| used____ | Hi. Anyone running devuan <any> on low end Ryzen? A320 etc chipset? Is kvm working okay? | 15:13 |
|---|---|---|
| used____ | kvm in qemu and or vbox context | 15:13 |
| Jookia | vbox doesn't use kvm, only qemu does | 15:14 |
| used____ | Well it uses something else which works instead of kvm. | 15:14 |
| Jookia | it has its own virtualization kernel module. kvm isn't working for you? | 15:14 |
| used____ | I am looking at exchanging an unwanted present for a wanted one. Getting a low end A320 Ryzen box for headless ops which include android sim and qemu work. | 15:15 |
| Jookia | kvm is a kernel module so it doesn't matter what distro you're on provided it ships the module | 15:16 |
| Jookia | it's CPU and BIOS dependent | 15:16 |
| used____ | <vaguely related> What would be the differences between Intel ClearOS linux and Devuan, for AMD support optimization, and what devuan packages could be relevant to this? https://www.forbes.com/sites/jasonevangelho/2020/02/12/surprise-amd-recommends-intels-clear-linux-for-best-ryzen-threadripper-3990x-performance/ | 15:17 |
| Jookia | compiler options | 15:17 |
| used____ | Jookia: CPU BIOS and motherboard sometimes. I have crippled motherboards from the 2010s which have virt but it does not work due to bugs. | 15:17 |
| used____ | Jookia: that all? | 15:17 |
| Jookia | used____: yeah, that's part of the BIOS | 15:17 |
| Jookia | used____: maybe some patches to set compiler options | 15:18 |
| used____ | And would those be "portable" to a remastered devuan system? Would they make a big difference / be worth it? | 15:18 |
| Jookia | most distros now use the amd64 ABI which is getting dated, clearlinux doesn't | 15:18 |
| Jookia | they're portable but break compatibility with old hardware | 15:18 |
| used____ | Ah. Interesting. Do you have a link handy about this? I would like to read more. | 15:19 |
| used____ | But, for Devuan: is there something specific one can select in packages or options, to maximize performance and support for Ryzen ? | 15:19 |
| Jookia | probably not, probably not for most distros | 15:20 |
| used____ | Interesting pie chart in that link of mine. I assume they did not tweak the test set to look good themselves ;) | 15:21 |
| Jookia | there's current discussions on how to proceed to provide support for newer CPUs in linux | 15:22 |
| Jookia | ie instead of compiling for x86-64, compiling for x86-64-v2, x86-64-v3, x86-64-v4 | 15:22 |
| used____ | Ah, nice to know. That mostly affects the kernel and high performance programs I assume? So 99% of user land can stay "unoptimized" safely? | 15:24 |
| Jookia | no, all programs | 15:24 |
| used____ | Sure one could compile all with relevant optimizations, but would it make sense? | 15:24 |
| Jookia | yes | 15:24 |
| used____ | Okay. | 15:24 |
| Jookia | compilers are pretty smart | 15:25 |
| used____ | I mean, I am not sure I need busybox to be cutting edge speed... but the kernel and kvm and probably heavyer lifting servers yes. | 15:25 |
| Jookia | but right now amd64 ABI means no features part 2003 | 15:25 |
| Jookia | past* | 15:25 |
| used____ | I see. | 15:25 |
| Jookia | programs have to add manual support for them | 15:25 |
| used____ | Interesting AMD benefits from *Intel*'s OS optimizations. I though those two are at eternal odds with each other. | 15:26 |
| Jookia | well it's the same ISA | 15:26 |
| Jookia | if you've ever tried to run linux on a 32-bit it's the same situation | 15:26 |
| used____ | More or less. I guess silicon wars are behind us? | 15:26 |
| Jookia | there's linux i486, i586, i686, pentium4 ABIs | 15:26 |
| used____ | I run linux on 32 bit all the time and dropping support for i386/i486 meant I had to scramble a little. | 15:27 |
| Jookia | yes, so we'll see that situation with amd64 in the next few years | 15:27 |
| Jookia | clear linux just drops support for older CPUs | 15:28 |
| used____ | Interesting, they do not advertise that... | 15:29 |
| Jookia | it says you need a CPU with SSE 4.2 for example | 15:30 |
| used____ | Yes, but I do not immediately associate that with no support for older cores... | 15:30 |
| Jookia | well, unless your older cores magic up SSE 4.2 support it won't run | 15:30 |
| used____ | Obviously. | 15:32 |
| Afdal | hmm you know maybe my issue with this audio popping business | 18:44 |
| Afdal | and why I never experienced it on Ubuntu, | 18:44 |
| Afdal | is that my audio wasn't handled by the snd_hda_intel driver on Ubuntu? | 18:45 |
| Afdal | This certainly isn't an Intel chipset or Intel hardware that I'm using now | 18:45 |
| Afdal | Could Devuan be using the wrong driver for my audio by default? | 18:45 |
| Afdal | I'm not even using a sound card | 18:46 |
| fluffywolf | selection of such drivers is handled by the kernel, not by devuan, and it is __very__ unlikely it would be using the intel driver unless you have an intel card. it could be you're not using the intel stuff, or it might not even have plugs if it's built onto a motherboard but unused or something, but it's there. | 18:47 |
| Afdal | It's definitely using snd_hda_intel | 18:48 |
| Afdal | but when I edit options snd_hda_intel power_save=1 | 18:48 |
| Afdal | err, power_save=0 | 18:48 |
| Afdal | my audio popping goes away immediately | 18:48 |
| fluffywolf | what sound hardware are you using? | 18:48 |
| Afdal | *It's definitely using snd_hda_intel because when I edit /sys/module/snd_hda_intel/parameters/power_save to 0, the popping goes away immediately | 18:49 |
| Afdal | Uh, just an AMD motherboard with some integrated audio jacks | 18:50 |
| Afdal | lemme see if I can get more specific than that | 18:51 |
| fluffywolf | ok, so that motherboard is using an intel hda sound chip, or compatible... and it's buggy and pops when going out of power save. you probably have the correct fix for it. a wider-scope fix might be asking the kernel people to detect that buggy motherboard and disable power save by default on it. | 18:51 |
| Afdal | Ah it's a Realtek ALC887 audio chip | 18:51 |
| Afdal | Yeah what I don't understand is why this only became an issue when I switched to Devuan | 18:51 |
| Afdal | Never given me trouble before | 18:51 |
| fluffywolf | that's an intel hda compatible chip, which is why it's using the intel driver... | 18:52 |
| Afdal | And I've looked at the various alsa/pulseaudio fixes for it and Devuan ships with the same settings as Ubuntu used | 18:52 |
| Afdal | Which leads me to think Ubuntu used some sort of kernel-level configuration to avoid this | 18:52 |
| fluffywolf | it's probably a different kernel version. I don't think devuan does anything with configuring sound other than let the kernel do its magic. | 18:52 |
| fluffywolf | ubuntu could have something that disables power saving by default. I've never used it... | 18:53 |
| Afdal | Whatever Ubuntu must do, it isn't a pulseaudio or modprobe setting | 18:54 |
| Afdal | Or it may just be a major Linux kernel change that happened recently | 18:54 |
| fluffywolf | the root problem is buggy hardware. it shouldn't pop coming out of power save. heh, I have a behringer equalizer I won't use because it pops when you turn it on. crappy hardware design. | 18:54 |
| Afdal | as you said | 18:54 |
| Afdal | Meh, you say buggy, yet this has never been an issue in the past for me with this hardware... | 18:54 |
| fluffywolf | my sound system here in my living room has three rackmount amps, six speakers, and enough watts to dim the lights. a pop makes the walls rattle. everything is pop-free. :) | 18:56 |
| fluffywolf | eight speakers | 18:56 |
| Afdal | D:> | 18:57 |
| fluffywolf | people give me crap for my choice of audio hardware... an ancient soundblaster extigy. despite being consumer hardware, it actually has exceptionally good output. lol | 18:58 |
| Afdal | gah am I really such an incompetent dummy that I can't figure out how to create a file in a high-permissions directory :c | 19:56 |
| Afdal | sudo blahblah > myfile.conf says permission denied :c | 19:56 |
| fluffywolf | I have no idea how redirects work with sudo. su - and then try? | 19:57 |
| Afdal | pff, like I can remember my su password <.< | 19:57 |
| Afdal | How can I do this without changing directory ownership or permissions :/ | 19:59 |
| critr | become the user that owns them | 20:00 |
| fluffywolf | sudo touch myfile.conf first? | 20:00 |
| Afdal | the file doesn't exist | 20:01 |
| Afdal | I want to make it | 20:01 |
| Afdal | How do I simply make a file in a root-owned directory >:/ | 20:01 |
| fluffywolf | touch will make a file, then you can sudo chown it to yourself, then you can redirect to it. | 20:01 |
| fsmithred | sudo -i | 20:01 |
| fsmithred | touch /root/whatever | 20:01 |
| fluffywolf | or use sudo to run a shell. :P | 20:01 |
| fluffywolf | I don't use sudo, not familiar with its tricks. | 20:02 |
| Afdal | okay, touch worked | 20:03 |
| fluffywolf | I would assume the problem is the redirect is being done by your shell, not sudo, so you only have your permissions. so use sudo to run a shell and have it do the redirect, or create the file and chown it to yourself first so you can write to it. | 20:03 |
| Afdal | why is bash so fussy about > | 20:03 |
| rwp | The bash > redirect is happening as non-root because the bash user is you not root. | 20:04 |
| Afdal | :} | 20:04 |
| fluffywolf | isn't that what I just said? :) | 20:04 |
| rwp | And then the su or sudo command is running as root *after* the redirection and using the redirection set up by you the bash user. | 20:04 |
| rwp | fluffywolf, Yes. But alternative ways of saying things is often useful for understanding. | 20:05 |
| Afdal | hmmm | 20:05 |
| fluffywolf | I'm rather against sudo. I think it reduces security, not improves it, in the vast majority of setups. | 20:06 |
| rwp | I am also somewhat afraid to ask what file would need to be user edited in a root owned directory. (me makes fearful face) | 20:06 |
| Afdal | /etc/modprobe.d/audio_powersave.conf | 20:07 |
| rwp | I feel su & sudo are two side of the same box. Similar in generalities but different in specifics. | 20:07 |
| Afdal | fixing my audio popping issue once and for all | 20:07 |
| fluffywolf | basically, virtually every program can, if ran as root, be used to give you full root access, in some fashion or another. an admin thinks they can safely let users run a couple things as root, but they're actually giving users full control of the system without realizing it. while giving out the root password forces admins to think about it harder. | 20:08 |
| critr | i just delete sudo after install to simplify matters. | 20:08 |
| rwp | Ah... Very good. That seems less scary than my imagination thought it might be. :-) | 20:08 |
| fluffywolf | for example, any program capable of writing a file, if ran as root, can give someone full root access. | 20:08 |
| rwp | Normally one would leave that file owned by root though. But it seems less acceptable to continue as you are doing with it. :-) | 20:08 |
| rwp | Sorry. But it seems acceptable to continue as you are doing with it. :-) | 20:09 |
| rwp | I was typing in less for something, got distracted, returned, and realized I had left that word there mistakenly. Changing the meaning entirely! | 20:09 |
| Afdal | Uh, I made it with the same permissions and ownership as other files | 20:09 |
| Afdal | the problem was simply making it | 20:09 |
| Afdal | 644 with root owner | 20:10 |
| rwp | A typical vim user might have run "sudo vim /etc/modprobe.d/audio_powersave.conf" and then edited the file that way. | 20:10 |
| Afdal | that's how files in /etc/ should be... | 20:10 |
| fluffywolf | vim users are never typical. :P | 20:10 |
| Afdal | Yeah I could have just done it with sudo nano myfiletomake.conf as well | 20:11 |
| Afdal | but I wanted another way | 20:11 |
| rwp | fluffywolf I agree that both su and sudo might allow the user to escape to root. But... For most people they are the solo user on their own system. So it's okay. | 20:11 |
| fluffywolf | rwp: sudo's big advertising point is that you can selectively let users run certain things instead of giving them full root access. but it's extremely hard to do this safely. | 20:12 |
| rwp | The use of sudo gets conflated both with using it for "su -i" types of things (sudo -i) and also with people trying to configure sudo for *limited* access. | 20:12 |
| fluffywolf | holy fuck, pigs are flying, hell is ordering ice skates, etc. the JWST launched. | 20:12 |
| rwp | Most of us are saying use "sudo -i" for just getting a root shell and continuing on. Or using "sudo $EDITOR /etc/somefile" and that's in the full access from the start side of things. | 20:13 |
| rwp | I do use in my environment sudo set up with restrictions and limitations. But honestly I impose them 1) upon other people not me and 2) as "training wheels on a bicycle" to give them a safety net so they don't have quite as many accidents. | 20:14 |
| fluffywolf | as I said, if you can sudo _any program that can write a file_, you have full root... | 20:15 |
| rwp | Yep! But "sudo -i" already allows full root. So... Doesn't seem like a problem to me. :-) | 20:16 |
| fsmithred | Afdal, after 'sudo -i' you can set your root password to something you remember. | 20:34 |
| Afdal | lol root sure sounds useless... | 20:35 |
| rwp | root superuser is critically important! | 20:36 |
| Afdal | not when I have sudo :^) | 20:36 |
| fsmithred | bbl | 20:41 |
| rwp | I think that is like the old quote, "No man's life, liberty, or property are safe while the legislature is in session." | 20:51 |
| critr | without root account, sudo doesn't work. | 21:39 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!