Sup guys, i really wanna use FreeBSD but my Realtek Wireless Card is not supported, RTL8821CE...

you could get a supported wifi card

im poor :(

how much can they be? 40$ 40€ ?

you could run virtualization and ssh into your freebsd vm

In Brazil for my laptop, like R$250-300

usb card? use wired ethernet until you get a card ?

This ASUS laptop dont have ethernet support

XDDDD

you don't _need_ the internet, once you have the system set up

Internet is a very useful thing to have.

btw: bugs.freebsd.org/bugzilla/show_bug.cgi?id=248235

Title: 248235 – rtw88: RTL8822CE 802.11ac PCIe support (WIP)

don't we have some software that automatically passes thru the NIC to a Linux bhyve which then routes the traffic to the host?

that's like saying 'you won't feel this hole in your foot.... if you cut your leg off.'

yes

Title: WiFi/Rtw88 - FreeBSD Wiki

could be worse

Q: This fails and prints out errors.

A: Do you have more than 4GB of RAM? See currently known issues

oh it's in current

then just run current

here's our Swiss army pocket chainsaw (some assembly required)

meena: sounds like a quintessential realtek driver,

That's every Realtek driver.

I don't know much about real hardware / drivers.

beware of programmers who carry screwdrivers

I own a soldering iron.

I dream of owning one (only to collect dust if actually did)

I have done debugging with an oscilloscope.

meena: it's a mess.

especially with modern hardware.

same, but i had to declare defeat. we couldn't find out why the circuit was doing the wrong thing

designing hardware that works is apparently Too Hard, so all the hardware bugs have to be worked around in the driver

RhodiumToad: what, you're not fond of a computer having ten different CPUs in it?

Oh, yeah working around hardware bugs in software isn't "modern" though.

I think that's just another way of saying QA at design stage is too expensive, let's fix bugs in software

2004 was when re(4) came out, and Realtak is the company famous for doing exactly that.

At least back then the WNICs were what we now term SoftMAC.

Now it's not just the WNIC, it's most devices you connect and even some that are on-board.

And that doesn't even cover things like SMM.

That got introduced with the 80386SL in order to try and make suspend/resume work.

I need to look up one word per debdrup sentence to keep up

(i use word loosely here)

Is this good, neutral, or bad? :P

I'll try to make things better if I can.

WNIC SoftMAC SMM 80386SL

wow. SMM is from the 1990??

1990, actually

Yep, it's very old.

The 80386 was a laptop CPU and suspend/resume was even thought of as a way of getting around the waiting time of starting up a laptop every time you had to use it.

Err, 80386SL.

The funny thing is, that problem wasn't even solved by the time Windows Vista came out, because it could take minutes to boot that system (and this was the point at which various system integrators started introducing quick-boot environments to add a cd/dvd player, web browser, et cetera).

Vista also had a bunch of other peoples that exacerbated this, among other things it changed the driver API interface for a bunch of things, which meant drivers had to be rewritten which introduced a lot of new bugs (and thereby, a lot of new sources of BSODs).

Mobile computing sucked for a very long time.

last work laptop I had ran MS-DOS.

oh ffs

the reason my port builds keep gronking out is that the VM came up without any swap space

RhodiumToad: any idea why?

can't tell yet until it reaches a convenient point for me to stop the VM and see what commands i might have mistyped

meh. I see why

wrong device names

swapon: /dev/label/vmswap0: No such file or directory

hey all! I am considering buying a used laptop and found good deals on a Thinkpad P52 from 2018 and a Thinkpad L15 Gen 2 from 2023. Leaning towards the P52 since it has a Core i7-8850H and Quadro P3200 and I believe there's more life left in it than the L15 with i5-1135G7 and no discrete graphics.

My first laptop ran NetBSD (of course), a 486 Thinkpad

I have a X230, my first non-apple laptop and I like it but it's starting to show it's age.

any1 watching devsummit freebsd day1?

link?

Title: DevSummit/202305 - FreeBSD Wiki

"more information to come" about live streaming

heads up... while building stable/13, [nit-pick]: there is a majority of "Building ..." messages and quite a few building messages displayed. a recursive sed(1) should probably be had at some point just to make them comfortative. s/building/Building/g

most of what i seen if within the first 10-15 minutes of buildworld on x86-64

err.. scary messages. I see in my log, "ZFS[6160]: catastrophic pool I/O failure, zpool=$m" ... and ALL my datasets seem to be going to hell at once.

ghoti: Are you getting messages from CAM?

Why would I be able to see mount with `mount -p` on datasets that don't even show up in `zfs list`?!

debdrup: lemme grab some errors. A few of them don't make any sense to me..

What sort of pool layout do you have?

There are three on this box, all a bit different..

NAME SIZE ALLOC FREE CKPOINT EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT

a 4.53T 3.89T 655G - - 50% 85% 1.00x ONLINE -

'a' is a 2-device stripe, 'm' is a mirror, 'z' is a single vdev on ada0

m 12.7T 2.57T 10.1T - - 0% 20% 1.00x ONLINE -

z 117G 43.3G 73.7G - - 17% 36% 1.00x ONLINE -

ghoti, Please use bsd.to to paste the output

Title: dpaste

Sure, it was just 4 lines so I thought it would be short enough..

Unless it's a single line, or can be condensed into one, don't ever paste into #freebsd.

debdrup: I've been backing thing up from a and z to m, but I'm worried now that `zpool status` reports "errors: 114 data errors, use '-v' for a list".

okie dokie.

I need the output of zpool status; zpool list is pointless.

The only thing zpool list is useful for is the free space fragmentation; I think all other values can be had via per-dataset properties.

The machine seems to have locked up again, one sec while I check the console...

I mean, that's not surprising..

Catastrophic pool I/O failure is nominally an "oh shit, one or more storage devices have decided they no longer want to store data, but instead be used for hardware bowling.

I just realized that the one hardware change before all of my disk errors was the insertion of a pci-e USB card (so that I could add a device to a bhyve instance). The card takes SATA power. Is it possible that multiple errors could be caused by a power drain?

(Sure, anything's possible. I guess "plausible" is what I mean.)

ghoti: not unless you were already too close to the limits of the PSU to be hitting them during high system loads.

i cannot run virtualbox on freebsd 13.2 i get Access Denied when try to run it.

Listening to karels@ talk at the devsummit is great. :3

His name is all over the SCCS repo from CSRG.

worked nice on freebsd 12.xx

Mel1965: as root or...

Mel1965: Are you a member of vboxusers?

that's the one

dch: you use rsyslog, right?

dch: how do you deploy it? do you cut images with it preinstalled and syslog disabled?

meena: nope, its a cursed tool. got suckered into it on debian, switched on good advice to syslog-ng

but I deploy it in 2 ways: poudriere-image with overlay packages installed, or on new systems, via ansible once it boots

and i get it to drop a unix domain socket into jails, 1 less process in the jails to run

dch: i dropped my review reviews.freebsd.org/D27411 because of… well, a good argument

Title: ⚙ D27411 add altlog_jaillist to syslogd's rc script

yes I saw that. i think in general thats a reasonable tradeoff. but in my case I am happier to have less processes in a jail, and solid logging that can't be tampered with

Is an ESP partition generally supposed to be FAT16 or FAT32? (or even exFAT?)

_xor: IIRC, and I try not to remember too much about UEFI, it's supposed to be FAT16 (for now?)

either way, I've got some code that's kinda ready to test: canonical/cloud-init #4119

Title: cc_rsyslog: Refactor for better multi-platform support by igalic · Pull Request #4119 · canonical/cloud-init · GitHub

Nice coincidence, because I'm actually refreshing up on cloud-init to possibly use.

I've used it before on Vultr cloud VMs, but never on bare-metal systems.

That's why I asked about UEFI, because I'm reading the poudriere source for producing system images. Want to be able to iPXE boot into an image, raw-write it to the disk, and then reboot and let cloud-init handle the rest.

the basic idea is to get it working on par with Linux

and sometimes that's a bit tricky…

_xor: ask dch how to poudriere image

I haven't quite figured it out yet, but then again, I was trying to get it to do to add my pkgbase repository

I used to use it to build jails and I have used it before to build system images (though not in a long time).

Reading the source because I'm going to script it for myself without having to use poudriere-image.

poudriere-image, as far as I can tell, doesn't support PkgBase (in that it will extract a base distribution first and then install packages), when it's not necessary.

_xor: also, if you hit any bumps in cloud-init let me know

they've now moved issues to github, which might make outside contribution easier

I just need to install PkgBase packages, any regular app packages, add an optional overlay, and package it all up into a raw-writable disk image with GPT + efi + freebsd-swap + freebsd-zfs.

(basically, I'm trying to do the refactor which will yield benefits, but in between, i try to prioritize real issues people are having rn)

I remember last time I built a system image with poudriere-image, it required a fixed size (which I set to 4GB). That part was fine, but I wanted to expand the freebsd-zfs partition to take up the remaining space on which host it's written onto. zpool autoexpand=on is obvious, but I'm reading up on cloud-init to figure out growpart and resizefs.

Actually, I guess that's a good question. Does the FreeBSD port of cloud-init support expanding the partition at boot to fill up the remaining space? I know it runs in the init stage on Linux.

_xor: cloud-init supports zpool growing

but! yesterday I implemented cc_growpart just calling service growfs onestart ;)

tricky bit with that: it adds swap if none exists

you can disable that if you don't like it

Hmm. The image comes with swap space, though it would be nice to be able to modify that on a per-host basis.

I've not gotten around to touching up disk_setup for BSD

cc_disk_setup

Maybe I'll just build a freebsd-zfs partition image instead of a whole-disk image and then create the initial GPT + efi + freebsd-swap partitions from the iPXE-booted installer, then create a freebsd-zfs partition that takes up the rest of the space and let zpool:autoexpand=on handle the rest at first boot.

I'd be curious to hear what you think has the best tradeoffs

Taking brainstorming notes on it right now actually. Here, I'll make a gist for it...

Title: FreeBSD Automated POST Provisioning & Boot Configuration · GitHub

Hmm, apparently Github Gists don't recognize org-mode. Meh, whateves for now. That's where I currently am with my notes. Going to experiment with mfsBSD in bhyve here shortly to see what happens.

_xor anything to do with autoprovisioning, ipxe etc I'm in

freebsd now has growfs in base it does what you want for zpool expansion

the hard bit IMO is getting from ipxe-world into bsd world, and figuring out all the network and filesystem bits

use gpt labels everywhere (then device naming and order isn't important)

and ifconfig_DEFAULT=DHCP gets you a long way for network stuff, but you need to kldload the right bits in loader.conf already

anyway _xor not this week but any other time feel free to mine my thoughts to deal with this

debdrup: weird, I haven't had a single error since I disconnected the power on the USB card. :(

I wonder if I cann find a lower-power USB2 PCI-E card. Any thoughts as to how else I might access a USB Sonoff Zigbee Dongle from within a BHYVE VM?

Hi, currently seeing segfaults when I try to start anope -- I'm on 13.1-RELEASE. Neither Anope nor inspircd's logs point to any problem. Anope successfully links and immediately dies, according to dmesg

bunch of "pid xxxxx (services), jid 0, uid 65534: exited on signal 11"

Using the inspircd package from the freebsd repos as well

I am connecting over localhost, though that's probably a given

s/connecting/linking, for clarity's sake, actually

also I've an unrelated question while here: what cipher/pbkdf are used by geli?