@rtprio the ARM single core 6Gb?

part of the free plan?

no, in my datacenter there were no ARMs available

oh! I just created the ARM one, as part of the free plan

I'll start testing tomorrow

but i have two of the tiny amd64 ones

which are free

has anyone here written any embedded networking software?

How do I get Elan touchpad drivers working

sh change in 13.1 in how space in arithmetic expression brackets is parsed: set -x ; x=$(( 2 - 3 )) ; y =$( ( 5 - 4 ) ) ; set +x -- last one causes error "command not found: 5"

s/y =/y=/

If anyone is on 12.x or 13.0, would you please note if you see "y=1" in output?

are you sure it was a change in 13.1?

it would seem its opening a subshell within a captured stdout shell

which seems normal

its also why $(( )) is old hat now that $[ ] is around (in bash, etc. at least)

nacelle, Well not quite. On existing 13.0 the arithmetic expression does use no space on between; was misremembering. Thanks for the nudge

nacelle, Does FreeBSD /bin/sh support "$[]"?

Do not see "\$\[" in "sh(1)" manual page

parv, In 12.4-RELEASE I see "sh: 5: not found" due to the "$( (" space present there and ") )". Same for bash.

rwp, Thanks. Must have been my own error

nacelle, $[...] is a deprecated construct in bash. It wasn't around long before being replaced by $((...))

I don't recall sh ever having supported it and recall it being a bash specific feature from before $((...))

can i set SRC somewhere else ( not /usr/src ) in make.conf ?

Yes you can

i wasn't able to find how, in man pages or looked in wrong ones

Set "SRC_BASE" to wherever; see "src.conf(5)" manual page

Nope, not there

( WTH! Where did I pick "SRC_BASE" ... )

angry_vincent, From Makefile.inc1 seems like the variable is SRCDIR; but a better reference (even a comment about in a Makefile) would be more appreciated

parv, "sys/conf/kern.post.mk:# SRC_BASE is how the ports tree refers to the location of the base source files"

V_PauAmma_V, Oohh! Thank you! I will note that.

parv: i guess SRCDIR not the one as well, use by build system to locate it's files.

Hmmm.. (create the symbolic link?)

it could work, but i wonder whether such possibility to have defined location of /usr/src exists at all

how do I upgrade from 12.3 to 12.4 ?

Binary upgrade?

yeah the freebsd-upgrade command

If a binary upgrade, see -r in the freebsd-update manual page.

thanks

angry_vincent: /usr/src/ is "just" the location where FreeBSD expects it to be, but nothing's stopping you from putting it in /path/to/directory iirc, so long as you don't have things like ports that need the SRC_BASE environment variable

debdrup: i see. i thought, there could be some variable that user can set to control the location and that's it

angry_vincent: I mean, there is - where you build it is where it's being built

Does FreeBSD also use there mnt_vnodecovered field as root node for this mountpoint? Or do they just not see it as "root node"

debdrup: i see. it can be anywhere you want, one just need to be aware for ports or 3-rd party modules that require src path

lockna: Each mounted filesystem has a "root vnode". "mnt_vnodecovered" points to the vnode (representing the directory) where the filesystem is mounted on. This vnode belongs to the mounted-on filesystem, and is thus different than the "root vnode" of the mounted filesystem. The root filesystem has NULL in "mnt_vnodecovered".

Ahh, alright, I understand. Thank youuu

some cirrus CI freebsd builders seem to set localtime to PST timezone or something, which breaks some amazon S3 interfacing code. What's the most robust way to set timezone to GMT?

maerwald: cp /usr/share/zoneinfo/Etc/UTC /etc/localtime && adjkern -a ?

tzsetup Etc/UTC # should also do it

add -s just to be on the safe side, even if you're setting it to UTC

is it possible to use qemu to run amd64 binaries on aarch64?

wcarson: given that we're raising Aarch64 to be a Tier 1 platform, then, yes, you should

if not, it's a bug

oh really? that would be very useful for me. i have been unlucky so far finding any guidance on how to do it. everything seems to be the other way around

wcarson: it should be the same thing, just using a different qemu binary

hmm ok

hmm yes i see emulators/qemu-user-static provides bin/qemu-x86_64-static

github.com/freebsd/freebsd-ports/bl…qemu-user-static-devel/Makefile#L15 hrmmm..

Title: freebsd-ports/Makefile at main · freebsd/freebsd-ports · GitHub

welp

so much for that :D

and no bugs, github.com/qemu-bsd-user/qemu-bsd-u…issues?q=aarch64++OR+arm64+OR+arm8+

Title: Issues · qemu-bsd-user/qemu-bsd-user · GitHub

wcarson: maybe open a bug, and lead with the Tier 1 Platform thing

is there a way to find out when a new version of a port will be built and put on pkg.freebsd.org?

edited the Makefile to include aarch64, we'll see what happens heh

(wondering about audio/ardour 7.2)

has anyone ever ran into packets being seen by tcpdump, but not actually going out an interface?

rwp: heh, I had to read the release notes to see that. they never actually deprecated $[] beyond verbal jousting. someone vile that I used to work with convinced me that 4.0 had $[] as a change for $(())... silly me for believing that jackass over documentation.

wcarson: yes, when doing some things with tagged vlans

hmm interesting. as far as i know i'm not doing anything with vlans

wcarson: so tcpdump on the other side of the switch shows nothing?

is this your qemu thing that's missing packets

1) correct, 2) sort of, it's a linux binary i'm running just using freebsd's linux emulation

wcarson: if it builds and produces something useful, you'll have to open a ports bug, and if it doesn't, you'll have to open an issue with the project

ok, thanks meena

can freebsd's linuxulator handle /dev/video* / v4l / v4lv2 ?

seems not

That used to work. The API has deliberately been kept identical.

meena: no dice, compilation failed. trying to see if qemu upstream even supports t

wcarson: aye

pertho, how are you testing? A Linux web browser?

cracauer: installed Ubuntu 20.04 for the linuxulator and instaled Signal Messenger.. text chat works but audio/video calls do not

just wondered if I was missing a package on the Linux side or whether it was something else

You can probably make audio work by running pulseaudio. Video I dunno, maybe I have some time later to check.

what happens if pulseaudio running both on linux side and freebsd side?

You should only need one. The connections to it should work from both envs.

Which instructions did you follow for your ubuntu?

I used this guide: micski.dk/2021/12/21/install-ubuntu…freebsds-linux-binary-compatibility

Title: Install Ubuntu base system into FreeBSD's Linux Binary Compatibility

and: micski.dk/2021/12/22/install-and-ru…gnal-desktop-application-on-freebsd

Title: How to install and run Signal desktop application on FreeBSD

linux: jid 0 pid 44276 (ThreadPoolSingl): linux_ioctl_fallback fd=21, cmd=0xc0205648 ('V',72) is not implemented and a lot of: linux: jid 0 pid 47746 (Chrome_ChildIOT): possibly incorrect MADV_DONTNEED and linux: jid 0 pid 48366 (ThreadPoolForeg): unsupported socket(AF_NETLINK, 3, NETLINK_ROUTE)

pulseaudio is installed on the linux side

Does anything make sound in pulseaudio?

FreeBSD binaries using pa should work, too.

on FreeBSD pulseaudio works and sound works.. on linux.. if I try to run pavucontrol on Linux it segfaults..

This is a FreeBSD or Linux pa demon?

freebsd runs pulseaudio server

the linux daemon for pulseaudio shows lots of errors when I run it

is Ubuntu supposed to be running any daemons or services? Or does running the binary just directly translate to the FreeBSD side?

For pulseaudio I strongly feel that it should run on the FreeBSD side.

Same for jackd.

yeah video works audio doesn't.. in signal.. microphone and speakers both say "No devices available"

the unsupported socket(AF_NETLINK, 3, NETLINK_ROUTE) must be some network connection it's trying to make but is failing

I think it is trying to reconfigure something about the network.

making calls results in the destination seeing a message about a missed call but nothing rings.. so something with the network is broken with it

s/destination/receiver/

bugs.freebsd.org/bugzilla//show_bug.cgi?id=258975 <- looks like this bug

Title: 258975 – unsupported socket(AF_NETLINK, 3, NETLINK_ROUTE)

Maybe it'll work on 14

which should be due out soon?

unless the netlink work for 14 gets backported? Not sure what gets done in this case.

pertho: you don't have to wait for 14 to be out to run it

not keen on the idea of upgrading my wife's laptop to bleeding edge :D

pertho: i ran into the same thing actually, and upgraded to 14-current which fixed at least part of that bug

wcarson: did it fix the network side but the audio side is still broken?

i wasn't testing any audio, just network stuff

ah i figured out the broadcast thing - or at least figured out why it isn't working. it isn't settings the broadcast mac address

Title: 252596 – ping to 255.255.255.255 does not set broadcast MAC

yesssss

it works

you'd think that's straightforward enough to just commit

these are computers, nothing is straightforward

rtprio: if we threw them out a high up window, would they actually fall and hit the ground

pertho: in my experience, current does not bleed that often.

just that one time where they removed my raid adapter from the tree without warning, other than that, no problems

wow, this is so great. i'm so happy

if I install docker on freeBSD, can I run regular linux containers in that?

seems like it on the wiki

yourfate: docker or podman?

my very limited testing with podman was mixed, at best

yourfate: from what i remember, docker support isn't great. podman is currently getting a lot of attention tho

ok, thy

rtprio: running FreeBSD or Linux?

not having much experience with docker, it sounds like I should not start to learn it on BSD :p

running a couple of random linux ones

podman is a compatible replacement for docker from what I know, right?

right

i don't know if i have the patience to build freebsd images, at least until i can get it into a CI system

yes, but unlike docker, it's receiving attention from FreeBSD developers

does anyone know the status of WINE in FreeBSD these days? A while back I was told the dotnet## is no longer needed and we should just use wine-mono and the 64bit and 32bit stuff was being merged into one. I think it was one of the WINE developers who hangs out on here.

Any idea why I would get a billion lines from sockstat -l for bsddisks ? "seqpac (not connected)" it seems to be counting up fds

bsdisks even.

Appears to be something from a jail

pertho: the 64bit and 32bit stuff isn't merged, to workaround the problem that they conflict, instead on a 64bit system you install wine from the i386 repo into a user directory and the wine from the normal repo can then use that, but the version has to match and if you have your own package repo you have to remember to build i368 too (at least one llvm in the dependency chain), no idea about mono

and dotnet except that mono isn't fully compatible, but the linux-dotnet port is old or even ancient, so could be that mono can replace that

hmm..I thought microsoft open-sourced .net?

.net core maybe

yep, dotnet core is opensource but doesn't support freebsd currently

pertho: I have my webcam running is Linux-chrome (where netflix also works). No sound for now.

s/is/in/

cracauer: sound is overrated

cracauer: yeah video works in Linux-Signal.. just no sound and netlink errors when it tries to make/receive calls

anyone have some experience with npm? why would a port package the npm cache rather than just having npm fetch it at build time?

or is that just how ports do it?

it would have to fetch at fetch time and npm is annoying so in most (all?) cases ports fallback on the workaround of putting that stuff in some archive and fetch that

i'm trying to fix multimedia/navidrome and i see why it's failing but not how to fix it

are there any obscure caveats with configuring zvol-backed iSCSI targets for Windows clients? I prepared one for a Windows user a while ago and they say they're seeing occasional file corruption, but the zvol and ctld setup are pretty basic and I don't see what the problem could be from my end