lw, you around?

Hi, configuring a new workstation, and the same problem as usual arises. The NVIDIA NVS290 card only works for the first monitor of the dual head. Any tips? I have installed the correct driver for the NVS290 and it works without any issue with one monitor, the other one is empty

did you try to run nvidia-xconfig?

should not be necessary but...

but more importantly, don't forget to use xrandr do acutally "enable" or "setup" the 2nd screen

I can not use the generated xorg.conf, it says use the -ignoreABI option

look at xrandr -q

is that a flag to add to xorg.conf

you wanna have something like this:

xrandr --output DP-4 --primary

xrandr --output DP-0 --left-of DP-4 --rotate left --mode 2560x1440

is https

I am not sure it is actually detecting DP-0 and DP-1

xrandr -q does not seem to detect the screens

adjust DP-0 and DP-1 in my example for whatever xrandr -q shows you

check the bpa.st link, the problem is that it does not show any DP

is the display actually plugged in and turned on and set to the correct input source?

yes it is

it is an old display from Dell, only one VGA connector, everything is connected, I have used this card in other workstations

could it be bios settings? like I have to disable the internal video adapter?

I don't think so

did you check whether the nvidia driver you're using is actually supporting the video card?

I check it carefully

it is the 340

the nvidia web page indicates that is the driver supporting the NVS290

aye, I don't have any recent experience with < 500 drivers, sry

is there any way to output to a file the current xorg.conf ? (there is none, so it is generated on the fly by xorg I guess)

my idea would be trying to extend the screen to twice the width and see what happens, maybe the nvidia card presents both screens as a single one

I think it might be the combination of the nvidia driver and the X version :(

This server has a video driver ABI version of 25.2 that this driver does not officially support. Please check nvidia for driver updates or downgrade X

this looks like the NVS290 is no longer supported in FreeBSD14

hmm

NVS290 is from 2007

I have added an option IgnoreABI true in /etc/X11/xorg.conf/30-nvidia.conf and now it loads the driver nvidia, but still no luck, I see the log it says "nv" can not be found

ok now I used nvidia-xconfig and the generated xorg.conf works like a charm with the added 30-nvidia.conf disabled option

good, so I hope my 2007's nvidia will last for another bunch of years (I have 5 cards)

the IgnoreABI is documented as a workaround for the bug in some linux forums

l

Hello, all. Is there a way to enable/disable /etc/rc.d entries without actullly moving the files in and out of the directory?

they all have options usually

ketas, I don't know. They are services, and are started automatically, taking options from rc.conf, right? I installed powerdxx (because there is no regular powerd package) and want to disable rc.d/power_profile to avoid conflicts.

services should read a variable <name>_enabled to decide if they should do anything

that thing had none option i see

nimaje, not necessarity from rc.conf? OK. So I check $power_profile_enabled, and see it is not set, and can rest assured the power_profile service is not used.

wait

Nope: devd runs the power_profile script. It is not a service.

might need to change it

iirc no way to override

'it should' and 'it does' are diffrent things, I didn't look at it and the default setting are in /etc/defaults/rc.conf but if some _enable variable isn't set, then it should default to NO normally

I wonder why such a non-service script as power_profile is located under rc.d... It is essentially a handler of devd events, not any kind of service.

Can anybody explain: is there any technical/architercutral reason to implement a devd event-handler as a nostart service: /etc/rc.d/power_profile ? devd invokes it as `service power_profile $notify`. One advantage is not have to specify the full path...

...so: in order to "dispable" the power_profile pseudoservice one can simply rename its event handler under /etc/devd/ so that it does not end in `conf'.

well, you would also have that advantage, if it were somewhere on PATH, like service …

nimaje, You see, I am trying to defend this desing decision, but cannot...

ant-x: several non services run in rc. like networking. ifconfig and route is ran, and then... not ran again

hi folks, how you maintain security or hardening freebsd?

thank you, remiliascarlet

ant-x, AFAICT there is no need to disable power_profile when running powerdxx as powerdxx will take over and set the power values as it determines. I am simply running powerdxx and everything seems to be working well that way.

The main use of power_profile is when NOT running powerd/powerdxx and then power_profile can be used to set something like performance or economy in a fixed way.

Hi, I have a fresh FreeBSD 14.1 installation. I would like to use python virtual environments. I install py311-virtualenv, so far so good.

I use python -m venv venv and it creates the folder venv

then you are expected to use source ./venv/bin/activate but source is not available in FreeBSD's POSIX shell

any help/tip is welcomed

use .

I already tried that -previously you need to apply +x - but I see no changes in the prompt, and evn shows nothing that might lead you to thing that you are in the python environment

env*

why would you need +x? just tested I didn't need chmod +x, it should change PATH for example

does your activate has the +x already there?

oh

I was doing ./venv/bin/activate instead of . venv/bin/activate

what does the . do?

about the same as source, there are small diffrences, but I can never remember them, iirc something about were the shell looks for the file

Using "." works in sh, ksh, zsh, but "source" works in bash. Portable use is ".". (csh use is "source" but csh is incompatible syntax. But bash gets the "source" word from csh usage.)

When a file is "sourced" that means that the current shell process reads that file and sets any variables from it into the current process.

so it does not execute the file like when you do ./venv/bin/activate, it just reads the variables and set them back to the current shell

Normally when a file is executed, using the +x bit, then the current shell forks and execs a new shell process to execute it. Meaning that no variables will ever be set in the current shell since the stacked executed process will do it and then exit.

I see

". somefile" can change variables in the current shell. "./somefile" can never change variables in the current shell.

both execute whatever code is in there

indeed

?

I see

thanks for the explanation

And also "changing variables" also means current working directory too. Sourced files can change the pwd but executed files can only change it for that process until that process exits.

Inline in a shell script "dosomething" will do it in a subshell. "(dosomethign)" in parens will do something in a subshell. With the same results. Subshells evaporate when they exit and no changes can return to the parent that invoked it.

I just screwed up that explanation. "dosomething" is in the current shell maybe but (dosomething) is always in a subshell.

if "dosomething" is "VAR=foo123" then that is in the current shell. "(VAR=foo123)" is in a subshell and evaporates.

understood

This allows useful things like "tar cf - ./dir1 | (cd /tmp/dir2 && tar xvf -)" which puts the right side into a subshell which is a separate process with a separate current working directory. Which evaporates at the end of the command. The original shell prompt is still in the original directory.

That's why I never see the need for the -C DIR option that I see with tar and make. "make -C /over/there" but why not "(cd /over/there && make)" without anything more than the basic features? No idea.

well, even if you're using csh, note that the python venv actually creates an activate.csh script as well

In that case then in csh "source activate.csh" should work. :-)

well they seem safer options if you are unsure

uskerine, But no were near as much fun as commanding the full power of the system! :-)

rwp: maybe because you don't want to use a shell for everything, when spawn() or fork()+exec() is enought

But adding this -C chdir-path paradigm to each and every program is a waste because it is a duplication of effort.

It's like saying (in my mind) that the shell should stop expanding * file globs and programs should do it individually instead.

Instead we have the shell do things like expanding * file blobs and programs don't do it. And for me for the specific feature of -C chdirpath I think it should be the same. The -C is creeping featurism.

uskerine: when i need to use a python venv, i pkg install bash and create another user and give that user the bash shell

shell expansion is one of common attacks too, so cutting shell out is good

among other things

it would insane to have shell for just cd

and it would assume that human runs those commands

having cwd in program is still useful

In what way does this assume that a human is running those commands? Let me state that it does not.

File glob expansion can be used as an attempted attack. But if used correctly then there is no attack success.

Sure when I am a student at university I can try to social engineer the superuser to do something stupid with files in my home directory. But hopefully superuser is more careful than to allow such simple things.

And we will just have to disagree about -C chdir-path being a useful feature in random programs.

Hello community, I need to know for sure if the following options are enough to remove 32bit support for the entire system:

In the custom kernel configuration file:

nooption COMPAT_FREEBSD32

In the src.conf file:

WITHOUT_LIB32=yes

This is enough? And if so, will it bring stability problems to the system?

Hello community, I need to know for sure if the following options are enough to remove 32bit support for the entire system: In the custom kernel configuration file: nooption COMPAT_FREEBSD32 In the src.conf file: WITHOUT_LIB32=yes This is enough? And if so, will it bring stability problems to the system?

GeoBarRod, if you don't want 32bit support just don't install lib32.txz when bootsrapping the system

on how to remove it? i don't know but you could probably start by listing all the files in lib32.txz for the release on freebsd your on and deleting those

i ran my laptop without lib32 just fine for a year but only ran into problems when trying to install wine since most windoze program are compiled for i386 for some crazy reason.