anyone else upgrade to 13.2-RELEASE in a digitalocean droplet and now more than once on a reboot you have to access the recovery console to manually `service digitaloceanpre start` and `service digitalocean start` to finish the startup procedures, which include getting an IP address (and therefore normal SSH access)?

might be an easy fix, but all i know is i would think the /usr/local/etc/rc.d/ files would all be run on their own. i see they both require `cloudinit` which i don't see anywhere .. hmmm

hm.. i wonder if this is a `pkg install py39-cloud-init` away... though it's odd that this issue didn't exist before, and i sort of think cloud-init and cloudinit are likely different

i can only guess that it's the case that this is a result of DO no longer officially supporting FreeBSD, so they don't have some sort of injection of the cloud-init package into their alternative base system, or something like that.

DigitalOcean are not actively supporting FreeBSD anymore AFAIK.

they aren't. i suspect they had a way to install cloud-init automatically, a way that only went away in 13.2-RELEASE, which is a little confusing because they stopped supporting FreeBSD before 13.1-RELEASE IIRC

hi

meena: ^

ARG i think i'm having wifi + ssh IPQOS issues again. i keep logging in and then rapidly hanging and disconnecting. what's the right qos flags to stop this madness?

scoobybejesus: i haven't tested digitalocean lately, but yes, aside from no longer supporting cloud-init, they also stopped supporting the digitalocean data source

scoobybejesus: see canonical/cloud-init #4130 / canonical/cloud-init #4172

Title: Move DigitalOcean before ConfigDrive datasource in default list by dermotbradley · Pull Request #4130 · canonical/cloud-init · GitHub

scoobybejesus: I removed the digitalocean services long ago and just put the network conf in rc.conf because it wasn't reliable

Does anyone know if vt|stty changed something or if vim 9 did as in console "Backspace" key works as "Delete"? FWIW "stty" shows "erase" to be "^?". "Backspace" key in (n)vi seems to be working as expected

I think I saw that on some mailing list c 2019-2022 (perhaps?) but a search for the issue bring up nothing that recent

parv: you think it could be a regression?

meena, I really do not know

I ought to compile+test vim 8 version

Oh crap! User error seems as "vim -u NONE" has the "Backspace" key working as expected *ugh* (also relief)

parv: good thing you tried that before going off compiling a pile of code

meena, Thanks for the prodding

On that note, later ...

Freshly installed FreeBSD 14-CURRENT, when i install pkg, i get "Newer FreeBSD version for package pkg" -> termbin.com/s6ve

does this mean there has been a new nightly? or 14-CURRENT version of the kernel released while i was installing my system?

if so what should i do here?

you should probably be building it from ports. to me it looks like you are installing a prebuilt pkg

current will most likely always be ahead of packages that are built for that system.

Ignore the mismatch and continue? [y/N]: Y

if it does not work then (cd /usr/ports/[...]/pkg && make install)

handbook will cover updating ports and packages

/usr/ports/ports-mgmt/pkg

CmdLnKid: That's much clear now. Thank you. Will do, i will try to build from ports, since the main reaosn to install 14 was to try as much as i can debug PR 271938.

271938 – Excessive heating on Lenovo Thinkpad E450 even with CPU @ 100% idle bugs.freebsd.org/bugzilla/show_bug.cgi?id=271938

sozuba, good luck man

thanks :)

added info pciconf -lv may lead you to the right path to load some modules that may tie into your heating problem

notice lines that start with "none0@"

where loadable modules may be available in /boot/{modules,kernel}/

id suggest looking at i2c and smbus mods

CmdLnKid: should i add the output of pciconf -lv to the PR? is that what you meant?

you can yes that would probably help but inspecting that output for you may lead you to solving that problem yourself

ive always found ibm derived hardware has quite a bit of complexities to solve by loading modules

and yes, I believe that this oculd be the GPU heating that's affecting everything, including the GPU. But i am just a noob and have no experience sprting such issues. Anyway,nothing wrong with that approach right? :D

nope

its just how we all learn

hrm. so wireshark does show tcp retransmits during ssh session, so maybe its not qos.

but... hanging for up to a minute at a time. wonder what i can tune to make that retry faster

demosthenex, ifconfig IF -rxcsum -txcsum possibly

possibly also on lo0

but unfortunately i couldn;t find the right tool to output my GPU temp like in linux. I am still searching. One of the issues in testing this was having both i915kms and amdgpu loaded in kld_list, which caused a pacnic and crash, as in PR 271897. Follwoing a comment that this is drm-610-kmod issues and the necessary commits is already implemented and is available for packages built for 14

If you search freebsd tuning you'll get a nice website that does some nice network tuning

If you are on irc you gfx card is fine

CmdLnKid: so its bad wifi, it's likely legit that i'm having retransmits

yeah thats def another thing to evaluate

Buy another device test and return

move the access point across the yard

;-)

Buy a wifi range tester channels as well

i had a cable, puppy broke it. wifi thru 2 ww2 quality brick/plaster walls

My router has a spectrum analyzer and move the wifi to a non congested channel

Or do what info and don't use wifi

Must be a strong puppy

i jsut wanted to make ssh a bit more resilient, i'm ok if it pauses a few seconds, but whole minutes looking like it terminated is a problem

no just teething

Puppy power

indeed

demosthenex, lol yeah that would be frustrating

Best way to fix a broken cable is repass or those telephone snap thingies

demosthenes.org/tmp/bjorn/20230607.jpg can't be mad at that face

i'm already purchasing replacement cables and finding a new cable route

My dog chewed mine twice

in the mean time, i'm on wifi. i'm just trying to find some tunables to improve the behavior a bit

Just patches it

as long as you have the tools

I do networking as a hobby so I have all the tools

use some pvc pipe to contain it

Or get the good stuff with conduit and use for outdoors

Makes the puppy chew it and still runs

Or you can paint it with a tubing bitter

spray it with some nasty tasting stuff like jalepeno juice

;-)

Yeah

unfortunately this brick flat has no cable runs, i can't access any plenums, the walls are solid, and i can't even find a gap along the steam pipes between rooms

Stop bitching and get to working

rofl

impact drill will work

so i'm laying cable under my carpet tiles to a door, under the door (replacing that with a flat cable due to door dmg), up a set of stairs, and over a wooden divider. that's where he caught it

In that case I would use wifi

Hah

its a 30m cable, because the ISP entrance is on the lower floor :P

your dog loves to watch you work

Put wifi on third floor whole house get some decent range and speed however I did some analyze and chose channels no one was using

indeed

yep, i did channel analysis, np there.

but 5g can't pass thru the walls, only limited 2g

Router is in my room pcs are in my room

i had a powerline ethernet repeater,but its not reliable and overheats ;]

yikes. im in a 110 occupancy complex, wifi here is pretty much shit all the time.

I tried that and got dialup speeds

Oh I own my home so yeah

i may go RV in the not so long future and become the traveling network systems guy

hate being locked down to a single place

thankfully i never thought id have a wife that would be onboard with that ... turns out she's the one heading it up now

ofcourse on the side of the RV ... its going to say "Powered by FreeBSD"

hehe nice

That's stupid

Should be GasBSD

hahahaha maybe if the FreeBSD foundation wants to get into fossil fuels

You didn't get it

Wtf man

lol

Freebsd foundation would .. well let's not go there

CmdLnKid: some trivia for you

colonel sanders in the kitchen with the tea bag!

lol

Whatever floats ya boat man

KFC was good as a kid now it's garbage

too much optimizing

how can i make chicken taste even more like chicken

Add more chicken

To the chicken

and then spray it with chicken

thats a sad definition of the linux kernel

And slap it with chicken paste and rub

Yeah reading git commits logs from Linux is funny

But it still has better hw support so I still use it

<- slackware since '94 soooooo it'll most likely always be in my toolchain

Waiting on some things for a new machine

Have Slackware in bhyve on my list

Still need a nve and a y cable for two ssds in a zfs raid1

virtualbox here atop macos

all other serious equipment has freebsd

Apple ...

yeah yeah, great client when compared to ms

I like Ms it's my main os

Or make doom eternal work in wine

with bsd behind the scenes

Userland

I have a iMac that sadly will need to be stuck on the current or replaced

As the new os dropps support for 2017 models

I blame CmdLnKid

lol

hackintosh that shit

its worthwhile especially if it includes iTerm2

Mom said she would buy a new one

So I said ok

Then I said enjoy your aeg

My what...

lol

Got this Lenovo for 299

Main os drove will be nve with 13.2 and backup will be 2x500 in whatever the equi a raid 10 is in zfs 32 gb of rams and a y cable for the extra drives

And it should fit in a book in my closet

nook

hi

Sup

Mr Blue Sky

Is it just me or does the latest Vagrant box not work? - 'freebsd/FreeBSD-14.0-CURRENT' version '2023.06.15

sorry havent been a vagrant fan

Yeah no

Yeah still no

$$ may change that though

🤔

hahaha

*, why use vagrant on BSD instead of jails? Porting?

SamuelMarks: good luck with that

avlaxis - I'm having fun supporting everything from MS-DOS to FreeBSD to SunOS. My host changes, but currently I'm on Linux

avlaxis, cause it has very good repeatable results

great for devops type of shit or web designers trying to understand what they are deploying on

docker is more popular than vagrant

why use less popular tool? convenience? familiarity?

jails is probably the least known but it's been around forever and just works

Familiar

id rather deploy on docker if needed but vagrant though

everyone has a niche tho

There are FreeBSD docker images?

Web host just added support to docker

I was looking at some other Hashicorp product a week or two ago and gave up part way through due to support tickets flying in

Boundary that's it

supposedly some ztna secure access solution

freebsd/FreeBSD-13.2-RELEASE failed also

OpenBSD I just tried; works like a charm

as always :D

this is biting me again

ld-elf.so.1: /usr/local/lib/gcc10/libstdc++.so.6: version GLIBCXX_3.4.30 required by /downloads/tailwind-freebsd-x64 not found

ldd says libstdc++.so.6 => /usr/local/lib/gcc10/libstdc++.so.6 (0x15c663fdc000)

I'm guessing I want to re-order ldconfig

to use gcc12 instead of 10

and I should use /etc/ld-elf.so.conf for this

you're running a binary compiled by someone else?

again you are not reading the error message

how so?

pastebin the following since people get upset when you post it here

file taild-wind...

and ldd tailwind..

the problem seems quite clear

ldd says it's loading gcc10/libstdc++.so.6

gcc10/libstdc++.so.6 does not contain a version GLIBCXX_3.4.30

(because that version is from gcc12+)

ah

this implies the code was linked against gcc12's libstdc++

I was waiting for dch rto reply not you

so yeah he had it right

Title: Snippet | IRCCloud

so the issue is I'm building this cursed tailwind nodejs abomination for other people

and telling them install gcc12 you should be fine

and you would be correct

but on *my* system I have gcc10 as well

did you use -Wl,-rpath=/usr/local/lib/gcc12

however getting others to install something is hard

and I don't know yet how to order ths

RhodiumToad: sadly it is not possible to build tailwind's embedded nodejs stack with that

(as specified in the install message for gcc12)

why not?

its truly an abominations

because npm is an enormous pile of shit

that downloads all of nodejs, builds it again with custom flags

tried getting ghost blog to work

and it is not practical (in terms of my time and frustration levels) to build it and figure out where and how in the cascading nest of downloading garbage

have to hack the hell out of it to make it work

to find the precise point to patch it

yeah

gave up after editing it 5 times

however its a fundamental tool for my startup, so ...

either fix this or use linux

what I'm not clear on are 2 things

1. what happens when gcc x or y version is installed, wrt the ldconfig settings

2. can I control this order, using /etc/ld-elf.so.conf somehow?

what exactly is this you are wanting to install ?

I want tailwind-freebsd-x64 to just run without complaining about glib

the overall project is a web project with esbuild + tailwind ui + a pile of other stuff

was looking morwe towards "this app name and link to website"

there is no site to linjk to yet, its a bespoke app at this m

*point

your typing is worse than mine, please get off the PC, take a coffee break and come back, thank you

cpet: whatever. I have a 2 inch cut in my middle finger here. I have made ... 2 typos in a paragraph.

dch: what happens when you install gccN is that it creates a file /usr/local/libdata/ldconfig/gccN

so no github link either

ldconfig looks in that dir to find entries to add to the library lookup path

cpet: I'm not being obtuse here. I just don't want to waste your time while you go from zero knowing nothing about how elixir + phoenix use and build web apps with esbuild, tailwind, and a bunch of other stuff.

Title: Releases · tailwindlabs/tailwindcss · GitHub

cpet: I can build tailwind on freebsd, as this weird packaged binary, but as soon as there's multiple gcc available, it doesn't run. And I'm hoping to fix that

cpet: please, don't waste your time here. you will simply end up back staring at the issues on github that I have raised there. full circle.

it looks like the rc.d/ldconfig script sorts the directories contained in those files when adding to ldconfig_path

RhodiumToad: so these 2 ldconfig files are just the paths that the libraries live in

wow so much for "someone got off work early today lets play with some stuff"

but ok you do you

dch: yes

cpet: my last reply. I'm not being facetious, this represents 6 months of hacking to get something a lot of people really appreciated. You're welcome to stop responding if you don't think you can offer something constructive.

dch: stopped caring after you replied with "dont waste your time"

dch: I don't see any obvious way of influencing the order

I was hoping that it would add gcc12 first, then gcc10 in the hope that the newer version has the right bits

what it actually does is this:

_files=`find ${i} -type f; if [ -n "${_files}" ]; then ldconfig_paths="${ldconfig_paths} `cat ${_files} | sort -u`"

in /etc/rc.d/ldconfig it just loops over the dirs and does a `sort u`

so the resulting list will be ordered as by a text sort of the resulting paths

I will see what `${ldconfig} -elf ${_ins} ${_LDC}` prints

I guess you could put this in rc.conf:

ldconfig_paths="${ldconfig_paths} /usr/local/lib/gcc12"

so it ends up executing `/sbin/ldconfig -elf /lib /usr/lib /usr/lib/compat /usr/local/lib /usr/local/lib/compat/pkg /usr/local/ffmpeg4/lib /usr/local/lib/alsa-lib /usr/local/lib/compat/pkg /usr/local/lib/gcc10 /usr/local/lib/gcc12 /usr/local/lib/mysql /usr/local/lib/mysql/plugin /usr/local/lib/perl5/5.32/mach/CORE /usr/local/lib/qt5 /usr/local/lib/qt6 /usr/local/llvm11/lib /usr/local/llvm14/lib /usr/local/llvm15/lib /usr/local/llvm16/lib

/usr/local/share/code-oss /usr/local/share/ungoogled-chromium`

which as expected sorts gcc10 before gcc12

and if I swap over gcc10/gcc12 there, run ldconfig again with that

it works

cool

that'll work until the next thing does service ldconfig restart

so I can see about reordering that a bit

which most package installs do

yeah exactly

but at least my understanding has advanced

maybe safer to do ldconfig_paths="${ldconfig_paths} /usr/local/lib /usr/local/lib/gcc12"

oh, not needed

I tried btw `patchelf --set-rpath /usr/local/lib/gcc12 ...` but this breaks the nodejs embedded garbage sadly

maybe it has other rpaths already?

so you'd need to add the new one to the list rather than replacing what's already there

maybe theres a tool I can use to dump this stuff. unfamiliar territory for me.

readelf or objdump

^ thanks RhodiumToad

no thank you for not allowing me to waste my time

cool so wombling through rtld(1) shows me I can use `env LD_LIBRARY_PATH=/usr/local/lib/gcc12 /downloads/tailwind-freebsd-x64 -h`

this is not ideal but is at least usable for end-users of tailwind

could of told you that if you would of 1. allow me to find what you were using and 2. give me a stie to download it or src instead of "dont waste your time"

LD_LIBRARY_PATH has its drawbacks, but yes that is one solution

Title: Snippet | IRCCloud

AFAICT that is the last build (at sourcehut) that produced a working artefact

I tried many different ways unsuccessfully to convince tailwind to either build node statically, or at least to include the specific rpath into the build

and in version 3.30+ upstream tailwind switched build system again so I gave up

uh

or install a linux distro in bhyve run that, run what ever dumb programs arent that well portable

and boom

if you're putting your own cc in ~/bin to make the build use it, you could put a script in there that invoked gcc12 -Wl,-rpath=foo "$@"

I can list the various things I tried, maybe there is something I missed

or port it to clang and stop using gcc ;/

thank you meena, vkarlsen, and grahamperrin. i will probably check those DO init scripts for anything interesting and then just put it directly in rc.conf since the provisioning details should never change anyway

for nodejs in general we can build with `if (targetPlatform === 'freebsd') { args.push('--fully-static'); args.push('--enable-static');` but thats insufficient here

also tried `export LDFLAGS='-Wl,-rpath=/usr/local/lib/gcc12'`

which is supposed to do what RhodiumToad proposed above, I will your trick tonight

scoobybejesus: that's not entirely true. Azure recently switched to having the network config run on every boot

ah, good to note, thank you again

Azure still supports a BSD ?

putting LDFLAGS in the environment is still subject to the vagaries of whatever the build decides to do

hence why I suggested putting it in the script

cpet: FreeBSD, at least

that is a better solution

maybe thats possible to get it to build statically even that way

if only it didnt take an hour ++ to build each time

ok to recap what I learned, assuming I get it right

- installing both gcc10 + gcc12 causes gcc10 to come before gcc12 in ldconfig because the rc.d script has `sort -u` in there

- it seems unwise to fiddle with something so fundamental to many programs just to fix on1 annoying nodejs binary

- the LD path is only required for binaries that have not got rpath embedded inside

- ports tree obviously handles this already for us

(at compile time)

ports tree does indeed try and handle that as automatically as possible

setting an env var is a terrible workaround but may do as a last resort

- best option is to fix the build (will try RT's trick later)

- next best is to post-build try to splice in the rpath

when I try using brandelf, we get this weird output

Title: Snippet | IRCCloud

AFAICT this vercel/pkg script, which is used to both build a custom version of nodejs, and to append the bytecode of the globbed js files on the end, doesn't like the size changing unexpectedly

there are good ways to embed bytecode in executables that don't rely on assumptions about the size of the file, I guess it's not using those

the method I've used before is ld -r -b binary -o data.o somedatafile

yeah

it's possible to do something similar with objcopy

nice yes

Hello all, quick question: I bought a 7900XTX and I'm trying to get X to work. I installed drm-kmod, did "sysrc kld_list+=amdgpu" and as per the handbook, I specified the driver in /usr/local/etc/X11/xorg.conf.d. All I get is a mouse pointer in the center of the screen but no X, any ideas would be greatly appreciated

indeed, thats how this was worked around for nixos

Title: now-cli: init at 11.4.6 by brendan-hall · Pull Request #48193 · NixOS/nixpkgs · GitHub

Title: now-cli: init at 11.4.6 by brendan-hall · Pull Request #48193 · NixOS/nixpkgs · GitHub

wxsabi: how are you starting X, and what's in the logs?

RhodiumToad: I set sddm to load in /etc/rc.conf(this was working with my old 3090ti). the logs say "No screen found"

pastebin the logs

(trying to load X in rc.conf is a pet peeve of mine; it does NOT belong there. the right place is /etc/ttys)

(and if sddm doesn't work in /etc/ttys, that's a bug)

RhodiumToad: I just noticed it also says "Failed to load module radeon. Driver does not exist". But lemme see about pastebin the logs. I'm not used to the console so it's gonna take me a minute

oh, try removing whatever you put in xorg.conf.d

RhodiumToad: ok

for drm-kmod drivers, autodetection should be sufficient.

RhodiumToad: now it's saying that it cannot run in framebuffer mode, I double checked and drm-kmod is indeed installed. I'm checking /etc/rc.conf, is it supposed to be kld_list="amdgpu" or kld_list=amdgpu?

and the simplest solution, if nodejs allows it, is to build this against gcc10 from quarterly

wxsabi: the quotes are usual

wxsabi: technically it's shell syntax so an unquoted single word is ok, but it's discouraged

wxsabi: pastebin the logs?

is sddm itself not starting, or is it failing after you log in?

wxsabi: is this current by chance ?

RhodiumToad: pastebin.com/uLr0QQh2

Title: GNU nano 7.22940.895]Ivar /log/Xorg.0. logX.Org X Server 1.21.1.7X Proto - Pastebin.com

cpet: 13.2-RELEASE

install xf86-video-ati

wxsabi: is /dev/dri empty?

RhodiumToad: sorry for the delay, pastebin was difficult for me. sddm is not starting that I can tell, all I get is a mouse pointer but the rest remains in console mode if that makes sense

for some reason I get this when I have two vid cards

and works fine when I take one out

wxsabi: easier to use e.g. termbin.com using cat file | nc termbin.com 9999

2940.898J (WW) VGA arbiter: cannot open kernel arbiter, no multi-card support

dpaste.org can be pasted to using curl

would have to force a BusID

or change the topic to one that works

wxsabi: do you have multiple cards?

RhodiumToad: /dev/dri no such file or directory

so y'all can cry at how bad my hack is, but irccloud.com/pastebin/CQ7wSWYy/hexdump is the nodejs code that handles unpacking stuff

Title: Snippet | IRCCloud

as expected its not obj/elf format aware at all

RhodiumToad: not multiple cards, I used to have a 3090 ti, sold it and now I have an AMD 7900XTX

wxsabi: amdgpu loaded ?

wxsabi: what (if any) messages were displayed on the console when loading amdgpu?

card is rather new so good luck with it

and calculating length of both payload and prelude before/after fixing rpath, its 4096 bytes extra

using textproc/bbe from ports (a binary-like sed)

bbe -e 's/44141048/44145144/' -e 's/47838014/47842110/' tailwind-freebsd-x64 > tailwind-abomination

"fixes" this

RhodiumToad: seems to like to ask the same crap I ask for so ill just go back to fixing shell issues

cpet: I added it to /etc/rc.conf, is there a way to check that it DID load?

I will try recompiling with the suggestions above tonight, maybe I can produce a cleaner binary

wxsabi: you rebooted after? take a look at the output of dmesg

RhodiumToad: Is this one of the messages that appear during boot?

RhodiumToad: yes, I rebooted

paste the dmesg output to termbin

thanks for helping me work through this one

dch: yw

RhodiumToad: termbin.com/trche

wxsabi: so i don't see anything there from drm

RhodiumToad: so it's not loading?

oh wait, here it is: [drm] amdgpu kernel modesetting enabled.

is there anything in /dev that belongs to the video group?

RhodiumToad: lemme see

RhodiumToad: this is what ls /dev returns htpps://termbin.com/cvkg

find /dev -group video -ls

RhodiumToad: https:termbin.com/1zj1

RhodiumToad: termbin.com/1zj1

what user is sddm trying to run as?

is xf86-video-amdgpu installed? if not, install it

RhodiumToad: good question... my guess is that as root since it's in /etc/rc.conf

RhodiumToad: ok, I don't think it's installed

RhodiumToad:ok, it's installed, should I reboot and test?

ya

ok

hello everyone, i'm new to freebsd and ports collection and wanting to compile emacs, do i need to uninstall the existing binary one first?

crzjp: Nope.

you don't need to uninstall it before building, but you will want to uninstall it before installing the one you built

thanks for the responses, ill compile it now!

and in my final terrible incantation:

perl -pe 's/(var PAYLOAD_POSITION = .)(\d+)/$1 . ($2 + 4096)/e; s/(var PRELUDE_POSITION = .)(\\d+)/$1 . ($2 + 4096)/e;' < tailwind-freebsd-x64 > tailwind-abomination

I can't think of another tool that would let me combine regex substitution with eval'd code without shame

RhodiumToad: still no luck, but, should I add the conf file to /usr/local/etc/X11 as per the handbook?

RhodiumToad: we deleted it a little bit ago

wxsabi: shouldn't need it.

wxsabi: do you see the display mode change during boot?

also what;s the output of pciconf -lv

RhodiumToad: termbin.com/cjoi

RhodiumToad: no change that I can see

RhodiumToad: what baffles me is that there is a mouse pointer, doesn't that mean that at least some level of graphics is working? I can move it and select text

did you check whether x is running on another vt? use alt-Fn on the console to switch

mouse pointer works in text mode if moused is active

RhodiumToad: x is not running

I have to go for a while, I will be back later (hour or two)

RhodiumToad: sure thing, I greatly appreciate your help!

back

wxsabi: any progress?

RhodiumToad: nope, I tried Xorg -configure but that created a huge config file that didn't work, so I deleted it

RhodiumToad: I wonder if the card is just not supported? my old Nvidia card "just worked" following the instructions in the handbook

you said fbsd 13.2 ?

what does kldlist -v show

and what exact kms package versions are installed

RhodiumToad: yeah, fbsd 13.2-RELEASE, kldlist -v says command not found

kldstat -v sorry

what command can i use for kms check?

probably pkg info -d drm-kmod should work

that displays the dependencies of the drm-kmod package (which is a meta-package that just installs other stuff)

termbin.com/mrsi

termbin.com/9u4c

RhodiumToad: i think those work without the https?

they work well enough, don't worry about it

ok. let me have a quick gleg at the source to check if your pci ids seem to be supported

RhodiumToad: thanks a lot! I really appreciate you helping me

hm.

seeing a distinct lack of mention of either 0x744c or Navi 31

looks like the GPU is too new for this drm-kmod release

I was afraid of that :-(

but it is what it is. Vesa should at least give me a desktop, right? or don't even think about it? XD

one would hope; but I'm way out of practice with fiddling with X in the absence of a usable card

ok, I'll play with it some and see what I can come up with. But thanks a lot though

wiki.freebsd.org/GraphicsOld/SCFB <-- that might be relevant

Title: GraphicsOld/SCFB - FreeBSD Wiki

(I actually got this desktop, which has the Skylake intel video, before freebsd had support for that; but I avoided that issue by ordering a low-end nvidia card with it)

Seeing as pciconf -lv shows the iGPU attaching via PCI, I don't see why it shouldn't work.

Intel GVT-g, which is what lets you pass a single iGPU to multiple VMs, doesn't work yet - but a recent commit was pushed, that at least mentioned it in passing (though didn't do anything to change whether it works or not).

I'm also very unclear how one iGPU with a set of outputs can work with multiple VMs in terms of switching outputs, but I guess it's up to the hypervisor to handle that?

It also only works for a very small subset of devices, and it was removed from the newest generation of chips, so I'm not sure working on it now is the best use of resources.

Title: Ice Lake support? · Issue #126 · intel/gvt-linux · GitHub

126 – newsyslog isn't run by default bugs.freebsd.org/bugzilla/show_bug.cgi?id=126

Confusingly, Xe seems to require SR-IOV - which is what Intel should've started using in the first place, some might argue. On the other hand, there's no signs Xe will be a very long-lived product line, so *shrug*

there was nothing indicating that amdgpu had attached the device; it showed up only as vgapci

oh you mean my GPU passthrough desire?

i am new on freebsd but already i am giving a hard time for myself. :( My system is unbootable and i use liveCD option now. i have to mount my /boot/efi which is on /dev/ada0p11 but i can not as getting Invalid filesystem. gpart show does see the partition but does not show the device path. like /dev/ada0pXX . no such geom...

what's the actual output from gpart show?

even if i use mount -t msdosfs

(if you can get any networking from the livecd then you can paste to termbin using blah | nc termbin.com 9999

RhodiumToad: it does list all the 13 partitions.

and index 11 shows as type "efi" ?

file -s /dev/ada0p11 shows what?

but i have diskID and label not showing the /dev/ada0__XX. it should show ada0p11

I can't interpret what you mean by that last

the thing is i was installing it as manual zfs guide and having efi on p11 partition and freebsd on p12 partition. i have other efi partitions also for windows and linux systems. so i dont want create new partition table with gpart.

you didn't already have an efi partition?

(you can't have more than one on a disk)

yes. it does show type efi

i always had three efi. one for windows and one for linux systems (they are sharing) and one for freebsd. it was working well

ada0p3 ada0p4 and ada0p11 are efi. i dont use grub for windows and freebsd but dedicated entry in BIOS. thats why.

so how can i mount ada0p11 ?

file -s /dev/ada0p11 shows what?

can not open (no such file or directory)

do you have any networking enabled in the livecd?

i want to reinstall bootcode but first i should mount it somehow...

RhodiumToad: uh...no

i was importing my pool called root but no network

can you do so without too much trouble?

i can. but i sont know how...

do you have a wired connection or only wireless?

netif?

wireless

eh. too much trouble

but i can get wired if really neccessary...

it'll be a lot more convenient

usually all it takes is plugging it in and doing dhclient re0 or whatever the interface name is

(assuming you have dhcp going already)

ok then. just a bit complicated as i have mobilewifi with tplink M7650 and the router is on bridge mode....but it does work. i did use it on arch-chroot also and was working. just i need few mins to set it up

sorry for trouble

ok. i have pluged ethernet cable now. but after service netif start still cant ping

service netif start won't do anything unless you have interfaces defined in rc.conf

probably easier to just use dhclient

on liveCD how to check /etc/rc.conf ?

cat /etc/rc.conf

it may not exist

yep

no exist

what's the network interface name?

alc0

and if you do dhclient alc0 what happens?

DHCPDISCOVER starts. but i cant get IP

how to list all interfaces?

ifconfig -a

you might need ifconfig alc0 up

no working leases

just double check. i have only two interface. alc0 and lo0. so alc0 should be the one.

but ping gives no route to host.....

if dhcp didn't work that's not surprising

did you try dhclient again after doing ifconfig alc0 up ?

and etc/resolv.conf also does not exist....but i have the pool...

let me try again

how to restart dhclient? service routing restart??

if you started it manually, then kill it and start it again

don't mess with the service commands unless you have a fully filled out rc.conf

debdrup: i hope I live for a long time :(

now DHCPDISCOVER switching the interface up and down....weird

no DHCPOFFERS received...

you're sure it's plugged in on a network with working dhcp?

yes

netstat -in shows any non-zero counters for the interface?

let me check

no. doesnt

not even for Opkts?

opkts 63

and Ipkts 0?

189

so my guess is no dhcp server

I was gonna say, can you try a different dhcp client, but, not getting any network might make installing packages a bit difficult

not trying to install packages, just want a way to get output onto termbin

might make everything difficult

could you configure a manual IP on it?

i have got issue with this when my system was healthy. as i could not get proper dhcp values. as i have two wifi network. one is huawei with 192.168.8.x and the other one is 192.168.0.x range. so all the time i had uncomment wpasupplicant to get proper dhcp...

right but this isn't wifi

but wired never tried. this is the first time.

i know...

does arp -an show anything?

but dhcp should be the same. 192.168.0.1 should be on resolv.conf. but cant check it...

arp -an no output

i donno what is that btw

you could try something like ifconfig alc0 192.168.0.123/24 up

192.168.0.126. the mobile wifi router was giving ip already. just freebsd does not get it...

i was checking just now...

linkstate ether media: Ethernet autoselect