ok...it's back to connecting to the wrong wifi at boot agian

glenny1: I have no idea if this is your problem, but I once had a similar-sounding issue caused by the default wpa_supplicant.conf containing a rule to connect to unsecured networks

wpa_supplicant should be properly configured

I see. Best of luck, then

i don't know what's happening. it's like it's not even reading wpa_supplicant.conf

when you don't know what's happening you're in a bad place, and should turn to logs or tools that could help you understand what's going on under the hood

if i have a bastille jail running a postgresdb, is it "safe" to bastille clone the jail using zfs ?

now dhcp just loops and the system refuses to boot. i'm in single user mode hoping to edit rc.conf but it won't let me

i'm just going to reinstall and try again. i have no idea why things are so broken

reinstall fixed whatever happened

*something* is making the EFI-lua loader on a certain system set "boot_serial" to "YES" (according to the "show" command at the loader prompt when booting) and I can't figure out why. I've combed through the /boot config files and I don't see what could be setting it.

sphex: we derive defaults from the ConIn/ConOut efi vars

`efivar -d 8be4df61-93ca-11d2-aa0d-00e098032b8c-ConIn` may be enlightening

kevans: Ah. Ok. Hmm. I see the loader makes use of EFI variables like this, but not when "boot_serial" gets set somehow... Anyway, is there a way to disable this?

I set "console=efi" (also tried "vidconsole" value) in loader.conf but this doesn't stop "boot_serial" from being set somehow.

sphex: there's a lot magic involved, part of it is in sys/kern for parsing howto flags (RB_SERIAL) and converting them into boot_* vars

But I see "boot_serial" being set directly in the loader (with the "show" command). Hmm something is wrong...

yes, by the magic

what practical problem is this causing?

ohhh this sys/kern file is compiled in the loader too...

kevans: I want it to stop using the serial console.

did you trry boot_serial="NO"?

huh no

hehe. so it's going to be this simple? alright let's try.

my recollection (can't get too distractd from what i'm working on at the second, sorry) is that we rebuild howto flags from boot_* foo right before kernel exec, so it should override the earlier logic and do the right thing

Ok. So yeah it just worked. I just had to set to "NO". Huh I should have tried it before. I let the source code confuse me I think. Alright thanks kevans...

BTW the loader menu is a little bit confused by this. The menu script only checks if boot_serial is set or unset, it doesn't understand "no".

sphex: there are a few other problems with how the menu interacts with console/boot_* vars, haven't had the time to straighten it out

for instance, if you set a console= in loader.conf that doesn't match what we'd infer from ConIn, the menu won't take that into account (iirc)

i began to install a package using pkg but it is pulling from FreeBSD-ports for the package and its dependencies. the FreeBSD documentation says it is best to not use ports and binary packages by pkg on the same system. am i correct here or is the FreeBSD-ports packages different from what the handbook is refering to?

glenny1: are you building packages on your local machine as well.. or just using pkg install <some package> for everything on this system?

if the answer is you are NOT building packages, you should be okay.

i don't even know how. i just ran "pkg install librewolf" and it started pulling everything from FreeBSD-ports

you are all good then :)

ok, thanks

packages are just ports that have been pre-built, if that helps...

it does. thanks

glenny1: FreeBSD-ports is just the name of the repo of the offical prebuilt ports.

yeah, it was decided that 'FreeBSD' as the name is ambiguous once a pkgbase repository was introduced

breaks precedent? sure. probably less confusing for future users? yes

I caught the repos name change and I think the new names make things more obvious and less confusing.

i am getting no sound. pavucontrol says "No cards available for configuration". module for soundcard is loaded. there is no audio group. user is in operator group

it does list volume controls that appear to be receiving audio. audio bar also jumps for integrated microphone which means it hears me

dvl: what does it mean if fresbports doesn't show that nice HTML grid of architectures vs build version per package?

I guess that some dependency didn't get built, so there's no versions available at all?

ugh. so the amd64 ZFS raw VM-IMAGE does not have 4K aligned partitions...

Please stand back. I'm running pkgbasify on a 2nd host...

:)

and now, I wait, remotely, for the reboot. gist.github.com/dlangille/e4ec0f3205617c531a2884a527c6316a

legend says the machine never came back and dvl still waits patiently :-p

kevans: It came back, and then I did a third host.

dvl: nice!

on to the 4th!

OK, that's FreshPorts prod converted. :)

i still have no audio on my system. i've tried every suggestion i've found online. none have worked. pavucontrol shows the audio devices exist and they are receiving sound. nothing is coming out of my speakers

all packages are up to date

Does running `mixer vol=1.00:1.00 && mixer pcm=1.00:1.00` give you sound?

nope

i want to mention that running beep plays no sound either

having /var/empty readonly via zfs was a fun adventure: dan.langille.org/2026/06/30/pkgbasi…-empty-is-a-readonly-zfs-filesystem

This is something that could be really cool: probablymarcus.com/blocks/2026/06/2…native-graphical-shell-for-SSH.html

dvl: I added this check to pkgbasify after running into this issue... imo having /var/empty enforced by zfs is a Good Thing

glad it helped somebody else

" beep

glenny1: my x1's hardware mute still mutes but mixer software cannot disable it;

rtprio: how to unmute?

in my case, it's fn-f1 so the mute light goes off. not sure if you're sporting similar hardware,

dch: I don't know where that concept arose from, I have it on two hosts.

See also dan.langille.org/2026/06/30/pkgbasify-authentication-error

dvl its from IIRC 9.x or 10.x era FreeBSD default zfs dataset layout

figures, these are my oldest hosts.

rtprio: none of my Fn+FunctionKeys seem to do anything

well, in 'not freebsd' it was also mute. you know the blue label with a speaker

glenny1, is your Esc/FNLock key's LED lit?

I have a 1-wire USB device (controlling lamps and measuring room temperature)

CrtxReavr: it is now

I read it via owserver

but sometimes it seems to disconnect from USB, and I can't persuade it to re-attach

CONNECT: ow_tcp_read.c:(109) TIMEOUT after 0 bytes

even unplugging the USB device, waiting & re-plugging it

is there some other USB hub or something that I might need to reset on it?

glenny1, with it lit, you should be able to hold down Fn and work the F1,2,3 keys to mute/unmute and move the volume up & down.

yeah, it recognizes the keys now. still, no sound

What does 'mixer -a' show?

hdaa0 is default (play/rec) and volume levels for vol, pcm, speaker ogain are 1.00:1.00

rec is 0.37:0.37 and monitor is 0.67:0.67

pcm0:mixer lists those values

beep gives nothing

user is in operator group

i have tried specifying the pins in devices.hint

dmesg doesn't appear to give any errors

the only error message i see is: hdac0: Command 0x2043b000 timeout on address 2

however, a forum post said that message shouldn't affect anything

dch: Do you know what /etc/mail/aliases updates are done by pkgbasify? It modifies the file (mtime changes) but compare to a snapshot, there is nothing output by `diff`

dvl: pkgbasify itself does nothing, so this is probably pkg itself deploying an update.

and if the file contents haven't changed IIRC it compares the 2 and then skips it

so mtime only seems sensible

you can always diff /.zfs/snapshot/before/etc/aliases /.zfs/snapshot/after/etc/aliases

Norway wins!!

lol. Almost scored again.

Almost again lol

glenny1: would you be able to pastebin the output of your mixer -a

alias tb='ncat termbin.com 9999'

mixer -a | tb

CrtxReavr: yeah, gimme a few minutes to boot the system. also dealing with another issue

Yay! After being stuck for a brief period termbin.com is back online again. I can recommend it to people again.

termbin's best feature is the simplicity with doing what I showed above.

I like bpaste.net (and the Pinnwand software behind it) but simple posting from stdout is a lot more tedious.

ncat isn't installed on the system. what is ncat

netcat

ahh, that

..and non-gnu netcat installs as nc on FreeBSD iirc.

There's gnetcat and netcat.

Hmm - there should be equiv syntax for 'nc' which should be in the base system, but I'm unsure what it is.

The command syntax is the same, yeah

CrtxReavr: termbin.com/ws8a

volume is on and should be working

boru, 'cept it's not.

netcat wasn't installed by default

What's not?

echo test | nc termbin.com 9999

both seem to be [OPTIONS...] [hostname] [port] if I understand the man pages right, maybe some options are incompatible, but the general syntax seems to be the same

For whatever reason, I don't seem to be able to alias with nc

What shell are you using?

why would you even use an alias instead of a function?

Good point. When sprunge was still up, I wrapped it in a function.

Shit's behaving weird.

Though now it's working.

Hint: netcat is called nc and installed as part of base in /usr/bin/nc available on every system. (Unless pkgbase has it in an optional package?)

aliases have the tendency to lead to mysterious problems, because the user forgot one of the finer points of using aliases, like where parameters come from (aliases don't have any parameters, as they are just dumb text replacement)

aliases also do not work when called from a script.

Depends(TM)

CrtxReavr: did you need any other info? i'm going to shutdown the machine if not

Got distracted by nc vs. ncat (vs. user issues)

looking at your output.

What if you run: `mixer vol=0.50:0.50 && mixer pcm=0.50:0.50`

Is that getting reflected in 'mixer -a'

?

CrtxReavr: yes, it is

did you try cat /dev/random > /dev/dsp already? Also against the other dsp devices?

still can't hear beep command

nimaje: i did it with a file

nothing

dmesg is also listing no errors

This is a ThinkPad, correct?

OS: Linux ('Gentoo' '2.18') (6.18.35-gentoo-dist) CPU: Intel(R) Pentium(R) Silver N6000 @ 1.10GHz MEM: 4 GB GPU: Intel(R) UHD Graphics (JSL) (Vulkan) UP: 16 hours 55 minutes 21 seconds ago

that's the basic specs. it's a Lenovo Ideapad 1i

And it works in Linux?

pciconf -lv

yeah, i have audio on this machine no problem

the only thing i can think to do is boot a livecd and see if audio plays through the speakers. just to verify the hardware itself works

however, considering others have had my same issue with the same hardware, i'm not too concerned with trying this

i also haven't tried headphones--either via USB or headphone port

Well, USB headphones would be their own sound device.

yep

the point is to verify there is at least audio of some kind

I haven't seen your 'pciconf -lv' output, but otherwise, it seems like the softare is seeing the hardware and doing everyhing it' supposed to do, so I'm left with questioning the hardware.

CrtxReavr: want to termbin that as well?

want me*

sure

Also maybe your /var/run/dmesg.boot

CrtxReavr: termbin.com/f6mm

i'm going to go afk for a bit. i haven't eaten all day and need food. i'll be watching IRC

I don't see anything there that looks like an audio device.

Oh, wait .

yeah, it's there

'Jasper Lake HD Audio'

yeah

Now do /var/run/dmesg.boot

so, i don't feel comfortable doing dmesg. however, the devices are recognized and associated with no errors

/var/run/dmesg.boot would just be your boot messages

one thing i did notice is that mixer speaker volume isn't able to be set, but it is when boot has completed

termbin.com/0sx1 <== mine from this vps

Woo-hoo! Look at the security advisories fly!

"using GLM-5.1 from Z.ai"

(3 of them)

Yep. AI is finding bugs left and right these days (which is good.)

i'm surprised we've gone this long since the last patch

It's fun to update machines based on the first three advisories, only to see more incoming after the updates are complete!

I'm a bit confused about the discription of the execve one, it mentioned a seperate process may access via procfs/linprocfs but doesn't mention what happen when procfs/linprocfs is not mounted

and other debug capabilities might also be possible

GLM is a nice model. Actually the chinese models are great. Cheap and affordable. Open weights if not open source. Essentially the opposite of all these VC-funded US companies.

I got my hosts onto pkgbase just in time for the patches today.

unlucky for me im maintaining a custom branch so i have to merge all these changes manually and build from source.

oh boy patch tuesday

ok, back

CrtxReavr: is there anything specific you'd like me to look for in dmesg?

Does your sound device show up in there?

CrtxReavr: yes

'Cause pciconf shows what the PCI bus can see, but it doesn't really verify a driver has the device initialized.

Indeed, updating a patch level within the same release is easier with pkgbase. However, until I do a clean release version update to my test server, my production machines will keep using freebsd-update!

does the zfs update need a zpool upgrade?

CrtxReavr: termbin.com/cb4f

plugged in headphones via USB and i can hear audio

hdac0: Command 0x2043b000 timeout on address 2

dutch: according to forum posts that shouldn't matter

i've already tried that. it doesn't fix anything

well, USB headphones work at least :(

plugging in through the 3.5mm jack gives no audio

i do want to mention that when the USB headphones were plugged in VLC chose the device immediately. not so for any other outputs. it requires manual selection and even then still outputs no audio

hm. freebsd update huh?

emails still trickling through :)

yeah. updates send out like 50 emails lol

dkeav: "does the zfs update need a zpool upgrade?" - Nope.

danke

but zfs feature flags though

Bitte schön.

Macer: I just updated one of my hosting servers (non-production) and there is no zpool update required.

Err.. zpool upgrade.

there hasn't been a zfs/zpool upgrade in what feels like a long while

You are not required to upgrade your zpool. But you may if you want to. Or if you want any of the newer features available. But if the existing zpool is working then you are not required to upgrade.

If you do upgrade then be sure to upgrade your bootcode so that it can boot the next time.

As for not having an upgrade in a long time, that's not correct because release 15 came with new zpool features.

The documentation is actually conflicting on zpool upgrades. The release notes or the handbook, I forget which, recommend against upgrading. But zpool status nags about it suggesting that you do upgrade. That's an interesting conflict. But I think it is only that the zfs authors think upgrading is desired since they added the feature and the documentation writers are tired of people forgetting to upgrade their bootcode so recommend

against.

having used a rescue cd to fix the bootcode, i fully understand

woops

Been there! I don't touch any of my zpools, unless there is a very good reason.

With 15 I have been working through and upgrading my zpools. Figured it was time to get everything sync'd up. And the code seems to be stable for it.

Whoops! I did a dumb!

I have learned so much from my mistakes. I'm thinking of making a few more.

ek, Don't tease us like that! Share with us. :-)

sometimes adventures in fixing things can be fun too

especially if you have backups :D

ran sysctl dev.hdaa and the speaker is listed as [DISABLED] for some reason

rwp: I used a server list in my for loop instead of the VM list for a reboot for-loop. :(

Has nothing to do with an FBSD derp. Just a personal derp.

I haven't finished looking into it yet, but my zpools are from FreeBSD 12 era, and I believe you can set a compatability level when you upgrade, so if all my boxes are on 15.0 now, then i can set the compat to 14 and rest easy that I'm not accidentally boxing myself out. but still, any zpool upgrade would almost certainly need updated bootcode