libera/#maemo-leste/ Saturday, 2025-02-08

dsc_freemangordon: nice :)02:03
Wizzupfreemangordon: ping, I am available if you wanted to talk about h-a-m13:44
Wizzupok, so at least the latest droid4 chimaera image boots fine13:49
arno11Wizzup: do you have time to have a quick look @u-boot stuff in daedalus n900 ?14:06
Wizzupyes14:10
Wizzupdownloading now14:10
WizzupI wonder if it is daedalus related or somehow tmpfs related (which I added)14:10
arno11ah ok14:11
arno11btw assistant said it could be permission or directory related14:11
arno11(from irc logs)14:12
Wizzupha14:15
Wizzupok14:15
WizzupI need to make a call, after that I will look in more detail at your msgs14:16
arno11ok14:19
Wizzuparno11: looks like the last chimaera build also had this error: https://phoenix.maemo.org/job/leste-image-n900/225/console15:29
arno11oh15:33
arno11(dd daedalus img again)15:34
arno11maybe something went wrong yesterday15:34
WizzupI am dd'ing now as well15:35
arno11ok15:36
WizzupI am adding a mkdir -p for the u-boot.img15:36
Wizzupbut let's see15:36
arno11yep15:36
arno11trying to boot15:38
arno11oh15:39
arno11it seems boot logs are not activated15:40
arno11lol15:40
arno11it works, after waiting one min15:40
arno11it is quite fast15:41
arno11but console says 'cannot change locale en_GB, no such filr or dir15:42
arno11settings menu is...weird15:43
arno11it is emmc swap15:43
arno11but no issue with it at the moment15:44
arno11almost no software15:45
arno11no wifi, no clock, almost no settings15:48
arno11so i can't test anything else15:49
Wizzupok, let me check as well15:49
Wizzupwhat image did you dl, just to confirm15:49
arno11but the UI itself is fine, i mean H-D is very responsive for a first boot15:50
arno11last one15:50
arno112025020615:50
Wizzupit might be a tiny img, I'll check15:55
Wizzupcertainly sounds like it is15:55
Wizzupbrb, 20 mins15:55
arno11ah yes ok15:55
arno11ok15:55
arno11yeah indeed, there is almost no usual software in the img16:36
arno11but at least, it boots fine16:37
arno11H-D, xorg and gpu seems ok16:38
arno11and 6.6.53 with emmc swap is ok with no freeze...16:38
Wizzupstrange, it should not be a minimal image17:24
Wizzupchecking17:24
arno11ok17:24
Wizzupaaah17:25
WizzupI forgot to push a commit I think17:25
arno11ok17:26
Wizzupsilly shell mistake17:27
Wizzupbut yes, it also boots for me17:27
Wizzupit should do wifi though17:27
arno11wifi 'works' apparently but the menu is not usable17:27
Wizzupyeah the tiny images are not meant to be super usable but to be < 700MB17:28
Wizzupso I suspect this is also btrfs then17:28
Wizzupah, no, it is ext417:28
Wizzupinteresting17:28
Wizzupmaybe that was mz617 only17:28
Wizzupanyway, I will build a new image now17:28
arno11ok cool, you should check @d4 img too, as it seems similar17:29
arno11@tiny img, it could be useful to test trixie imo17:29
Wizzupwell, there's a lot more ot build before we can do that17:32
Wizzupeven the tiny image has a lot of deps17:32
arno11ah yes ok17:32
arno11git submodule seems to timeout again btw17:40
arno11even with timeout=3017:40
arno11scratch that, it works17:59
Wizzup:)17:59
WizzupI bought two 2TB SSDs to run this stuff on17:59
Wizzupbut then one of them dropped off the sata bus and I am ~2000km away from the machine18:00
arno11omg18:00
WizzupI will try to get someone to be close and restart it, and then we can move it to ssd again rather than hdd18:00
Wizzupso now it's on hdd, not ssd18:00
arno11okay18:00
WizzupI didn't want it to be on a single ssd18:00
arno11indeed18:01
WizzupI managed to bring the ssd back with a sata reset / rescan18:16
WizzupI'll check later today if I can move some ephemeral stuff to it and see if it doesn't brek aagain18:16
Wizzupbbiab18:16
Wizzupbreak again18:16
arno11ok, btw now building seems ok, i see conversations, calendar stuff18:24
arno11bbl18:25
sicelothat uboot*.img thing looks wrong. we don't need to build uboot for n900. we only need a boot.scr (generated via `mkimage`)18:39
siceloyes looks like it's building uboot. it shouldn't. in fact n900 was dropped in uboot upstream18:44
Wizzupsicelo: I think we build a specific release19:28
Wizzupand we do need a u-boot for people who want to flash it19:28
Wizzupdirectly19:28
Wizzupwith 0xFFFF19:28
Wizzuphttps://maedevu.maemo.org/images/n900/tools/ although I guess we have u-boot here20:04
freemangordonWizzup: sorry, I am doing other (non-leste) things now, maybe tomorrow20:16
Wizzupnp20:19
Wizzupfrom my pov, we mostly need to sort out upower, otherwise we're ok, at least for daedalus images and dist-upgrade20:20
freemangordonyep20:21
freemangordonjust to think about, re HAM: what about if on every update I distribute packages without domain to the respective one? based on where the package was installed from.20:23
WizzupI'll think about it, but from my pov I want to minimise the amount of 'hacks' required for ham to work like apt20:24
freemangordonWizzup: no change in apt20:47
freemangordonby 'update' I mean 'HAM update'20:47
Wizzuparno11: we probably also disabled parallel apt updates20:49
Wizzupthat I expect also really helps with usability20:50
Wizzupsee the latest leste-config changes20:50
arno11yes i saw them20:51
arno11btw n900 stays usable during update/install when H-D, mce, xorg and systemui are renice'd (without overclock)21:02
arno11(i'm currently installing jib, no slowdown during the process)21:04
siceloWizzup: already experiencing issues with upower?21:12
Wizzupsicelo: sure, I have to disable it on mz61721:16
Wizzupit immediately seems to force a shutdown21:16
Wizzupit's sad to be honest :D21:18
Wizzupbut we can build a newer version I guess21:18
Wizzupthe cpio performance on loopback that is backed by tmpfs is surprisingly slow21:24
Wizzupand it failed, huh21:26
siceloyeah, guess for daedalus we need to fork the latest upstream upower, and ship a customized UPower.conf21:33
arno11Wizzup: argh, build failed21:36
Wizzuparno11: yes, but the good news is that I am looking into the IO issues21:38
arno11ok :)21:38
WizzupI think somewhere the no cow (copy on write) bit got lost so the performance just went to hell with btrfs raid21:38
WizzupI'm defragging the file and will set +C21:38
Wizzup(which disables copy on write)21:38
arno11ah ok21:38
Wizzuptl;dr I probably shouldn't run this on btrfs with copy on write on for the file, but it's what I have raid wise21:41
Wizzupbut I think I cna make it work ok21:41
arno11cool21:43
WizzupI hope things will fly after this, but we'll see21:43
WizzupI suspect that this is also why git was timing out21:43
arno11ah yes21:55
Wizzupalmost done22:22
Wizzupand yes, filefrag showed a rather unfortunate fragmentation22:22
arno11ok cool22:29
Wizzuphttps://phoenix.maemo.org/job/leste-image-n900-daedalus/8/console22:51
Wizzuplet's see22:51
arno11yep22:53
arno11no more big timeout on git submodule :) so good start22:55
Wizzupyeah, I bet it'll be much faster now23:07

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