uvos | sec just rebooted | 10:42 |
| uvos | yay working ts | 10:43 |
| Wizzup | sicelo: try dpkg --configure gconf2 | 10:43 |
| Wizzup | uvos: dpkg -l | grep leste-config | 10:44 |
| Wizzup | ? | 10:44 |
| uvos | i deleted the offending file | 10:44 |
| Wizzup | sicelo: maybe also dpkg --configure dbus-x11 | 10:44 |
| Wizzup | uvos: sure but dpkg -l will still tell you :D | 10:44 |
| uvos | haha | 10:44 |
| uvos | leste-config-common/now 1.113-1+m7 all [installed,local] | 10:45 |
| uvos | leste-config-mapphone/now 1.113-1+m7 all [installed,local] | 10:45 |
| uvos | leste-config-mz617/now 1.113-1+m7 all [installed,local] | 10:45 |
| uvos | leste-config-droid4/now 1.113-1+m7 all [installed,local] | 10:45 |
| Wizzup | huh | 10:45 |
| uvos | both d4 and mz617 is installed | 10:45 |
| Wizzup | very strange | 10:45 |
| Wizzup | uvos: do you also get an error probing the soundcard btw, on mz617? | 10:45 |
| uvos | Wizzup: no | 10:45 |
| uvos | yes | 10:45 |
| uvos | i mean it dosent probe | 10:45 |
| Wizzup | [ 41.545867] mot-mdm6600-codec 4806a000.serial:modem:audio-codec@2: ASoC: error at snd_soc_component_probe on 4806a000.serial:modem:audio-codec@2: -110 | 10:45 |
| Wizzup | [ 41.605468] asoc-audio-graph-card2 soundcard: ASoC: failed to instantiate card -110 | 10:45 |
| uvos | es | 10:45 |
| uvos | yes | 10:45 |
| Wizzup | [ 41.619659] asoc-audio-graph-card2 soundcard: error -ETIMEDOUT: parse error | 10:45 |
| Wizzup | [ 41.632446] asoc-audio-graph-card2: probe of soundcard failed with error -110 | 10:45 |
| Wizzup | this | 10:46 |
| Wizzup | ok | 10:46 |
| Wizzup | my thinking was that this might be responsible for not hitting RET | 10:46 |
| sicelo | thanks. that helped | 10:46 |
| uvos | Wizzup: btw the samall image is also lacking h-i-m | 10:47 |
| uvos | which makes it pretty hard to start wifi | 10:47 |
| Wizzup | uvos: I thought him was on there specifically | 10:48 |
| Wizzup | maybe it doesn't start properly, or it really doesn't exist? | 10:48 |
| Wizzup | hm you're right | 10:48 |
| Wizzup | hildon-input-meta is in hildon-meta, not in hildon-meta-core | 10:49 |
| uvos | yeah its there | 10:49 |
| uvos | but it dose you no good bacause its disabled by default and the controll pannel applet to enable it is not there | 10:49 |
| Wizzup | ok, but once you enable it, it works? | 10:49 |
| uvos | mybe i dont know the gconf call on hand | 10:50 |
| Wizzup | ok | 10:50 |
| Wizzup | I suppose you just plugged in a keyboard | 10:50 |
| uvos | not yet | 10:50 |
| uvos | also the power button dosent work | 10:50 |
| uvos | which is strange because it dose work | 10:50 |
| Wizzup | the power menu might not be present perhaps | 10:51 |
| Wizzup | hildon-meta-core is pretty trimmed down | 10:51 |
| uvos | doubleclick dosent work either | 10:51 |
| uvos | but power key contributes to inactivity timeout | 10:51 |
| uvos | strange | 10:51 |
| Wizzup | if you have a way to type btw, you can install the applet from apt | 10:51 |
| Wizzup | assuming there's any space left :D | 10:51 |
| uvos | i have 100mb | 10:52 |
| uvos | but catch 22 i cant go online | 10:52 |
| uvos | to get the applet to go online :P | 10:52 |
| Wizzup | hildon-meta-core has the systemui for tklock, powerkey and mode change | 10:52 |
| Wizzup | so it -should- be ok | 10:52 |
| uvos | and i cant use usb-net since then i cant type via serial | 10:52 |
| Wizzup | does it show the usb dialog if you plug in usb? | 10:52 |
| Wizzup | ah | 10:52 |
| Wizzup | well, usb net should just work no? | 10:52 |
| Wizzup | (for ssh) | 10:53 |
| Wizzup | freemangordon: I'm considering getting rid of the libosso/src/osso-mem.c stuff at laest for daedalus, it seems to try to do some userspace oom prevention by hooking malloc, but it's only used by 1 program (hildon thumbnailer iirc) | 10:56 |
| Wizzup | there is the lowmem state check but I don't think that exists on any of the modern kernels | 10:57 |
| Wizzup | see https://github.com/search?q=org%3Amaemo-leste+osso_mem&type=code | 10:57 |
| Wizzup | uvos: how did you make the img btw? | 10:58 |
| uvos | i copied the contents of both mz617-tiny-bootstrap partitions to a singe parition image | 10:59 |
| uvos | then i manually installed a newer kernel | 11:00 |
| uvos | and fixed fstab and boot.cfg to point to p18 instead of p20 | 11:00 |
| Wizzup | ok | 11:01 |
| uvos | what was the flag again to create an ext4 partition that the stock kernel can mount? | 11:02 |
| uvos | something with no 64bit file offsets | 11:02 |
| Wizzup | googling shows -O ^64bit but idk if that is right | 11:06 |
| Wizzup | https://serverfault.com/questions/950704/mkfs-o-64bit-metadata-csum-t-ext4-in-2019 | 11:06 |
| Wizzup | uvos: for the image we also want no partitions right, no partition table | 12:06 |
| Wizzup | mz617 tiny I mean | 12:06 |
| uvos | both mz617 tiny and regular | 12:14 |
| uvos | tiny gets installed on p18 "cache" via fastboot | 12:14 |
| uvos | and is then used to install the full image to p20 "userdata" | 12:14 |
| uvos | there is no room to have a partition table anywhere on a mapphone emmc ofc | 12:14 |
| uvos | since that is part of mbm | 12:14 |
| uvos | idealy we would build single parition images for all mapphones | 12:15 |
| uvos | since then you would not need a sdcard to install leste to those devices | 12:15 |
| uvos | you could just fastboot flash leste to userdata (on the lest resticitve bootloaders anyhow) | 12:15 |
| Wizzup | right, agreed | 12:16 |
| Wizzup | I'm trying to make the changes to libdevuansdk now | 12:16 |
| Wizzup | it doesn't support not having a partition table but I'm adding that now | 12:16 |
| uvos | a problem with doing this atm is btw that kexecboot requires boot.cfg and the kernel to be in /boot | 12:16 |
| uvos | while we mount boot to /boot | 12:16 |
| uvos | so the packages install stuff to /boot/boot | 12:16 |
| Wizzup | yes | 12:16 |
| uvos | we could just have the image builder link /boot/boot to boot for now i gues | 12:17 |
| Wizzup | or have a different /etc/fstab | 12:17 |
| Wizzup | we can just not mount boot | 12:17 |
| uvos | no we cant | 12:17 |
| uvos | the package installes to /boot/boot it will do that in the single partition | 12:17 |
| uvos | but thats no use | 12:17 |
| uvos | because kexecboot will open that partition look at /boot not find /boot/boot.cfg and continue on | 12:18 |
| uvos | it only works atm because /boot on the root partition is a mount to a different partiton where /boot/boot.cfg dose exist | 12:18 |
| uvos | so kexecboot finds it there | 12:18 |
| uvos | so kexecboot finds it there | 12:19 |
| Wizzup | ok | 12:29 |
| Wizzup | uvos: ok, think I'm mostly there with the image builder | 13:20 |
| Wizzup | yeah, got an image without partitions, size 747M | 13:33 |
| Wizzup | btw, I think we should try to upgrade our raspi images to whatever debian does on raspi | 13:36 |
| Wizzup | uvos: what size should I make the image again for mz617? | 13:43 |
| Wizzup | 23:09 < uvos> the only partition flashable via fastboot that is large enough is cache at 850 ish mb | 13:55 |
| Wizzup | ok | 13:55 |
| arno11 | Wizzup: nice @raspi. btw i tried several img's few weeks ago and that's totally broken iirc | 14:11 |
| Wizzup | yeah probably | 14:12 |
| arno11 | back @home in 2 weeks, then i can try new img if needed on both pi3 and pi4 | 14:14 |
| Wizzup | /dev/loop0 800M 450M 319M 59% /mnt/extern | 14:20 |
| Wizzup | btrfs zstd:15 compression :) | 14:20 |
| Wizzup | vs ext4 | 14:21 |
| Wizzup | /dev/loop1 771M 747M 0 100% /mnt/extern_1 | 14:21 |
| uvos | /dev/mmcblk1p18 1732608 3694591 1961984 958M Microsoft basic data | 14:40 |
| uvos | /dev/mmcblk1p19 3694592 5128191 1433600 700M Microsoft basic data | 14:40 |
| uvos | /dev/mmcblk1p20 5128192 31047679 25919488 12.4G Microsoft basic data | 14:40 |
| uvos | (Device Start End Sectors Size Type) | 14:41 |
| uvos | Wizzup: so netxt problem i have is that the locale is all wrong | 14:47 |
| uvos | Wizzup: freemangordon: please kick the kernel | 15:32 |
| Wizzup | ok | 15:37 |
| Wizzup | uvos: experimental or what | 15:37 |
| uvos | devel | 15:37 |
| Wizzup | uvos: so shall I aim for 700M or 958M ? | 15:37 |
| Wizzup | with btrfs compression 700M is easily doable | 15:37 |
| Wizzup | merlijn@gentoo-x13 /mnt/extern $ sudo du -sh . | 15:37 |
| Wizzup | 751M. | 15:37 |
| Wizzup | merlijn@gentoo-x13 /mnt/extern $ df -h . | 15:38 |
| Wizzup | Filesystem Size Used Avail Use% Mounted on | 15:38 |
| Wizzup | /dev/loop0 800M 450M 319M 59% /mnt/extern | 15:38 |
| Wizzup | seems like 700M might be better | 15:41 |
| uvos | cache is the 900MB parititon | 15:49 |
| uvos | p18 | 15:49 |
| Wizzup | uvos: right, but I also saw a 700mb one | 16:18 |
| Wizzup | uvos: ok I will investigate why the droid4 leste-config was there and send oyu an image if you're up for it | 17:18 |
| Wizzup | issuing kernel build | 17:19 |
| uvos | the 700mb is the other flashable partition: cdrom | 17:47 |
| uvos | obviously we want kexecboot on the smaller of 2 flashable partitions | 17:47 |
| uvos | Wizzup: sure i have leste on p20 now so im open to trying things on p18 | 17:47 |
| sicelo | tysm @kernel build | 17:51 |
| sicelo | those mapphones (i assume mapphone?) don't have sd card slots? | 17:51 |
| uvos | no the mz61x variants either have sdcard or a modem but never both at the same time | 17:52 |
| sicelo | i don't know much about mapphones ... they didn't really land in these parts. only motorola of those days that i think i ever saw was the photon q, or some other that looked similar | 17:53 |
| sicelo | Leste's D4 that I have is probably the only D4 in all the 9 or 10 SADC countries :p | 17:54 |
| sicelo | ah yeah, it was Motorola RAZR. Not sure which variant specifically, since I see a few on gsmarena | 17:57 |
| Wizzup | uvos: doesn't my mz616 have both? | 17:58 |
| uvos | Wizzup: nope | 17:59 |
| uvos | Wizzup: mz615 has sdcard and mz616 has modem | 17:59 |
| Wizzup | well this tablet has a sim slot and a (working) sd card slot\ | 18:02 |
| Wizzup | I haven't tested the sim slot I suppose | 18:02 |
| uvos | hmm | 18:03 |
| uvos | at | 18:03 |
| uvos | the very least no lte version also has sdcard | 18:03 |
| Wizzup | right | 18:04 |
| Wizzup | uvos: so don't hit RET either on mz617 right? | 18:11 |
| Wizzup | you don't* | 18:11 |
| uvos | no | 18:11 |
| uvos | it never has for me | 18:12 |
| uvos | (also minimal linux root) | 18:12 |
| Wizzup | ok | 18:12 |
| Wizzup | my suspicion is the audio but I'm not sure | 18:12 |
| uvos | Wizzup: do you have any mz617-32 variants? | 18:32 |
| uvos | or any mz61x-32 | 18:32 |
| Wizzup | I will check | 18:45 |
| Wizzup | (after dinner) | 18:45 |
| sicelo | fun, the charger uevent isn't working properly in the leste kernel, but works 100% in 6.10 that I was using while testing it. | 19:55 |
| sicelo | uevent is generated alright, but here, isp170x is much faster ... and finishes everything before bq2415x finishes updating itself | 19:56 |
| sicelo | i think our kernel is correct. anyway, no more time to play with this for a while. i will have a quick check to see what quickest solution is - probably blacklisting bq2415x again | 19:58 |
| sicelo | what's a good, simple way that i can get myself setup to build leste kernel locally, as a deb? | 20:00 |
| sicelo | i know how to build it the 'plain' way | 20:01 |
| uvos | sicelo: in theory https://wiki.debian.org/Multiarch/HOWTO and then add the arch to the dpkg-buildpackage commend (not sure what flag use help) | 20:06 |
| uvos | i have never done this except to build i386 on amd64 | 20:06 |
| uvos | but should work, kernel has should be extra easy | 20:06 |
| uvos | or just build in armhf quemu | 20:07 |
| uvos | please avoid submitting patches that you have not tested against the kernel in question | 20:08 |
| Wizzup | I just use a normal cross compile fwiw | 20:13 |
| Wizzup | no need to build the .deb | 20:13 |
| sicelo | Wizzup: then moving modules manually between pc and device? | 20:15 |
| Wizzup | uImage and all modules yes | 20:16 |
| Wizzup | just run make modules_install with the var set so install the mods to some dir | 20:16 |
| sicelo | uvos: sure. anyway, you won't need to revert the existing patch. it's already an improvement since we're getting an uevent when plugging in charger ... just not good enough yet | 20:19 |
| uvos | you can also just directly install to the device by mounting it with sshfs | 20:19 |
| uvos | which is what i do | 20:19 |
| Wizzup | guys, shall I also make trixie setup now, or does that make no sense? I worry it will be hard to keep sw synced with three repos, but I can do it ofc | 20:27 |
| Wizzup | I suppose it can't hurt to just make it work at least | 20:31 |
| sicelo | three? i'd say two at a time. i.e. bookworm, so we switch to it, and as soon as switch is complete, start with trixie. | 20:32 |
| Wizzup | ok | 20:33 |
| sicelo | s/so/until/ | 20:33 |
| sicelo | anyway maybe others have different idea :-) | 20:33 |
| Wizzup | uvos: so then shall I call these mz617-emmc-bootstrap, mz617-emmc en mc617-sd or something? | 20:35 |
| Wizzup | s/ en / and / | 20:35 |
| uvos | mc617-sd? | 20:39 |
| uvos | sure sounds sane otherwise | 20:39 |
| Wizzup | mz617-sd or mz617-sdcard | 20:41 |
| Wizzup | or just mmc I suppose | 20:41 |
| uvos | but mz617 has no sdcard | 20:41 |
| uvos | and mz615 will probubly need a different dtb | 20:41 |
| freemangordon | Wizzup: sorry, can't look at this ATM | 20:42 |
| uvos | so im not sure what you want with that | 20:42 |
| uvos | on mz617 the lack of hw video decoding really stinks | 20:42 |
| uvos | since its sutch a nice platform for viewing videos :( | 20:43 |
| Wizzup | freemangordon: at what? | 20:44 |
| freemangordon | that osso malloc stuff | 20:44 |
| Wizzup | uvos: you're right, sdcard makes no sense | 20:44 |
| Wizzup | freemangordon: ok, I might just #if 0 it and continue for now | 20:44 |
| Wizzup | since there's basically no users | 20:44 |
| Wizzup | then we can discuss later | 20:44 |
| freemangordon | yes, please do, I'll have a look when I am back in BG | 20:44 |
| Wizzup | take it easy :) | 20:44 |
| freemangordon | sure | 20:45 |
| freemangordon | arno11: so, that pastebin, there is no presence-ui button in h-s-m right? if that's the case, that means that by that time there are no telepathy accounts enabled/reported. I will prepare a version of the plugin with traces to see what exactly fails. | 20:46 |
| freemangordon | but, it is not h-s-m to blame | 20:47 |
| freemangordon | IIUC | 20:47 |
| Wizzup | uvos: it's still with leste-config-droid4, but: https://wizzup.org/maemo-leste-1.0-armhf-mz617-20241006.img.xz | 21:00 |
| uvos | Wizzup: ok cant test today | 21:09 |
| arno11 | freemangordon: indeed, there is no presence ui btn in hsm | 21:11 |
| Wizzup | uvos: no worries | 21:12 |
| arno11 | freemangordon: accounts seem not reported yet since it works if hsm starts later | 21:15 |
| sicelo | freemangordon: i suppose you read above that I'm back to my misery. but now i think i have better understanding of that charger driver, and at least some wq basics. i'm thinking that I need a second workqueue, specifically for running the `power_supply_changed()` function. will schedule it with a 2s delay, to match the timing characteristics of the device. | 21:38 |
| sicelo | the existing workqueue is specifically for kicking the timer every so often. i could do the power_supply_changed() inside it, but that'll send excessive uevents. | 21:40 |
| Wizzup | sicelo: if it works on in 6.10 but not on 6.x, should that give you some hint too? | 21:40 |
| sicelo | there are no differences whatsoever in the isp170x and bq2415x drivers in all those versions ... actually that's why i just expected things to work the same | 21:41 |
| Wizzup | annoying.. | 21:45 |
| sicelo | yeah, it has taken time that i really didn't have. anyway reward is having learned something | 21:46 |
| Wizzup | :) | 21:57 |
| Wizzup | forgot that maemo-system-services uses python-gconf, will need to port it to calling gconftool-2 | 23:21 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!