what's the issue with just issuing pw multiple times?

handling lists of arguments always adds complexity

I assume that using pw is something that happens either extremely frequently or with a list of groups that exceeds several elements.

yeah, there's some extra validation if you start allowing lists

e.g. it probably doesn't make sense to change multiple group names at a time

it makes sense for member lists

rockyh, I think it was missed in all of the chatter that groupmod modifies the group but usermod modifies the user. "pw usermod -n rwp -G operator,wheel,staff,guest" just added me to that list of groups no problem.

Note that the user is removed from any groups not specified. It must be a complete list.

I just connected my FreeBSD router to my iPhone for USB tethering and it just worked

I cannot believe I have a line that says "nat on ue0"

in my pf.conf

lw, ping

jbo: hi

lw, I read some stuff by now - did mean to check with you whether non-lower case PORTNAME is required

"by now" not meant to sound negative. Let's say "in the meantime" :)

jbo: i don't think it's required, but other ports do it like that (net/py-GeoIP2 for example) - is that discouraged?

lw, wiki.freebsd.org/Python/PortsPolicy

Title: Python/PortsPolicy - FreeBSD Wiki

> PORTNAME MUST exactly match the registered PyPI package name, including casing.

so I guess that :D

that should be now it is now

s/now/how

yes indeed - did not mean to complain :)

although i think this means audio/sublime-music needs to be renamed to audio/py-sublime-music

I did not check that port yet

but yes, as a committer it's sort of my motivation to point submitters towards the resources necessary to increase the likelyhood of a port submission to be as accurate as possible :)

although net-im toot doesn't follow that convention... which imo makes more sense for end-user applications, but the guidelines say explicitly that you can't do that

in case something is only an end-user application you can drop that py- prefix

nimaje: PortsPolicy disagree, it says you can only do that if it's "NOT present in PyPI AND not likely to ever be". maybe it's out of datE?

i'm not opposed to including the py-, it just seems like existing package contradict the policy there

i guess i will email -python⊙. tomorrow :-)

there are plenty of things in the ports three that does not adhere to guidelines. that is partly because of oversight and partly because guidelines and ports infrastructure evolves over time.

How do I do the equivalent of parted's "parted /dev/sdX set 1 boot on" on freebsd? I have an image on the disk already I just need to make it bootable

How do i change the console (vt) resolution? "vidcontrol -i mode" lists no modes!

Axman6: Maybe check "man gpart" to see if there's some flag?

there's the bootcode command, but I think I just want to set the device to bootable, not change the bootcode... but I'm not sure

Axman6: I think you want `gpart set`

(bootme / bootonce attributes)

or active, I guess, depending on the scheme

I'll give that a go

Is there seriously no libpng package in the base system? I have to install that from ports?

Ok, so why does this page not mention portsnap even once? docs.freebsd.org/en/books/handbook/ports

Title: Chapter 4. Installing Applications: Packages and Ports | FreeBSD Documentation Portal

ok... I gave in and just used an ubuntu vm. wish me luck!

SomeRandoUser, portsnap is being deprecated.

What's the replacement? Just doing a git clone?

yes

Does the value of 2000 in the property "dev.cpu.0.freq_levels: 2300/2000 ..." means miliwatt per cpu core ?

occ: man.freebsd.org/cgi/man.cgi?cpufreq

Title: cpufreq

"Values are in units of MHz and milliwatts."

so the total power usage of a 4 cores cpu is 2*4=8 Watt, is it right?

That's how I'd interpret it.

assume all cores's freq_level is 2300/2000

that doesn't mean right, since it means my Ryzen 5800X3D would consume 30W at full base clock (3400/3740), actual TDP is >100W. i wonder if this is supposed to be mW but is really a made up number

s/mean/seem

My mouse (PS/2 touchpad on laptop) isn't working in the console or X. How do I troubleshoot it?

how do you make unattended bsdinstall set up a password for zfs "Encrypt Disks"?

SomeRandoUser: is moused running?

yes

elirco: Thanks for the advice earlier regarding "hw.snd.default_unit" . It works! But is there no way to change VLC's output on-the-fly? Is the list of output devices in VLC  just for show? :)

dmesg shows that the mouse is detected. "psm0: model Elantech Touchpad, device ID 3"

and I have a /dev/sysmouse file

But the touchpad just does nothing

an external USB mouse does work, though.

SomeRandoUser: you might need a moused for the touchpad as well

I do have one running

ps -A shows /usr/sbin/moused -p /dev/psm0 -t auto

hrm

alepzi, Welcome to the main channel! It's good to see you here.

alepzi, Your best option might be to do the install by hand. wiki.freebsd.org/MasonLoringBliss/ZFSandGELIbyHAND

Title: MasonLoringBliss/ZFSandGELIbyHAND - FreeBSD Wiki

Which I am going to always refer to as RTFM Read The Fine Mason docs. :-)

parv: regarding the Intel Arc GPU testing with drm-61-kmod, I can't practically do any while using CURRENT atm. By comparison with stable/14, package builds with poudriere have been randomly breaking and getting in my way. I would rather wait for drm-61-kmod to be MFC'd to stable/14 (if that ever happens) :(

boot environments really saved my ass though

kenrap, Alright; no worries.

Just wanted to update you :)

kenrap, As I mentioned earlier I do not have the GPU; may or may not have in future. Appreciate the update in any case

rwp no that's way too complex. i almost have everything i need in installerconfig. i just need to add encrypted zfs disk pw protection

Anyone able to check if ccache -sv is broken on 14 with reporting local storage stats?

I get size and file count but nothing for hits/etc...

hello, can I ask if the freebsd packages are signed so that if an attacker modify it, the os refuses its installation ?

mystic: the default FreeBSD repo's packages are signed. If you take a look in /etc/pkg/FreeBSD.conf, you will see how.

hi all

hey all, I need to kill a bhyve VM properly. it's booted with wired memory

and the guest is stuck

do I kill the bhyve process

or do bhyvectl destroy?

any tips are appreciated

antranigv: Tried this? vm poweroff -f name_of_your_vm

hi adilix

tercaL trying now

How do I track progress on bugs I reported? I reported 2 bugs 3 weeks ago, there are patches posted relatively shortly for both bugs on bugs.freebsd.org but nothing else happened since.

(275870 and 275871 are in question)

dautor: if there's nothing on the bug then there's probably been no update

lw: so, it is reported back to the tracker once progress is made / bug is fixed?

dautor: if the patch is committer, you'll get a comment on the PR, yes

committed

dautor: if it's your bug, it might be helpful to test the suggested patch and report back

if only because this confirms jah@ reproduced the same bug you found

lw: ok; I will try testing the patch that once I set up a build machine.

thanks for info

dautor: also, this isn't the case for these bugs, but sometimes a patch will be reviewed on phab (reviews.freebsd.org) in which case the review URL should be in the URL field and you can check there for updates

Title: FreeBSD Dashboard

this is what it looks like when the patch is committed: bugs.freebsd.org/bugzilla/show_bug.cgi?id=276119 (in this case, it was committed to main, there should be another update later when it lands in stable)

Title: 276119 – mail: typing ^C twice during compose causes SIGBUS

I see. Now things are much clearer. :) thank you, once again!

how do you make unattended bsdinstall set up a password for zfs "Encrypt Disks"? or where's it documented how to do it plz?

I'm getting errno = 37

that's a permission denied right?

when I do the poweroff / bhyvectl --force-poweroff command

like if pf is blocking network

maybe i'm wrong sorry

errno(2) says it's EALREADY

rwp: thank you. For me `pw groupmod' or `pw usermod' are both suitable. But what if you were already belonging to group `wheel' and you just wanted to join also `operator', `staff', `guest'? Would `pw usermod -n rwp -G operator,staff,guest' do this?

jbo: it's arbitrary, but I find quite boring to issue `pw groupmod' multiple times: it would be more efficient if a single issue could perform all the tasks

rockyh: that will remove you from the wheel group. -G overwrites the existing group list

lw: ok, so even this command doesn't do exactly what I was looking for

not that i know of (other than 'for' or whatever)

how do you make unattended bsdinstall set up a password for zfs "Encrypt Disks"? or where's it documented how to do it plz?

Hi guys, a good new year 2024 for all of you.

to you too Aedil

I just installed freebsd on an old computer with a Radeon 9550xl AGP video card. I installed the drm-kmod package and put kld_list="radeonkms" in /etc/rc.conf. It does boot and give me a high resolution console, but the boot process stalls for at least a minute at "Loading kernel modules:" when I have radeonkms enabled. Any way to find out what's causing it to take so long?

SomeRandoUser: boot -v might produce more interesting output, although drm is very chatty anyway so i'm not sure if -v adds much

I added kern.msgbuf.show_timestamps=1 to /etc/sysctl.conf, but that doesn't help since it only takes effect after the filesystem and init are started.

lw is that the same as selecting the Verbose option from the boot menu?

i've never really used the boot menu but probably, yes

The verbose option gave me more output, but there's still a very long pause at "Loading kernel modules:" with no output.

Is there some way to set a sysctl early in the boot process before the root fs is mounted?

ok it turns out I can put kern.msgbuf_show_timestamps=1 into /boot/loader.conf

I booted without radeonkms, then ran "kldload radeonkms" manually after logging in, which resulted in a long pause before it finally changed my resolution.

Something about loading that module is very slow.

oh great... startx results in a blank black screen. I can't even switch out of it with ctrl+a

lt+F1

best part is that the Xorg process is in the uninterruptible D state, so it's impossible to kill it!

DOes the freebsd autmated install process have a name? Like RHEL is kickstart, Solaris is jumpstart. Is there anything more specific than just "automated freebsd install"?

unattended, headless, or scripted bsdinstall

Cool, thanks.

i'm learning it now too

the main file is installerconfig btw. man bsdinstall talks about it

it gets even better. Doing "kldunload radeonkms" results in a kernel panic!

seems like FreeBSD really hates this GPU, even though Radeon 9550 is listed as "Works". wiki.freebsd.org/Graphics/AMD-GPU-Matrix

Title: Graphics/AMD-GPU-Matrix - FreeBSD Wiki

SomeRandoUser: probably regressions since no one uses these any more

SomeRandoUser: Is it AGP card?

yes, it's AGP

SomeRandoUser, You might just need to tolerate the one minute pause at boot time for a while. Might be easier than alternatives. Then debug it over time.

i've been meaning to debug why switching the loader to a 4K GOP resolution causes a ~30 second delay in boot, that's kind of annoying

Using the "vesa" driver works, but there's no OpenGL acceleration and I'm limited to 1024x768 max resolution

Is there any way to "reset" the vt terminal driver?

Xorg leaves me with a black screen when it exits, and switching to another tty doesn't fix it. Wondering if there's some way to reset the video card state.

I think I'm going to give up on FreeBSD with this computer. Anything that tries to use /dev/drm/0 gets stuck in an uninterruptible state and can't be killed.

SomeRandoUser: if it works on linux, you might want to file a bug against the freebsd drm-kmod port, since this code comes from linux mostly unmodified (so it's probably a freebsd issue then)

Ok, I will.

welp, I just accidentally hosed my ufs partition trying to resize it

on FreeBSD, I have a php script that makes use of semaphores and it keeps locking up

ipcs -a shows one key with a large negative number, can that be correct ? or is it a sign of an error somewhere ?

ex: s 430833667 -2043517800 --rw-rw-rw- www www www www 3 11:12:53 11:12:53

SomeRandoUser: exciting! how is it hosed? any chance to fix it?

how do you make unattended bsdinstall set up a password for zfs "Encrypt Disks"? or where's it documented how to do it plz?

alepzi-: fwiw you might try asking on the freebsd-question@ mailing list, if no one here knows

thx! if i don't figure it out in a day or 2 i'll try that. really don't wanna bother ppl if avoidable

alepzi-: I don't think you can, unfortunately not everything can be automated with bsdinstall. I've had to dig into bsdinstall shell scripts a bunch of time to find out how to force it to do my bidding without any interactivity, but I've found every time can't be done.

dh is there any way to add disk encryption after install manually then?

Basically it's like this - if the automation aspect is not described in bsdinstall man page, it's very unlikely that it can't be done

dh are you sure it's not covered in man bsdinstall though? ZFSBOOT_GELI_ENCRYPTION and ZFSBOOT_GELI_KEY_FILE are in man bsdinstall and *sound* right, but there's no docs

Yes, but you asked about supplying password, which according to man page can't be supplied through environment.

Read /usr/libexec/bsdinstall/zfsboot and see if there's undocumented way to give password through environemnt and skip the dialog UI interaction, I doubt there is though

part of the problem is i don't know how to say what i want in the right way. using bsdinstall manually, i enable disk encryption for zfs then it prompts me for a pw before committing. then when i reboot, i have to enter pw to boot. i want to configure the same but with unattended bsdinstall

so i don't know certainly that i want to give installerconfig a pw, but i know i want disk encryption to ask me for a pw on each boot

what i'm hoping is that it's like the pw utility, where i can manually set a user to have a pw hash, that i got from another program where i gave the user's password

so i *hope* i can enter my pass into a util, get a hash, add that file to my distribution, and give path to that file in ZFSBOOT_GELI_KEY_FILE

alepzi-: how many installs are you expecting to automate with the same passphrase?

just 1

this hardly seems like an effective use of time

ok well that's your opinion

reproducible systems aren't a good idea to small brains

if i was going to do it, i would script it but not use the bsdinstall at all

or unravel bsdinstall source to figure out how to make it do what you want

but the fact that it has to be entered on the terminal should give you a hint that you're out in the weeds with regards to "reproducible systems"

unravelling bsdinstall's zfs script is what i probably have to do

So I did manage to get X running on this Radeon 9550 card with dri support. Seems like it's a regression in FreeBSD.

I managed to get it working on Linux I mean

i thought radeon was better supported than nvidia because nvidia is so hostile to FOSS?

oh

This is an old AGP card from 2004

yes, in genera

l, nvidia sucks if you're not using windows

I am using a fanless AMD HD 5450 Radeon from like 10+ years ago driving 2x 1920x1200 display and it has been very good for me.

used to be the other way around, but since open source radeon drivers, i'd never buy another nvidia card

(my desktop has an RX 6800 XT, works absolutely fine in freebsd with wayland)

My other box had a modest nVidia running Linux and I used it with the nouveau driver until about two years ago when the driver maintainers started thrashing the driver, making it unstable, and I suffered frequent graphics freezes and kernel panics. I gave up and pulled the nVidia card out swapping to the AMD and I am never going back to nVidia again.

Is the entire freebsd graphics stack basically the same as Linux?

yes, it's literally just a copy of the linux drm/kms code with a shim layer to make it load as a freebsd kld

+ wayland, X, etc., is identical to linux

dirty but i guess it makes sense to shift all the real work to linux

nvidia is different, they have a native freebsd driver and X.org module

then just piggyback it

(although i would assume the closed source part the freebsd nvidia driver is probably identical to the linux one with different shim layer)

yeah, I doubt freebsd has the resources to develop something custom

It just makes sense to share the work. No need to be NIH and ignore it.

especia

lly with hundre

ds of gpus out there

it kinds of sucks that it's all GPL but no one who does proprietary graphics stuff cares about this, e.g. Sony wrote all custom stuff for PlayStation

goddamn enter key

alepzi-, Since you are in the channel, continuing previous discussion, I learned that the AES-NI driver on 13 is no longer needed to be loaded as a module as it is apparently now built in. That's why it was not needed to be loaded as a module anymore and I didn't see a performance problem not having it.

rwp yep i found the same, thx!

(well, no one really knows what sony does... but the basic OS is freebsd and they clearly can't include any GPL code so they must have written their own drivers, unless perhaps AMD did that)

I avoid hardware that I know does not have good clean free software drivers in the mainline kernel. It's just not worth the hassle hacking around the fringe trying to make shady hardware work.

pretty sure that graphic driver stuff is under multiple licenses so that you can use it even if you don't want gpl stuff

I see claims that the Nintendo Switch runs a variant of FreeBSD. Is that true?

nimaje: the Linux DRM code?

lw: yes, but not sure where I read that

hmm interesting

you may be right, i've never really looked into this

SomeRandoUser: no, but it seems like their user space networking stack is derived from freebsds networking stack (the nintendo switch uses a micro kernel architecture and has most drivers in user space)

lw, another PR committed

I think there are other OSes that derived their networking stack code from FreeBSD too, like Haiku

jbo: thanks! i haven't got around to fixing sublime-music yet (been working on someone else today) but at least all the dependencies are in now, i'll try to get that done tomorrow

uh.. s/someone/something

jbo: i think #275962 (multimedia/mediaelch) should be ready to commit too, that one doesn't have any python

(no rush, just mentioning it)

I think the sockets API that most operating systems (including windows) use came from BSD (not FreeBSD)

and that's where the confusion comes from

SomeRandoUser: nah, people analysed the Windows NT IP stack and concluded that the actual implementation (not the API) came from BSD. but it turns it it was an unrelated commercial implementation

not that loads of people haven't used the various versions of the BSD IP stack of course... MS just isn't one of them

Where can I report this GPU driver bug?

unless it's an upstream bug: bugs.freebsd.org

Is there any way I can find which driver version I'm using so I can compare it with the Linux one?

pkg info

lw, Since MS-Windows code is proprietary and the BSD license does not require it I might say that the real problem is that we are prevented from being able to know if Windows includes something or not.

why is that? part of the beauty of the BSD license is that a commercial entity can "just use it"

uh oh, I'm sensing a copyleft vs permissive debate coming...

SomeRandoUser: include your drm-kmod pkg version (and drm-XXX-kmod version if any of those are installed) and freebsd version

rwp: it was only a problem for the people who spent decades accusing MS of 'stealing code' :-D

(even though the license says you can do that, so... who cares?)

kenrap, Nope. I am not going there.

Thank you :)

lw, They were not stealing it. "Stealing" is the wrong word as that must deprive the owner of the use of it and that isn't the right thing for software. And also the BSD license allows it. Plus the community encourages it. That's a far bit from any illegality.

rwp: i didn't say they were stealing it. i said _other people_ said that. i'm well aware of how the BSD license works

And honestly we all thought it was a good thing anyway. Can you imagine how messed up the networking would be if MS had to create it out of nothing from scratch themselves? Everything would be broken and it would be a bad situation for the entire network!

basically people were mad that MS used the BSD IP stack without acknowledging that they did. then it turned out they didn't anyway. so the whole thing was extremely silly

kenrap, I would be going to the advantage of a free software operating system. One where we can all see all of the code and know what is in there and how things work and find and fix bugs. I wonder where we might find one of those? :-)

lw, Do we have a reference describing the situation with MS networking code? If we can't look at it then how will we know anything about what they have in it?

rwp: over there, "A wild permissively-licenced BSD OS appeared" :P

Which is really the main problem and why most of us are avoiding Windows code as much as possible. We can't audit it. We can't fix the problems in it. It's a nuisance!

rwp: gotta hack'em all :)

rwp: on the same note, I would like to spend some time getting familiar with the FreeBSD kernel/KPI (and brushing up on C), so I can solve my own problems when they arise

Not a high priority for me, but a nice to have

kenrap, +1 and me too on that at about the same level. I need to expand my skill set. But a few other things are in the queue ahead of it.

|Backup of ZFS information for all imported pools

|

|-- End of daily output --

i'm confused why daily send me this email when i have info output disabled

cron job ?

yes, periodic runs from cron, but i have informational output disabled in periodic.conf

maybe you didn't restart your cron daemon ?

or reloaded it ?

you don't need to restart cron after editing periodic.conf, but in any case i haven't changed it for several reboots

the message is from 223.backup-zfs, it seems like it sets rc=1 if /var/backups/${poolname}.bak doesn't exist, but that file still doesn't exist after running the script

if i run the script by hand it exits 0, hm

i guess i'll see if it happens again tomorrow

jbo: how do you become a ports committer?

i don't think i have any interested in being a src committer but ports might actually be useful

There must be twenty-something control variables that are needed to disable all of the (useless?) things periodic does routinely. I am still trying to understand any utility of some of them.

rwp:

# wc -l /etc/periodic.conf

101 /etc/periodic.conf

How is it helpful to know how many hours the different accounts have spent logged into a system in a month?

0 dalmations^Wvariables

rwp: for a multiuser system, a high amount of cpu usage (i don't think it's hours) could account for someone using an unfair share

but for a single user system, it's silly

i really think periodic should be updated for modern standards, no one wants all this crap emailed to them by default

let them turn it on explicitly if they want it

but you know if you propose this you'll get a bunch of people saying "i've been using UNIX since PWB and i depend on these emails for my critical infrastructure!"

this is a funny user-agent: User-Agent: Mozilla/5.0 (X11; Linux x86_64; FreeBSD amd64) AppleWebKit/537.36 (KHTML, like Gecko) QtWebEngine/6.6.1 Chrome/112.0.5615.213 Safari/537.36

it's linux! but it's also freebsd. and it's mozilla, and chrome, and safari

In order to log into chase.com I have to override my User-Agent and lie saying it is Linux (or Windows) or it refuses to allow me access.

rwp: i'm suspcicious that having FreeBSD in my UA is why eBay permanently banned me

it's a bird, no it's a plane, no it's fingerprint-resisting :)