<remiliascarlet>
SiFuh: That headline photo alone is hillarious!
<remiliascarlet>
Woman holds man who's yelling at everything on the right (literally), then an obese woman is kicking a man in the stomach, who seems to fall asleep over it, and a 3rd man just spectates with a smile on his face, and then a man and a woman are staring from above going like "I have no idea what the hell just happened".
<remiliascarlet>
s/everything/everyone
<remiliascarlet>
But yeah, soy-based generation at full display!
<remiliascarlet>
Not only in the sense of how they fight, but also in the sense of the need to "brawl" because they couldn't agree on stuff.
<SiFuh>
remiliascarlet: I went camping over the weekend at a recreational campsite. I slept outside on a Tarpaulin near the fire using my duffle bag as a pillow and a towel to stay dry from the odd bit of rain. It was pretty cool. Lots of moths though. Worse than mosquitoes. They kept landing all over the place. On me and my stuff.
<SiFuh>
Anyway when I was sleeping, I kept a machete beside me. There were 3 dogs sniffing around stealing food and the Muslims kept chasing them away. When the male dog came to my camp site it started to bark at me and then as I looked to see what was barking it started to growl and then came charging towards me. Should have seen the look on the dogs face when I began to growl back and then started barrelling
<SiFuh>
towards the dog at top speed. It hit the skids, flung it's body around and took off as fast as it could go. Didn't even look back or stop. Just over the hill, across the creek and up the other side of the bank and over the top towards the main road where you can't see it anymore. It never came back after that.
lavaball has joined #crux-social
emmett1 has joined #crux-social
lavaball has quit [Remote host closed the connection]
<farkuhar>
SiFuh: according to CRUX-MUSL.txt you modified four core ports to remove linux-PAM. Of the four ports listed, only libcap appears in your Documentation/musl/ repo. Did you intend to upload the PAM-free versions of openssh, shadow, and util-linux too?
<SiFuh>
farkuhar: No, because it is super simple
<farkuhar>
I'm specifically interested in util-linux, because a naive attempt to build it with different configure options resulted in a footprint mismatch. Not only did the footprint fail to match the package on your ISO, it also was missing three binaries that appear in the official core port: /usr/bin/{col,exch,waitpid}
<SiFuh>
Okay okay okay
<farkuhar>
The only flag I deleted from the meson setup command was "-D pamlibdir=/lib/security" but other than that, I kept the same build() as the official core port. Yet on the CRUX-MUSL system, the resulting footprint differed from what had been installed by the ISO.
<SiFuh>
That would be dumb to enable pam in sshd_config and not compile in pam
<SiFuh>
Oh no, I missed that part on the second full build. Damn!
<SiFuh>
My first ISO build before the second ISO build I did remove it.
<farkuhar>
A quick comparison of the official core port and your no_pam version of util-linux ... official version disables cryptsetup, newgrp, vipw while your version doesn't. Your version disables nls, asciidoc, and python but the official version doesn't. I'm guessing one of those differences is the reason for my footprint mismatch.
<farkuhar>
Or it could be the difference between autotools and meson. Back when they tried to recover from the libuuid breakage, they switched to autotools temporarily, and now they're using meson again. Your version still relies on autotools.
<SiFuh>
farkuhar: Yeah well. You got a powerful machine to build an updated ISO?
<SiFuh>
I am a bit annoyed because when I went to KL I left my truck with the mechanic to do stuff. I gave them a strict set of rules on how to do it. What torque settings should be done and how to reset the suspension. They have done none of what I said on how to do it and my truck drives like shit because everything is out-of-wack. Every mechanic should know this yet they seem to live in la la la land where
<SiFuh>
a rattle gun is fine for everything. So now I have to loosen everything myself and adjust everything. Worse still everyone knows drive shaft in and and out must have same angle! I have two drive shafts with two different in and out angles.
<farkuhar>
4-core AMD Ryzen 3 3200G with 16GiB RAM and a 20GiB swap partition. It's at least three years old by now.
<SiFuh>
farkuhar: You could probably whip a new ISO faster than me then.
<farkuhar>
I suspected as much, when you cited 5 days as the time it takes to build a new kernel.
<SiFuh>
I was exaggerating about the kernel
<farkuhar>
Speaking of mechanics tasked with fixing heavy vehicles, did you hear about the Boeing mechanics who looked at the Iranian air fleet and refused to undertake the necessary repairs, for fear of being held liable when the machines would eventually break?
<SiFuh>
farkuhar: So anyway, you caused me to find the first error on my ISO. Openssh Pkgfile adds PAM into sshd_config
<SiFuh>
No, but that is funny shit because we all know Boeing is like the new 'made in China'
<farkuhar>
It reminds me of the scene from Back to the Future where the 1885 Doc Brown scoffs at an electronic component marked "made in Japan". Marty McFly has to correct his prejudices, informing him that "all the best stuff is made in Japan now"
<zorz>
Boieng is an insult for made in China :Pp
lavaball has joined #crux-social
<SiFuh>
farkuhar: I think about that one often actually
<SiFuh>
zorz: maybe
<SiFuh>
farkuhar: By the way, I have a problem with building my third ISO. To explain the process. I used CRUX-glibc ISO and in a chroot environment I installed all of the ports from emmett1. Then after I was able to get it functioning. I booted into emmett1's version. Then I built the first bootstrapped CRUX-musl ISO. Then I booted from that ISO and installed it onto the hard drive and booted into that and
<SiFuh>
began building the second ISO. Which is what is available now. I then booted into that ISO and installed CRUX-musl from it and after that for the third time I began to build the ISO again. However, it dies at gcc build and I haven't done much since because I was busy. My plan is to do it all over again but I am not sure when. Also, I will use updated ports.
<farkuhar>
Apropos of "made in Japan", I watched Highlander (the 1986 film) for the first time this past weekend. I didn't quite catch the reference to swordsmithing techniques too advanced for the accepted chronology (as if secretly brought to Japan centuries earlier by Sean Connery's character, the immortal called Ramirez).
<zorz>
guys just doanloaded cvs openbsd src tree, going to compile the kernel with OPTFLAGS=-O2 -march=znver3
<zorz>
MAKE_JOBS=16
<zorz>
farkuhar: Highlander was a great movie for 1986
<SiFuh>
farkuhar: In the Ancient Aliens series, they explain how the Aliens came down to Earth and taught swordsmithing to the two Japanese brothers
<SiFuh>
Two brothers that redesigned the sword thanks to a visit from a UFO. Was laughed at when they gave it to the soldiers. They said to try and count how many broke. When the battle was over they counted and very few swords had broken
<SiFuh>
Season 3 Episode 9 if I remember correctly
<SiFuh>
Yes I am correct. It is indeed Season 3 Episode 9. 'Aliens and Deadly Weapons'
serpente has joined #crux-social
<SiFuh>
Ooooo pretense is here.
<SiFuh>
I mean serpente. Must be dyslexic
<SiFuh>
farkuhar: What do you mean from 1986?
<zorz>
realease year
<SiFuh>
Most of the best stuff was done in the 80's and early 90's. Not to mention pre-80's stuff as well. But movies started going downhill in the mid 90's. Actually they focus more on special effects these days then actual real stories or new stuff. Just regurgitating old stuff and adding special effects.
<serpente>
hahahah
<SiFuh>
Wasn't the latest Top Gun movies considered in the top 10 because it was just like a return of the past and had no bullshit it in and true patriotism?
<serpente>
sucked
<SiFuh>
What's up preteens?
<SiFuh>
I mean serpente
<serpente>
im ok Fukiuh
<SiFuh>
Top Gun: Maverick was good
<SiFuh>
The _pretense_ of Serpente's _preteens_ smells of _terpenes_. :-)
<SiFuh>
Hahahaha
<zorz>
lets see how much time will take
<SiFuh>
zorz: For what dude?
<zorz>
SiFuh: it says -O2 -pipe but no where znver3
<zorz>
SiFuh: i am compiling the base system, cannot open now.
<SiFuh>
Why are you doing that?
<zorz>
learning... why not....
<SiFuh>
I did that with OpenBSD 3.X
<SiFuh>
Never done it again
<SiFuh>
Waste of time
<SiFuh>
Although it was a Toshiba Tecra 510CDT with 32MB of RAM
<zorz>
Toshiba!!!!
<farkuhar>
ukky: it appears that PATH is unset when /etc/rc.single runs, so every command is written with absolute paths. By copying verbatim your loop over `kmod static-nodes`, I basically created a fancy no-op. Interestingly, changing your loop to use absolute paths had the opposite of the intended effect: the soundcard went unrecognized, at least until I manually rmmod and modprobed the offending modules.
<zorz>
all laptops used to be Toshiba,,, and some compaq that time.
<SiFuh>
farkuhar: Everything should be scripted with absolute paths if it is to be run as a super user.
<SiFuh>
I even use and it is good practice, absolute paths in /etc/doas.conf
<farkuhar>
I think I've settled on the right combination of /etc/modprobe.d options (no mucking around with /etc/rc.single to pre-load any modules). The soundcard reliably appears in the `aplay -L` output now, either from a cold boot or a warm reboot.
<SiFuh>
rc.modules <---
<SiFuh>
Don't waste time with /etc/modprobe.d directory
<SiFuh>
Are you tweaking a module or loading?
<SiFuh>
Because I am talking about loading.
<farkuhar>
Just tweaking a module, to compensate for CoreBoot not giving the kernel enough information about which drivers need to be loaded.
<SiFuh>
So why mention rc.single?
<SiFuh>
rc.modules is run very early in the rc script in CRUX
<SiFuh>
Line 104.
<farkuhar>
SiFuh: I was following ukky's suggestion to insert a loop over the kmod static-nodes, in between mounting /sys and /proc, and starting the udev daemon.
<SiFuh>
To auto load certain modules?
<farkuhar>
ukky explained the reason for this idea a few days ago. Since the only place that satisfied the "in between" criterion was rc.single, that's where I inserted the for loop.
<SiFuh>
I am not scrolling back.
<SiFuh>
From personal experience. The kernel doesn't load anything you don't need or it can't see at that time.
<SiFuh>
So basically if you want 'loop.ko' you'd use rc.modules to load it or you do it manually. (Not in reference to your loop script)
<SiFuh>
If your sound card is detected and loaded then there is either a conflict which the kernel can't resolve, or it just can't be seen at that particular time.
<SiFuh>
Probably guessing that is why you are trying to run a loop script
serpente has left #crux-social [Killed buffer]
serpente has joined #crux-social
<SiFuh>
And I'd never touch rc.single with anything. Because rc.single is suppose to be single user mode regardless. Who wants audio in recovery ;-)
<farkuhar>
Remember how I asked whether emmett1's patches for kmod on CRUX-MUSL might have side-effects on the auto-probing of needed modules? It was a failure to automatically load iwlmvm, the module that accompanies iwlwifi, that prompted me to ask. Then I discovered another example: snd_hda_codec_hdmi was not being loaded before snd_hda_intel_icl, resulting in the failure to initialize the internal soundcard.
<SiFuh>
Can you unload snd_hda_intel_icl then load snd_hda_codec_hdmi and after snd_hda_intel_icl?
<SiFuh>
Do you even need snd_hda_intel_icl? And if you don't then blacklist it
<farkuhar>
But ukky then pointed out that CoreBoot might be giving the kernel an incomplete device tree, which could explain the difference in behaviour between this laptop and another device with the original (proprietary) firmware.
<SiFuh>
Also with iwlmvm. There are so many issues with loading, and conflicts amongst intel Wi-Fi that there is so much info on the internet now to help you figure out what to disable and enable in the kernel
<SiFuh>
farkuhar: Possible, but I doubt it. It seriously sounds like you have two devices that are conflicted with similar drivers/firmware
<SiFuh>
A cheap example: Kernel sees the beginning of something and loads the appropriate driver without and the device doesn't work because it didn't need that driver, it needed another similar driver.
<farkuhar>
So the solution I settled on ... populate /etc/modprobe.d with some hints about the "soft dependencies" among the various modules. For example, "softdep iwlwifi post: iwlmvm", or "softdep snd_hda_intel_icl pre: snd_hda_codec_hdmi". With those rules in place, all hardware is successfully initialized now.
<SiFuh>
I'd start by blacklisting shit
<farkuhar>
If I'm building a custom kernel anyway, why bother to blacklist anything? Just don't build the module in the first place!
<SiFuh>
You know like NVIDIA vs nouveau
<SiFuh>
The point of a fully modular kernel is to understand what your system wants, conflicts with, and what it doesn't need.
<SiFuh>
From that you can start making the foundations of the perfect kernel.
emmett1 has quit [Quit: ]
zorz has quit [Quit: leaving]
<ukky>
farkuhar: It was just for a quick test. Usually all commands in my runit startup script use absolute path, even 'cp' and 'echo'.
<SiFuh>
ALL super user commands should be run by direct path!
<SiFuh>
Even if it is a quick test :-/
<SiFuh>
It's bad practice otherwise
<SiFuh>
Unless it is a built in shell command then you have different rules to play with.
<lavaball>
you mentioned troonix, and you get "hidden".
<farkuhar>
To be fair, ukky's suggestion used absolute paths for every command except cut -d' ' -f3, on the receiving end of a pipeline. We don't set -o pipefail at the beginning of /etc/rc, so I'm not sure what the expected result of that command substitution would be. In any case, the modules are all loading correctly now, so I don't intend to investigate the suggested for-loop any further.