2021-10-17.log

+ savely (~savely@2a02:2698:2807:219c:692:26ff:febe:c52d)00:27
savelyhi there, so i've finally installed arch linux on mnt reform, seems to work pretty fine, currently don't see any difference with debian-based in case of quality of life00:30
- mjw (QUIT: Quit: Leaving) (~mark@gnu.wildebeest.org)00:30
savelybut still wanted to ask about mesa, do i really need this patches? or they already was merge?00:30
chartreuseI think almost every patch is mainlined now, just a couple missing when I updated my kernel to 5.1400:31
chartreuseFor the kernel, no clue on mesa00:31
savelyoh, even kernel patches was mainlined?00:31
savelyi thought it is a long process  00:32
chartreuseThere's a few that have made it in/fixed00:32
chartreuseThere's still a few left, the patches are for the imx8mq the processor module, not really custom stuff for the reform00:32
savelyso has someone a repo to build the latest kernel with only needed patches?00:33
chartreusehttps://source.mnt.re/Chartreuse/reform-system-image/-/tree/main/reform2-imx8mq00:35
chartreuseThat'll build 5.14, I haven't tested later yet, that's what I'm running00:35
savelythanks! didn't know about forks00:35
savelyalso seems like i kill at least some of my batteries, had a vaccation, didn't know that a should pull off the batteries...00:36
chartreuseIt at least built correctly for the git 5.14 around the 5.14.2 release so should be fine00:36
chartreuseAh yeah that issue was recently helped with a new LPC and keyboard firmware to reduce the draw on the batteries when off00:37
chartreuseIf you have an external lifepo4 charger you can probably bring them back from 0v, to a point the reform will charge them00:37
savelyhow much fast it discharges while off now?00:37
savelychartreuse yep, i already ordered a charger... but anyway i didn't expect this, modern electronics teach you to forget about such things...00:39
chartreuseI don't have anything to measure it properly, and I think mntmn might be working on getting tools for that. But in quick test I think I lost 0.04v in around 23hr, which if the drop was linear would equate to half a year or something, though that's not an accurate way to test00:39
vagrantcit's much better with the updated firmware00:39
chartreuseYeah the MNT really should have had a low-voltage protection IC installed00:39
chartreuseBut it's definitly much better since both the LPC and keyboard controllers enter deep sleep mode now while the system is off, only briefly waking to see if the user is attempting to turn the system on00:40
chartreuseBefore those two were running all the time, so uA draw vs mA draw really00:40
savelyis there space for improvement btw?00:41
chartreuseNot sure there's much else on top of that to gain, even low-voltage battery cutoff ICs draw some uA all the time00:42
vagrantccould the firmware be written to stay in deep sleep longer when the voltages are low ? e.g. oh, your battery is very low, not even going to bother to wake up as often00:42
chartreuseMaybe, I haven't looked at the changes mntmn made to the LPC firmware in depth, but the keyboard controller is in the arduinos deepest sleep more, and only wakes up for a few cycles every second to see if the circle key is being pressed (It's on a pin that can't directly interrupt the CPU)00:43
chartreuseIf we had something like a micro-current to measure actual draw, it could be tested if there's ways of configuring the pins to save a little more00:43
chartreuseBut I think with the changes it'll probably last over a year without going down to 0v00:44
chartreuseNot sure if the charge circuit can be convinced to work with fully discharged cells or not00:44
vagrantcyou want to keep each cell above 2v to not wear them down badly...00:44
vagrantcand really 2.5v00:44
chartreuseYes, and I think it'd last a long time before hitting that, though someone will need a micro current or similar device to measure the current draw at the batteries in shut-down00:45
chartreuseEven the low-voltage cutoff chips still draw current from the battery in the cutoff state typically00:45
chartreuseAnd of course cells also self discharge slowly too00:45
vagrantctrue00:46
savelyalso maybe there is a way to somehow patch firmware to charge batteries even while they have so low voltage?00:46
chartreuseIdeally with the changes I'd expect in the <10uA range of draw, but I have no clue about the LPCs behaviour in its low power mode00:46
vagrantcthough lifepo4 cells discharge pretty slowly00:46
chartreuseYou'd have to ask mntmn about that, I'm not quite sure how the battery charging works00:47
vagrantcindividual lifepo4 cells was one of the main points that convinced me about the mnt/reform ... that and it can take a voltage input directly out of my solar battery management controller :)00:47
chartreuseYeah it's a neat feature and at least they're reasonably priced if they do get killed somehow00:48
savelyFor sure... I didn't expect to order batteries after a month since I get it...00:49
chartreuseCharging is managed by the LTC4020 chip00:49
chartreuseYou should be able to bring them back up, just need to get them above maybe 1.5-2v. If you have a current limited bench PSU you could charge them with that00:50
chartreuseSet the voltage limit to say 3v, and current limit at 100mA, and that'll get them to a state that the reform will be happy with00:50
chartreuseAnd update the reform LPC and keyboard firmwares, the update also improves the battery balancing00:51
savelybtw maybe i really have it, will check tomorrow... thanks for the tip00:51
savelyyep, i will update it now, happily reform works fine even without batteries00:52
chartreuseUnless you just got your reform maybe 2 weeks ago at most, it'll be on the original firmware00:52
chartreuseI'm just another owner like you so I don't have all the answers 00:53
savelywell, i wouldn't be in this situation if i had a new firmware00:53
savely'=D 00:53
chartreuseHave you tried just leaving it plugged in but off with the dead batteries? I'm reading through the datasheet of the LTC4020 to see how it handles the battery voltage being low00:55
savelyi could try to leave it for a night or even longer00:56
savelybut currently i pulled them off00:56
chartreuseI think it might actually be controlled via the LPC which is what can sense the voltage, so it might not be turning on the DC-DC, I'll have to read its source00:57
+ freakazoid12345 (~matt@2603:9000:cf0f:80e3:cdc5:a33b:1284:64f8)00:57
savelyi am afraid that in this case they will continue to discharge00:58
- freakazoid343 (QUIT: Ping timeout: 265 seconds) (~matt@2603-9000-cf0f-80e3-cdc5-a33b-1284-64f8.inf6.spectrum.com)01:00
chartreuseIt's possible, but with it plugged in it'll draw from the external unless the load is greater than the external can supply01:01
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:cdc5:a33b:1284:64f8)01:02
chartreuseLike it sounds like it should be charging 'slowly' even when it detects undervoltage cells01:03
chartreuseAs long as the LPC is running01:03
vagrantcwonder why the acrylic bottom rather than a metal case all around ...01:04
chartreuseIt doesn't seem to ever 100% turn off charging, just greatly limit it01:04
chartreuseI think the reason was aesthetics, to show off the insides01:05
chartreuseAlso benefit of not having a chance to short on the battery contacts I guess01:05
- freakazoid12345 (QUIT: Ping timeout: 268 seconds) (~matt@2603:9000:cf0f:80e3:cdc5:a33b:1284:64f8)01:05
chartreuseI think if I'm understanding this right even with a fully discharged cell the charge controller should still be putting out ~1% of full charge current01:11
mntmnsavely, chartreuse: with the new firmware, iirc there is a better chance of undervolted cells to be recharged01:57
savelymntmn i will try, but currenty trying to build it (currently cannot because arch arm does not have arm-none-eabi-gcc i guess)01:58
mntmni see01:58
savelyi built reform2-lpc-fw successfully at x86 pc, but i am not sure is this cross-compile build for aarch64 or not...02:01
savelymanaged to install arm-none-eabi-gcc on arch, lpc firmware builts fine, but cannot build keyboard on both x86 pc and reform...02:33
mntmnsavely: keyboard is avr02:44
mntmnsavely: apt install gcc-avr02:44
- vagrantc (QUIT: Ping timeout: 264 seconds) (~vagrant@2600:3c01:e000:21:21:21:0:100b)02:46
savelymntmn i installed it ofc, i am getting this error: https://paste.sr.ht/~lord/940756f217e2dc12abf6190137b3f7642a8f5fde02:50
mntmnwat02:51
mntmnthis is with reform repo from source.mnt.re, master branch?02:52
savelyi've just updated lpc firmware and it started to show battery voltage! it's currently low for sure, but i hope it started to charge batteries!02:52
savelymntmn yes!02:52
mntmnsavely: that's really weird, i will look into it tomorrow, it is sunday 3am here02:52
savelybut i use arch, maybe newer tooling has another behavior...02:53
savelymntmn sure, thanks for the help!02:53
mntmnsavely: if you don't have the kbd firmware updated, you won't be able to easily wake up lpc from sleep. the solution is to connect the laptop to power then02:53
mntmnsavely: this could be. try to compile the fw on reform itself. that's anyway the most comfortable way to update the kbd fw02:54
savelymntmn yes, but my reform also on arch currently...02:54
mntmn(you need a usb keyboard or a trick to do the actual flashing of the kdb fw once it is in programming mode though)02:54
savelyand i am getting the same error02:54
mntmnsavely: oh, ok then. did you integrate the systemd shutdown hook? or made otherwise sure that the laptop is turned off properly after shutdown?02:55
mntmni mean, powered off02:55
savelyit was planned, i saw this in reform-system-image repo, but currently not02:56
mntmnsavely: reform-tools deb includes that script02:56
mntmnsavely: ok, then you have to manually turn off reform after shutdown using the keyboard combo02:57
mntmnotherwise a lot of stuff stays on02:57
savelyyes, i notices this behavior02:57
mntmnok good02:57
mntmnjust wanted to make sure02:57
savelybatteries seems to charge02:58
mntmnthat's good news02:58
+ freakazoid12345 (~matt@2603-9000-cf0f-80e3-cdc5-a33b-1284-64f8.inf6.spectrum.com)03:01
- freakazoid343 (QUIT: Ping timeout: 268 seconds) (~matt@2603:9000:cf0f:80e3:cdc5:a33b:1284:64f8)03:04
- freakazoid12345 (QUIT: Read error: Connection reset by peer) (~matt@2603-9000-cf0f-80e3-cdc5-a33b-1284-64f8.inf6.spectrum.com)04:05
+ freakazoid12345 (~matt@2603-9000-cf0f-80e3-cdc5-a33b-1284-64f8.inf6.spectrum.com)04:06
savelymntmn: so i finally flashed keyboard (using Debian 11 VM to build fw). but i found the bug, in case of holding circle key keyboard became totally not responsive and the display starting to blink, nothing except pulling out batteries helped, tried to press any keys04:17
chartreuseDid you flash the LPC as well? Not sure what that would be04:28
savelychartreuse: yes, flashed both04:29
savelyit works okay if i am just pressing it shortly, but it requires few clicks (to get into right the moment of polling i guess?)04:30
chartreuseFor waking up from off you need to hold it for up to a second, 04:31
chartreuseI polls every second to see if it should wake from deep sleep04:31
chartreuseNot sure why constantly holding it would do anything, it doesn't do anything while running normally04:32
savelyhm, maybe i am really did something wrong while flashing LPC04:33
chartreuseWell if it went wrong you shouldn't be able to boot at all 04:33
savelyrechecked, dd definitely uploaded a new firmware.bin04:36
+ verpolter (~verpoler@dyndsl-085-016-201-138.ewe-ip-backbone.de)05:08
- verpoler (QUIT: Ping timeout: 245 seconds) (~verpoler@host-091-097-112-026.ewe-ip-backbone.de)05:10
- freakazoid12345 (QUIT: Remote host closed the connection) (~matt@2603-9000-cf0f-80e3-cdc5-a33b-1284-64f8.inf6.spectrum.com)05:37
+ freakazoid12345 (~matt@2603:9000:cf0f:80e3:cdc5:a33b:1284:64f8)05:37
- robin (QUIT: Ping timeout: 265 seconds) (~robin@user/terpri)05:50
+ robin (~robin@user/terpri)05:55
- robin (QUIT: Ping timeout: 245 seconds) (~robin@user/terpri)06:06
+ robin (~robin@user/terpri)06:32
savelychartreuse for reason cannot build valid 5.14 from your repo. latest patch mnt5000-imx8mq-import-HDMI-driver-and-make-DCSS-compatible.patch returns an error "error: corrupt patch at line 3202". without this patch i cannot get display work :(07:10
savelychartreuse: made i patch for a patch :)  https://paste.sr.ht/~lord/6fc291d8621e2c2a45a93b03bb848219877ab90307:39
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:9d79:7459:1d29:717f)09:33
- freakazoid12345 (QUIT: Ping timeout: 268 seconds) (~matt@2603:9000:cf0f:80e3:cdc5:a33b:1284:64f8)09:37
+ Christoph (~Christoph@p54bf6bbc.dip0.t-ipconnect.de)11:30
+ mjw (~mark@gnu.wildebeest.org)12:03
- rasmus-mk (QUIT: Remote host closed the connection) (~rasmus-mk@c80-217-132-63.bredband.tele2.se)12:48
+ rasmus-mk (~rasmus-mk@c80-217-132-63.bredband.tele2.se)12:48
- verpolter (QUIT: Quit: Leaving) (~verpoler@dyndsl-085-016-201-138.ewe-ip-backbone.de)13:00
+ doctorhoo (~hanno@194-18-252-127-no2005.tbcn.telia.com)13:42
doctorhoohej!13:43
doctorhooI have been following the project for a little while and have been increasingly impressed by the final product but also everything around it (including the community!) -- so I just placed my order :)13:45
doctorhooI will continue to linger here why I practice my patience ;)13:46
doctorhoojust a small note to mntmn: on the shop's order confirmation page, the pdf invoice link is broken, it only leads to a page saying "Not Found". That might just be the unique combination of configuration parameters (NoScript et al) on my end.. the email came through in any case.13:48
mntmndoctorhoo: thank you! sorry for the pdf issue, normally the pdf will be available after a few minutes14:08
doctorhoono worries, mntmn, I just thought I'd flag it -- I can do without the pdf :)15:27
- mjw (QUIT: Quit: Leaving) (~mark@gnu.wildebeest.org)15:31
- doctorhoo (QUIT: Ping timeout: 252 seconds) (~hanno@194-18-252-127-no2005.tbcn.telia.com)15:31
savelyafter night it i cannot turn it on with circle button, even holding it for a few seconds15:55
savelyonly without batteries and power from wall it restarts 15:56
savelysomething definitely wrong15:56
savely@mnt15:56
savelymntmn: btw, i have a question about kernel size and kernel config. I am still having an issue with WireGuard, probably because of unusual kernel config. i tried to compile Image with default Arch ARM kernel config from their repo, but Reform don't starts. Is because kernel size limitations or maybe Reform requires some special kernel config lines?16:00
mntmnsavely: i suggest you compare your kernel config with the one in reform-system-image (template-kernel/kernel-config), also suggest you get a UART adapter to check the serial output during boot16:20
+ S0rin (~S0rin@user/s0rin)17:15
+ freakazoid12345 (~matt@2603-9000-cf0f-80e3-9d79-7459-1d29-717f.inf6.spectrum.com)17:29
savelywireguard definitely works, it is really something with process substitution that for some reason disabled or broken17:32
- freakazoid343 (QUIT: Ping timeout: 268 seconds) (~matt@2603:9000:cf0f:80e3:9d79:7459:1d29:717f)17:32
- savely (QUIT: Ping timeout: 256 seconds) (~savely@2a02:2698:2807:219c:692:26ff:febe:c52d)17:41
+ mjw (~mark@gnu.wildebeest.org)17:54
+ savely (~savely@2a02:2698:2807:24b4:692:26ff:febe:c52d)17:57
savelywell, i managed to get it work with iptables (with nft it cannot for some reason): CONFIG_NETFILTER_XT_CONNMARK=y CONFIG_IP_NF_RAW=y CONFIG_IP6_NF_RAW=y18:03
savelyworks only for ipv4, ipv6 seems to be broken18:04
+ bibliocar (~quassel@pool-74-106-16-80.syrcny.fios.verizon.net)18:04
savelybut it's usable for now at least!18:06
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:ecd7:c30b:133e:765f)18:17
- freakazoid12345 (QUIT: Ping timeout: 268 seconds) (~matt@2603-9000-cf0f-80e3-9d79-7459-1d29-717f.inf6.spectrum.com)18:20
mntmnsavely: oh interesting. i now remember i might have some notes about this already18:22
savelyalso i reflashed LPC and seems like bug with keyboard has gone, at least i can now hold circle key safely18:24
savelymaybe the order of flashing really matters18:24
mntmnsavely: hmm, i doubt it but maybe lpc was in a strange state18:25
mntmnsavely: but glad it works... how are your battery voltages now?18:25
savelyeverything ok! currently every battery has 3.2 (just plugged it off)18:26
mntmnthat's cool18:27
mntmndo you remember the lowest voltage they were before?18:28
mntmni'm curious what it was able to recharge18:28
savelymntmn: yes, i some has 1.1, but the lowest was 0.25-0.318:28
savelyi used multimeter18:29
savelybut after LPC update Reform showed similar values18:29
mntmnwow ok18:30
- freakazoid343 (QUIT: Remote host closed the connection) (~matt@2603:9000:cf0f:80e3:ecd7:c30b:133e:765f)18:35
+ freakazoid343 (~matt@2603-9000-cf0f-80e3-ecd7-c30b-133e-765f.inf6.spectrum.com)18:35
savelymntmn: do you have any ideas why reform sound card cannot be initialized randomly for some reason? i see it in pavu with 20% chance after boot, and it works in that cases just like with Debian-based image. where i can dig?18:40
mntmnsavely: it fails with a "wm8960: can't issue reset" or similar in dmesg right18:41
savelyFailed to issue reset, yes18:42
mntmnsavely: i don't know what the issue is, yet. i wonder if adding some code in the driver to retry resetting it with a delay could work. maybe it is a race with initializing the rtc chip which is on the same i2c lines.18:42
mntmnmy hunch is that it is i2c related18:43
savelyso it's "known bug"? 18:43
mntmnyes18:43
savelygot it18:43
mntmnbut for me it happens rarely18:43
mntmnmore like it works at least 80% of the time18:43
savelywith arch it happens as i said in almost every case :(18:43
mntmnyeah the distro should have no effect here, but the kernel18:44
savelyhm, you are probably right, i didn't test it much with 5.12, mostly used chartreuse 5.14 kernel18:45
mntmnsavely: you could try to build wm8960 as a module (.ko) and load that later18:46
mntmnsavely: that way you could also retry it if it fails... with rmmod/insmod18:47
savelymntmn thanks, will try!18:47
mntmnsavely: there's also a reform audio patch on lkml from lucas stach that sets up wm8960 a bit differently, with different clocks. haven't tested it yet18:48
mntmnmainline has a bunch of mnt reform support now, but missing display and pcie118:49
savelymntmn: i got excited early, holding a circle key still dangerous in my case. i tried to turn it off and on 10 time at least, got this bug 3/1018:57
savelyi guess i can easily record a video18:58
- savely (QUIT: Quit: Client closed) (~savely@2a02:2698:2807:24b4:692:26ff:febe:c52d)19:15
+ savely (~savely@2a02:2698:2807:24b4:c6e9:84ff:fe1d:3728)19:21
- savely (QUIT: Quit: Client closed) (~savely@2a02:2698:2807:24b4:c6e9:84ff:fe1d:3728)19:27
+ savely (~savely@2a02:2698:2807:24b4:c6e9:84ff:fe1d:3728)19:28
- savely (QUIT: Client Quit) (~savely@2a02:2698:2807:24b4:c6e9:84ff:fe1d:3728)19:30
+ savely (~savely@2a02:2698:2807:24b4:692:26ff:febe:c52d)19:31
savelyuploaded video here, lazy to setup mp4 streaming, so: curl -LO https://krasovs.ky/files/reform-bug.mp419:46
- freakazoid343 (QUIT: Read error: Connection reset by peer) (~matt@2603-9000-cf0f-80e3-ecd7-c30b-133e-765f.inf6.spectrum.com)19:46
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:ecd7:c30b:133e:765f)19:47
- S0rin (QUIT: Quit: WeeChat 2.3) (~S0rin@user/s0rin)19:51
+ S0rin (~S0rin@user/s0rin)19:55
- wiedi_ (QUIT: Quit: Textual IRC Client: www.textualapp.com) (~wiedi@2a01:138:a015:15:f9d5:3929:ef93:d98f)20:00
- freakazoid343 (QUIT: Read error: Connection reset by peer) (~matt@2603:9000:cf0f:80e3:ecd7:c30b:133e:765f)20:02
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:ecd7:c30b:133e:765f)20:03
+ wiedi (~wiedi@2a01:138:a015:15:ed6e:1b0d:1f97:9f87)20:16
- bibliocar (QUIT: Ping timeout: 265 seconds) (~quassel@pool-74-106-16-80.syrcny.fios.verizon.net)20:24
+ nerd (~quassel@pool-74-106-16-80.syrcny.fios.verizon.net)20:25
- S0rin (QUIT: Ping timeout: 252 seconds) (~S0rin@user/s0rin)21:22
+ S0rin (~S0rin@user/s0rin)21:26
- nerd (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~quassel@pool-74-106-16-80.syrcny.fios.verizon.net)21:41
vkoskivI rejuvenated an old iBook battery with fresh cells yesterday:22:08
vkoskivhttps://teensyimg.com/img?i=IbdNwPj0XA22:08
vkoskivWhile spot welding is extremely satisfying to do, I very much look forward to the 18650 slots in my Reform :D22:08
vkoskiv(the suspicious looking image hosting site is one I built a few years ago)22:11
vkoskivI still use it personally. No ads, no tracking BS :D22:11
+ freakazoid12345 (~matt@35.136.210.185)23:23
- freakazoid343 (QUIT: Ping timeout: 265 seconds) (~matt@2603:9000:cf0f:80e3:ecd7:c30b:133e:765f)23:26

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!