RhodiumToad: out of interest. On void linux I had to enable the agetty-ttys0 service

so I am trying to do the pci passthrough for my wifi card

i dont really understand what I am doing. Got wireless working in the bhyve vm

what do I need to know BSD side?

would this work? bsd.ac/nqdaavi

heh, cute: "error: Fssh_kex_exchange_identification: client sent invalid protocol identifier "GET / HTTP/1.1""

Probes. Endless probes. Sigh. It's the dumpster fire that is the Internet. (shrug)

can ZFS do snapshotting?

does ufs do snapshotting?

I just deleted some crap due to symlink

usr and home

ufs

any way to go back 4 hours?

UFS? AFAIK does not support snapshots. But even if you were on ZFS which does you would have to have had something making snapshots for that to be a help.

When on ZFS a best practice is to install zfstools which includes zfs-auto-snapshot. Install a crontab as directed by the package to create snapshots periodically.

agggkgkgkk

why does home map to usr home

whats up with that

Hmm... Doing some searching I found mksnap_ffs(8) which talks about snapshots on UFS. I know nothing more about it though.

At one time I knew the answer to /usr/home but at the moment I only vaguely recall that it had to do with either dataset inheritance or with boot environments. I remember thinking it made sense to be that way.

concrete_houses, wiki.freebsd.org/ExampleUfsSnapshots

Title: ExampleUfsSnapshots - FreeBSD Wiki

Also man.freebsd.org/cgi/man.cgi?query=snapshot

Title: snapshot

concrete_houses: funny enough, it was recently decided that it needs to move back to /home

uh guys.. any idea why fbsd 13.2 doesn't see my 2nd monitor? nvidia driver is installed, 1st monitor is connected via hdmi, 2nd to dvi-d - but doesn't get detected. Tried running nvidia-xconfig to generate new xorg.conf - nothing. the nvidia panel "detect displays" also doesn't see it.

Hi, I had a zfs pool composed of only one 10tb drive. Then yesterday added a 2nd, 12tb drive. Now I have a raidz with two drives (10Tb total size).

As the objective was replacing the old 10Tb drive, now I want to extract it, leaving only the 12tb. How can I do that?

that was easy, first offline disk (ada1 in this case), then detach it.

rwp: ufs does indeed support snapshots

really?

i thought only zfs did that

for a long time you could have _either_ snapshots _or_ journalling, though

but that was recently fixed

kevans: a victory for sanity! (/usr/home was always silly)

morpho: did you get it working?

RhodiumToad I'm up to the last bit

I have a Linux VM with wireless working

Thanks for asking. I don't know how to pass that connection to my bsd host

you'll have to use the linux host to do NAT for the freebsd one, I think, because of how wireless usually works

so configure the linux host as a NAT router (I don't know how to do that, but it should be an extremely common setup)

I THINK I even got that working

and then just make sure that the freebsd system's default router points to the linux IP on the bridge

How do I do that? What I tried caused "service NetIf restart" to hang

I think the way I'd do it is to not attach the host physical nic to the bridge, and put a private network IP on the bridge interface

have you configured the linux host to have a static private IP on its vtnet interface?

on the freebsd side it'd look something like this in rc.conf:

ifconfig_bridge0="inet 192.168.1.2/24 addm tap0"

ifconfig_tap0="up"

defaultrouter="192.168.1.1"

Yes I have something like that in my rc.conf

this is assuming that the linux host is statically assigned 192.168.1.1 on its vtnet interface

I have a static resolv.conf on my bsd host. I THINK this is also causing issues

that shouldn't be a significant problem

I'm on my mobile right now as my laptop has no internet outside of the vm

Thanks for heping. I will try again and get back to you

Is there commnd to let me see my bhyve guest network information from host?

the host knows almost nothing about the guest network config

Greetings

you can look at the arp tables to find what IP the host thinks it is using on the bridge, in most cases

*the host thinks that the guest is using on the bridge

RhodiumToad all i get is ping unknown host

doing what?

first thing to test is whether the host and guest can ping each other by IP address

thehn try and ping the guest's wifi interface IP from the host

anyone else seeing scrambled (looks like binary) in /var/log/messages ? restarting syslogd seems to have 'fixed' it but why would it happen in the first place?

recent 13-stable

scrambled how exactly?

i'll paste 1 line of output

ideally, pass a chunk through the `hd` command and paste to termbin.com or dpaste.org

it looks like how you'd see a binary file opened in a test editor. it's not the entire file either, just messages from this afternoon UTC

is it just the messages, with the timestamps still readable, or is it all binary?

also what filesystem setup do you have

timestamps and host are readable

zfs

ive checked zpool health

it's happened before, too. on april 17th

please show a chunk of the affected data hexdumped as I requested above

trying to fijnd the most recent

RhodiumToad: dpaste.org/2FDGP

Title: dpaste/2FDGP (Python)

wait, the actual messages file has <F1>~u^H etc. in it? literally like that?

yes

this is 13-stable, arm64 (it's a rpi4/8GB) hd is usb3

is syslogd listening on any sockets?

it's launched with -ss

also this machine is firewalled really tight

only sshd is available

www3 is available but only locally

sorry, www

weird

relaunching syslogd fixes it for weeks

might be a bug in syslogd then, but hard to say

lol these are the most frustrating type of problems

ok - worth making a PR?

if your syslogd isn't very active it might be interesting to ktrace it

yup. lots of space for the logging too

and leave it traced, then if/when it happens again you have something to look at, which will help determine whether syslogd was sent bad data, or received good data and logged it as bad

ok

how would i launch it and with what permissions? ktrace i mean. i think it might want root

you can use ktrace to enable tracing on an existing process

so just ktrace syslogd as root ?

ktrace -p somepid will suffice, that'll put the trace file as ktrace.out in the current dir, use -f /path/to/file to put it elsewhere

the ktrace program does not remain running, all the tracing is done from the target process

ok thanks. as root or will user do?

as root

ok

users can trace their own programs, but usual debugging visibility rules apply

ok hopefully next time ill have something to go on! thanks

the trace file is binary, but you can read it using the kdump command

yep just tried that with -f it's working normally

just remember not to use ktrace -C because that turns off tracing for all processes if done as root

[/usr/home/freebsd] $ uptime

2:47PM up 1400 days, 2:57, 33 users, load averages: 0.79, 1.09, 1.18

:D

if your uptime is that high, you're not updating often enough :-)

correct

RhodiumToad: is the patch from D24324 still applicable to today's tree, or does it need any adjustment?

I can regenerate it if need be but it's not like it's complicated

also there's been a bug open for depends-list.sh that I don't think has been addressed

just waiting for git to tell me if there's a conflict

looks like the patch should still apply cleanly enough

yup, it does

cool, thanks

bugs.freebsd.org/bugzilla/show_bug.cgi?id=269982 <-- that's the depends-list one, but the patch there doesn't address my last comment

Title: 269982 – Mk/Scripts/depends-list.sh: Default flavour for dependencies are overriden

(it breaks things like make fetch-recursive and other recursive targets)

I can put up my local fixes for those if it would help

I'm having issues with getting NFSv4 acls going on some fbsd virtual machines - here's one example - fbsd 13.1 vm running under bhyve also under fbsd 13.1. i have the rw,acls option set in fstab.Yet when trying to setfacl on anything, i get: branding mismatch, existing acl is posix 1.e... OR acl_set_file failed; operation not permitted - anyone

care to entertain what's going on? Documentation is confusing, to say the least, and to complicate matters, the virtual machines under bhyve are in their own zfs filesystem so i get snapshots of the vms as desired ...

I'm going to simplify: hypervisor = FBSD 13.1with zfs dataset in which below vm runs, i.e. zfs list yields /vms/fbsd/virt_machine_1

guest under this hypervisor = FBSD 13.1 with rw,acls in /etc/fstab

only POSIX 1.e acls are useable in this vm ... how to get it to use NFSv4 acls?

...do I need to enable acls in the hypervisor fstab?

actually, strike the last statement because use of zfs on root - no entry in fstab ...

RhodiumToad, As I found out and posted those links to the documentation for it. :-)

RhodiumToad: can you elaborate on your last message?

RhodiumToad im still trying to set up the NAC

when i service netif restart it hangs

hrm, any ideas? dpaste.org/eOp0c/raw same command gives different return codes when run from subshell

Demosthenex: that first line doesn't do what you think it does

sh -c expects ONE argument after the -c which is the whole command string, and any additional args after that go into the positional parameters starting from $0

RhodiumToad i am still stuck with using my linux guest as a NAC

stuck how?

well I cant restart netif without it hanging

don't focus on that

i think everything is working in Linux

i just cant ping from my host

RhodiumToad: yeah, puppet's calling it this way, gotta tweak it

can you ping the guest from the host?

how do i do that?

what IP did you give the guest's vtnet interface?

Im not sure. I have defined eth0 (my guests ethernet interface) using wpa_supplicant

with the address 192.168.6.2

is that the passthrough wireless nic though?

I am using iptables to pass the wireless to the ethernet i think

no

that is impossible

davidschlachter.com/misc/t480-bhyve…3.0,ac%20on%20Linux%20and%20Windows.

Title: FreeBSD WiFi PCI Passthrough: Intel Wireless 8265, bhyve, 13.0-CURRENT

followed this guide and am at the end of it now

RhodiumToad: indeed, puppet was missing a line of quotes!

morpho: ok, so the linux host is configured to do NAT, which is what we want

morpho: can you ping 192.168.6.2 from the host?

ping: sendto: Host is down

ifconfig bridge0 says what? paste the output to dpaste.org or termbin.com

note you can paste to termbin using | nc termbin.com 9999

sorry

i had to do that from my linux host, no internet in freebsd so far

and ifconfig tap0

also the output of arp -a

you might need arp -an

what does linux say about the state of the eth0 interface?

inet 169.254.120.217/16 <-- that's not 192.168.6.2

so whatever you configured so far on the linux side, you haven't successfully configured the interface

RhodiumToad iv got linux tech support on the line :s

thanks for your help

RhodiumToad i can pink the guest

ping* i cannot ping the world

ok. is forwarding enabled on the guest?

is there a monospaced block in mdoc for an example snippet? I'm showing a long output piped to column(1) but the man page in pdf it not aligned

i ran this script... termbin.com forwarding in enabled on the linux kernel

Title: termbin.com - terminal pastebin

incomplete url

rdg: maybe look at the source for the man page for tbl.7 ?

(seems to be using .Bd )

RhodiumToad: I'm also using Bd, but in pdf the result is not monospaced

MANPAGER=mupdf man -T pdf tbl

morpho: this is getting more into linux than freebsd

what do i need to google

dunno.

can we assume this works?

are you SURE you enabled forwarding?

the voidlinux irc say so

net.ipv4.ip_forward is set to what, on linux?

1

the fact that your FORWARD chain says 0 packets is suspicious here

not really

why not?

its just a vm doing nothing

yeah, but you tried to ping the outside from the freebsd host

what is the default route set to on freebsd

use netstat -rn to see that

i think i have multiple

no, you don't have one at all

is defaultrouter set in your rc.conf ?

don't try and restart any services, just tell me if it is set

yes

to 192.168.6.2 ?

yes

do this on freebsd: route add default 192.168.6.2

then try pinging the outside

i can ping the outside

on freebsd

progress!

does that means it works?

basically yes

i am getting 41Mbps in freebsd firefox

well that sounds pretty good

thanks so much man

you'll want to reboot everything at some point to make sure you got the config files right.

yes!

the linux driver bin

its reproducible, i created a few scripts to automate it all

the only script you should need on the freebsd side is starting the VM

i am disabling wlan0 before i start the vm

I don't think that can work without a reboot

if the wifi hardware is being used for pci passthrough, it can't be used on the freebsd side at all

i.e. if you didn't boot with pptdevs=whatever in loader.conf, you can't do passthru, and if you did boot with that, freebsd won't see the wifi

i am doing that in a script that loads the bootloader

loads the linuxvm

how?

hmm

what?

that looks reasonable

RhodiumToad: now I have backlight keys, power saving, 802.11ac wireless

Its a shame I didn't document this journey more as some of the guides for setting up laptops are a few years old now

thanks for the help

yw

bit of a shame that getting the wireless required using a linux VM, but...

(I have a printer that has no freebsd driver, but that doesn't require using a VM since i can run the linux printer driver under the linux ABI)

I wanted to learn bhyve anyway

so it gave me an excuse to learn some new things

i just use my partners Mac for printings haha

I think OpenBSD supports 802.11ac on this wifi card