Where do I find the source code for the program described by this manpage? man.freebsd.org/cgi/man.cgi?query=l….2-RELEASE&arch=default&format=html

Title: lint(1)

If there's a link on that page to the associated package/port/whatever-the-BSD-nomenclature is, I'm not seeing it.

Looks like man.freebsd.org/cgi/man.cgi?query=l…tion=1&manpath=FreeBSD+11.2-RELEASE is the more canonical link for that page.

Title: lint(1)

I tried searching the git repositories at git.freebsd.org, to no avail. I'm much more familiar with how Debian is organized than I am with any of the BSDs.

Ok, I think I found the source for that file. I cloned git.freebsd.org/git and checked out the release/11.2.0 tag. It appears the troff source for that manpage is at usr.bin/xlint/xlint/lint.1

I think what was throwing me off was that I was also trying to find the source of lint(7) which is referenced in that manpage. I tried searching for a file named lint.7 but it actually comes from the shell script usr.bin/xlint/lint1/makeman which does little more than cat out the manpage from a large block quote in the script.

I guess even an unresponsive IRC channel makes a decent rubber duck

plasma41: 👍

Man! zfs-destroy & zsf-release on different snapshots c 2021 of same dataset (~50 TB) for ~30-90 GB took worryingly long time for relatively low space. Finished without issue thankfully.

"zsf-release"! Oh Noes!

plasma41: for built-in tools (userland etc) then `whereis $thing` tells you the manpage, binary location, and src code

`whereis jail` returns `jail: /usr/sbin/jail /usr/share/man/man8/jail.8.gz /usr/src/usr.sbin/jail`

parv: a friend once was busy destroying snapshots that were taken every 15 minutes over 2 years or so

it took him 4 days

Damn! Interesting.

Remilia, Well, that certainly puts the expected time to finish in perspective

I would expect zfs channel program to do that much better

delphix.com/blog/delphix-engineering/zfs-channel-programs | klarasystems.com/articles/understanding-zfs-channel-programs

Title: Understanding ZFS Channel Programs | Klara Inc

its perfect for that sort of stuff

dch, delphix URL says 404.

Am aware, but have not looked deeply into that. Will do soon, thanks to your nudge, dch

how odd. delphix.com/blog/proposed-zfs-feature-channel-programs is not the one I bookmarked, but its a reasonable intro

Title: Proposed ZFS Feature: Channel Programs | Delphix | Delphix

dch, Thanks

for context I have some automated CI stuff that ends up with 10_000 snaps if I'm not paying attention, and doing a zfs send of that is basically unusable.

so I trash the older snaps with a channel program

can I *clear* the bootfs property on a zpool?

I have 2 zpools (a split ex-mirror) while I convert a 14.x compat pool to a 13.x compatible one

and only want 1 zpool to boot

dch zpool inherit bootfs

tsoome: only zfs-inherit, unless you have some illumos trickery that hasn't trickled upstream yet

there's no inherit for zpool? try just zpool set bootfs= pool

tried that too

I'm rebooting now and will use the loader to switch

works for me (tm)

my bad, I tried none but not `= `

good tip, now lets see if I can get back from the loader prompt :D

well now I am indeed in the shit

OK set vfs.root.mountfrom=zfs:zboot/ROOT/default

and also with zroot (because the original half of the mirror)

neither can load kernel

why?

that is a good question "no bootable kernel" is all I get

ls /boot/kernel ?

show currdev ?

ooh I only set vfs.root.mountfrom, not currdev

lszfs zboot shows the expected bits

set currdev=zfs:zboot/ROOT/default:

tsoome: whats with the trailing : there I would never have guessed that

man 8 loader_simp has the info

a new manpage for me

all device names end with : :)

It Is The Way

ok, lets see if we can get 13.2-BETA2++ to work on this

dch: loader_simp wasn't in 13.1

otis: oh good, then its not just me forgetting/ignoring stuff :D

dch_simp

dch: in 13.1, loader(8) contains that info

all that just to reboot into 13.2-BETA2 ... phew

tsoome, yuripv thanks for the tips

is this covered by handbook?

yw

hi

whats the best method of importing a freebsd instance in azure to another site into bhyve?

f451, not sure if 'best' but dd would work, just boot it in single user mode and poke out a con via ssh or netcat to something big enough to hold the instance

from there you can boot it via bhyve or qemu and do any method you like

or just use the image potentially

could dd to a zfs vdev or w/e you want

zvol*

daemon: ok, thanks. i thought it'd involve dd. fortunately there's (non-zroot) a large zfs disk, i can dd to a file there then compress it as much as possible

yep

there is also possible a solution with dump and restore

but I always found using dd 'just always works'

i was concerned with the 'disk' (the ufs one) being /dev/label/rootfs

Hi everyone. I have a very strange problem with my freebsd desktop. After a reboot my my network seems down. I cannot reach anything internal nore external. I've tried lots of things but cant figure it out. The strange thing is i'm getting all the ip settings from my DHCP server. I even tried to bind the mac to a different IP in the dhcp then restarted netif and routing and i'm getting the new ip

but still cannot reach anything.

daemon: thanks

also +1 to the 'dd just always works' ;)

yup the old tools are the best :)

was a but worried cos the 'disk' is vxd or something, cant remember

tried setting hw.vtnet.csum_disable="1" in /boot/loader.conf tried setting static ip, changed from DHCP to SYNCDHCP in rc.conf . changed dns and doublechecked default gateway etc. nothing

f451, it can be whatever it likes, that is the base format dd does not care it cares what bits and bytes the drive says it has, so perfect for throwing around different backin volumes

and never mind. tried to disable ipfw now and suddenly it works *sigh*

xlnt.

f451, oh one warning ... remember that when you get it running in bhyev the device names or uuid's may change; if that happens boot it up with cd or memstick image mount / read/write and edit fstab to be what it should be

oh yeah, thanks

mdconfig is our friend here

i'll make a bhyve swap partition on a seperate image too

I need to swap out some components in my computer. Motherboard/RAM/CPU. Discovered some heavy memory problems...

Trying to decide if I should go with intel or amd.. Is there something to be aware of before purchasing cpu/motherboard when it comes to support... or is the hardware 100% supported (ish)

go POWER. tailos has a cool board

no idea if bsd works there

but it'd be worth many internet points

Any ideas, why the memset() doesn't cause segfault while the malloc() returns NULL: paste.debian.net/plainh/aa9f52ab ?

mrtnt: optimized away perhaps

compile it with cc and have at it... then it will segfault, when added some extra prints to the mix =D

aah, behaves the same. sorry

mrtnt is this PPC specific

?

drobban: it's strange that for example "memset(NULL,'x',10);" segfaults, but "memset(malloc(-1),'x',10);" does not

paulf: not to my knowledge, but i can try on x86-64

paulf: I tested on x86-64 machine running Linux and it behaves the same: "memset(NULL,'x',10);" segfaults, but "memset(malloc(-1),'x',10);" does not

Hmm

On Linux I get a segfault from your paste example

paulf: I just wanted to write, that I get the segfault both on PPC FreeBSD and x86 Linux if I compile with cc

Which compiler is cc?

gcc on Linux

I think only one Linux distribution uses Clang and that one has not yet officially been released

paulf: "FreeBSD clang version 13.0.0" on FreeBSD machine and "cc (Debian 10.2.1-6) 10.2.1 20210110" on Debian 11 machine

yep that is gcc 10

With or without optimization? With optimization It may see that malloc(-1) is UB and optimize it away

on FreeBSD/amd64 I get a segfault from that code if I compile with -O0

paulf: yes, I ran "gcc -O1 -g -Wall ..". Sorry, I should have mentioned this. Without the "-O1" I get the segfault even when compiling with gcc

any higher and it gets optimised away and never prints anything

I cannot get it to print anything with -O1 and up

mrtnt: you should probably look into disassembly and see what it is actually doing

The compiler can also see that the memory allocated by malloc is never used

0000000000401140 <main>:

  401140:    31 c0                    xor    %eax,%eax

  401142:    c3                       ret

(clang 14 on amd64 RHEL 7.6)

if you're lucky, -O0 won't mangle it

paulf: 31c0 is basically a nop

when you do not touch eax afterwards, that is

it's return value though so not a nop (setting it to 0)?

oh right, it is also used as mov eax,0

to shorten the instruction

What do you guys say. Should I stay away from AMD Ryzen 9 - is there any known issues running freebsd with such a machine?

drobban: I am running Ryzen 9 5950x with no problems on 14-CURRENT

buildword with -j32 takes ~20 minutes

roughly a year ago there was a little bit of stuttering on desktop systems, that it was fixed

drobban: typically the CPU / architecture isn't the confounding factor, but the almost unbounded set of peripheral devices that get attached on motherboards.

debdrup: thats my primary source of concern as well

Am I making a misstake going for something new in general?

drobban: the answer to that is typically to have a look at dmesgd.nycbug.org

debdrup: thanks

There's even a fortune that explains how to submit the information, see `fortune -m dmesgd /usr/share/games/fortune/freebsd-tips`

=)

bsd-hardware.info might also have some information, but I'm not as familiar with that site.

Title: BSD Hardware Database

Hi, dmesg show that my wifi card in theory seems to be supported but when doing "ifconfig wlan0 create wlandev iwlwifi0" I always get: ifconfig: SIOCIFCREATE2 (wlan0): Bad address

 any ideas ?

im trying to mount a usb flash drive and it's not showing up. no errors in /var/log/messages. so i tried fsck /dev/da0p1 and it's asking "ADD SUPERBLOCK CHECK-HASH PROTECTION? [yn]" what's that mean tho?

file -s /dev/da0

Also: file -s /dev/da0p1

nbari: try `ifconfig wlan create wlandev iwlwifi0` instead of `wlan0 create`

CrtxReavr termbin.com/50we

i haven't answered yes or no to the add superblock check-has protection question

it will probably ask you next time if you say 'no' now

so i should say yes? what even is that feature? searched web didn't find anything

Well, that tells me /dev/da0p1 is a uffs2 partition. . . so it shoudl be safe to say yes.

And let fsck do its thing.

and add inode check-has protection is yes too?

sorry i've just never encountered these before

ok answered yes to that, it ran, fs marked clean now it mounts fine

i wonder wtf happened

ufs2 doing its job as good as it's able

I would just: fsck -y /dev/da0p1

searching for that "ADD SUPERBLOCK..." reveals a few commits about ufs

And let 'er rip.

perhaps that happened between creation and mounting it now

Not sure just passing -y to fsck is the best idea..

debdrup, well, the alternative would probably require a hex editor and a pair of tweezers.

So. . . now what do you think?

CrtxReavr: or pressing 'y' a few times

c'mon man

What.

From my experience, it can mean pressing y *A* *LOT*

So press y, 'til you get tired of it, control-C out of it, then restart it with -y

That's an all too common workflow.

>=]

we don't all play fast and loose with our filesystem; and for somthing like this, it's proably one or two confirmations and gtg

Who knows what he did.

Besides, it's up to the user to have backups.

rtprio, you've never run fsck where it's thousands of confirmations?

yes, but i make sure it's going to need thousands of confirmations before running -y blindly

13.2 is coming next month i'm so excite

gonna finally learn about jails and how to containerize my deployments

lol imagine still running 12.x

might as well install debian or smth

It's very rare I see a OS feature that makes me track the latest releases.

Name a feature 13.x has that 12.x doesn't an how you make use of it?

it's not for everyone

"But but but the number is bigger!"

anyone around that's deploying as containers? curious what the 50k ft view looks like. i guess based on jails? i really like the idea that i can take a $2 vps and basically subdivide it for as many totally isolated projects as i want

instead right now i gotta either make projects share a vps or dedicated 1 vps per project

jails can work, but if you're on a $2 vps you'll have to have some proxy for all of them

what's the bit about a proxy?

and let's call it a $10 vps

how many ip addresses do you get with that vps? 1

rtprio, exactly

or you're paying $10 more for a couple of ips

ipv6 are free :)

if you only get one IP for it - then you need to do some magic on the host node in order to distribute traffic to loopback devices

for http(s) lets say

polyex: sure, if you only want nerds to access your services

and cheap vps servers usually don't have many cores

well the price does not really matter - if you only get one :-) ipv4 or six

you have to do the same hackery to make it work

polyex: ok, routing the ipv6 to your jails would absolutely work

k so i'll only deploy as many containers to a host vps as i have separate ips

You can also use vnet and NAT.

With the right set of sysctl(8) OIDs, I think it's even possible to make it so that your service provider can't detect (based on network traffic, at least) that there's any address translation going on.

I say get rid of virualization just use ha-proxy and werc.cat-v.org

debdrup: do some care about that?

get rid of databases

polyex, depends - you need to have as many ips as hosts you plan to have public internet access at least I think. Say for hosting http(s) - else you have to start running stuff on other ports per debdrup say 8080 and nating

get rid of MBA and project people

have king and free houses

how is freebsd > netbsd?

freebsd actually gets security updates

netbsd used to run on everything

I haven't followed it for years

just because it runs doesn't make it usable

for practical uses - no but for experiments - it does

rtprio: I have vague memories of needing it, but can't remember the context.

way back in the day I had it running on my amiga 2000

I would buy an amiga again just because old times - but damn the mc68xxx series cpus have become expensive

should have kept my 68040

a used 68060 in the right rev is so expensive it will make your eyes water

have anyone ever tried the actual commodore unix?

did not know it existed until about a couple of years ago

think it was a system V derivative

I'd like to try a System V some day but they're generally made difficult to use and run.

using pkg how do i extract tmux from 'tmux-3.3a'

i swear i get tripped up on this every single time

pkg query %n tmux

or %o if you want origin

%n, yeah; i somehow find myself in a situation where i always get the full name-version and what i need is the short name

(for pkg_cutleaves exclude)

hi, what's the best way to share files from the host into bhyve vms? NFS?

i would probably use nfs

Agreed

thanks

nothing like docker's "volumes" exists for bhyve then?

also, hi michaeldexter

we've been chatting on mastodon :)

There is a simple 9p server in bhyve that can be used with Linux VMs. Hi!

You CyberPickle932?

not me :)

I'm a filthy openbsd dev

That's a thing? :)

haha

big respect to openbsd

Likewise.

"If you want to hate on OpenBSD, remove SSH and related tools from your system"

michaeldexter: we spoke about supermicro ipmi and zfs recently

OpenCON/London?

michaeldexter: correct

HEY!

HEY indeed

disks for new home server arrived today, and it's getting freebsd this time

zfs just makes sense

i asked this in the openzfs channel, let's see if we get any other answers

But it's a laying violation! (Every OBSD Dev)

vext01: Ping me here or better on Skype as needed. I am not a cool kid with a persistent IRC history. :(

if you do mirrored zfs on root, what do you do to keep the system bootable should the boot disk die?

vext01: Both sides of the mirror should be bootable. FWIW, have you looked at TrueNAS CORE? It does lots of housekeeping for you.

michaeldexter: I like to do it myself

that way I understand it

Amen

Are you separating boot from storage?

michaeldexter: yes

FreeBSD does not assist with creating the storage array, though the boot configuration is good/good enough. I suggest one script the array configuration. How many disks are you planning for?

vext01: I need to run to a call but you can reach me on several other platforms by this exact same user name. Except bsd.network where for once I got the name I wanted.

vext01: generally you'll want to put partitions on all disks that'll be part of the pool, and partition every disk with an identical bootblock (gmirror(8) is good for this)

If one disk dies, the firmware should start from the one that isn't dead.

michaeldexter: thanks

debdrup: sweet

vext01: this, of course, assumes that disks don't lie - but since they do, you can get into a situation where the firmware will think that the disk is fine, so you'll manually need to switch boot disks.

debdrup: when i have the second root disk in my hands I may prod you for advice on gmirror etc, if that's ok?

vext01: there are some notes on it here wiki.freebsd.org/MasonLoringBliss/ZFSandGELIbyHAND

Title: MasonLoringBliss/ZFSandGELIbyHAND - FreeBSD Wiki

Ah, I see that's been switched away from using gmirror - still should work fine, though.

Where can I find the source for the FreeBSD icon? (Specifically just the icon portion of the logo, not the full logo with wordmark.)

plasma41: There's freebsd.org/art

Title: FreeBSD Art | The FreeBSD Project

plasma41: For the logo itself: freebsdfoundation.org/about-us/about-the-foundation/project

Title: Project | FreeBSD Foundation

That includes usage restrictions and guidelines.

mason: I tried downloading the zip file on that second page, but it only includes images of the full logo, not just the icon by itself.

looking

plasma41: Hm, it includes it when I download it here.

The website favicon freebsd.org/favicon.ico is of the icon alone, but I would like something that is higher resolution.

mason: Which directory?

plasma41: Download the archive again, and look for FREEBSD_Logo_Vert_Pos_RGB.png

Archive/Vertical/Full Color/Positive/RGB/FREEBSD_Logo_Vert_Pos_RGB.png

Thanks to: find FreeBSD\ Logo\ Archive/ -name '*.png' -execdir xv {} \;

There might be another more suitable but that's the first I found that was just the logo.

mason: That file has the wordmark underneath the icon.

ASCII logo appears at least in these source files: stand/forth/logo-orb.4th | stand/forth/logo-orbbw.4th | stand/lua/gfx-orb.lua | stand/lua/gfx-orbbw.lua

I could cut it out with gimp, but I'd prefer to find the source file that was used to compose the image of the full logo.

Oh, interesting, xv didn't show the words.

mason: Does xv have a black background?

  /boot/logo-beastiebw.4th and /boot/logo-beastie.4th for nice ASCII ones ;-p

Mm, not seeing the graphic by itself. If you write to the Foundation folks, they're friendly/helpful and can probably point you to the right thing.

mason: Ok, thanks for your help.

plasma41: freebsdfoundation.org/about-us/contact-us

Title: Contact Us | FreeBSD Foundation