There it is

scoobybejesus_tl: thaks for tip

thanks

jails are so cool

cgit.freebsd.org/doc/commit/?id=319…2a75b6c8c37f3ebe4bf28b60e3703d95bab High-flying FreeBSD release.

my bhyve vm is getting shutdown (when i go to sleep) on 15-RELEASE

not sure its related to 15-REL, but only now this is happening

hernan604: That sounds like a HV setting on what to do on sleep. save state or shutdown?

skered: hmm could be maybe ? it also coincides with my recent switch to kde-plasma after 15-REL install... and i dont know if kde-plasma sleeps or does some weird thing...

i only used i3/dwm for decades

Oh, we got the announcement a day early. Unless they meant December 2nd UTC.

mason: I think everything is UTC.

Makes sense.

That's how we do it in amateur radio. So all the logs match up.

Thanks for another release

so, 3 or 4 days ago, i upgraded from 15-RC2 to 15-RELEASE ... how an i know i did get the final 15-REALEASe ?

hernan604: You should be able to use freebsd-update for it.

RC2 should be known.

There are checksums in the announcement. use sha256sum and or sha512sum to check the image you downloaded matches.

Oh, never mind. I misread.

hernan604: "freebsd-update -r 15.0-RELEASE upgrade" should do it, IIRC

hernan604: Even better: freebsd.org/releases/15.0R/installation/#upgrade-binary

mason: hey man how you doing?

The 15.0-Release release notes are live.

scottpedia: Doing well! Got to go snag dinner, and we're going to take in a podcast on a system we recently moved to FreeBSD. We're using FreeBSD for more workstations here, which is nice. That it's suitable for that nowadays is great.

right good for you man mason

The jail(8) command now supports the zfs.dataset parameter to attach a list of ZFS datasets to a jail. e0dfe185cbca wow awesome

It sure is!

i was using nullfs

I still use nullfs, but there are times when that would come in handy, for sure.

i kept trying to force it to use dataset. before i rtfm

Is there a way with pf to basically say "listen on port 443. if it is SSH traffic, redirect to localhost:22. If it is HTTPS traffic, check the SNI and redirect to the appropriate jail/ipaddress". I want to replace what SSLH does because it seems to be crashing lately, leaving me remotely locked out from the home system, and I can't seem to find out why SSLH has been crashing.

mns: is there any reason as to why you can't use separate ports?

DarkUranium: port 22 would be blocked from work, so I use 443 to connect to home

not just at work, but when I'm travelling, many places block of most ports and just keep 80 and 443 open

I don't think pf is gonna be able to distinguish between protocols like that, you would want to write a service tha does the switch for you

like SSLH...

mns: closest thing might be an authenticated https proxy, then configure your ssh client to use it

LxGHTNxNG: sslh or haproxy would be the other choice.

zi: might look into that as well

Nice artice on Phronix about Sylve. :)

I have not messed with that one yet.

vm-bhyve works fine for my simple setup. Huge user base for point and click though. I love how slim Sylve is while providing a GUI.

rtj: no you dont

cpet: hi

hi

hey all, I'm just introducing a young programmer to the joys of IRC. anyone around to say hello?

anybody running a mastodon server on freebsd? is that even possible?

Macer: you should check out bsd.cafe

and bsd.network

oh. yeah i know they have their own fediverse servers.. but that doesn't necessarily mean they're running them on fbsd. i just figured there would be a bunch of awkward requirements that you may only find in linux

that seems a bit dated

TIL of Sylve, I'll have to try it.

I assumed that bsd.cafe is on freebsd, and bsd.network is on openbsd. And illumos.cafe is smartos in a freebsd bhyve vm

josephholsten: not sure

maybe on freebsd.. in bhyve... running linux :)

i wonder what version is in the ports tree now

mastodon-4.3.6_1

so i guess they must be... albeit a slightly older version

mastodon lifespans for minor versions last forever though. i am pretty sure they still do updates for 4.2 which is fairly old nowadays

Countdown to the end of life for Mastodon 4.1.x - this version will no longer receive updates (including for security issues) after April 8, 2025.

i guess 4.1 was obsoleted in february

is using freebsd-update to go from 13.5-RELEASE-p7 to 15.0-RELEASE expected to work? The relnotes say "Systems running 14.3-RELEASE can…", does that suggest other versions cannot?

congrats on 15R!

Macer: I run my Mastodon on FreeBSD.

Although from source, not from the port.

˜/39

entrop: my gut is telling me.. that would be a big leap and would recommend not doing.. especially if it does not say it explictly

Hm. If I have a webcam and webcamd is detecting it, but it's not in /dev/sndstat, is there a proper way to deal with that?

Weird, another FreeBSD box, and it shows up. Going to compare everything I can.

entrop: Upgrade to 14, then to 15.

Figured it out. It was the cargo cult zapping me. devmatch_enable="NO"

With devmatch enabled, the camera is able to bind to both ugen *and* uaudio, whereas with devmatch disabled it only bound to ugen.

entrop: you have to go from 13.x to 14.x to 15.0

ok I file 4 zpools on this machine. I boot the 15.0 installer and use manual partitioning where I get a shell and run zpool import and none of my zpools are listed

odd

why not just update it using freebsd-update, or src

cpet, I don't want to update, I want to create a whole new fs and install on that. I want a totally clean install with no artifacts, and I want to be able to look at the old root filesystem from within the new one (I also do one FS for / /usr /var, all my data is on a separate fs)

the installer will create a new pool for you

I don't want a new pool, I want a new fs on an existing pool

why

15R is feeling smooth here. upgraded and reinstalled to a bunch of machines no problems

kerneldove: had some sisues with firmwares but that was resolved by removing all gpu-* and having fwget install it

did an upgrade or fresh install?

src install

ah

i build world in like 25 mins so

i rather build than use freebsd-update

I don't want a new pool I have enough pools and I have space on an existing pool and I don't want a new install to have left over files from a previous install that weren't touched by this install, my root fs is throwaway, all my data is elsewhere

Well then. 15.0 has some bugs. I'm taking notes and will start opening them tomorrow.

you may need to install updated boot code or upgrade the pool before you can use it on 15.0

figured it out, when importing, the / partition on the zpool was mounting over the installer root

look at you doing some googling

nope, it just dawned on me

yeap google can do that as well

Pirates can work around that.... ARrrrrrrr! (-R)

i like these videos that make building a computer like rocket science when I buitl my first computer at around 12 years old back when you actually had to know a little about the computer and imply adding in ram pretty much failed each time

thing that gets mne every time are the power, and IO connectors always backwards ;/

so, direct 13.5 -> 15 with freebsd-update worked fine

living on the edge good for you man

In general probably the most important thing would be to update the bootcode so that it will boot from an upgraded storage array especially if it is zfs.

rwp: i gave up after he kept on saying i want i want and i want heh

when doing an install what are the "optional" packages?

none really kde was rtemoved

you can install vim

that is if you want '

I'm not talking about the packages you can install after installation but when you install systems components like base/debug/devel/kernel-dbg/lib32/optional/src/test

all of them are clear except optional

heh doesnt say much

You can download the other optional components from download.freebsd.org/releases/amd64/14.3-RELEASE and then unpack them.

on a 15.0 system ?

Do I have to do all of the work? Sigh. download.freebsd.org/releases/amd64/15.0-RELEASE

pkgbase optional packages differ than the ones in that ftp dir

<grammar> s/differ than/differ from/ </grammar>

why do we need 15 FreeBSDs? wouldn't 2 or 3 be enough? i don't understand why the project is endorsing such conspicuous consumption

^ Good alliteration

Does FreeBSD support UAS/UASP for external USB hard drives?

ant-x: no

ant-x: there were some concerns around patents that discouraged implementation of UASP, meanwhile, no one implemented UASP

Not even Linux?

?

i mean no one implemented it on FreeBSD

we can't take code from Linux, the license is wrong

OK.

Any other BSD?

no idea

OK. In my case, a UASP HDD works with FreeBSD, but quote slowly, and takes about 30 seconds to mount.

if it works with FreeBSD, it's using umass(4), and taking 30 seconds to mount is a bug in something not related to UASP

you should post to a mailing list about this, probably fs@ or stable@

A fallback protocol.

yes, most UASP devices can fall back to USB Mass Storage

The same UASP HDD is taking 30 seconds to mount on my Windows XP box.

but that doesn't mean they take 30 seconds to mount, that's clearly wrong

OK, will report a bug.

unless it waits 30 seconds before offering USB mass storage, which... i suppose could be a thing...

Several times I have observed it lose data: shall I umount it before ejecting, unlike usual USB sticks I use with the Automount udev daemon: <github.com/vermaden/automount> ?

ivy, re: wait before defaulting to USB mass storage: seems likely.

i have never heard of a USB storage device which does that but vendors do do weird things so i wouldn't be surprised

Hardly the same bug in FreeBSD and Windows XP.

(on different hardware)

it seems when adding: echo 'FreeBSD: { url: "pkg+https://pkg.FreeBSD.org/${ABI}/latest" }' > /usr/local/etc/pkg/repos/FreeBSD.conf on aarch64 freebsd-update -f fails

on 15-RELEASE

CrypticSquared: can you be more specific? how does it fail? freebsd-update doesn't usually care about your pkg configuration

pkg: Repository FreeBSD has a wrong packagesite, need to re-create database

that's normal, just run pkg update

that's what i get when i run pkg update

it prints that ever time you run it? please paste the full output somewhere

in a sec. just running it again

incidentally, you should not have a repo called 'FreeBSD' in 15.0

the repos are called 'FreeBSD-base', 'FreeBSD-ports' and 'FreeBSD-kmods'; see /etc/pkg/FreeBSD.conf

i'm not sure if that's the cause of your issue, though

sorry, FreeBSD-ports-kmods, not FreeBSD-kmods

Hrm. I got disconnected from my colo server. No biggie. SSH back in, but... screen -r doesn't find any screen sessions.

Even though I can see it in ps

Alver: did you upgrade from screen 4.x to 5.x? this is normal behaviour (according to upstream), you should have killed your screen 4 session first

CrypticSquared: right, the problem is you're trying to set the URL for 'FreeBSD' but there is no such repository in 15.0

ivy: ah - that might be it, yes. Let me check

ivy i'm following the directions in the handbook for adding latest repo

CrypticSquared: then the handbook is outdated

CrypticSquared: look at /etc/pkg/FreeBSD.conf, find the repo you're trying to modify, and use the actual name

ivy ok

Right, indeed. Nice. Okay, nothing to see here, carry on. :°)

CrypticSquared: if you're trying to switch from quaterly to latest, then you probably want 'FreeBSD-ports' as the repo name

(still no tor .21 available - that is not good)

Alver: bit shitty behaviour though, especially since they don't warn you beforehand

ivy: somewhat. But eh, it is what it is. I'm already happy it's not voodoo happening.

I do wonder why there is no updated package for tor yet - it hit ports a week ago, I believe

it usually takes 2-3 weeks for the builder to rebuild ports

i think the amd64 builder did get upgraded recently, at least for 15.0

but still, a week is not very long

Ah, heh. Okay

I was imagining that once a package hits ports, it would take a day or two for builders to churn the pkg out

heh, if only

serious freebsd users either 1) build their own ports (using poudriere) or 2) do not use ports

this is an ongoing issue that we currently have no good solution for

Not enough builder capacity? Or is it more complicated than that

I always end up needing something that wasn't enabled :-(

Alver: it's somewhat more complicated in that we can only scale the builders so far before you literally can't put any more CPUs into the computer

Alver: the real solution will be to make builds distributed, but that requires someone to actually have the time to do that

I was just going to say. :°)

If it's single machine then yes, you hit a wall somewhat quickly

i personally think we should prioritise builds for the 5% of ports people actually use and let the 95% of other ports lag behind a bit

but again, that requires someone to actually implement that

like, it should not take 2 weeks to publish a fixed security update for Apache, or whatever

True.

Or yes, at least make a distinction between "urgent for CVEs" and "version bump without further ado"

yeah, but it's more complicated than that because the build tool can only take a full ports tree and build it

I figured. Easy it is not

Could always ask around if anyone has a spare SGI UV to donate. :°)

Is anyone working on FreeBSD support for the Framework Desktop?

I found this: github.com/FrameworkComputer/freebsd-on-framework/tree/main

OK yes so looks like various pieces missing, some of which are being actively worked on.

Are there any major changes in 15.0 that require an upgrade in the boot code/ EFI partition?

s2r: ZFS was upgraded, if you upgrade your root pool, you should also upgrade the EFI loader

and it is not a guarantee it will boot, even after bootloader update

4.19T resilvered, 77.02% done, 23:46:25 to go

well... at least i finally made it to the same day lol

expand: expansion of raidz2-0 in progress since Thu Nov 27 22:35:46 2025

12.2T / 69.2T copied at 34.2M/s, 17.63% done, paused for resilver or clear

i still have that lingering in the background too

Alver: I was gonna say I could try to donate a machine and then I realized its 15 years old lmao

SGI UV is probably a ridiculously high priced collectible

Is there a RISC/V port of FreeBSD ?

yes, freebsd was the first operating system with a riscv port

risc architecture is going to change everything ... hackers (1995)

i don't know how it didn't work out yet. they had p6 chips that were triple the speed of the pentium.

Every time I hear the word Pentium, im reminded of that time I put Linux on a really old laptop and sent a system info screenshot to my friends and one said "blimey! A Pentium!"

if im running 14.3 without pkgbase, should i run pkgbasify before upgrading to 15.0 or after?

been on a pkgbase install through the betas. now i get "pkg: No trusted public keys found" ... not sure if that's a problem, or if the pkg servers just haven't all synced up yet for the release

there's a note about it

thanks

Macer: bro that film is quite unrealistic

ultra-superficial

i have zfs running raid10 over 4 SAS drives. zpool status reports all online. but just to be sure everything is ok, what smartctl command can i run to check if any of the drives have failed or are experiencing errors?

is it smartctl -a /dev/drv0 ?

my freebsd VM is happily on 15.0R now :D

first time doing a release upgrade

great job folks!

ariadna, was it difficult, or did you follow the Handbook to the letter and succeed?

I followed the handbook

I mean, I'm not as fluent in this OS as I am in Linux

I just can't upgrade the ports though

hello

Anyone know what happens between "Loading splash ok" and "---<<BOOT>>---" during a system startup? Since the last RC4 update (not RC4 itself, but the last pkgbase update to it), and now on 15R, my system hangs after the splash message.

I was wondering if it is possible to upgrade from 14.3-RELEASE to 15.0-RELEASE with pkgbase? do I need to run the pkgbasify command?

ok, it's solved, I just had to remove the reference to the 'latest' repo and revert to 'quarterly'

ahah .. Belgium is "complicated"

I'm in that funny spot where the question is: what's simpler/faster? troubleshoot my issue, or just reinstall and restore ... hmm

markmcb: use bectl and go back?

markmcb: boot with verbose and look for more info

i'm in a be now, so that works. when i ran into the RC4 issue, i booted a BE and copied the RC4 kernel from the installer image and that seemed to make things work ... i'll try verbose mode, haven't done this before but sounds easy

I'm on 14.3, mod_perl with p5-libapreq2-2.17 and I'm getting ld-elf.so.1: /usr/local/lib/perl5/site_perl/mach/5.42/auto/APR/Request/Apache2/Apache2.so: Undefined symbol "apreq_handle_apache2"

markmcb: if you troubleshoot the issue, you are guaranteed to learn something, assuming you have the time/energy to dedicate to it.

it's never worked on this VM, I've trying reinstalling and even compiling from ports, no joy

markmcb: ah yeah, EFI isn't covered by bectl. i always rename current boot images to .old, before replacing

just tried "boot -v" from the loader ... nothing new.

Demosthenex: yes, I renamed to .old before copying the new one over

mage: that's an impressive understatement there :°)

so if I read and understand well a 14.3-RELEASE should run pkgbasify before attempting pkgbase upgrade to 15.0-RELEASE.. correct?

does pkgbasify will be part of BASE someday?

mage It's not mandatory to use pkgbase on 15 so you don't _have_ to pkgbasify 14 before upgrading to 15

perhaps it will come to ports tree firts

cavokz: yeah I know :) I have tons of jails and I was awaiting pkgbase with great hope of speeding up upgrades

CrypticSquared, you got it working right? It was in the release notes, like 10 lines from the bottom.

having to freebsd-update -j ... -r xy-RELEASE on many many jails is painfully slow

mage: when i tried 14.3 -> pkgbase -> 15 i did not have a good time

but then i didn't see a documented update path

rtprio: I am not a big BSD expert. I went from 12.1 to 14.3 using this last ~7 years: paste.rs/6aeyy.txt

how long does buildworld usually take

approximately :)

it never broke my system, never failed to boot...

Thank goodness for mobile tethering... The WiFi decided to not cooperate after upgrading to 15 (not blaming fbsd here, I probably had a PEBKAC moment)

JetpackJackson: its likely freebsd

wifi breaks all the damn time

hell i'm running ghostbsd on a thinkpad E440 (based on 14.2 iirc) and the wifi stopped cooperating after suspend & resume.

good learning experience before installing freebsd though \o/

Ah

polarian: on thinkpad t480 with i7 cpu buildworld ( from scratch ) takes around 6 hours. with META_MODE and CCACHE it takes significantly less with consequent builds. unless llvm bumped or/and FreeBSD version bumped then buildworld is being rebuilt again fully. ( but with CCACHE again it is much faster.

more cores = faster builds as usual. there is no definite time for it.

angry_vincent: ah fuck me

T480 is probably 2-3 times faster than my laptop

so I should expect around 12 hours

it is not like that

maybe I should look into building on a build server and moving it over

angry_vincent: I know :P

it is not linear thing

I am aware

maybe I should have kept a core to use my laptop while it compiles :P

mehh scheduler should sort it out

wonder if I could tarball the obj and then move it over :P

then I can buildworld on a more powerful device

yes, you can

as well as share it over NFS

angry_vincent: link to docs?

oh you mean yes I can just tarball obj and copy it :P

dvl: the question is right, I was also curious if such a scenario is doable, please let us know if you will find the soltion

mzar: The multiple zroot boot problem? dan.langille.org/2025/12/02/how-to-…ver-when-booting-to-multiple-zroots

yes, this one

Hmm looks like the WiFi driver is crashing

JetpackJackson: which driver

Uh

One of the rtw

ah realtek

cant help ya

Ah

#freebsd-wifi on efnet might be able to though

or submitting a bug to bugzilla

How come the channels are on two different networks? Historical reasons?

freebsd is a fan of efnet lol

idk

the docs channel is there too

Ah ok

How can I check my specific WiFi chip before I go yap in another IRC channel lol

Oh fun I rebooted and it can't find the kernel

I think I botched my pkgbasify

At least this is a test laptop

Hmm can't tether

Shucks

JetpackJackson, I'd "grep rtw /var/run/dmesg.boot".

Moot at this point cause I can't access the system

Time to boot into an installer image and use the live option?

Oh true

Was gonna just reinstall

Yep. Should be a quick, easy recovery.

God I hope

Hmm

Lemme try a different usb

JetpackJackson: You should be able to recovery from that even without a live boot. Just using single-user mode.

Boot the old kernel and make your changes and try again.

OK booting, a bunch of modules say mismatch in multi user

Assuming the loader can find the old kernel. That's not clear from the report.

Single user can't find kernel

Wait now both can't find kernel

Wait I had to press 6 a couple times

One sec

Can't log in. Lemme try again

OK I'm here. What next

JetpackJackson: Mount your FS and make your changes.

What changes do I make to get it to see the kernel

.eou

Also I can't WiFi tether so I can't do anything that needs internet

JetpackJackson: Are you using bectl to create snapshots?

Well, are you making any snapshots at all?

No... But Theres nothing Important on here

Its just for testing

Maybe I shouldn't have used pkgbasify since I managed to mess up somehow

:( mad at myself for being a dumbdumb

"can't see the kernel" -> that's a little concerning

Should I boot a live ISO and try to fix things thru there?

you can still fix things from single-user

define 'cannot find the kernel' -- loader fails to load it entirely?

is it trying to load a `kernel` kernel, or?

Yeah

Kernel kernel

(I thought I had fixed it to not propose `kernel` if it didn't exist, unless you have a custom `kernels` setting in loader.conf)

I did something dumb on my end dont worry

Youre fine

well, probably not, but

I'm in the live system iso

What should I do next

unfortunately I have to run, but for now you could set `kernel="kernel.GENERIC"` in loader.conf to at least get loader doing the right thing out of the box

To fix my messed up 15+pkgbasify mess

Ok

as for the wifi driver, hmm

you probably want to use devmatch blacklisting to avoid loading it, but you have to grab the name first

I'm not worried about that now. Wanna fix the whole not booting thing

OK mounted drive in live system to mnt and chrooted in

How do I fix the kernel

/boot/kernel is empty

/boot/kernel.old/ has a bunch of stuff

did you look in /boot/loader.conf? and set the thing as suggested above?

OK now I did

JetpackJackson: Check out docs.freebsd.org/en/books/handbook/boot

Specifically, the "15.2.3. Stage Three" section.

Should give you an idea of what to look for as far as unloading any current kernel and booting an old (or different kernel).

I'm also quite sure pkgbasify does create a snapshot (if you're using ZFS). Which you should be able to put right back into place and reboot to the old snapper.

I'm.not using zfs but I should have

I think I accidentally Ctrl c'd at some point. I shouldn't have even asked in here cause its my fault

I would certainly suggest ZFS. It's quite handy.

I wanna fix it but I can't convert ufs to zfs and I dont wanna give up and reinstall

Cause I should fix it instead of giving up

When you boot into single-user mode, can you not "unload /boot/kernel" and then "load /boot/kernel.old" and then "boot"?

Well, I suppose that'll depend on kernel names. But, I'd guess you understand the gist of it. It should be simple if you have the kernel names.

OK booting old kernel

In single user

How do I fix the kernel now so that things go back to how they were

Just (re)move /boot/kernel and copy /boot/kernel.old to /boot/kernel.

that's proven to work

Again, I'd suggest ZFS in the future. :)

Yeah it was just a test

I agree. ZFS is worth it.

ZFS is better, but TBH we have not official way of downgrading kernel

OK rebooting and getting module errors

Live and learn! Good decisions come from learning from mistakes. Mistakes come from bad decisions. There's nothing wrong with learning.

Um ok I can't log in

JetpackJackson: Might need to edit /boot/loader.conf to not load modules that aren't built for the current (old) kernel.

yes, we always realn from mistakes, that's righ and efficient method

You can still boot single-user and boot kernel.old (provided you just copied it to /boot/kernel). Actually, you can likely just boot single-user and boot the current kernel and make your changes to the modules.

remove the spurious module

it's also proven to work, 100% reliable duct tape fix

does it bring panic ?

Its a bunch of modules

that's odd

But j can't log in.what module would do that

I botched my pkgbasify thats how

I'm trying to not let this get me into a bad mood but its hard

I kinda wanna reinstall with zfs and figure out snapshots

This is why I backup my files before any big update...

your mood is important, please take it easy, it's only OS

Is it ok if I just reinstall

JetpackJackson: Of course it is. As long as you're not losing anything.

mzar: it also doesnt help that the video game I'm playing lately is kicking my butt lol

There are a few different things you can do to recover. But, they're take time. A re-install takes a few minutes.

ek: yeah I'm not losing anything. This is the spare laptop

Out of curiosity, what part of the pkgbasify got messed up?

I did a dumb and Ctrl c'd but it kept going and I think I missed a step or smth

Or I Messed up the upgrade to 15

One of the two

ha..

no worries, pkgbasify will get polished, it will reach the ports tree at some point too

Yo Ctrl alt delete restarts? Nice

JetpackJackson: Ah. Gotcha'.

you are allowed only to press Ctrl T

JetpackJackson: Of course!

Thats really nifty

mzar: CTRL+T works for status in almost everything I've used it with. Love it.

Running installer now

yes, it's harmless

What vdev do y'all use for one disk of 100 ish GB?

Stripe?

Oh

Nvm I see the disk counts

You can choose whatever you'd like. 1 disk, obviously stripe is the only choice. 2+ disks, you can choose mirror, z2 or z3.

It's a pretty easy breakdown.

Hmm wireless is being strange

One sec lemme restart

JetpackJackson: here's the answer lists.freebsd.org/archives/freebsd-stable/2025-December/003641.html

I think its not happy

JetpackJackson: When you re-installed, did you choose traditional or package?

Haven't gotten there yet

WiFi is being weird

See images

Ah, dang. I'm not terribly versed in the wireless stuff. :(

Ah

I see where you're at though. Trying to set up the connection just to install.

Just kinda sucks cause I *can* tether but its annoying

Wait.. You can tether to what? The same wifi driver works for tethering to something else other than your router?

WiFi tethering thru android

Gonna do zfs and regular packages

Ah. With the Android physically connected.

Yeah

Idk why the WiFi is broken now tho

Yes. I'd suggest ZFS with the package install option as opposed to traditional.

Oh

Pkgbase?

Are you booting the 15 installer?

Yeah

Then, yes. Use the package installer as opposed to traditional.

It'll automatically put pkgbase into place for you.

OK one sec

Gotta reboot cause its giving no disks present to configure when I restarted the ksntaller

That'll happen depending on how far you get into the installer. Sometimes, you gotta go into the terminal and use gpart to destroy the bootloader/partitions to start fresh.

But, if you're just installing from scratch, who cares? A reboot and re-install is the same thing.

Yeah

Now it says updating repositories failed

Oh there we go

OK theres gotta be a bug with this WiFi or something

Unless I corrupted it somehow

This is the fresh install

Following handbook for WiFi config

JetpackJackson: Is there no SSID or is just hidden?

Also, when doing the install, you should be able to choose the wireless interface and just choose DHCP (or static) and enter your login information. Does that not work?

I tried at one point but the module broke

So then I tried to get the pictures of the output

Searching bugzilla now

Technically I'm in a zoom class right now but I wanna figure this out

hmm has anyone managed to install chromium with pkg on freebsd 15.0, looks like I cant find it :( when searching after it with pkg search..

JetpackJackson: That's no good. I wonder if there's another driver you can try. But, reporting the bug is certainly your best bet.

nwe: There are still some packages being built for 15.0-RELEASE. Might take a few more days for everything to be available.

I'm trying to see if its been reported

Especially if they (or a dependency) failed for whatever reason.

ek: this is what I have installed

ek: ah, thanks for answering :)

JetpackJackson: Is that config and pkg install from the original installation or did you install the pkg and do the rc.conf config manually?

Manually

The firmware was installed by the installer tho

During the install it would fail the same way?

Yeah

Ah, okay.

Well at least I'm on zfs now

Even if it has been reported in bugs, just report it anyway. They'll either link or close it. The more reports, the better (usually).

I can't seem to find my issue specifically

Like error codes and such

I should check dmesg

JetpackJackson: It might also just be a module that needs to be compiled against the new 15.0 kernel.

Ah

How do I do that

Should I? This poor laptop is underpowered

So, you could try installing net/wifi-firmware-rtw88-kmod manually in the meantime until the package is updated/available.

Ah

Obviously, that would require the ports tree to be installed.

But, that's easy.

Or, you could set up Poudriere.

Hmm installed it

Didn't ask to build

Just installed

Oh

That might've worked

Lemme reboot to test

Nvm it broke

Oh I need to load the module on boot

Hmm

This is a problem for later I think. I'm tired

You should be able to load the module at any time. But, it should be automatic with devmatch enabled.

It shows up in ifconfig.

5 hours in, and still buildworld, I think my 10 hours estminate was correct

that seems like a long time.

isley: I am compiling on an ivy bridge i5

i reguarly buildworld on an old ass 2015 era xeon and it doesn't take that long

okay then, that might be right.

welp I am 5 hours in,

no clue how long

did you launch multiple jobs? (make -j)?

15R running great!

isley: yes, all cores hit

ah nice. looks like the resilver is finished with the scanning part and is only issuing so it picked up some speed

might be done by tomorrow morning lol

Macer: whats resilver?

polarian: it's rebuilding the raidz2 after a disk failed

ah right

and unfortunately i put a drive in that was smr so it's been going for about 5 days now lol

hope you dont use another disk in the process :)

yikes

whats the capacity?

even for smr 5 days sounds like a lot if its not high capacity

isley: ah shit

I think I know why

actually no I dont lol

6 hours now :)

jesus 6 hours of downtime ?

cpet: not downtime, just compile time :)

90C on all cores on my laptop for 6 hours \o/

making a grilled cheese?

where would i start on trying to figure out what's gone sideways here? umbrellix.net/~ellenor/polaris12-huh.txt

90C for a CPU is way below its specs

depends on the CPU

As long as you are not on battery for that 90C compile run.

MelMalik, What version of the OS are you running (freebsd-version -kru)? What version of the gpu kernel module are you running (pkg info |grep drm.*kmod)?

I can't imagine being on battery or not matters, as long as the power endurance at that load is sufficient.

cpet: yeah but its hot enough that the chassis is uncomfortable on my lap, and the exhaust is almost hot enough to burn me

MelMalik: the first thing that went wrong was using an amd gpu :)

not constructive, Macer.

polarian: aww thats too bad

sorry. couldn't help it.

cállese.

I don't have a laptop with a battery that can endure 6 hours so far at 90C with probably the same in the future to finish.

indeed

mine will last maybe 1 hour

it can do 2 with normal usage

if its a gaming laptop they are knwon to get hot

extended can do ~3 hours

gaming laptop battery life is shit though

if you are compiling and using the full extent of the CPU and then comaplining that your laptop is hot

well

no comment

( a lot of curse words here)

MelMalik: seems like the wrong fw?

I don't think it was a complaint. I think it was simply a commentary on the situation.

I am really regretting running buildworld on my laptop now

MelMalik: just use fwget

I should have used another device, even if it had a weaker cpu, and let that burn

tarball /usr/obj and copied it voer

over*

cpet, it reports nothing to do

polarian: nfs and build it on something that "burn your lap with excessive heat cause thats what laptop manus want"

MelMalik: when updating to 15.0 I have to remove all gpu- pkgs and just use fwget

cpet: why nfs

i actually removed the gpu-firmware already and reinstalled it (building from ports for Reasons™), and that's what let me even get this far.

Reasons™: before going further I should probably check against plain my-upstream and then plain FreeBSD itself as I'm on a fork of a fork. but uname -KU is 1500502 1500502, drm-kmod is 6.6.25.1500502_6 and gpu-firmware-kmod (including polaris12 in particular) is 20241114,1 . if it works on my upstream, then somehow I've managed to change something in my branch, which I find unlikely as I've

really not touched much; if it works on plain freebsd then I need to politely yell at my upstream

err

20230625.1500502_2

anyway

polarian: I would build world then mount nfs and then mount nfs on the laptop and do a make installworld

yeah you can kinda just ignore all this spam

since when is asking for help spam?

cpet: why mount nfs to /usr/obj

why not just sftp it over

i mean i just have the source tree generate tarballs on the powerful server and copy those over usually

so you need it to be secure on your own network

heh

polarian: is distcc not a thing nowadays?

cpet: you trust local networks are a secure, I dont

Macer: sure, but it breaks some compile jobs afaik and theres no point distcc

I dont need it built quickly

just wish I compiled it on my server

polarian: so you dont trust your own home network ?

cpet: well my router is OpenBSD, but no I dont 100% trust it

lol

ok

always assume your network is compromised :)

you know how those openbsd guys are

Macer: sadly yeah

sadly?

OpenBSD folks are cooL!

so are freebsd guys, and girls

the fact that you said cool

makes me nbot trust you

see what I did there

polarian: do you download things then check the checksum to make sure what you downloaded is what you dowenloaded ?

checksums are not for security but integrity

but yes I mostly validate signatures

is integrity not a part of security

sure

should i try retrying loading the kmod, and then if that fails rebooting

i would like to feel how to live that paranoid

that did something odd... umbrellix.net/~ellenor/polaris12-huh2.txt

cause I just download, install and youtube

i DIY

MelMalik: reboot

cpet, already issued poweroff

oh this is a VM

if this is a VM you need to enable passthrough

or else to bhyve you will only have the emulated gfx

this is not a VM

i powered the machine off, and then powered it back on by hand

normally when you issue a power off its either a VM or IPMI

or whatever the equivalents are for other manus

i literally just said «poweroff» in a root shell, and then walked back to the computer when it'd died and pressed the power switch

so, a reboot the long way around

can this problem be caused by efifb being in use? i know long ago there was a problem with some gender of conflict from s.th. like that

nope

pastebin your sysctl.conf

not everybody uses a "pastebin" service to produce links to text files ...

what ever it is you want to use

but anyway, as i said, i'm pretty sure what I'm doing here is basically spamming, given my unusual circumstances

i yield my time.

not sure how asking for help is spamming but ok

MelMalik: atleast you didint call me a jackass when I tried to help you

so kudos to you for that

I... can see why the person who did that did, however

feels like talking to 50 grit sandpaper.

30 grit

MelMalik: back in the day asking for help was almost a 99% RTFM

MelMalik: now you mention google it's like eating cereal with soy milk

acmetool is weird

this isn't even in the manual

it should be if its gfx related

where would i find a manual on the various error codes the hw_init routine can produce and why

the error is pretty self explainitory

its trying to find the fw and fails

then finds it

now if you want to dig into the drm and driver code

thats not going to help you fix that issue

it's trying N+1 different paths, and ultimately succeeds after 4 attempts

i had a bunch of erros when I upgraded to 15.0 for some reason removing everything and using fwget didint say anything didint error but then I looked at pkg info and it installed 4 pkg's for both my cards

and then I did a startx and BAM!

kde loaded like magic

the package fwget would install does not install firmware to the first-try locations under /boot/firmware/amdgpu/, but to the fourth-try locations where it ultimately found the packages

er, the firmwares

it wa slike that movie anti trust when the guy goes i think I cracked it man i think I cracked it