2022-02-07.log

- Ar|stote|is (QUIT: Ping timeout: 256 seconds) (~linx@149-210-32-252.mobile.nym.cosmote.net)00:04
+ Ar|stote|is (~linx@149-210-32-103.mobile.nym.cosmote.net)00:09
- mark_ (QUIT: Quit: Leaving) (~mark@gnu.wildebeest.org)00:12
- mtm- (QUIT: Ping timeout: 256 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)01:03
- mtm (QUIT: Ping timeout: 256 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)01:03
- Christoph_ (QUIT: Remote host closed the connection) (~Christoph@p54bf6bef.dip0.t-ipconnect.de)01:16
+ mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)03:08
+ mtm- (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)03:08
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:20)04:02
- sts-q (QUIT: Ping timeout: 250 seconds) (~sts-q@91.200.108.233)04:54
+ sts-q (~sts-q@212.53.219.232)04:57
+ FreXxX (~FreXxX@240f:104:ad8a:1:f083:3de6:7741:a06a)05:27
FreXxXHello everyone! Does someone ( mntmn ?) know if the standalone keyboard will be available again in this amazing green?05:28
FreXxXOr does one have to order in black for now and maybe get a green case later?05:28
- sts-q (QUIT: Remote host closed the connection) (~sts-q@212.53.219.232)06:04
+ sts-q (~sts-q@212.53.219.232)06:20
- erlehmann (QUIT: Ping timeout: 256 seconds) (~erle@ip5f5bd57a.dynamic.kabel-deutschland.de)08:50
+ erlehmann (~erle@ip5f5bd57a.dynamic.kabel-deutschland.de)09:03
+ MajorBiscuit (~MajorBisc@2a02:a461:129d:1:193d:75d8:745d:e91e)09:04
- MajorBiscuit (QUIT: Ping timeout: 245 seconds) (~MajorBisc@2a02:a461:129d:1:193d:75d8:745d:e91e)09:32
+ MajorBiscuit (~MajorBisc@c-001-017-058.client.tudelft.eduvpn.nl)09:34
- FreXxX (QUIT: Quit: Client closed) (~FreXxX@240f:104:ad8a:1:f083:3de6:7741:a06a)09:55
- V (QUIT: Ping timeout: 268 seconds) (~v@anomalous.eu)10:12
joschmntmn: I wanted to send you a ping about /ramdisk/builds/ on the CI runner. I guess I forgot to clean up something and the disk is full? Error message here: https://source.mnt.re/reform/reform-debian-packages/-/jobs/60911:09
- GNUmoon (QUIT: Remote host closed the connection) (~GNUmoon@gateway/tor-sasl/gnumoon)11:17
+ GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon)11:18
Boostisbettermntmn: have you ever thought about using the work you've put into the BMC on the Reform in a standalone power bank device? If you look across the industry you'll see that most are designed to not be user repairable and that after the batteries in it wear out, you need to buy a whole new bank. If you made one with the circuirty you've already designed, it could be low hanging fruit, in that it sho11:27
Boostisbetteruldn't take much work to create an enclosure, and make a complete product. If you could get something milled in the same material as the Reform itself, you would make a power bank, that like the Reform, was designed to last. I think it would sell like hot cakes. Just have it use 8 18650 lifepos like the Reform, and you have a powerful relatively compact power bank that can last as long as the cus11:27
Boostisbettertomer needs. 11:27
+ Christoph_ (~Christoph@p54bf63e6.dip0.t-ipconnect.de)11:47
mntmnBoostisbetter: interesting idea, thanks12:09
mntmnjosch: looks like we don't have enough space for both builds on the ramdisk12:11
mntmnjosch: i would suggest i delete your build dir, disable the runner on your repo and you do a MR on the main repo, lets see if it gets built in the same old directory12:13
joschmntmn: sure!12:32
mntmnjosch: ok, i already did step 1 and 2 ;)12:32
joschmntmn: the MR already exists here: https://source.mnt.re/reform/reform-system-image/-/merge_requests/36 Is there anything else to do on my end?12:33
mntmnhmm > Merge blocked: merge conflicts must be resolved12:39
joschah yes, you pushed "Update to Linux Kernel v5.17-rc2" into sysimage-v3. In the future, could you put the kernel patches here instead: https://source.mnt.re/reform/reform-debian-packages/-/tree/main/linux/patches12:42
joschBecause we want to install the kernel as a Debian package so that we can upgrade it using "apt-get upgrade" instead of building a uImage as part of reform-system-image12:42
joschI already put the v5.17-rc2 changes from reform-system-image into reform-debian-packages12:43
mntmnjosch: yes, alright12:45
mntmnjosch: btw how does the kernel update mechanism work? i.e. how does it find/mount the boot partition that contains the kernel?12:45
joschI didn't do any additional partitions so right now there is only one partition for /. Would you prefer an additional partition for /boot?12:46
joschThe package that does the magic with u-boot in Debian is called flash-kernel. On every kernel package upgrade it will place the kernel image and dtb in a known place where uboot can find it. This is what the following patch to flash-kernel does for the reform: https://source.mnt.re/reform/reform-debian-packages/-/blob/main/patches/flash-kernel12:47
mntmnjosch: ah, i was only thinking about the "boot from sd/emmc, system on nvme" case12:49
mntmnjosch: sorry, i should have said boot device, not partition12:49
joschYou mean a scenario where you still use the SD card to boot but the system is on nvme?12:50
mntmnjosch: yes, because the SoC can't boot directly from nvme12:50
joschThat would be done by putting the nvme partition in /etc/fstab and then re-running flash-kernel.12:51
mntmnjosch: i boot u-boot and kernel from SD card, then reform-init mounts the nvme (with luks support) and then pivots the root to that and runs init from the nvme12:51
mntmnjosch: ok12:51
joschSo your sd-card would contain /boot and not just u-boot?12:52
mntmnjosch: also, you have a new mechanism for single/dual display selection (a file called disable_hdmi that is checked by u-boot). did you already integrate this with reform-display-config?12:52
joschnope12:52
joschI'm trying to figure out how to have flash-kernel manage multiple dtbs for the same board.12:52
joschWhich dtb will be submitted for kernel upstream integration? Both?12:53
mntmnjosch: reform-display-config just overwrites the dtb on the boot device12:53
mntmnjosch: the one for the single display, because the HDMI driver is not in mainline12:53
mntmnjosch: a version of that dtb is already in mainline, it's only missing pcie internal refclock support IIRC 12:54
mntmnand sleep/wake support12:54
mntmnyeah so our remaining patches are really only to cover dual display+HDMI12:54
mntmnand a hack to support sleep via the vendor version of TF-A12:55
joschI'll have to put a lot more work into making the duo-dtb setup work because that's nothing other boards do.12:55
joschAnd currently the dtb is also managed by flash kernel because the dtb gets upgraded with the kernel itself.12:55
mntmnjosch: the only drawback of the dual display dtb is that the internal display is run by a less performant display engine (LCDIF instead of DCSS) and the panel is run at a lower freq12:56
mntmnso we can't just default to it12:56
josch:/12:57
joschso things will stay hacky for another while12:57
mntmnmaybe we could solve it with 2 debs that are alternatives or something?12:57
mntmnso you could switch with apt? only an idea12:57
mntmnyeah, things will stay hacky because we need to work around hardware limitations12:58
- mtm- (QUIT: Ping timeout: 250 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)13:02
- mtm (QUIT: Ping timeout: 250 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)13:02
joschI rebase MR !36 on top of the current sysimage-v3 branch. Next I'm working on making u-boot work with a boot.scr.13:03
josch*rebased13:04
mntmnjosch: thanks! i'm gonna merge so we can see if the pipeline works13:15
joschGreat. If something breaks i'll open another MR.13:16
joschYour fix of /ramdisk/builds also has worked and the reform-debian-packages pipeline is running again.13:17
mntmnok cool13:18
joschI moved both reform-debian-packages as well as reform-system-image from unstable to testing. In the past three weeks I was unable to work because unstable contained changes that made building images fail. This should not happen with testing.13:18
joschFor example right now the libwacom transition is underway (thus we cannot install anything that uses Qt5) and the perl5.24 transition is happening (hence no perl)13:18
mntmnohhh ok13:19
+ V (~v@anomalous.eu)13:50
+ mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)15:09
+ mtm- (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)15:09
joschmntmn: could you bump the CI pipeline timeout by a bit? The sysimage-v3 just failed because it reached the 1h timeout. Maybe bump by another 30 minutes to find out how long the job needs on that machine? Thanks!15:51
mntmnohhh ok... 1h is a lot! before it took like 45 mins16:05
mntmnincl. the kernel16:06
mntmnany idea why it would need this long?16:06
joschyes16:06
joschThe reason is genext2fs which turns the tarball into the image. Doing that takes half an hour so there is a bottleneck somewhere. I'm currently trying to find out whether genext2fs can be made faster.16:07
- qbit (QUIT: Quit: WeeChat 3.3) (~qbit@ns2.suah.dev)16:14
mntmnjosch: what was wrong with mke2fs ?16:32
joschmntmn: it cannot produce reproducible output16:33
+ qbit (~qbit@ns2.suah.dev)16:36
mntmnjosch: if you can make it faster, that would be great16:39
mntmn:D16:39
joschYup. I'm currently profiling. It looks like there is some O(n²) hidden somewhere...16:40
Boostisbettermntmn: it failed to parse everything I said:16:52
Boostisbetterin that it shouldn't take much work to create an enclosure, and make a complete product. If you could get something milled in the same material as the Reform itself, you would make a power bank, that like the Reform, was designed to last.16:52
Boostisbettermntmn: the last little bit: 16:53
BoostisbetterI think it would sell like hot cakes. Just have it use 8 18650 lifepos like the Reform, and you have a powerful relatively compact power bank that can last as long as the customer needs.16:53
Boostisbettermntmn: for the first time I have gotten two LPC freezes. I had to disconnect the batteries to get it to restart, but I did see that there is a reset button next to the power connectors on the mainboard. 16:55
Boostisbettermntmn: is there something that you are already aware of that could be triggering an issue? When the LPC freezes the LPC menu is up, but it is flickering with other display noise, and does not respond to commands or the arrow keys. 16:56
- erlehmann (QUIT: Ping timeout: 256 seconds) (~erle@ip5f5bd57a.dynamic.kabel-deutschland.de)17:52
+ erlehmann (~erle@ip5f5bd57a.dynamic.kabel-deutschland.de)17:55
- MajorBiscuit (QUIT: Quit: WeeChat 3.4) (~MajorBisc@c-001-017-058.client.tudelft.eduvpn.nl)18:39
- mjw (QUIT: Quit: Leaving) (~mjw_@2001:1c06:2488:1400:9e5c:8eff:fe8f:a440)18:58
+ mtm_ (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)19:08
+ nppvxexhf (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)19:08
- mtm- (QUIT: Ping timeout: 256 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)19:09
- mtm (QUIT: Ping timeout: 256 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)19:09
+ stjohn (~stjohn@cpe-68-174-13-238.nyc.res.rr.com)20:20
+ verx_ (~verx@matrix.16bit.dev)20:31
+ sbp_ (~sbp@2001:19f0:6800:1102:5400:ff:fe11:39a1)20:36
- jackhill (QUIT: *.net *.split) (~jackhill@kalessin.dragonsnail.net)20:37
- ex-parrot (QUIT: *.net *.split) (~fincham@user/ex-parrot)20:37
- sbates (QUIT: *.net *.split) (sid451432@id-451432.ilkley.irccloud.com)20:37
- verx (QUIT: *.net *.split) (~verx@matrix.16bit.dev)20:37
- kfx (QUIT: *.net *.split) (~kfx@wopr.sciops.net)20:37
- sbp (QUIT: *.net *.split) (~sbp@apache/doge/sbp)20:37
- bleb (QUIT: *.net *.split) (~cm@user/bleb)20:37
- TadeusTaD (QUIT: *.net *.split) (tadeustad@psi.koof.com.pl)20:37
* sbp_ -> sbp20:38
- sbp (QUIT: Changing host) (~sbp@2001:19f0:6800:1102:5400:ff:fe11:39a1)20:38
+ sbp (~sbp@apache/doge/sbp)20:38
* verx_ -> verx20:41
+ jackhill (~jackhill@kalessin.dragonsnail.net)20:42
+ ex-parrot (~fincham@user/ex-parrot)20:42
+ sbates (sid451432@id-451432.ilkley.irccloud.com)20:42
+ bleb (~cm@user/bleb)20:42
+ TadeusTaD (tadeustad@psi.koof.com.pl)20:42
+ kfx (~kfx@wopr.sciops.net)20:55
+ mjw (~mark@gnu.wildebeest.org)21:14
- GNUmoon (QUIT: Ping timeout: 276 seconds) (~GNUmoon@gateway/tor-sasl/gnumoon)21:59
- sts-q (QUIT: Quit: Mezzano-OS running Pet IRC client says goodbye soon.) (~sts-q@212.53.219.232)22:03
+ jryans (~jryans@2001:470:69fc:105::1d)22:36
+ scops (~scopstchn@2001:470:69fc:105::8da)22:38
+ Lewis[m] (~lewislewi@2001:470:69fc:105::e363)22:39
+ maxim[m] (~maximtchn@2001:470:69fc:105::1f39)22:39
+ nio (~nio@2001:470:69fc:105::172d)22:39
sigridugh... power cord looked like was full in, but not in enough :/ now i have a bunch of overdischarged batteries22:42
sigrideh, maybe not. will see if it charges22:45
+ GNUmoon (~GNUmoon@gateway/tor-sasl/gnumoon)23:17
- mjw (QUIT: Quit: Leaving) (~mark@gnu.wildebeest.org)23:31
- q66 (QUIT: Quit: q66) (~q66@q66.moe)23:39
+ q66 (~q66@q66.moe)23:39
joschmntmn: could you refresh https://mntre.com/reform-debian-repo/ ? The last pipeline for reform-debian-packages was successful and I want to make sure that everything that was built works as expected.23:51
+ vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:20)23:53

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!