ivy: then I should change mine to 16-CURRENT as I like even numbers better

uh, sure, if you like

BIGGER Number = Better

Don't CURRENT if you don't need to. I use CURRENT due to an edge and uncommon stuff and working with also HEAD drivers.

Going to gloat here - Just got a wifi device sending back USB callbacks.

First time ever on FreeBSD :)

farhan: client-side?

sorry? client side?

wifi device callbacks.

I don't follow what you mean by client side?

Fromt he wifi device or the FBSD wifi device?

from the wifi device

Nice!

as in, the wifi device is receiving frames from the air and sending them to the USB stack, which my driver is receiving.

That was *not* easy

Yes. I get it. Very cool!

packets are going into the bit bucket ?

bit bucket?

farhan: /dev/null

Trash

Yeah, if packets are getting dropped. You put down the bit bucket to catch them

rigt now they aren't processed at all

The next step is to pipe them into ieee80211_input

Perfect. Forward is progress.

farhan: 1st time writing a driver?

2nd

but I didn't really know what I was doing the firs ttime.

sounds like a real adventure

its a lot of reading and reading...

I would like a wiki account to write my notes there too

farhan: see wiki.freebsd.org/LinuxKPI to get ideas on stuff needing to be ported :) Different stack, may be less boring than this driver stuff.

I have no problems with that layer, but its an example of some inconsistencies in how to approach FreeBSD's current position vis-a-vis Linux

There's no vis-a-vis.

I don't follow?

farhan: if you want a wiki account, please see wiki.freebsd.org/Wiki/About#How_To_Create_an_Account

farhan: I commented directly on below your "vis-a-vis" comment. What's there to follow.

I wonder how much effort is involved in doing the "drm" stuff that comes from Linux. They are on like kernel 6.6 and I need it for like kernel 6.14 (current stable)

probably no small task, and that is why they tend to lag behind so far

SponiX: github.com/freebsd/drm-kmod ?

SponiX: amdgpu + drm work flawlessly for a few years now on FreeBSD

regis: Yes, my RX 5700 is working very well. But I would really like my 9070 XT GPU working in FreeBSD also

It only got support in Linux kernel 6.13.4 if my memory is correct

Yeah, it is for sure a process I'm not ready to take on

SponiX: I think master on drm-kmod is approaching 6.9

mason: going back to the earlier conversation. i mostly use containers in proxmox too. i want to migrate those to fbsd jails and hopefully they carry over properly. but yeah it would be nice to have a webui similar to proxmox to admin the containers and vms (i have 2 vms)

realistically i only run 1 vm... it's a hackintosh vm to host bluebubbles lol

i'm not sure if bhyve can pull something like that off

it depends on the hardward hackintoshed mac os wants

bhyve might be missing support for doing some lies, but that shouldn't be too hard to add

i'd have to see. not having imessage was a big caveat of moving away from macos in favor of running linux on my ancient 2013 macbook pro

i really wish freebsd had a more advanced version of gnome/wayland

i woudln't mind running it on my "slightly modified" macbook pro if possible but i prefer gnome

i swapped out the broadcom wifi module for an intel one and the ssd with an nvme using adapters

i guess there were some attempts to make bsd derived out the box user workstations but they obviously don't compare to the linux flavors

how do I know if post-reboot `freebsd-update install` (upgraded to 14.2 from 13.2 just now...) is hanging or just slow?

ok nvm it was just slow :3

hello

what is the correct way to make a kind reminder about this? bugs.freebsd.org/bugzilla/show_bug.cgi?id=286587

shall I send an email in x11⊙Fo or freebsd-x11⊙Fo?

bitblt: try asking in #freebsd-ports

I get bug report emails, have for years, but cannot figure out where to turn them off.

I've been through bugzilla a few times...the email says at the bottom I get them because "You are on the CC list for the bug". but these are new bugs I didn't sign up for.

and that I'm an assignee.

any ideas on how to turn that off?

You select "Preferences" at the top and then select "Email Preferences" and uncheck boxes

all "I want to receive mail when" are checked. All "but not when" are checked.

Click "Disable all mail" and then only enable the stuff you want

Done - I still get them :/

I'm on the CC list - not sure how/why

I think I got it...thank you GenAI - My general preferences somehow had me on the CC list.

one project at a time, but if usb modeswitching is a common thing, it perhaps should be in usbconfig.

one project at a time? You mean like one project and one idea a day? Right

[tj]: haha :)

Think of small projects as snacks between meals

[tj]: I have a personal list - there's another driver I want to bring to FreeBSD after my current one.

Things arrive best small and often

Drivers kind of break that as you need a lot of driver to do anything

All my src contributions have been adding chip IDs to existing ethernet or sound card drivers.

neat

CrtxReavr: thank you for that!

Look at the OS, or any OS, like an ecosystem. You start with tinkerers, then edge-case users, developers, and then production users -- and obviously some people jump order in the heirarchy

If I'm a brand new FreeBSD user and on day 1 my wifi card doesn't work and I'm told "buy hardware that works" I very well might just go ack to Linux.

but if some other arbitrary feature doesn't work, like graphics acceleration, I might not necessarily care on day 1.

These were instances of "WTF! This is Intel Gbit! Why doesn't it work?"

Someone suggested it was just some minor tweak, and just adding the chip ID to the driver may "fix" it, and they were right.

Worked multiple times.

what do you think, how linux got those cards working?;)

I have hunch that it was not "I very well might just go back to Windows" ;)

tsoome_: There were a few factors, but basically they focused on getting any and all devices working on Linux and spent time writing guides on how to do it.

tsoome_, from my experience, Linux device driver quality is very poor.

it is.

Every flip the bit on verbose SCSI logging on Linux?

I used to hang on an IRC channel with Alan Cox.

I once told him I thought the Linux IP stack sucked. . . and he agreed with me.

I've seen stuff that makes no sense. For example, their implementation ot realtek 8188 PCI has (had?) a function A that called B, that called A with another condition that then did whatever it needed.

(Alan Cox wrote the Linux IP stack, if you were un-aware.)

They have conflicting naming conventions. For example, BSD will do rtwn_usb_attach and rtwn_pci_attach.

Linux will have the same function name.

Linux is often stealing code from BSD. . .

The BSD license sadly allows that.

and vice versa...that's fine.

but the users reward that

that's an Open Source vs Free Software discussion

instead of taking BSD as a base and actually improving upon it, linux just adds support for everything imaginable, all the tiniest dumbest variations in hardware

ring0_starr: that's a good thing.

so the result is that linux is this giant bloated pile of shit and bsd has old code for sure but at least it's nice

no, it encourages hardware vendors to do crap like that

BSD probably does not drive hardware vendors in general.

there used to be standardization efforts

tech is already caving in on itself with complexity

ring0_starr: feel free to disagree with me, but I do think *every* device should be supported, at least entire classes of devices.

who designs those devices like that

we have things like SMBus, PCI, EISA, and so on from industry forums, but then you have these chinese SoC vendors who each do their own thing and it's all a mess

so complicated enough as-is. i can't even get bsd to run consistently on this single board computer i have. when it panics, i can't even get a dump because it freezes

ugh.

sorry to hear that.

and that's the main problem, is that the world is running at a million miles per hour without getting any of the fundamentals right

ring0_starr: I appreciate that perspective, not just saying that, but I think BSD is over-indexed on that.

there is a level of "chasing shiny things" that should be done.

the hardware is random, the variations are random, the documentation is horrible

i was adding support for an SoC in some DMA code and for some reason linux shifted the physical address over by 2 bits

couldn't figure out why that is by reading the manual at all

the guy who wrote the linux driver for it explained and pointed to this one specific line where it said something about "word-sized addresses"

see if i didn't get told that by a person, i would've probably not figured it out

ring0_starr: we should document our notes to help BSD here.

empower others with documentation

Please.

some is "easy", like how Free/Open&Net differ on naming conventions

I fail to see the value of Free, Open, & Net remaining separate projects.

'Course. . . . the Net people would insist on keeping VAX & Sparc IPX alive, and hold things back. . .

I would say Net and Free need to merge, while OpenBSD can remain separate.

no, because then you're going to just kill off vax support

I'm genuinely not certain if that's sarcasm or not.

netbsd's thing is compatibility

(with jurassic junk)

with all due respect, I personally do not see value in that.

and that's the reason why forks exist.

you can focus on what you see value in

Look at the CPU cycles per kWatt.

openbsd is different...They are perfectly fine with trying radical new things and that's good/healthy.

Oh, and then there's DragonFlyBSD.

Which was basically bourne of a pissing match amonst FreeBSD kernel devs.

+g

hammerfs wouldn't exist

BSD is supposed to be a research operating system

That's what -CURRENT is for.

> BSD is supposed to be a research operating system

ring0_starr: This channel is FreeBSD-focused and you're talking about en.wikipedia.org/wiki/Berkeley_Software_Distribution

anyone has the right to fork this or any other OS. I might not like it, but its their choice.

regis: research, not production?

Did DragonFly ever get more than one FreeBSD dev? seems like a vanity project to me

paulf: a filesystem is vanity?

It kinda was. . . forked FreeBSD 4.x and built a sand castle with it.

Granted, 5.x had major issues.

farhan: I'm pointing out that the other commenter mentioned principles driving the "BSD" which is dead for 29 years now.

ahh, excuse me.

regis, so then FreeBSD is a fork of 4.4BSD that focuses on being production-ready

you get to focus on the aspects of OS-dom that are important to you :)

ring0_starr: I wanted to point out how "BSD is supposed to be a research operating system" outdated and irrelevant is where it comes to the OS this channel is focused on.

+1

okay..?

I think "BSD is supposed to be a research operating system"

throws back to the Bell Labs/Sun fiasco and related court cases.

Or what lead up to them.

Would you say that it's "difficult" to create / maintain software packages for FreeBSD?

They have a strong mentorship program and set of documentation.

so what's new and/or borked in current

no gdb 16 :-(

lldb?

not for me at least not for the foreseeable future

oh dear seems to be plenty of brokedness around ptrace and signals :-((((

I'd love to see a kgdb equivlanet in llvm or lldb or whatever..I'm not familiar with all this toolchain stuff