how the hell is it that mixer has to be invoked to set recording device again?

man says "=rec <recording device>" but doesn't say what a recording device looks like.

tried just using mic, but the output of that is even more confusing…

mixer: unknown recording device: mic

devices: vol, pcm, mic

like, "I don't know what mic is, but maybe you want to choose mic" :F

Interesting. Does 'mic' show up when you try 'mixer' and 'mixer -?'

also, "mixer recsrc" gives me nothing at all – and this is a soundcard that's essentially just 2 big XLR ports for mics and a few knobs… recording is literally what it's intended and designed for…

Oh, that makes me think you're set to use a different audio device. What does 'cat /dev/sndstat' show?

yes, tho "mixer -" isn't a valid command?

I said 'mixer -?' with question mark included :3

i did the sysctl thing to set it to the right one. i can play back on headphone out of the card.

aha

mixer: No match.

Oh, you're probably using csh, my example wasn't very good

sorry, lemme think

i also have fish, bash and the preinstalled ones like tcsh on this system. but all of those don't seem to handle this -? thing right

The point is to get mixer's error message, which does include a list of rec devices as the last line

yeah, including the one it didn't recognize.

devices: vol, pcm, mic <- pretty sure you're talking about this.

these are also the things a blank call of "mixer" shows.

this kinda looks like there's two different ways mixer queries for what it calls "devices" tbh…

I guess I have no idea how to solve it. You should probably paste the output of the six commands I've pasted here for my machine dpaste.com/7F3R8P8U3.txt

Maybe someone else can help, then :D

paste.xinu.at/scr <- uname, sndstat, snd.default_unit, mixer

Title: stdin

interesting, it doesn't see mic as a recording device

yeah, but i already tried "mixer +rec mic", too and get the same confusion from mixer… "i don't know mic, i just know vol, pcm and mic" ^^

oh, maybe important context: this worked before. but i yanked the volume with mixer up to 100 for pcm and vol and got extremely jarring distortion. unplugged the card while some process was still using it and couldn't get it recognized (because mixer10, possibly others weren't removed when i unplugged the card) and tried "kldunload snd_uaudio" at which point my machine spontaneously rebooted.

after reboot, the distortion is gone, but no signal for recording. i can still hear the mic in on the headphone out of the card itself tho – that's done without roundtrip through the pc, apparently

adjacent question: how do i remove a usb soundcard "safely"?

hello

MAKE_JOBS_NUMBER=1 and DISABLE_MAKE_JOBS="yes" on /etc/make.conf have the same result, right?

CyberCr33p: test it. set DISABLE_MAKE_JOBS="yes" and see if it results in make -j1

meena yes it's the same

CyberCr33p: cool!

phryk: removing safely more or less means, calling sync (which is only for file systems to tell them, to write from cache to the medium) and switching power off. The latter is not really needed, as the electric mass on USB is longer, and all should work fine. Hence: a USB sound card should not be damaged in any way by just unplugging it. And

FreeBSD should be able to handle this without being warned too.

phryk: you need to ensure that no application has /dev/dsp* open before removing a soundcard

Interesting point. Such an applikation would likely crash :-)

Trying to run debootstrap to install Debian bullseye in a jail dataset, it dies with: mkfifo: cannot set permissions of '/run/initctl': Bad file descriptor

Are there known issues with fifos and Linuxulator?

mason: if it's inside jail, maybe jail config is to restrictive?

meka: Wasn't in a jail yet - I ran it as root from the base system. This was just initial prep.

ah, I get it. My bad

No worries.

brb, errand, but I'd love to explore more ideas if they bubble up

I do have ubuntu chroot, though

this is what I've done: meka.rs/blog/2022/07/01/freebsd-linuxulator

Title: meka - Goran Mekic - hacker and a musician

morning town!

oh crap wrong channel. morning anyway, all of youse

meka: Useful blog, thanks.

mason: you're more than welcome and thank you!

meka: does sound work yet in Linuxulator? Seems broken in 13.1.. guessing it will be fixed in 14.0?

video does work

pertho: his blog post above mentions that it does.

well, pavucontrol picks up on the mic.. and sound playback works.. but signal messenger doesn't even after selecting the internal mic

but I think the biggest issue is the calls (sending/receiving) don't work. some sort of NETLINK system call issue

pertho: that's my experience, too. For example slack, I can hear notifications, but calls don't work

so, I just came by interesting observation, I have 2 FreeBSD instances running dovecot and they run in multimaster mode, so they replicate their stuff, I create a file on one and it gets replicated to another, in this example it's sieve. Now the interesting observation is that it not only replicates from A to B but it also sets the birth and ctime accordingly, even worse, the replica is sooner than the

original, so is there a way to tell the OS to create a file using specific dates for creation and birth ( actually falsifying them ) ?

nazdar gusto. man utimensat?

dockaj

you can set ctime/mtime/atime (even with touch)

For

file systems that support file birth (creation) times (such as UFS2), the

birth time will be set to the value of the second element if the second

element is older than the currently set birth time.

vyzera to byt ono

setting ctime with touch?

I know about atime and mtime, ( man ) but to my knowledge not ctime

(ok, no ctime with touch, but mtime yes)

seee

gustik: Have you verified time on the boxes?

yes

both boxes are synced with NTP

that was my first idea

but now it becomes clear with this utimensat he can modify all times, and it looks like the replica is taking over the original time and when it's done, the original overwrites his time with a later time (usually one or several seconds off) ... maybe to mark that sync was successful

mtime…modifie time, atime…access time, ctime…?

creation time

or change time

creation is birth

on linux both same

gustik: sounds like a dovecot question. it you check the source

it's create, or, as we say in Unix: creat(2)

yes, no one is home there, however, it's more unix related since the question was more about if an app is even allowed to do that

are you sure? stat(2) states st_atim access, st_mtim modified, but st_ctim status change

gustik: sure. look at the awful crimes rsync does. even curl will set the create time of a download to when the Webserver says it was created

(if it does that)

yes, as long as nothing prevents them from doing so

no: stat(2) states st_birthtim Time when the inode was created.

I see that it's even with utime(), ppl ask with stat() and then just copy that value by "=" to utime(path).modtime (or creation time)

One but only one of my -current machines refuses to ssh auth for localhost with the agent. `ssh localhost ls` asks for a password although the agent is there and the right keys inside. Any ideas?

cracauer: anything in the logs? any libs wonky or missing?

Logging into this machine not from localhost works fine with the same keys.

Let me check logs and then run sshd in debug mode.

Nothing new. Just says that it rejected all 4 keys in the agent. /shrug

Howdy! I'd like to acquire some FreeBSD stickers for my laptop. Of course Googling `freebsd stickers` gives me lots of options. Is there an "official" location whereby I might support the project with my purchase?

pedahzur: is there anything linked from freebsd.org or freebsdfoundation.org ?

(if not, then the answer is: no)

meena: I see marketing stuff and artwork, but no "buy stickers here" kind of link. :)

then, I'm afraid, you have your answer

(but I'll happily take several hundred per year in (virtual) Hardware to support alpha.pkgbase.live )

Title: Unofficial FreeBSD pkgbase repository

Bummer.

if i add devil's horns to that package logo, i can drop the Un in Unofficial. Right? Right?

well, looks like utime indeed can only updated atime and mtime (and that in turn updates ctime), but it can not go back in time and falsify ctime, that's where utimensat comes in place

from what I understand, utime only can see a struct/tuple of atime and mtime (and that's how it can overwrite them) but those new nanosecond utilities seem to directly write into the filesystem struct, so that's how it can falsify any timestamp

eh, again wrong

pedahzur, I could swear I saw som... ah. freebsdmall.com/cgi-bin/fm/bsdsticker?id=9W3LakjA&mv_pc=402 .

Title: Product Details

V_PauAmma_V: Thanks!

There's also a bumper sticker, if you own a car.

because the nanosecond functions are also a tuple of 2 objects, he only fakes the birth time when mtime is older than birth (exactly like stated in the man page)

hi. i run pkg upgrade on a jail that had Zoneminder installed. it upgraded php 8.0 to php 8.1, but the version of zoneminder available is only compatible with 8.0. how can i force it to install php 8 instead?

meena: so your rsync example would not be rsync but the OS itself, since rsync sets the mtime (correctly) to a date that is older than its birth and then he goes according to specified behaviour and fakes birth according to mtime

so I think we have solved it

gustik meena: stat(1) shows all versions well: `stat -f "%Sa | %Sm | %Sc | %SB | %N" somefile` (or without `S` for Unix time stamps).

stat only shows it, we were talking about modifying them. Of course I am using `stat -x` in order to check the results

hjf_, see pkg-lock(8). (You may have to choose between PHP 8.0 and 8.1 though.)

Mainly I ment, we can check the meanings well using stat(1) (including that c is changing e.g. the file flags, not the creation time, but b…birth ist the creation time). `stat -x somefile` is good, but does not show birth-time (%SB)

do `stat -x | tail -n 1`

How is APFS support in Linux? I'm trying to grab some files from a Mac Mini that refuses to boot.

I think it may be corrupted as plugging it into another Mac doesn't work - the Mac doesn't like it.

Specifically for Ubuntu 22.04 LTS

opps, thought i was in #linux

I am struggling to get two vnet jails talking to one another, they're both on the same machine, but attached to two different if_bridge(4) interfaces. I'm hoping I've just got some stupid pf rule missing that somebody might be able to help with

V_PauAmma_V: i did pkg lock php 8 but it insists in installing php 8.1?

hjf_: in parallel? or to replace the other?

rtyler: what's your config look like?

here's a snippet gist.github.com/rtyler/359579ad03ccebcc76203d9bd9c480a3

Title: jail.conf · GitHub

meena: on an old build of this system, I had this structure more or less working, but I cannot find a copy of my old pf.conf, which is where I am _pretty sure_ I was allowing the interconnect