10$

lw, your mediaelch PR is now the next in the queue

exciting!

i hope it's actually fixed this time

I hope so too because each run eats up like 2 to 3 hours of my build machine time :p

i wouldn't mind giving you a shell here to test it, except since this is the last one (other than sublime which is fairly quick) it's probably not worth it...

that's nice - thanks. but all good

it's just part of the job, nothing wrong with that :)

do we not have nv(9) in userland?

that library was so useful on solaris

meena: i have created a git repository for dladm (except it's called dlctl now). let's see what happens!

lw, unless the Solaris library is something different, that's a work in progress in Phabricator, IIRC.

V_PauAmma_V: yeah, it's the solaris library that (iirc) was imported for zfs

what is the connection between netlink(4), genetlink(4) and rtnetlink(4)? it looks like rtnetlink can configure interfaces, so if that's all i want to do, i should just use that? the manpages are a bit vague

I think rtnetlink is for routing?

you'd think that given the name, but the manual page says it's for "interfaces, interface addresses, routes, nexthops and arp/ndp neighbors" and it deals with creating/destroying entire interfaces

what i'm trying to do is get a kqueue event whenever an interface/address/route/whatever is modified, it's not really clear to me if rtnetlink even does this

i know this functionality exists though because things like BIRD use it

That's beyond my ken. My window into available APIs is constrained by (what I understand and remember of) the docs and manual page reviews I see pass.

ah, seems like you can use NETLINK_ADD_MEMBERSHIP with the various RTNL_GRP_* constants for this

what's this doing here /usr/include/netlink/route/common.h:RTNLGRP_DECnet_IFADDR,

did freebsd *ever* support decnet?

i know we had XNS at one point...

lw: wow just wow with this checkout and branch

screw up something, just go back to the main branch.. get rid of branch that is bad.. make a new one.. start again.. what has taken me so long to find just beauty

CVS? Subversion?

parv: i believe we're talking about git

voy4g3r2: wait until you find out about interactive rebase. messed up a commit? just make a new commit, then squash it into the commit you want to fix

lw: I was replying to rhetorical question of "what has taken me so long to find just beauty"

voy4g3r2: git's philosophy is very much about mutable history, so you can commit stuff, then edit your commits, or rearrange them, or whatever, so the final set of commits you submit to be merged is exactly what you want

lw: my horror started with cvs and that hell.. then i got "smart" and wrote our requirements and other "stuff" and gave it to someon with more patience than me

yeah, CVS was... fine for its fine. but i'd never go back to that from git

there was some microsoft product i use to use in like 2005ish time frame.. which just sealed the hell of source code repositories

Microsoft Team Foundation Server?

Title: Microsoft Visual SourceSafe - Wikipedia

this!

horrible just horrible

ah yeah, i never used that but i've only ever heard bad things about it

that is when is tarted the journey.. before i made a major change, i would just "backup" the tree.. to another directory and pray i did not break it too much

freebsd used perforce for a while as a replacement for git... i think a lot of video game developers still use that because it handles binaries better than git

s/replacement for git/replacement for cvs/

sourcesafe HATED excel documents, i remember that

yes, make fun of me now, i did a lot of vba coding and data analysis with access+excel

so route(8) called snl_read_message_dbg() which... doesn't exist? what am i missing here

voy4g3r2: when i was in high school i wrote a nethack-like RPG in Excelt VBScript :-d

oh, it's an inline function in netlink_snl.h

interesting.. a lot of this manual page stuff references.. mac os x uses blah.. i found something that is the other way around xlocale.. appeared in Drawin 8 and was incorporated into freebsd 9.1

interesting..

btowc_l has no man page but btowc does.. and the different is the data type it works with..

put good ole #include <wchar.h>

a category 4 man page referencing a category 8 man page..

tslog(4) -> boottrace(4) which is actually boottrace(8)

seems valid.. change going in

hrm.. dialog4ports(1) is only available through ports but a base package references it..

whoo! i can execute 'ifconfig wg1 create; ifconfig wg1 1.1.1.1/32; ifconfig wg1 1.1.1.1/32 -alias; ifconfig wg1 destroy' and my program sees all of these events via netlink

this interface is actually quite easy to use, it's just really badly documented

wireguard?

isn't 1.1.1.1 cloudflare dns?

yes, but i only added it for 5 seconds to test

the important point is i can detect interfaces and addresses being added and removed

lw, patches/pull requests/suggestions for improvement of the manual page(s) welcome.

i don't even understand the api yet so i don't think i'm the one who should be writing the manpages :-d

still waiting for my sctp(4) manpage updates to be merged anyway...

You're the one having trouble using them. I can't think of anyone better suited. :-)

haha

maybe if i finish this app i'll consider it

Fair enough.

i appreciate the impetus behind "you should fix this thing you're complaining about" but i have like ten thousand projects on the go at once + my actual work i get paid for

i do try to submit PRs for things that can be fixed relatively quickly

Fair enough. *eyes his own to-do list and grimaces*

... the reason i mention this stuff is because i try to fix things when other people mention this here. like i submitted my sctp(4) PR because someone here was (rightly) complaining that sctp(4) is missing a bunch of docs

so maybe someone will see my messages and do the same

i guess this can come across as complaining but the way i mean it is more like "hey, it would be cool if someone could fix this if they felt like this"

Ping me in a few days? I'll be happy to turn your pain points into better manual pages.

when i install freebsd zfs on root with bsdinstall, boot up disk select shows my 2 nvme drives but also a FreeBSD i can select. any 3 will boot freebsd. but doing an unattended bsdinstall only the 2 drives show in the boot device menu. where's the "FreeBSD" line?

alepzi: what does efibootmgr say? usually freebsd installs two copies of the boot loader, \EFI\BOOT\BOOTX64.EFI and another one called something like \EFI\FREEBSD\LOADER.EFI, possibly your firmware is picking up both

(i don't really understand why it does this)

sudo efibootmgr says boot to fw: false, bootcurrent 0002, bootorder 0002 0003 0001, boot002 uefi os, boot003 uefi os, boot001 windows boot mgr

alepzi: you can check the contents of the ESP to verify this, it's usually mounted on /boot/efi by default iirc

e.g. `find /boot/efi` ... if it has two copies of freebsd loader, delete one of them and reboot, and see if the mysterious third boot option goes away

although as to why unattended would work differently here, i have no idea

got /boot/efi/efi/freebsd/loader.efi and /boot/efi/efi/boot/bootx86.efi

bootx86.efi? are you on 32-bit x86?

no

amd 64

hmm

alepzi: please show the output of `file /boot/efi/efi/freebsd/loader.efi /boot/efi/efi/boot/bootx86.efi`

you said you had bootx86.efi, but your paste shows bootx64.efi

(x86 != x64)

lol i'm sorry it's 64

ah ok

so it's normal i guess?

alepzi: delete /boot/efi/efi/freebsd/loader.efi and reboot. that will probably make the freebsd option disappear from the menu. or, if you prefer, leave /freebsd/loader.efi and delete /boot/bootx64.efi

it doesn't show up already. i was asking why it didn't

hm, you said "boot up disk select shows my 2 nvme drives but also a FreeBSD i can select" - i thought that meant it was showing a third option you didn't expect, which might be .../freebsd/loader.efi

ya when i...

when you...?

nvm

well in any case i suggest deleting loader.efi because there's no reason for this to exist

why is it there then?

who knows?

shouldn't the freebsd devs remove it?

probably. feel free to file a PR and/or ask the mailing list

i always delete loader.efi when i do a new install

i do a new install about once every 10 years though so it never bothered me enough to do something about it

i do a new install for every point release like 13.1 to 13.2

then just freebsd-update patch levels

well, file a PR, someone will probably look at it. i do think this behaviour is wrong

i'm so mad that we can't write K&R func definitions anymore

they're so much more readable than this ANSI crap

uh, no, removing loader.efi is a terrible idea in general

kevans: why? the EFI firmware will load \EFI\BOOT\BOOTX64.EFI. there is absolutely no reason for \EFI\FREEBSD\LOADER.EFI to exist, it's just a redundant copy of BOOTX64.EFI

we don't want to claim bootx64.efi

but you *do* claim that

we don't if it already exists

bsdinstall puts both files onto the ESP

and we want to prefer the vendored copy, that's the entire point of the efibootmgr entry

removing it isn't going to remove the boot entry, it's just going to break it for no apparent reason

ok, for the 0.01% of users who have multiple EFI loaders on their ESP, this might make a difference

for the vast majority of users, this just creates two copies of loader.efi for no apparent reason

ideally we move away from bootx64.efi in general, but there's a lot of crap firmware out there

i stand by my advice that most users should simply delete \EFI\FREEBSD\LOADER.EFI as it serves no purpose if the disk only has freebsd on it

ok, but that doesn't make it good advice

ok, but installing two identical files isn't good practice either. your move...

lol

fwiw, this confused me a lot when i came to replace the loader after zpool upgrade

i discovered there are two copies of the loader. so what do i do? replace them both? but why are there two copies?

ideally you follow the steps in the release notes for this exact scenario

it gives precise steps to follow to determine what's in use

the release notes just say to update the current EFI loader as described in efibootmgr -v, but this is wrong. this will teach the user to replace bootx64.efi (if that's what the firmware loader) but ignore loader.efi

which will create issues if, for example, the disk is moved to another system that loads \EFI\FREEBSD\LOADER.EFI before BOOTX64.EFI

having two copies of the loader where one might be out of date cannot possibly be the correct behaviour

why would your new system magically use the vendored path? swapping out freebsd installs under a different firmware is kind of weird

ideally we'd have a tool to deal with all of this

kevans: i don't know why it would "magically" do that, but as we just discovered in this channel, some firmwares will notice the FreeBSD loader and display it as an additional boot menu entry

they do not notice it, no, we explicitly add the entry at install

so imagine you switch from a system that doesn't do that, to one that does, and your \EFI\BOOT\BOOTX64.EFI is up to date, and your \EFI\FREEBSD\LOADER.EFI isn't

rEFInd might notice it, yes, but all firmware I'm aware of will not

now when you select 'freebsd' at the boot menu, the system won't boot

i don't really care to argue about this anymore, but installing two identical copies of a file, and requiring the sysadmin to update both copies, and then not telling them that they have to update both copies, is bad practice

how do you determine what event happened on an fd from kevent()? does the kernel set ev.filter?

yes, the kernel fills in the good bits to describe the event

Hi folks! I am confused by this filesystem behavior. I mount_nullfs one folder onto another, and I get different outputs from "ls" for the same folder when accessing it from the concrete and the nullfs-mounted entrypoints to it: dpaste.org/vCom8

Title: dpaste/vCom8 (Markdown)

oh rubber duck debugging :)

is it just me or is freebsd terminal a lot faster than linux? when i run commands, even from a remote ssh, the output is like really fast?

or is it zfs?

heh

I was somewhat surprised to see it pick up all loaders on my ESP.

jbo: how did mediaelch go?

ok, so i maintain a port. i want to submit a patch. how do i do that? i don't see anything in the porter's handbook about this

lw: been ages since I did that procedure. But in short, bugs.freebsd.org/bugzilla/enter_bug…&component=Individual%20Port%28s%29. Summary: "<category/portname>: Update to v1.2.3", Add some description, attach the patch/diff from your local ports-tree.

Title: Log in to FreeBSD Bugzilla

An old ports-update I did bugs.freebsd.org/bugzilla/show_bug.cgi?id=256595

that's basically what i did bugs.freebsd.org/bugzilla/show_bug.cgi?id=276347 but i wasn't sure if i should set maintainer-feedback or not... the wiki TriageTemplates is a bit vague on that

Title: 276347 – net-im/toot: mark as BROKEN since devel/py-urwid is broken

this was basically, bump the version number in the Makefile, 'make makesum' (?) to get the new packages ( port tracked github releases of the matterircd service )

i'm going to regret this lexi⊙lo | + 1 Jan 15 freebsd-python+help⊙Fo ( 17) Welcome to freebsd-python⊙Fo

soon my entire life will consist of nothing but reading mailing lists

meena: how do i get this committed freebsd/freebsd-src #1057 should i just add bsdimp as a reviewer?

Title: service(8): direct user to rc(8) for a list of valid commands by llfw · Pull Request #1057 · freebsd/freebsd-src · GitHub

lw: that's mostly documentation, so I would bug someone who's does docs.

meena: this sounds like good advice. please give me @usernames of people who does docs :-)

lw: I just posted it in #bsdocs (EFnet) and #documentation (discord)

meena: #bsdocs on efnet does not seem to exist?

ah #bsddocs

meena: how do i get a src commit bit?

lw: yeah, I wonder too

heh

i had a weird conversation with j.bo about this the other day

the answer seems to be "commit bits are magic and not understandable"

Yeah, and, honestly, I don't like it.

although j.bo claimed, from what i remember, there are ~4,000 src committers and ~192 ports committers

which, by itself, is kind of hilarious

it makes sense when you think about the structure of the Org, but, yeah

I thought there was voting process after you've been a proven contributor for a while.

(this is not a criticism of j.bo, who i'm sure has nothing to do with who gets a commit bit)

I highly doubt that all 4000 of those are active…

_xor: who knows. Is it documented anywhere? If it isn't, everything we think we know is hearsay.

meena: i also find this unlikely

although, perhaps Netflix has 4,000 people working on freebsd. who knows

probably not though

netflix has at least two people working on FreeBSD.

passive-aggressive pr of the day bugs.freebsd.org/bugzilla/show_bug.cgi?id=276347

Title: 276347 – net-im/toot: mark as BROKEN since devel/py-urwid is broken

Title: Committer's Guide | FreeBSD Documentation Portal

_xor: that's not how to become a committer

I think it's delegated by people in those specific areas, and those people are probably appointed/elected by the core members. I'm pretty sure I've seen the details documented somewhere.

i feel like my various manpage updates are trivial and should just be committed. it annoys me that they aren't

Title: BecomingACommitter - FreeBSD Wiki

it's not like i'm comitting a replacement pmap or something

lw: re #freebsd-python: bugs.freebsd.org/bugzilla/show_bug.cgi?id=273122

Title: 273122 – lang/python311: backport netlink support

"As more and more companies leverage FreeBSD for projects" ah got it, i need to add Sponsored by: in my commits

lw: i documented a pmap!

Title: FreeBSD Project Members | The FreeBSD Project

Those two links should give a starting point answer.

wait there's a #freebsd-python?

yes

maybe I should subscribe to the ML…

meena: why did you link me PR #273122, idgi

273122 – lang/python311: backport netlink support bugs.freebsd.org/bugzilla/show_bug.cgi?id=273122

i mean i'm doing netlink stuff, but in C, not python

lw: just, take a read across the PR when you have some time, and nerves to spare

is this going to make me angry

probably

Hello. I can't access my micrphone in ungoogled-chromium

Could anyone kindly help with that?

microphone*

Installed devices:

pcm0: <Conexant CX20561 (Hermosa) (Analog 2.0+HP/2.0)> (play/rec) default

pcm1: <Conexant CX20561 (Hermosa) (Internal Analog Mic)> (rec)

meena: i can't comment on this because the other person is on this channel, i guess

which one is active?

this one is definitely not working

at least in ungoogled-chromium

how do i switch microphone in freebsd?

i have only one rec devices

i switched recording device to "dsp0" in audacity. i also tuned fossmixer a bit, though not sure if it was needed.

in me@x200:~ % sudo mixer =rec

mixer: no recording device specified

V-T60: i havent used pavucontrol but maybe it can help you vermaden.wordpress.com/2024/01/13/f…29-configuration-audio-improvements

Title: FreeBSD Desktop – Part 29 – Configuration – Audio Improvements | 𝚟𝚎𝚛𝚖𝚊𝚍𝚎𝚗

taking someone's advice to nag people about stuff lists.freebsd.org/archives/freebsd-python/2024-January/005480.html

Title: Re: devel/py-urwid and devel/py-setuptools_scm (PR# 274411)

but the chance this will actually fix the problem is like, zero, right

i don't really understand how a port can be broken for months and literally no one cares enough to do anything to fix it

why even have ports then

i don't understand how that might help

% pactl list sinks | grep 'Name: '

W: [(null)] caps.c: Normally all extra capabilities would be dropped now, but that's impossible because PulseAudio was built without capabilities support. Name: oss_output.dsp0

so i'm trying gpu passthru to win on bhyve but not having any luck. gfx card is detected in guest, but can't start

crest: reviews.freebsd.org/D41318 would be a nice thing to have IMO

Title: ⚙ D41318 Add the "missing" WireGuard rc.d script

maybe you can move it along a bit at some point

Bheam: there's a bhyve specific channel for this, you should post your bhyve command invocation there, and a bit more info (what h/w device, what guest OS, how its detected etc)

dch: the easiest way to move it along would be to have more reviews / tests / patches to it

more reviews but only magical know about phab

so you have to find magical people who like your commit

looks like this diff will never be committed because of bikeshedding, sorry, bye

Smartctl open device: /dev/nda0 failed: INQUIRY failed , how to get the total writen bytes of device /dev/nda0 ?

smartctl /dev/nda0

hi all

occ: do you have any '/dev/nvme*' devices?

norrland: yes,there is one

try querying the '/dev/nvme0' device instead.

use /dev/nvme ?

Title: dpaste/PHhtb (Bash)

yes , smartctl -a /dev/nvme0 gives information.

but smartctl -a /dev/nvd0 gives "Smartctl open device: /dev/nvd0 failed: INQUIRY failed"

Yeah, could be that it's not providing that interface on non '/dev/nvme*' paths.

smartctl -d nvme -a /dev/nvme0

arg, i really didn't want to recompile llvm and rust in a manually installed port

Has something changed with poudriere, "Inspecting ports tree for modifications to git checkout..." now takes ~1h for some reason.

poudriere-3.4.0

openzfs rpm for EPEL9 (I'd suspect all supported distro binary packages) has arcstat and arc_summary. I see them in the contrib src repo. Does FreeBSD not include them? locate can't seem to find anything if they were not in a binish dir.

I'd guess maybe the kmod port might include them?

oh it's all python. So I guess that would explain it... disabled by default in the port too.

Quarterly patches successfully installed (to the best of my knowledge).

lw: meena: the number of active src committers is exactly names still listed in cgit.freebsd.org/src/tree/access?h=internal/admin, which is... under 200

ports number is pretty close, and they have a tighter reaper policy: cgit.freebsd.org/ports/tree/access?h=internal/admin

Title: access - ports - FreeBSD ports tree

(doc and src requires one commit per 18 months to avoid reaping, ports requires one commit per 12 months)

hello! where can I see the package description of dmd?

ah, apologies, I think the question is for OpenBSD, not FreeBSD (doesn't seem to have a dmd package: freshports.org/search.php?query=dmd…t=1&casesensitivity=caseinsensitive)

kevans: i had to hand in my ASF membership formally, i wonder if i can still commit, if i can remember my password

what desktop files are neeeded to make the gdm greeter work? By default as on `pkg install gdm` there are none in the configdir

why there is 4.7G used and 7.7G free on a 14G volume?

dlfke: compression?

i am trying to start greeter on freebsd whose pkg manager puts no files at all in the gdm config directory, consequently greeter GUI does not start when i boot but GDM is running

what files are neeeded to make the gdm greeter work? By default as on `pkg install gdm` there are none in the configdir

no compression. it's 1.5 GB missing

Oh, right. Wrong way. dlfke: Snapshots maybe, then.

can I remove /usr/freebsd-dist?

using freebsd as a desktop system is not easy due to a crazy amount of junk that is installed by the package system

hu hum... I see now...

That looks fun, what are those relations exacly?

the dependencies between /etc/rc.d/* scripts, as they are parsed by rcorder(8)

that one looks much better: hostit.tfnux.org/obj/k2hkfkyta7snyrf7vsuxwmd66plz57dd/twopi.png

dlfke, space reserved for root?

Look at tunefs(8).

tsoome: this looks good to my eye, but you might wanna take a look, too freebsd/freebsd-src #1063

Title: vt.4: explain console fonts by concussious · Pull Request #1063 · freebsd/freebsd-src · GitHub

seems good, but its 01:21am here, so I'm rather sleepy.... :D

back to bed, mister!

tsoome: it's 23:28 here, and I just put my daughter to sleep