does anyone know what group one needs to be in for access to VAAPI?

I could of sworn it used to work fine, but now it doesn't init unless I run as root... hmmm...

I'm a member of 'video', for example

hmmm, I think my devfs rules perhaps aren't working properly?

Anyone have any tips on figuring out XkbModel?

X isn't recognizing my volume keys (xev isn't showing anything at all when those keys are pressed).

Hello. I was wondering if anyone here had managed to get FreeBSD to function on an ASUS C201 (or similar) ARM chromebook? It's ARMv7.

morning

what was the makefile syntax to capture a shell command's output to a variable?

!=, thanks

bleh, I'm losing my mind here

ok, so 'sudo vainfo -display x11' runs fine and produces the expected output

but 'vainfo --display x11' fails due to failed init

which suggests permission issues.... but I'll be damned if I can find out where :/

vainfo --display drm ..... that works fine, but using the x11 driver does not

edenist: the only thing that comes to my mind is "video" group

meka: thanks, yes I am in the video group. All the permissions in /dev/drm and dri are set to video too

it's weird that only the x11 driver seems to have this issue, I have no idea what has changed to cause this

edenist: O dpm

sorry

I ment to say I don't know how vainfo works, but you might need to open port for X? I'm just guessing, of course

When is FreeBSD's next quarter update? (R-13.1)

for packages

the quarter just started, the 2022Q4 branch exists in the ports repo, not sure if the first pkg build is running or already finished, probably finished for some pkg repos and still running for some others

nimaje: Thanks a lot.

I think we could permit creation of groups within a jail (both VNET jails and non-VNET jails) and simply rely on the existing prison_load_ip(4|6)/prison_equal_ip(4|6)() checks when binding a port. That is, processes can bind to a LB group so long as the processes either 1) belong to the same VNET jail, or 2) belong to non-VNET jails with compatible v4 or v6 address visibility. In the latter case, if two processes belong to the same jail, they can bind to a

group, or if one process is in jail J1 with child J2 that inherits the addresses of J1, then processes in J1 and J2 can bind to the same group. All of this is still subject to the effective UID check mentioned above.

btw, ty again freebsd. smooth upgrade today.

Affected? nitter.fdn.fr/nixcraft/status/1577599587858415616

Title: nixCraft (@nixcraft): "Heads up! Linux users with Intel GPU should avoid the Linux kernel version 5.19.12 from damaging the display lore.kernel.org/all/YzwooNdMECzuI5+h⊙ic" | FDNitter

The kernel bug, wouldnt it be part of the subsystem linux and fbsd share with relation to the video drivers?

AmyMalik: i'm not completely sure, but looks like current drivers are based on 5.10

right

Does anyone use FreeBSD and restream.io with virtual_oss? I get my own mic echoed back into my headphones. This does not occur with other video and audio recording applications in the browser.

The settings in restream (and other web-app video/audio applications) only show me 'Default' for my audio devices. I have only a single input and output configured in virtual_oss and things work seamlessly elsewhere throughout the OS.

ggb: may I ask what purpose is virtual_oss for?

That seemed the best way to get my setup working when I was first configuring the OS.

I have a USB DAC -> amp -> headphones and an XLR mic -> Scarlett audio interface -> USB.

ggb: wouldn't you just use the mixer to ... not hear the mic in the headphones?

rtprio: there is no mixer in restream and there is no problem in any other application.

Audio input is redirected to my output in restream. The latency makes me feel like it is not a network roundtrip, but something funky happening locally in the application.

No other video conference web app I have used exhibits the same behavior, i.e., nothing else mixes my input into my own output.

And no native apps give any trouble.

ggb: if scarlett is stereo interface, you probably don't need virtual_oss

also, if it's firefox, did you try other audio backends?

meka: my USB DAC is separate from scarlett, if that is what you are asking. I have two different PCM devices configured through virtual_oss 1. my DAC and 2. scarlett

ggb: ok, makes sense

I do not have any other known working local audio setups.

I am very open to changing it, but probably do not have time for extensive experimenting today, as I have to present in a couple hours.

Short term workaround is easy enough to just turn the volume on my headphones down when I am talking (easy enough with an anlog knob positioned next to my keyboard).

I will try some other browsers in a bit.

I suppose you're combining it into /dev/dsp using -P and -R. I mentioned firefox as I know how to switch audio backends with it (it's noted in pkg-message). It's the only thing that comes to mind

Spot-on for the options.

I will give firefox a try.

Firefox works with restream.io and virtual_oss. Chromium was only displaying a 'Default' device both in the browser permissions dialogue and in the app's settings.

Firefox shows a number of dsp devices, but also, critically, does not seem to mix my input into my output regardless of devices selected.

Some device selections obviously do not work, but selecting my virtual_oss /dev/dsp works a treat.

Thanks for the feedback, y'all.

Also, looks like I need to update my prior on 'Chromium tends to support heavy webapps better'.

How do I do `make buildworld` without it doing a clean first, I'm trying to debug a build failure.

aha, it's NOCLEAN=yes

Crocodillian: Also look at: wiki.freebsd.org/MetaMode if you are going to recompile the world/kernel a lot

Title: MetaMode - FreeBSD Wiki

thank you

so, considering the fsync issues on linux (wiki.postgresql.org/wiki/Fsync_Errors), there were some reports there on freebsd, but not with zfs. anyone knows how zfs handles that?

hello

what does it mean that kern.opts.mk is missing?

ahh I got it

Demosthenex, I have collected some URLs to investigate "later" but haven't gotten to them. Would you will find them interesting? Even if they are references over ten years old? christopher-technicalmusings.blogsp…up-freebsds-nfs-on-zfs-for-esx.html ateamsystems.com/tech-blog/solved-p…sd-zfs-backed-esxi-storage-over-nfs

Title: Technical Musings: Speeding up FreeBSD's NFS on ZFS for ESX clients

Be sure to look at the dates on those articles. And I know it is talking about NFS and sync rather than fsync() directly but it feels related to me.

21

Hi.  I have a laptop with Intel iwlwifi (AX200) and AMD Renoir graphics and Zen2 CPU (Ryzen 7 Pro 4750U). Does any FreeBSD 13 / 14 support it?

icarious: iwlwifi is under active development in 14

rwp: o

i'll have a peek

#freebsd, what is the status of LACP on FreeBSD? I've come to the conclusion that Linux just doesn't seem to be able to get it right.

meka : So, is it better to use "14" if I want to jump to FreeBSD from Linux based on the hardware configuration I mentioned?

What I'm asking is, does anyone in this channel use LACP regularly?

stable/13 has been getting updates to iwlwifi driver from 14 periodically. I started using 14 when stable/13 did not have any support for that (along with graphics/drm-510-kmod for Intel GPU driver)

I can't say which version would be more usable in terms for [GC]PU support

dch: do you use LACP, or am i misremembering?

s/for/of/

meena: yes I have several of those things but I cannot claim to be an expert. If you're thinking cloudinit we can find a way of testing this probably.

Good evening

I've a bit of a problem with a freebsd box using full route table and frr, in the sense that the route lookup is very slow

dch: nah, hexadecagram is asking

date; netstat -nr | wc -l; date => Wed Oct 5 22:57:49 CEST 2022 1041632 Wed Oct 5 23:02:42 CEST 2022

So about 5m for looking up 1041632 routes

dch: I think as far as cloud-init goes and LACP or similar, I'll probably tackle that some time next year

Is there something that can easily be tuned like a cache to make the lookup faster?

meena: I think the intersection set of people using LCAP & cloudinit is very small

dch: eh, i just want feature parity.

*LACP

hexadecagram: i use lacp..

hexadecagram: lagg(4) works fine.

dvl... new york calling...

gman999: here too!

lol..

#metabug

;)

I asked this last night but I'll throw it out again now while there are more people here ;-)

I've got a really weird issue at the moment with vaapi, specifically the x11 driver.

if I run 'vainfo --display x11' it fails, but 'sudo vainfo --display x11' works fine, which suggests a permissions issue

I am a member of the video group and have access to GPU hardware

'vainfo --display drm' works as non-root user, so there appears to be something up with the x11 driver

normally I wouldn't mind as most things use drm directly, but mpv really works best with the x11 driver

I might write up a post on the forum, I am suspecting there might be a regression somewhere, as it used to work for me a couple of months ago

edenist: #freebsd-desktop might be more helpful

if no answers here

thanks koobs I'll pop by there too ;-)

roger