meena: just a suggestion, don't stress yourself

Is it enough to host OpenVPN and DNS servers like Adguard and Pihole with 1 CPU, 0.5GB of RAM, and 10 GB of SSD on FreeBSD?

raspberry pi 4 is back in stock

docs.pi-hole.net/main/prerequisites -> Min. 2GB free space, 4GB recommended, 512MB RAM

Title: Prerequisites - Pi-hole documentation

Could 13.2 read, write 14-CURRENT ZFS datasets created c late 2021?

(It shooould, & obligatory: There Is Only One Way To Find Out.)

Thank you, Nixkernal

how hard would it be to remove the ipv4 stack?

we know freebsd has a CLAT so maybe there's some dependency between the v6 and v4 stacks...

Soni: WITHOUT_INET

Soni: see here man.freebsd.org/src.conf(5)

Title: src.conf(5)

we see

I've got a first gen framework laptop that mostly collects dusts, so i tried installing 14 beta 2 on it. i can complete the installer no problem, but after reboot it's unusable. if i drop to a shell in the installer i can see 3 partitions, but fdisk can interpret and i can't say "mount_msdosfs /dev/nda0p1 /tmp/boot"

the laptop works fine. installed arch just to confirm it's not a hardware issue

*fdisk can't

did the install with geli and uefi

(and zfs)

i know that's limited troubleshooting, but i thought i'd mention it. i tried dd'ing the drive with zeroes too just in case, but that didn't help.

markmcb: I'm about to disappear, but I would make sure you've picked the combined UEFI+BIOS install flavour, and re-try the installer

I *guess* they're UEFI only but I don't know for sure.

does it work without geli in the way?

worth doing a `gpart show -l` and pasting that in the channel, along with whatever ended up in /etc/fstab

dch: i'll try that. was definitely doing UEFI only.

also when you say its unusable after reboot, be a bit more specific,

e.g. does it get as far as the loader prompt, does it get stuck trying to find the root partition, etc.

bbl

dch: unusable as in nothing happens. bios error message appears stating no bootable devices

i have to boot from usb to do anything

and once i've booted from usb, if i drop to a shell i can't mount the efi partition created by the installer

Well Idk folks new FreeBSD14 mirror just keep dieing with some signature error after 10 minutes

I try to install the xfce with pkg

what kind of cdn is this :/

don't paste it or anything, everyone here is telepathic

I am so telepathic that i can tell — without consulting any documentation or searching the Web — that rtprio is being sarcastic

I just cant get the volume up

sysctl hw.snd.vpc_0db=1 did this and trying firefox

hmmm

are you using OSS?

concrete_houses: why are you ignoring what I've pointed out several times

Hi. Could i somehow start the ssh-agent and load my ssh private keys automatically? I got a pair for github, i need to run 'ssh-agent -s' and 'ssh-add <path-to-key>' every time before doing git pull / push

running ssh-agent automatically is the intended method; the way to do it depends on how exactly you're logging in

obviously running ssh-add requires interaction to specify a passphrase if the key is encrypted

KingShark: freshports.org/security/keychain helps with that, depending on… what RhodiumToad said

Title: FreshPorts -- security/keychain: User-friendly front-end to ssh-agent(1)

I source it in my shell startup.

i wrote something little in my .profile

Title: dotfiles/keychain.fish at main - dotfiles - Codeberg.org

Title: dotfiles/.ssh_agent at master - dotfiles - projects.17es.net

I really need a new profile picture. that i one is older than my daughter

github.com/google/cpu_features/blob/main/src/impl_x86_freebsd.c well

Title: cpu_features/src/impl_x86_freebsd.c at main · google/cpu_features · GitHub

who needs any of that tricky CPUID instruction assembly stuff when you can just.. parse /var/run/dmesg.boot

that is the second worst code I've seen all day (that wasn't written by myself)

sigh, __attribute__((target="")) only works in C++ mode in gcc

"And the C programmers likely prefer an explicit way using `__attribute__((ifunc(...)))` so nobody really wanted to spend time implementing this for C."

no, I'd like it to work as it does in clang, thanks

I love having order of magnitude performance improvements that never get used because someone might be using a 15 year old CPU

Freaky: is there any way to make this code less worse

meena: I'm sure that's as pretty as CPU feature detection gets

meena: the kernel probably works it out by parsing the same file from the previous boot tbh

Freaky: i… i want to doubt that

but i would think it's present in a sysctl

the kernel certainly does not do that :-)

afaik it's not in any sysctl.

I think apps that care about cpu features are expected to do CPUID themselves

Title: fast-bytecount/bytecount.c at addcc5468cf664d68f9a9ece3add5987f298a93f · Freaky/fast-bytecount · GitHub

RhodiumToad: well, it's in kern.msgbuf which gives you the same thing :D

it might not be in kern.msgbuf since that's of limited size

using lldb on a program I compiled with `-g O0`, but it seems not to be aware of the C sources it was built from

i.e. can't set a breakpoint to a given line, doesn't display C source of file

what exact commands were used to compile and link?

does `file` say that the executable is not stripped?

file says its not stripped, image <name> says it has a section (not entirely sure what that means)

Title: Snippet | IRCCloud

and you're running it in the same dir you compiled in? what clang and lldb version?

yup, all in the same dir, default clang/lldb from CURRENT a few days ago

also there generally won't be source info for libs that come from ports

this isnt a port, just local hacking

it's like cflags isn't being used here

let me do this by hand

so e.g. a command like 'list main' in lldb isn't working?

adding `-v` to CFLAGS should show more stuff, but its still silent

so maybe its not even being picked up

oh I see why

stupid me

ah

your second CFLAGS= is clobbering the first

ffs thats been there since day 1 of this project

ok, now lets see if we has the sources... nope

can't set any breakpoints either

does 'list main' in lldb work

RhodiumToad: ok, that does

so what did you try that didn't work

so I start `lldb -- ./zen ... args`

source list -> nothing

but `list main` does

before you actually run the program there's no current location to list

sure, but I get that result even after starting the prog

starting it how?

Title: Snippet | IRCCloud

ok so this is unexpected

the first `source list` is blank

you're still in the startup code

but if I do it again, we're suddenly showing L524

so I guess there's some libc init stuff that I *don't* see beforehand?

so I should breakpoint main, run it

then try to set more breakpoints

yes

TIL all the things

with shared libs there's a _lot_ of work done before main

since it's the rtld that's responsible for finding all the shared libs and linking them into the program

huh

there is a "gui" command in lldb

this is pretty nice

so in the end, 1 error, CFLAGS being clobbered, and 1 unexpected behaviour, need to get to main before source works

RhodiumToad: thanks for the rtld(1) info, I'll look at that later

if you have the base system -dbg files installed, you should be able to see the ld-elf.so sources

Freaky: that looks suspiciously like compile time, and not run-time.

moritz.systems/blog/lldb-support-for-fork-and-vfork how convenient

Title: LLDB support for fork(2) and vfork(2) | Moritz Systems

> posix_spawn(3) triggers events depending on the implementation. On NetBSD, it is reported as a single event. On the two other platforms, it is reported as the underlying vfork(2) and exec(3) calls.

that's interesting

I just used it, feeling very proud of my new skillz

ah, ok, I see the table now

meena: the #if's are compile-time, the __builtin_cpu_supports() are run-time

Freaky: aye

meena: as evidenced by the same binary working on this Zen 3 and an old K8 I have lying about

Freaky: consequence of me reading commits every morning: reviews.freebsd.org/D41920

Title: ⚙ D41920 local-unbound: remove stray config.h.in backup

basically, this is your fault.

meena: I like how 3 people were asked to review it

you know, just in case someone liked it

Freaky: i asked all the people who have touched that code in the last two years lol