💾 Archived View for gemini.ctrl-c.club › ~phoebos › logs › kisslinux-2022-06-03.txt captured on 2022-07-17 at 03:00:55.
⬅️ Previous capture (2022-06-04)
-=-=-=-=-=-=-
[2022-06-03T00:33:11Z] <dilyn> https://0x0.st/oB7u.png [2022-06-03T00:33:13Z] <dilyn> :trollface: [2022-06-03T00:36:01Z] <noocsharp> dilyn has completed his treachery [2022-06-03T00:39:25Z] <dilyn> it became too cumbersome moving files between my desktop, qemu, and my laptop [2022-06-03T00:39:48Z] <dilyn> but i'm managing my own kernel and there's no grub so I can dual boot easily :v soon, wayland-only ubuntu. once I have the time [2022-06-03T01:08:48Z] <noocsharp> how to do you keep stuff up to date on ubuntu, or do you not [2022-06-03T01:34:36Z] <dilyn> debian actually has a really large ecosystem for packaging and maintenance [2022-06-03T01:34:54Z] <dilyn> I mean, a ton of it is legacy cruft or things you wouldn't use to maintain your own system [2022-06-03T01:36:42Z] <dilyn> apt will fetch sources for packages or build dependencies for you, you can use dpkg-* scripts to automate building things (just the library, just the binary, etc). Modifying builds is just editing a makefile, you can directly push to a PPA pretty easily, you can have launchpad do the package build for you... [2022-06-03T01:37:18Z] <dilyn> granted it's a LOT to learn so the curve can be quite steep [2022-06-03T01:43:28Z] <noocsharp> makes sense [2022-06-03T01:52:50Z] <jslick> I have used junest + manual compile at work, but will probably switch to gentoo-prefix soon. keeps the stuff that I care about up-to-date (such as tmux, neovim, ripgrep) [2022-06-03T04:42:48Z] <testuser[m]> Hi [2022-06-03T12:13:54Z] <phoebos> > 1040 packages [2022-06-03T12:13:57Z] <phoebos> good going [2022-06-03T12:14:37Z] <phoebos> is that because everything is split into a -lib, -bin, -dev etc [2022-06-03T12:47:02Z] <dilyn> correct [2022-06-03T12:47:45Z] <phoebos> makes sense [2022-06-03T12:51:16Z] <dilyn> seems pretty heavy tbh tho: /dev/nvme0n1p3 98G 53G 40G 58% / [2022-06-03T12:51:20Z] <dilyn> df -h output [2022-06-03T12:52:44Z] <dilyn> that's a joke tho. It's because I've got about five full kernel trees in ~ :v [2022-06-03T13:19:38Z] <soliwilos> An alpine system I have for kodi entertainment uses a total of 5.5GB.. It's not setup for full desktop usage, but still, that would not add much I'd guess. [2022-06-03T13:57:53Z] <phoebos> hehe /dev/nvme0n1p5 38.3G 29.2G 7.2G 80% / [2022-06-03T13:58:08Z] <phoebos> 10 gigs of ccache, kernel sources, mail, chroots [2022-06-03T13:58:44Z] <phoebos> are they 5 different kernel trees you've got? [2022-06-03T13:58:50Z] <phoebos> if not, you're gonna love git branches [2022-06-03T14:43:46Z] <omanom> :/ ipv6 issues, what year is it?! [2022-06-03T14:59:10Z] <dilyn> phoebos: after a small amount of cleaning and moving... /dev/nvme0n1p3 98G 13G 81G 13% / [2022-06-03T14:59:15Z] <dilyn> not terrible [2022-06-03T14:59:24Z] <dilyn> and yeah they're full-blown git trees, each with at least three branches [2022-06-03T14:59:29Z] <dilyn> I've been doing a lot of kernel work lately:) [2022-06-03T15:01:09Z] <phoebos> cool [2022-06-03T15:03:36Z] <dilyn> you say cool but I say "I am not a kernel developer" [2022-06-03T15:06:22Z] <phoebos> have you been doing riscv stuff [2022-06-03T15:06:42Z] <dilyn> I have! [2022-06-03T15:06:51Z] <dilyn> I got Ubuntu Core to run on a $25 board [2022-06-03T15:07:01Z] <dilyn> that work is getting presented at embedded world [2022-06-03T15:08:02Z] <phoebos> that _is_ cool [2022-06-03T15:08:54Z] <phoebos> what kind of things did you have to do [2022-06-03T15:10:20Z] <dilyn> well I had to learn all about the boot process for riscv, which is actually quite sane when compared to arm [2022-06-03T15:11:14Z] <dilyn> had to build u-boot and opensbi, for my board specifically I had to take in some patches from someone elses work, and the kernel I used was a very barebones kernel with ubuntu's patches to get things like e.g. apparmor to work [2022-06-03T15:11:30Z] <dilyn> it was surprisingly simple, once you learned a bunch of prior art stuff [2022-06-03T15:28:49Z] <omanom> >actually quite sane when compared to arm -- good to hear! [2022-06-03T15:29:28Z] <omanom> part of why i stopped trying to get kiss on my rpi is because i didn't feel like dealing with that [2022-06-03T15:33:18Z] <dilyn> The pi is maybe one of the most opaque bits of hardware... [2022-06-03T15:33:44Z] <dilyn> Who'd have thought that Western Digital of all companies would drive standardizing the early boot stages of an entire hardware platform! thank you WD for making riscv nice [2022-06-03T15:34:34Z] <testuser[m]> <dilyn> "it became too cumbersome..." <- What about having uwuntu on external ssd or something and sharing /home on that [2022-06-03T15:34:37Z] <noocsharp> wtf is happening, a few days ago i talked to a guy in real life who worked on risc-v at wd [2022-06-03T15:35:11Z] <dilyn> they're pushing the most common implementation of SBI! it's great [2022-06-03T15:35:14Z] <dilyn> testuser[m]: ew [2022-06-03T15:35:19Z] <dilyn> I mean I could, but [2022-06-03T15:35:40Z] <dilyn> i've got oodles of space on this nVME anyways and I needed to do a full disk wipe already [2022-06-03T15:36:19Z] <noocsharp> yesterday i started trying to get a little bare metal program running on opensbi [2022-06-03T15:36:22Z] <noocsharp> no success yet [2022-06-03T15:36:40Z] <dilyn> is it a problem with loading it or the actual execution? [2022-06-03T15:36:59Z] <dilyn> you should just be able to build opensbi with a payload and Just Twerk [2022-06-03T15:37:16Z] <noocsharp> i'm pretty sure loading, might be loading things at the wrong address [2022-06-03T15:37:28Z] <dilyn> probably [2022-06-03T15:37:46Z] <dilyn> there are three different ways to get to execution and you use them all for different reasons :v [2022-06-03T15:37:51Z] <noocsharp> since i got it working in qemu virt without opensbi [2022-06-03T15:38:03Z] <dilyn> are you sure? [2022-06-03T15:38:07Z] <dilyn> on KISS? [2022-06-03T15:38:22Z] <dilyn> qemu carries around its own OpenSBI afaik so if you don't specify a BIOS it'll just load its own [2022-06-03T15:38:32Z] <noocsharp> no, i loaded my bare metal program as the bios [2022-06-03T15:38:37Z] <dilyn> ah i see i see [2022-06-03T15:40:13Z] <noocsharp> weird coincidence that you bring this up since i was literally reading about all this stuff yesterday [2022-06-03T15:41:27Z] <noocsharp> i sent a pr to opensbi since they were using find `-iname` when `-name` would work just fine in their Makefile [2022-06-03T15:43:22Z] <dilyn> nice! [2022-06-03T15:43:53Z] <noocsharp> we'll see if it ever actually gets merged [2022-06-03T15:48:24Z] <noocsharp> what board did you get ubuncu core running on? [2022-06-03T15:48:44Z] <dilyn> the sipeed lichee rv with the dock [2022-06-03T15:49:14Z] <dilyn> just a cheap $25 dev board haha [2022-06-03T15:49:29Z] <dilyn> there's a bigger and better version (the nezha) and my coworker is going to take my work and try to get Core on there [2022-06-03T15:50:21Z] <noocsharp> did you have to like port any programs, or was it relatively straightforward? [2022-06-03T15:51:07Z] <dilyn> luckily all of the main components (snapd, core20) are already built for riscv64, so I didn't have to build those myself (though it shouldn't have been hard...) [2022-06-03T15:51:21Z] <dilyn> building the kernel was tricky just because I wanted to try and get an ubuntu kernel [2022-06-03T15:51:36Z] <dilyn> there aren't many snaps available as of now tho so anything else I'd want to run I'd have to build myself [2022-06-03T16:06:55Z] <dilyn> sorry for the noise trying to get my mic and speakers working in Ubuntu... [2022-06-03T16:07:12Z] <dilyn> Everything is so much harder on distros that aren't KISS. God I feel like a boomer [2022-06-03T16:08:14Z] <noocsharp> noise? [2022-06-03T16:10:29Z] <dilyn> logging in/out [2022-06-03T17:42:09Z] <rohan> man [2022-06-03T17:42:15Z] <rohan> just a question [2022-06-03T17:43:15Z] <rohan> if i use the gentoo kernel with same configs and the linux default kernel in kiss [2022-06-03T17:43:22Z] <rohan> will be the same performance? [2022-06-03T17:43:43Z] <dilyn> does the gentoo kernel apply patches to the kernel? [2022-06-03T17:45:28Z] <rohan> yeah [2022-06-03T17:46:12Z] <illiliti> those are minor afaik [2022-06-03T17:46:49Z] <rohan> is because i dont know why John the ripper have less performance with kiss compared to other distros [2022-06-03T17:50:13Z] <dilyn> what do you mean it has less performance [2022-06-03T17:51:11Z] <rohan> more time to crack the same filr [2022-06-03T17:51:28Z] <dilyn> like, 50% longer? or 2% [2022-06-03T17:51:43Z] <rohan> like 30% i dont remember well [2022-06-03T17:51:56Z] <rohan> but is a good difference [2022-06-03T17:52:09Z] <rohan> not like 10 seconds and 11 seconds [2022-06-03T17:52:20Z] <dilyn> well [2022-06-03T17:52:40Z] <dilyn> is your CPU running at the same speed in both cases? [2022-06-03T17:52:58Z] <rohan> i think so [2022-06-03T17:53:07Z] <rohan> i set the same parameters in gentoo [2022-06-03T17:53:13Z] <dilyn> sure [2022-06-03T17:53:40Z] <dilyn> I'm just saying it's possible that your CPU is getting throttled at some point during this process, either from the governor or a pstate or thermal throttler... [2022-06-03T17:53:52Z] <dilyn> it's hard to know without... watching... [2022-06-03T17:54:14Z] <rohan> yeah maybe thats it [2022-06-03T17:54:53Z] <dilyn> you can certainly test if it's *just* how the kernel is configured/patched by doing what you suggested and using the gentoo kernel, but that isn't sufficient because it's possible some other part of gentoo is helping things (for instance, udev might load a certain module which you aren't doing on KISS) [2022-06-03T17:56:02Z] <rohan> udev maybe a point [2022-06-03T17:56:14Z] <rohan> i will do some more testing [2022-06-03T18:03:01Z] <illiliti> no, "udev" should not be a problem if you compiled kernel with builtin modules, which is recommended way for kiss [2022-06-03T18:05:26Z] <dilyn> I'm saying they didn't do that illiliti:P [2022-06-03T18:05:55Z] <dilyn> or they possibly enabled two governors that conflict with each other [2022-06-03T18:06:06Z] <dilyn> could be many things [2022-06-03T18:10:12Z] <illiliti> preemption, scheduler, "niceness" level [2022-06-03T18:10:24Z] <illiliti> what else? [2022-06-03T18:14:03Z] <illiliti> musl's allocator could be a problem too btw [2022-06-03T18:14:52Z] <dilyn> indeed [2022-06-03T18:15:21Z] <dilyn> might as well be comparing apples and elephants [2022-06-03T18:15:24Z] <rohan> why tão governors will be a problem? [2022-06-03T18:15:52Z] <rohan> all the time my kernel have more than one governor enabled [2022-06-03T19:09:49Z] <rohan> is bad for my processor using ondemand cpufreq policy governor and performance governor? [2022-06-03T19:18:03Z] <dilyn> those are the two I enable [2022-06-03T19:18:30Z] <dilyn> so if it is... we're both screwed:P [2022-06-03T19:31:47Z] <dilyn> anyone else fighting https://bugzilla.redhat.com/show_bug.cgi?id=1891286 ? [2022-06-03T19:31:48Z] <dilyn> I wreckon it's something wrong with my kernel (preempt/watchdog?) and/or device manager (smdev, shoddily built) [2022-06-03T19:41:01Z] <testuser[m]> dilyn: it always showed that for me [2022-06-03T19:41:09Z] <testuser[m]> Doesn't impact anything tho [2022-06-03T19:41:25Z] <testuser[m]> seem to* [2022-06-03T19:41:43Z] <dilyn> ah [2022-06-03T19:41:53Z] <dilyn> wayfire crashes on me and it's the only thing I notice being wrong [2022-06-03T19:42:16Z] <dilyn> perhaps disconnected then. Interesting... [2022-06-03T20:48:22Z] <noocsharp> i got that with my pinephone touchscreen sometimes when i was adding touch support to swc [2022-06-03T20:49:12Z] <noocsharp> think it means event processing in userspace is taking too long