hay when i load up lxdm how come there is no session dropdown section for fluxbox to login with? when i pass credentials it says failed to start session

thunderbolt networking supported?

I'm trying to decide whether to put FreeBSD on an M1 Mac or a 2020 Intel Macbook Pro... any thoughts?

intel

i haven't seen or heard any info on the m1 port in some time... i wouldn't use it

tho fbsd on any mac will test your patience and most things just dont work

would netbsd or openbsd work better?

could try openbsd but in all honesty, linux is a better choice

this is only because of the hardware in macs, propieretary wifi chips etc

linux just plainly has better support for the hardware, unless you have some 2010 macs lying around

I do, come to think of it

2012, rather

How can I enter the L2 mac address statically for vxlan unicast connection on freebsd?

Linux commad equivalent brıdge fdb appen aa:aa:aa:bb:bb:bb dev vxlan10 dst 172.20.20.20

should work the same way as assigning an L2 address to any interface?

I don't know RhodiumToad

ifconfig ifname ether aa:bb:cc:aa:bb:cc ?

hmm I'm try it

Great timing for me, I have an unrelated but still networking question: I want to be able to access my bridged adsl modem from my laptop (or any machine on the lan). Currently I do static configuration and plug directly into it. Is there a way to configure a route to it? The device is em0, but has no ip. The tun0 interface connects just to the other end. Should I configure em0 to have an ip and route

192.168.1.1 to that? (I'm asking here first because I don't want to kill internet - my wife is using it :-)

Am L2 MAC address?

Not an L3 MAC address?

the modem is running freebsd?

RhodiumToad: no the router is. The modem is just some consumer box, but it serves a web UI on the lan port even in bridged mode.

and you want to be able to access that UI? should be no problem

I figure if I can point a browser to it when plugged in directly, fbsd should be able to forward web packets to it. I just don't know how.

exactly!

(so I can see things like link quality, which is currently in the toilet based on my horrible upstream lag)

Is that idea generally correct, ie assign an ip em0, and add route 192.168.1.1/32 to <em0 ip>?

usually what you'd do is add an ip to the appropriate interface that's on the same subnet as the modem's private ip

RhodiumToad > thanks for advice is working.

RhodiumToad: I guess ifconfig adds the route? I did 'ifconfig inet em0 192...2/27 192...1' and ping works. Now why the http traffic doesn't... perhaps pf is interfering.

Hi! Is there a "standard way" of overriding the rc.d script bundled with a port? Directly modifying the service script doesn't work for long since it gets overwritten by a port update. (For instance, the rc.d script for caddy has an extra `reload` directive which reloads the Caddy config but not external certs. Adding `--force` to the directive works but gets removed when caddy is updated.)

I don't know if you want to deal with this, but it's not hard to build caddy with xcaddy. So you can rebuild it whenever you want, and your rc script will go untouched

tmtt: if it would benefit everyone, file a PR with patch? or introduce additional rc variable that would add --force

scoobybejesus: I guess it would work, but I don't think I want to go down this route. The caddy package built from ports is good-enough for my needs. Plus, there's another service (turnserver) for which I'd like to add a custom `reload` directive, too.

yuripv: I actually thought about doing this, but I don't think I know enough FreeBSD to modify, build and send the patch for a port at this time.

try asking maintainer about it, it could work :)

tmtt: there's no standard way to override stuff in rc.d scripts that aren't exposed as variables.

That's a nice idea. I guess I could send them a mail with my changes. Though, it could take a while before making their way to a stable FreeBSD release.

(assuming they get accepted lol)

tmtt: i never build my own ports, I just bug people until they give in

meena: I guess I'll just make a my_caddy rc.d script then

what are you trying achieve exactly?

world domination!

err wrong chat

>.>

dkeav: How could you tell?

:o

meena: Well, the caddy service has a custom `reload` directive which makes Caddy reload its configuration. The thing is, it doesn't reload external TLS certificates (e.g. certs managed by acme.sh) unless the configuration changed. To overcome this, caddy has a `--force` flag which makes it reload everything even if the configuration hasn't changed. Thus, I simply added `--force` to the reload command inside rc.d/caddy.

tmtt: what's the difference between reload --force and restart?

No downtime?

you could add a force-reload comman to the script, and propose that as patch

Why not

I just don't think I know enough FreeBSD to clone the latest port tree, modify the caddy port and send a patch to the FreeBSD mailing list

As yuripv mentioned, I could probably try and send an e-mail to the maintainer of the caddy port with my changes and wait until FreeBSD 14.1 for them to be landed in the ports tree

tmtt: git clone --depth=1 cgit.freebsd.org/ports ; cd ports/www/caddy ; <do stuff> ; git add … ; git commit -em "www/caddy: add force-reload command to rc script" ; bugs.freebsd.org/bugzilla/enter_bug…6%20Packages&short_desc=www%2Fcaddy

Title: ports - FreeBSD ports tree

The ports tree and FreeBSD releases aren't synchronized. You won't have to wait for a FreeBSD release for your change to make it.

meena: Okay lol thanks, seemed more complicated in my head

speaking of this kind of thing, I should get back to coding

Done! bugs.freebsd.org/bugzilla/show_bug.cgi?id=274085

Title: 274085 – www/caddy: Add forcereload command to rc script

yeah that took me a whole hour

tmtt: first time is the hardest :P

meena: For sure

Thanks again meena :)

I like that. It's nice and simple.

Hi, I've just managed to nix my /lib directory trying to remove a jail using a relative path instead of an absolute path.

Now nothing works, just lots of errors about missing libraries etc.

Before I switch off the box and lose it, anything that I can do to fix this? I'm running UFS btw

fraxamo: most useful tools won't work, but, stuff in /rescue might come in handy

Yes, I was thinking of /rescue

are you on a freebsd version that has /rescue/fetch?

depending on your version, /rescue might have fetch.

what kevans said

jinx

I'm on 13.2 p3

kevans: let's fight over who's the bigger nerd

hah

I'm looking at another 13.2 box and I'm not seeing fetch in /rescue

fraxamo: don't recall when it was added, just trying to execute /rescue/fetch should give you a usa-

hmm

Was it on an earlier version?

probably new in 14.0 then, maybe in future 13.3

fairly certain we didn't MFC that before 13.2's release

Shame

kevans: question: unreleated to my head-banging: Why don't we have tags for every release? tags don't cost anything. git isn't CVS

fraxamo: so, boot a live-boot medium, and do it from there.

meena: at the risk of sounding dumb, what do you mean? ew have release/* tags for quite a few

meena: Thanks, I'll give that a go

kevans: yes. And all of them are .0 — but what about all the -p "releases" ?

emaste: ^ *runs away*

(I suspect the answer is just that nobody has requested it since the switch to git; in the svn world IMO it was more of an annoying step to do)

it's really a question for secteam tho

right, secteam are the ones cutting patch releases

i bet if you fire off a request to do it going forward, they probably wouldn't mind too much. they probably don't really want to go back and hunt down historical patch-levels and tag those, though

i can see the value in ez-diff between patch levels

oh, yeah, that'd be terrible work that you'd need a dedicated nerd to do, and then they'd find shit we don't want to know

*nod*

why does sh start in /usr/home instead of /home

so it's a much better request when it's scoped down purely to "can we incorporate this into the process for maintenance of future releases"

is /usr/home hardcoded into it or is it getting that from somewhere

it starts in your $HOME, /home is usually a symlink to /usr/home (pre-14.x anyway)

bz: /usr/home is traditional for FreeBSD.

$HOME is set to /home though, not /usr/home

and its also /home in /etc/passwd

so im not sure where its getting /usr/home from

from following symlink, no?

/home is a

...

kevans: now for the head-banger question: Whom can help me with the virtio vsock module, by holding my hand, and slowly telling me it's gonna be alright, because I think I'm way too deep and I don't actually know how to swim here.

damn you're just throwing curveballs today =-p

im just curious about the weird hack required in .shrc to make it not display /usr/home on the prompt

bz: fwiw this is something being worked on in main, at least; see, e.g., reviews.freebsd.org/D40086, which fixes the zfs pool layout

Title: Login

kevans: speaking of curve-balls, I'd still like to have a FreeBSD-src package in PkgBase… bugs.freebsd.org/bugzilla/show_bug.cgi?id=268401

Title: 268401 – PkgBase should package the src snapshot it was built from

like by default, the .profile makes sh run the cd command every time you log in

in order to switch from /usr/home to /home

email to secteam is off, and now it's time to put the kid to sleep who is very very giddy…

if you comment out that line it starts in /usr/home, despite that not being the setting for pw or the $HOME environment variable

bz: yeah, my shell history on <14 machines is: cd ; tmux a ; or tmux

you could just destroy /usr/home and make /home a directory (or dataset), that's what i did everytime after install :D

now with 14.x (clean install though), /home will be /home, no action needed

next /usr/bin -> /in

/bin

solaris does symlink /usr/bin to /bin for a long time now :)

wow a 26 year old kludge

thats like netbsd level obstinacy

ok i just removed /usr/home and the symlink and created a regular /home

hopefully that wont break anything

emaste: what's the ETA on that to roll around?

Also, that would fix do many issues in rc (tho not rc itself lol)

meena: to move to /bin?

emaste: yes. what's the time frame for that?

emaste: yeah, because then we don't have to keep debating if a binary's available then move the damn thing

"your diskless setup now has a whole /bin k?"

👍

has anyone see email to @freebsdjournal bounce?

As far as I can tell they don't have an MX so it's odd to me that google is still trying to deliver

tmtt: i would approve that patch ;) but I'm not the maintainer.

crb: All you need is an A record to deliver, in theory. The MX record is not needed.

thumbs: hmm, didn't know that thank you

Now, there is no guarantee that the IP set on the A record will accept emails.

Or if gmail will play ball with the A record

postfix will work with it.

crb seems to be using gmail

yes I'm sending out mail via gmail

What does the bounce tell you?

meena: time frame right now, but ideally we could do it for 15

emaste: not on here, hackmd.io/@jhb/ByWrxQmr2

Title: FreeBSD 15.0 Planning - HackMD

but also, i just saw MINIMAL kernel is on that plan… and i have a patch for that.

Remove consumers of DES des?

heh

heh

yuripv: that was deliberate

a 20 year long plan

i forgot how many things I volunteered for in 15.0

could you please volunteer to port vmwgfx?

delivery incomplete, there was a problem

yuripv: what's that

kms driver for vmware svga adapter

kevans: could you please take a look at reviews.freebsd.org/D41947?

Title: ⚙ D41947 regcomp: use unsigned char when testing for escapes

yuripv: oh, I meant to say something on that

yuripv: netbsd had found some of these, but I hadn't yet checked if any of them were still applicable to our libc: NetBSD/src d5c6fca

Title: - cast GETNEXT to unsigned where it is being promoted to int to prevent · NetBSD/src@d5c6fca · GitHub

obviously at least two of these are

that looks more complete yes

kevans: got a response from delphij@ re tags, and it's basically: Nah.

how come when i run pkg search perl, the terminal never gets prompt back?

meena: really? interesting

i wouldn't have thought they'd care that much

i can paste the response if you want, and I can see how that makes sense.

i'm a bit curious, feel free to paste or forward to me

kevans: updated the review with fix from netbsd

yuripv: nice. Also, I cannot seem to mark my comments as done. which seems not ideal

yuripv: i'm happy, thanks

within a jail, bazel outputs an infinite amount of "Starting local Bazel server and connecting to it...", can it bind to an IP?