libera/#devuan/ Friday, 2021-09-24

junicchianyone tried plymouth on devuan?01:20
junicchiwondering if it'd work on my openrc setup01:20
n4dirthe info on the gentoo wiki seems to be too gentoo specific. Perhaps a bit can be found anyway01:25
golinuxjunicchi: Looks like it's in Chimaera and Ceres   https://pkginfo.devuan.org/cgi-bin/policy-query.html?c=package&q=plymouth&x=submit  But there are parts of it here too: https://pkgmaster.devuan.org/bannedpackages.txt01:30
junicchiyeah that's why i'm wondering01:58
junicchiit's both on repos and is banned01:58
junicchiand seems like no one tried so far01:58
n4dirfor no other reason but fooling around i have choose openrc on this machine. It seems to work very well (say 1 week later)02:03
n4dirboot time is very fast, RAM usage is very good. I don't really fiddle with init systems, so more i can't say02:05
ham5urgHave anyone tried kdump? I still have systemfreezes randomly on one of my machines after I switched it to Devuan. I can't locate the problem, neither the logs help me. Anything I can do after freeze is a hard reset.12:57
sadoon_albader[mWhat hardware are you running and what version of devuan?12:59
ham5urgBeowulf and the machine is a laptop with i5-8365U13:00
ham5urgI tried the standard-kernel as well as the one from the backports13:01
gnarfacesame freeze with both of them?13:01
ham5urg5.10.0-0.bpo.8-amd64 currently13:01
gnarfaceand you memtested it, right?13:01
ham5urgyes, same freeze13:01
ham5urgyes, memtest was good.13:02
gnarfacenothing in the logs at all?13:02
gnarfacenot even the xorg log?13:03
gnarfacethe .old one should be the one from the previous run13:03
gnarfaceso if it crashed and you had to hard reboot there should still be an error in there if it was anything to do with the graphical stack13:04
gnarfaceotherwise check /var/log/*13:04
gnarfaceparticularly syslog, kern.log and daemon.log13:04
gnarfaceeh i dunno13:06
ham5urgXorg.0.log.old is fine. Yes, I switched over to the console ctrl+alt+f1 and used VIM all day and next day it freezed 3 min after resuming suspend-to-ram.13:07
ham5urgI guess it is not X13:07
gnarfaceoh, is it always after a resume from suspend?13:07
ham5urgNo, can't say that.13:07
gnarfacewhich cpu governor are you using?13:08
gnarfaceyou might want to try setting it to userspace then setting a static frequency other than the max one13:09
gnarfacei've had some stability issues associated with the cpu governors under certain types of workloads13:10
ham5urg# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor13:11
ham5urgpowersave13:11
gnarfacehmm, that shouldn't change it either13:11
ham5urgWhen I use Ubuntu, I don't encounter system-freeze. Maybe they use a different governor. The battery lasts longer with Devuan.13:12
gnarfaceupstream default i think is schedutil now13:13
gnarface?13:13
gnarfacesomething liek that13:13
gnarfacei can't imagine powersave would be less stable than any others but it might be worth a try13:14
ham5urg# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors13:15
ham5urgperformance powersave13:15
ham5urgI will try13:15
gnarfaceoh you can get more governors13:15
gnarfacethe kernel modules are installed they just dont' all load by default13:15
gnarfacefind /lib/modules/ -iname 'cpufreq*'13:15
gnarface"ondemand" was the previous upstream default i believe13:16
gnarfaceperformance would definitely use the most battery13:18
ham5urgSeems "performance" is a built-in. All others are available as modules.13:18
gnarfaceit's strange to imagine it would be less stable than powersave, but i guess it would kick the fans in13:18
gnarfacedo you monitor temperatures on this thing?13:18
ham5urgNo, but it does not get to warm. I mostly type via vim.13:20
ham5urgSometimes a stupid site pushes the browser to warm up the CPU.13:20
ham5urgDo I set the governor permanently via GRUB at boot-time?13:22
gnarfacei think so13:23
gnarfacei forget the flag13:23
gnarfacewhen i bother to change the boot default i usually do it in the kernel build13:23
ham5urgI did a modprobe cpufreq_ondemand but in # cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors there is still only powersave and performance13:25
gnarfacehmm, weird13:25
gnarfaceis it for the right kernel?13:25
gnarfacemake sure it loaded13:25
gnarfacei usually put mine in /etc/modules13:25
gnarfacebut i use the userspace one13:25
tr-amarsh04Is it worth trying a later kernel e.g. linux-image-5.14.0-1-amd64-unsigned           5.14.6-213:27
ham5urgthe modprobe command is successfull, lsmod shows cpufreq_ondemand13:27
gnarfaceit might be worth trying an even newer kernel yea but 5.10 should really be recent enough for this i'd think...13:28
gnarfaceat this point it's between either that or enabling kernel debugging13:28
gnarfacedo we know the whole thing even freezes?  could it still be accesible by ssh after the gui freezes?13:29
gnarfacethat would explain a lack of any traces of errors in the log too13:30
tr-amarsh04I like having 2 pc's for that reason, being able to connect via terminal session from one to the other13:30
gnarfaceyea me too but the irony is not lost on me13:31
tr-amarsh04being familiar with the magic sysreq key sequences also helps13:33
gnarfacei'm still not 100% sure it's not a graphical issue since i've seen that nvidia drivers have an extra last gasp before such lockups that doesn't make it to any logs, and doesn't even make it to the screen unless you enable it by module parameter13:35
gnarfacetrying to connect via ssh or serial terminal, or better yet even having the terminal open and tailing a log file already when it locks up might be informative13:36
tr-amarsh04all good suggestions13:36
gnarfacei don't know if the other video drivers have a similar option to this one nvidia thing13:36
gnarfaceno_console_suspend=113:37
gnarfacekernel parameter actually13:37
gnarfacenot module parameter13:37
gnarfacei thought it was nvidia specific but now i'm not sure13:37
gnarfacebut i know if you get an absolute blackscreen lockup that will at least let you see on screen the final words of the video card13:38
gnarfacein just this one corner case on some nvidia card i was having trouble with13:38
gnarface(the ones with extra ram are never as well supported as the base models)13:38
ham5urgThe next freeze I will ping this thing, the ping works if not freezed.13:41
gnarfaceyea but then if you're running ssh you can get back in and find out what's wrong maybe13:41
gnarfacerestart xorg and it magically comes back sometimes13:41
gnarfacesometimes you gotta reload video modules13:42
gnarfacesometimes it's still too far out to lunch and needs a cold boot but at least then you know what's wrong because the offending module actually does lock up when you try to modprobe -r it13:42
ham5urgOk, I will install ssh13:42
gnarfaceor sometimes dmesg has output that doesn't make it to the filesystem13:42
ham5urgok, works13:44
gnarfacea really useful one just in case you don't know is tail -f [log]13:45
gnarfaceit just follows the end of a log file on screen for you13:46
ham5urgAny modprobe cpufreq* does not change the list of available governors.13:46
gnarfacethat's weird, i'm forgetting something... if you put them in /etc/modules then reboot does it work?13:47
gnarfacedo you have acpi_cpufreq loaded?13:47
gnarfacei started loading that one too and i forget why13:47
gnarfacebut this might have been why13:48
ham5urgIt is due to intel_pstate13:51
gnarfaceoh my bad13:51
ham5urgThis is the new replacement for acpi-cpufreq13:51
gnarfaceyea intel also has one thing of its own that's supposed to be fancier13:51
gnarfacei forgot about that13:52
ham5urgIs there a better place to blacklist this module as in /etc/grub/default?13:53
tr-amarsh04I put custom blacklists in /etc/modprobe.d/local.conf13:54
gnarface /etc/modprobe.d/*.conf13:55
gnarfaceyea13:55
ham5urgI've put an "install intel_pstate /bin/false" there13:58
tr-amarsh04by putting "blacklist name-of-module" in /etc/modprobe.d/local.conf the module doesn't get auto-loaded13:59
ham5urgMaybe some other module invokes a load and circumvent the blacklisting.14:00
gnarfacegotta reboot for that change too14:00
gnarfaceand it has to be "blacklist [module name]"14:00
ham5urgMaybe a non-loading of intel_pstate will help. Gonna reboot14:01
ham5urgintel_pstate is hardcoded14:10
ham5urg# cat /boot/config-5.10.0-0.bpo.8-amd64 | grep PSTATE14:10
ham5urgCONFIG_X86_INTEL_PSTATE=y14:10
gnarfacewell damn does it have any controls?14:11
ham5urg# find /lib/modules/ -iname '*pstate*'14:12
ham5urgshows nothing14:12
gnarfacehmm, looks like no14:13
gnarfacenot as a module option anyway14:13
tr-amarsh04kernel menuconfig shows X86_INTEL_PSTATE Depends on: CPU_FREQ [=y] && X86 [=y]14:14
gnarfacewait, so does ubuntu not use intel_pstate?14:15
gnarfacemaybe this is futile anyway14:16
ham5urgidk if ubuntu does not use intel_pstate14:16
ham5urgMust check14:16
ham5urgI will try to block pstate via grub and after reboot I check a live-ubuntu14:17
gnarfacesee if it exposes any controls in /sys/14:19
gnarfacei'm not sure if they're the same ones as the ones i'm used to using... they may be just in a different place14:20
ham5urgok14:22
ham5urgpstate can be disabled via GRUB14:26
ham5urgI installed acpi-support acpid acpi14:28
ham5urgI will try Ubuntu 21.04 now.14:30
ham5urggnarface, could not find a difference in intel_pstate in devuan <> ubuntu. It's unconfigured and uses defaults.15:33
gnarfaceprobably not the issue then15:34
gnarfacejust a wild goose chase, sorry15:34
gnarfaceworth trying to install lm-sensors15:35
gnarfacethen run sensors-detect as root15:35
gnarfacesay yes to all tests15:35
gnarfaceaccept default option to modify the /etc/modules file as long as it's not something already loaded15:35
gnarfacethen you should be able to just run "sensors" as any user15:36
gnarfaceand mind you the values will largely be inaccurate but they won't be wildly varying in their inaccuracy; you'll be able to tell temperature change relative to them... probably ignore the voltage stuff, it's not known for bioses to report this stuff super accurately15:37
gnarfaceand negative temperature values are probably just misparsed15:37
gnarfacepatches welcome15:37
gnarface(:-p)15:38
gnarfaceit's probably not overheating but as a sanity check we should make sure it isn't15:38
gnarfacei don't know where you'd get all this information from ubuntu exactly but if you could get it and check for any large discrepancies in fan speeds or temperatures that would probably be useful just as a sanity check15:41
ham5urgyou mean a little script which every 1/n of a second writes data to a file. E.g. last 5 min in 1/100s resolution?15:41
gnarfaceyou would really only have to run it by hand a few times at idle and a few times at load to have a good idea if something was wrong compared to the ubuntu measurements but by all means don't let me stop you from overcomplicating it15:42
ham5urgHehe15:43
ham5urgLet me see today and maybe the weekend if the intel_pstate disableing did something good.15:43
gnarfaceyou could run munin and get graphs15:43
ham5urgOk15:43
gnarfacedisabling intel_pstate might have actually been a bad idea, i recommend you get a second opinion on that15:43
ham5urgI'm using cpufreq_ondemand for now15:44
gnarfaceyou have acpid running too?15:44
ham5urgYes15:44
gnarfaceok good15:45
gnarfacefortunately it does have controls15:45
gnarfacelook in: /sys/devices/system/cpu/cpufreq/ondemand/15:45
tr-amarsh04silly question, is cpufreqd installed?15:45
gnarfaceondemand wouldn't need it, it's controlled by the kernel15:45
tr-amarsh04ah, ok15:46
gnarfacethe userspace governor is the one that uses a daemon15:46
gnarfaceany daemon, really... it's controlled through /sys too15:46
gnarfacei actually prefer that one but i use amd hardware so ymmv15:47
hagbardI get this one when I try to update kexec-tools to the version in daedalus: invoke-rc.d: initscript kexec, action "restart" failed18:23
EmanuelLoos[m]Hi, I have a few questions about Debian packaging of coding languages:18:49
EmanuelLoos[m]First, I don't completely understand what the exact requirements are.18:49
EmanuelLoos[m]I read about Kotlin and I don't think I understood this. If Kotlin is written in Kotlin how was the first Kotlin binary compiled?18:49
EmanuelLoos[m]What are the pitfalls with coding languages regarding freedom?18:49
Hydragyrum"if kotlin is written in kotlin..." the original compiler/etc. for kotlin was written in another language, and then used to compile the self-hosted kotlin compiler that's used there19:15
Hydragyrumgcc is also usually compiled with gcc for instance19:16
EmanuelLoos[m]<Hydragyrum> "gcc is also usually compiled..." <- So, what is the issue exactly? Is the Kotlin compiler non-free? The Debian page about Kotlin metions putting the original compiler in the non-free repos.19:29
HydragyrumI'm not personally that familiar with kotlin -- but if it's in the non-free repos, it sounds like it's either closed-source or under a restrictive license -- but if the self-hosted compiler is free, that's not an issue, as being compiled with a given compiler does not count as derivation or the like19:32
EmanuelLoos[m]What are the exact requirements that have to be met for a coding language to get into Debian?19:33
Hydragyrum"coding language" -- it19:34
Hydragyrum*it isn't the language -- it's the compiler or interpreter etc.19:34
onefangThis is all #devuan-offtopic, since this is the Devuan support channel.19:34
VallHello everyone20:42
Vallham5urg: IME memtest isn't worth crap. It misses all but the most egregious failures20:44
VallIf you want to really test a machine, you have to get it through a burnin for at least 72h.20:44
VallBurnin means a parallel run of at least mprime in torture mode, plus dledford's memtest (google for it) to *really* stress the RAM.20:48
nemoVall: I sent memtest dude $10 once 'cause he saved me a lot of pain on some bad ram. was then able to salvage it w/ badram mask21:11
nemothat test sounds impressive but probably more than I wanna bother doing21:13
nemoI think catching egregious is good enough for me ☺21:13
nemobut I guess if he's desperate to catch some ephemeral subtle error21:13
nemognarface: the idea of powersave being more stable sounds reasonable to me though if his CPU is marginal.  like... isn't that what CPU companies do? sell the marginal CPUs at a lower clockrate?21:15
nemothen overclockers spend a bunch of money on crazy cooling setups to try to coax a higher clockrate out of it without the game crashing?21:16
demo_Hello,I am new here!.22:21

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