nimaje: Taylor Swift has must disdain with black spaces too.

bugs.freebsd.org/bugzilla/show_bug.cgi?id=289220 still trying to figure out exactly why my FreeBSD only wants to boot up normal after a successful boot and "poweroff" command from Linux. Yeah, it won't even boot back up without hanging when doing the same 'poweroff" command from FreeBSD itself LOL

i'm curious, is there a way of modifying a gpt partitions index/partition number?

I have some disks that I used to habve 3 paritions on. now there are only two. but the third partition is still p3. I don't see something to do it at a quick glance of man gpart. not important but just .. curious :)

duskmoss: sounds like an interesting way to fool around and lose data

feel free to not answer if you don't know. but also I don't have data I could lose by losing one drive.

duskmoss: gdisk has that option. From the manpage: "s Sort partition entries. GPT partition numbers need not match the order of partitions on the disk. If you want them to match, you can use this option."

Consider the consequences!

im seeing some old code i have in a bsdinstall installerconfig, service bridge restart, but nowhere else do i do anything to configure a bridge0 interface. so what does a bare "service bridge restart" do?

SponiX: maybe a stupid idea, but have you tried to load the default UEFI/BIOS settings?

as already stated it looks like a problem with inter processor interrupt delivery

hyperthreading on/off **shouldn't** influence if the system starts or not

the usual suspects in such situations are buggy ACPI tables or byte code

crest: Yes, I've tried resetting to factory defaults in the UEFI. Someone else recommended trying Legacy BIOS booting a few times for giggles - I haven't tried this yet. It is the latest UEFI version for the board. I am going to search around for possible community written UEFI/BIOS also though.

duskmoss: sorry for being a smart ass. sometimes I just can't help myself. these are regular FreeBSD UFS formatted partitions? Not ZFS?

my old x99 board worked like a charm, but i don't have anything similar to test on

does it always boot with SMP off (not that it's useful in that configuration)

?

if so you may be able to dump the acpi tables in both situations if you feel like investing time and effort into the problem

crest: I will try booting it a few times with only SMP disabled in /boot/loader.conf -- I am fairly sure that will work consistently

so it looks like just running service bridge restart autocreates a bridge0 interface and adds the interface with the public ip as a member? how does it know which interface is the right 1 to add?

BUT, I also though just having Hyper Treading disabled would do the trick too for a while, and then I found out that wasn't the case either :P

in that case my recommendation would be to look for changes in the ACPI tables you can dump

or if you feel like shotgun debugging try disabling ACPI subsystems via loader tunables

i had a server that has a very annoying ACPI thermal zone bug

it will always complain about a failed cooling zone when that zone has no active cooling

crest: well that is the thing, if I turn off ACPI from the boot menu, the system won't even go past the boot loader LOL

disabling that took care of it spamming the dmesg output every 5 seconds

SponiX: that's to be expected

ACPI is required for SMP systems

you can only disable ACPI and boot if you have a single core system or disable SMP via loader tunables

because ACPI tables are required to setup the other CPUs interrupt controllers

so these days the option of disabling ACPI is kind of a vestigial feature in the loader

will I get a performance boost if I use ports instead of packages on i9?

lessless: potentially, it will require setting up a build system and then test and optimize.

lessless: by default when you build a pkg from ports it is exactly the same as the binary ones already provided. You can tweak build flags specific to your processor though when building ports for a slight performance boost. This is generally not recommended though, as it can break some things from building

sacrifice time / energy to build a package or just download the pkg and focus on other things, depends on your objective.

what SponiX also said

the performance gains are normally not very large. If I was building from ports it would be to get a newer version of something, probably for new app features or security reasons

yeah, i have recently foudn the best example is llama.cpp the amount of times they update this program.. the ports person can not just keep up

i would love to find volunteers to get numpy to freaking build on FreeBSD :) or jq package.. because python is just a pain to build out on freebsd

Yeah, I have had that for qBittorrent at times, I run it 24/7 and always want the latest version. So I build it from ports when possible, and also at times edit the port build Makefile and so on to do an even newer version than the port is ready to support LOL

I build my own packages, because I change options, some of them will understandably not become defaults, so …

yea another good point

you can do your magic lessless , just some points to consider.. if these work for you.. knock yourself out

nimaje: that is for sure another good reason. One that completely slipped my mind

Or remove options, although one should be careful that another package doesn't depend on those options

Wanted to install nut, but it tries to pull in avahi and whatnot, 200Mb of dependencies. Will go for a stripped down build in the local poudriere...

okay, so it's not worth it. fortunatelly, I didn't go to far with this install, so wiping it out is a 30 mins job

I find going down the ports rabbit hole can be tedious. I doubt there would be a massive performance increase though. Usually it’s done for adding options that weren’t default in the pkg bins.

Like znc used to not have python support and could only be added using ports as an option.

I think nowadays it does by default though.

Macer: thanks, it's good the understand the ports purpose

im seeing some old code i have in a bsdinstall installerconfig, service bridge restart, but nowhere else do i do anything to configure a bridge0 interface. so what does a bare "service bridge restart" do?

so it looks like just running service bridge restart autocreates a bridge0 interface and adds the interface with the public ip as a member? how does it know which interface is the right 1 to add?

did anyone had noticed that rust 1.89 is building longer than usual?

it's almost been 3 h

it seems that build time reduces as much as you compile the same version (ccache)

its packaging now :-)

rust is a massive beast

i'm procrastinating here

on rust and go

ketas: yeah but there are more massive ones like electron, chromium, webengine

ketas: do you want to working on these langs?

firefox?

oh, its no big deal, 1h for me

llvm

what were the biggest ports

cmake too?

nevermind trying that on older hw

it works tho

i also have armv7 qemu running where i observe 2 days of cmake but that's extreme host too

and no i don't want :p

cmake is taking under 10min for me, i don't think its big for kinda recent hw

termbin.com/bcbc here, logs of llvm19 and firefox

build time reduces with ccache as much as you build the same version

someone was battling in mailing list with cached build

im grateful that i just only need one llvm version: 19. in the past i have to had build multiple llvm versions

i mean it's fast

until it breaks

i made it break after using same ccache dir for both amd64 and i386 jails, seperated them, no problems but i dont use i386 jail since a while, idk

% du -shA /ccache

37G /ccache

and it needs some disk space :p

what is your full buildworld buildkernel?

oh, i am using a release version

it's under 10h at c2d

i did use stable and current in the past but yeah it takes some time too but reduces with meta mode and ccache

and old box

:p

an

i have r3 4300ge 4c8t and 16gb ram

ketas: why would you procrastinate? building them or learning them?

i guess building

oh yes building

i sometimes do that for learning too

eg i only now tested poudriere out

seems like i don't need to patch it

im using rust for a few years now and love it. i know it's trendy but it's actually great

builds and runs great on fbsd too

hi, I flashed freebsd on a sdcard for my pi3 and when I try to setup pkg (or do anything related to openssl like using fetch), I get these errors paste.opensuse.org/pastes/74aa6f4a0f0a is that already known? or is that specific to my install?

rrahl0: what's the date/time on the rpi?

divlamir: hm +17h so that's wrong :S

divlamir: thanks that fixed it :)

Enable ntpd, the rpi has no bat to keep the clock

yeah, just did that and initiated a sync