Is anyone else getting core dump saves after turning your system back on either from a cold boot or restart on 13.1-p4? I just confirmed it's that as I did a rollback to p3 and no more coredump saves. The system acts perfectly fine during reboots; as in no appearance that a panic occured.

13.2-p4, sorry

14.0-rc1, someone broke firefox XPCOMGlueLoad error for file /usr/local/lib/firefox/libxul.so:

Shared object "libdav1d.so.6" not found, required by "libxul.so"

I just restored it from a snapshot

good morning

hello all if I wanted to dialout on 4/5g (mobile network) what type of card or adapter would I require?

or would I need a nic into some type of hardware conversion box and then dialout via mpd5

I'm using a ME906s with the u3g driver (and pppd)

Ok dokey so I would want a cellular modem, ... anyone have any suggestions for an internal one with a type N (male) connector that would be compatible with freebsd

ah so they do show up as a dialout device, perfect

now just need to find one that also has a male type-n aerial ext_port :)

huh

you really want to loose 2.5 dB in one connector ?

loose?

its just what the aerial I have specifies for its connector type

'N (female)

Hello community. In my opinion it would be great if a similar application existed in the *BSD world, I mean the new software technology "bpftune" (github.com/oracle/bpftune). Have a great day!

hm, the monitoring part should be possible via dtrace and then you have to check if the tuneables that are changed are correct; I wonder why oracle didn't use dtrace for that as their linux has dtrace support, dtrace was build for monitoring stuff and they could use it as marketing for their oracle linux

@nimaje I am convinced that software technology would be improved in the BSD world. I hope the community is encouraged and that soon we will be able to enjoy an auto-tuned system. It would be great!

So, is the intel driver coming for new gen intel 12th in freebsd 14 ? i tested in 14-RC1 and X wont start.. i guess the only solution will be sfcb

?

GeoBarRod: so are you going to contribute code?

linux has this it's open source. just copy the code /s

@Demosthenex I wish I could contribute code in C language. I have only learned to program shell scripts to better manage my servers. My contribution to the community in terms of programming has been "KDE-Services" (opendesktop.org/p/998464), which by the way I still need to adapt to the BSD world.

Title: KDE-Services - OpenDesktop.org

GeoBarRod: conveniently there is reviews.freebsd.org/D41745 right now which should allow you to exec dtrace(1) and read the output fairly easily from other languages if you want to try and "port" the idea of bpftune without writing C or using FFIs

Title: ⚙ D41745 dtrace: Add the 'oformat' libdtrace option.

need to run firefox under lldb

`lldb cannot launch firefox: operation not permitted`

env G_DEBUG=1 lldb -- /usr/local/lib/firefox/firefox --profile .

is what I'm trying

hoping there's an obvious thing I'm doing wrong here

ok, its not specific to firefox

sysctl security.bsd.unprivileged_proc_debug=1

thanks babz!

Is there recommended open source software I could use from my freebsd machines that would allow me to VPN from my mac on the road to my local network?

crb: wireguard, zerotier, openvpn

duh, thanks

dch, thanks

np

how would i match the different versions of linux-nvidia-libs to my graphic card? honestly no idea what 535 / 470 / 390 and so on means

should i just go with the latest / highest?

I usually look at the year my card was released, but I remember there's a table somewhere as well

but if it's newer than 2018 it's pretty safe to say you'll want 535

not necessarily

KingShark: there are tables of supported cards on the nvidia website

Title: Stefano Marinelli: "Today, something odd happened to me, slightly con…" - BSD.cafe Mastodon Portal

arg, this is a stupid question. but with pkg i wanted to see the full description of a package and maybe the source site. pkg info doesn't do that

am i just that thick?

i get that info works after the install, i was trying to see if this is the right package, so i wanted before install ;]

Demosthenex, Have you tried: pkg info -f rsync

rwp: that only works if it's installed

Are you sure? Let me find something not installed...

something like debian's "apt-cache show pkgname"

Sorry. You are right. Try this: pkg search -f 0ad

I knew it was one of those but got it wrong. My bad!

rwp: ok, i just missed it was a search flag ty

yeah, my muscle memory is still updating! ty =]

I am right there with you! I have my own mapping sheet of equivalents and explanations.

hi. Can you explain me how to manage usb storage devices, like pendrives? let me explain

i am trying to write iso image on my pendrive. Two commands, lsblk and gpart says that my pendrive is da2

so I use dd with of=/dev/da2 option but I get invalid input

i checked /dev and there is no da2 but da2%

is LS command broken?

i did ls /dev and there is da2 device

but if I do (( ls /dev/da[TAB] )) ls suggest da2% da2s0% da2s1%

can anyone explain it to me because it is thing I can't understand under freebsd

at least i can say that % characters are not part of file names. with dd (and in general) you should trust what ls command says

ok but I have no idea why I cant use da2

i can't mount it and I can't dd to it

"mount -t ext2fs /dev/da2s1 /mnt/usb" gives me invalid argument error

what lsblk says about the file system of /dev/da2s1?

"fuse-ext2 /dev/da2s1 /mnt/usb" does nothing, no error, no mounting

lsblk and gpart say it is linux-data

I don't actualy need to check what is there so I did "dd if=Downloads/bla.iso of=/dev/da2" and I got invalid argument

dd command is exactly the same I always used under linux to write iso images to pendrives

i notices that under freebsd i generaly have issues with mounting partitions and using storage usb devices. I can't find anything about these issues and I am not sure if my knowledle is correct or not

ok, fixed dd error :| it isn't 100% the same as under linux. There is no bs option

freebsd's handbook has a following example command: dd if=FreeBSD-13.0-RELEASE-amd64-memstick.img of=/dev/da0 bs=1M conv=sync

i removed bs= argument and dd is writing

i always set bs=100000, under bsd got argument error

so what is a problem with -> mount -t ext2fs /dev/da2s1 /mnt/usb?

asking because these error are wrong very often

mount has most incorect error messages I have ever seen

`dd if=/dev/zero of=/dev/null bs=100000` works for me *shrug*

is it possible it didn't work 10 minutes ago but work now somehow?

unlikely, but we won't know if you don't paste the exact command an error

a did. commands are 100% the same as I paste here

now I can;t do anythng more because dd is still writing

make tea

always a good excuse

i will try anther usb after dd finish

and we will see, ok?

status=progress -argument would show the progress...

too late :)

but thanks

never checked for progress info

btw, another question

I noticed one more thing

you can send USR1 signal to the dd process

it will dump its progress on stderr

when I use any da device like da2 and unmount it and remove it and then I insert another usb then there is no da2 but da3

i quess it is wrong :|

somehow da2 becomes reserved and new device use da3

i wonder what will happen now

also, reasonably certain our dd doesn't have a SIGUSR1 handler

oh your're right

the conversation's a little rough to follow, since it starts off talking about lsblk

but then there's reference to "operating like on linux systems"

works fine

new usb device has the same name, da2

USB devices do normally attach as /dev/daN where N is a number. They are attached as SCSI devices.

SATA drives normally attach as /dev/adaN devices, which is convenient because it keeps them in a separate namespace.

SAS drives attach as SCSI devices so spinning SAS drives attach as /dev/daN as well. Don't confuse a USB storage device with a spinning SAS drive!

I have not found the FreeBSD lsblk command to be useful. It's not the same as the lsblk from linux systems. Instead FreeBSD has "geom -t" which I find very useful.

However if I am trying to identify what is on a USB storage device using "file -s /dev/daN" will often be most effective. The -s tells file to read the special device even though it is a special device. And then say what it finds there.

To write a USB storage at /dev/daN I will use a command like "| dd of=/dev/daN bs=1M iflag=fullblock oflag=sync,direct" to do it. This has several advantages.

iflag=fullblock says that if dd reads a partial block it will read again and pack up a full block for writing. Sometimes on a loaded system this will be needed.

The oflag=sync,direct says to open the output file with O_SYNC | O_DIRECT which means writes wait until the storage says it is written and the file system buffer cache is bypassed since we are going to eject the USB device in a moment anyway it is not useful to cache anything.

Wouldn't it be nice to have a nice progress bar? That's easily done using the port "pv" progress viewer command. Install pv.

pv FreeBSD-13.2-RELEASE-amd64-disc1.iso | dd of=/dev/daN bs=1M iflag=fullblock oflag=sync,direct

That will write the data from the input with a pretty reasonable progress bar to dd which will write the data using O_SYNC|O_DIRECT. When the command terminates it is safe to remove the USB storage device.

i've been meaning to see if pv is really worth it. i've never felt like I need more than what status=progress/SIGINFO shows, but maybe that's just because I've never seen the 'greener grass'

Try it! I think you will like it. It's a lot nicer output than dd's status=progress output.

Lastly I am often in an iteration cycle and doing this more than a few times. I am certainly comfortable with root using sudo but wouldn't it be nice if I could avoid root entirely? It would avoid any possibility of confusing da with ada and causing an accident.

I do this additional thing. Which I would be interested if people thought it was okay or otherwise. I add a rule to /etc/devfs.rules saying this.

[localrules=10]

add path 'da[0-9]*' mode 0660 group wheel

And of course I am in group wheel so that I can use su so this means that USB drives are attached as /dev/daN and are writable by group wheel. Meaning I do not need sudo to write to them.

On a SATA desktop the spinning drives are /dev/adaN and this keeps them in separate namespaces. Making this much safer IMNHO from accidentally doing something in a slip of the finger.

(And in that same file I add "add path 'cd[0-9]*' mode 0660 group wheel" so that I can then read cdroms, usually for me with "cdparanoia -B", as me as well.)

wikan, ^^ Perhaps this helps? Hopefully useful hints.

Oh! I almost forgot another excellent hint! BSD tar uses libarchive and will read most types of archives and compression methods. It's truly a marvel.

rwp: i am reading your messages...

Which means that I can attach a USB storage device with, say, a Devuan netinstall ISO image. That attaches as /dev/da0 here. Then I can list the contents trivially with "tar tvf /dev/da0"!

That makes it really easy for me to |less there and browse an ISO image. And to extract files from it, such as base.txz really trivially without ever mounting it.

With something like "tar xvf /dev/da0 '*/base.txz'" which easily extracts files from it without needing to mount it.

nice, great hints rwp

thank you very much :)

wikan, The rule is see one, do one, teach one!

device rules are awesome I will use it for sure

i always warry about of= :D

i am checking command many times before I hit enter

A caution in the case of actual scsi devices like SAS drives. A system with SAS drives uses the SCSI driver and mounts those disks as /dev/da0 too. I wish USB used a different namespace!

and even that I am scared :D

i dont use scsi I think

But most desktops have SATA drives, either spinning or SSD, and those use /dev/ada0 naming.

btw, do you know why my LEDs and keyboard are da0 and da1?

i have usb leds string attached to my monitor

Think of my stress when accessing USB devices on my SAS storage array!

Huh? What? Huh? Your LEDs and keyboard are /dev/da0 and /dev/da1?

no

I bought leds string (about 30cm) and it is connected to usb

so leds have their own da device

One would think and hope that LEDs like that would only use USB for power and nothing else. No? That's awful!

but why da, have no idea

exactly.Leds only use power

another question if I may

It is probably a manufacturer who used a chip that presents itself as storage, may actually have some storage, but is also using power.

as I wrote before ls /dev gives me output where I can see da0 da1 da2 da2s1 etc

I suggest using a USB power isolator in series to allow LEDs to prevent the data lines from being active to prevent this. I expect it not to affect the lights power.

Though if it uses some type of programmable feature it would block being able to program them dynamically.

but when I type "ls /dev/da[TAB]" then autocomplketion suggest me not da but da0% da1% da1s0% etc

What command line shell are you using? echo $0

i use bash

i have no information about programming possibilites

it has only little controller attached to cable to change colors and modes

If you type in "ls --version" do you get a GNU version string or a BSD ls option error?

--version gives me unrecognized option error

All good so far. I was just trying to identify the environment in order to diagnose it further. But this one is stumping me.

Try this: env -i HOME=/ /usr/local/bin/bash

That will initialize the environment to nothing before setting HOME=/ and then invoking bash. Then from that shell try echo /dev/daTAB and see what happens.

If that does NOT produce % characters at TAB expansion then I would claim that the problem is somewhere in your environment or files in $HOME that has customized bash to emit % characters on device files.

well gives me da0 and da1 correctly now

If that still does then I would claim it is something in termcap. Oh! Good. Then it is something in your $HOME in files there.

~/.bashrc and ~/.shrc are 100% my own scripts

Temporarily move all of your custom files off to the side in a sub-directory for safe keeping. Then see if the problem is there. Expect not. Then carefully add things back in until the problem is located.

let me check

Bash would not be reading .shrc but would be reading other files .bash_profile, .profile, .bashrc and also possibly other files listed in the "man bash" FILES section.

my bash does read .shrc

i source this file

some option I set in .shrc

and becouse I don't want repeat commands I source shrc from bashrc

If you are explicitly sourcing a file then of course it is read. But with your other custom files moved off to safe keeping bash would not read it on its own.

BTW... I use a "PortaPow USB Data Blocker" which I would recommend for your LED power isolation. Here is one example: amazon.com/dp/B00QRRZ2QM

shrc has the issue

checking

may it be "command -p ulimit -c 0" ?

there is no other possibility in my opinion

exactly

well

weird

it works now

i didn't change anything and it works now

oh wait

You have stumped me with regards to "command -p ulimit -c 0" as I cannot see how that can do anything. Since ulimit must be a shell builtin in order to affect the currently running process.

root is broken not my user :D

Oh! Well that will be csh instead. And I get that behavior too now that I try it. It must be native csh behavior.

echo $0 gives me _su

shel is csh

correct

you are correct

Reading the man page for csh says "% Character device" so the % is designating the name as a character device.

so % does nothing. it is just something like an information flag

thanks

now i understand what is happening here :D

TIL that BSD csh has a builtin "ls-F" command which behaves similar to "ls -F". (eyes wide in amazement)

I feel certain that is new since I ran csh as a daily driving command shell.

i will change shell for root to bash

it is nonsense to use two different shells

Well... /bin/csh and /bin/sh in base and upgrade as part of the OS when base upgrades. /usr/local/bin/bash are from ports and upgrades as a secondary step.

it is not a problem

I have locked myself out of my own system by changing my root shell to bash, just as you are thinking. Been there, done that! And all was good. Until I upgraded from FreeBSD 12 to 13.

wut>

why?

And then I forgot I had done this and rebooted and found myself unable to log into any available login on the system, as all were bash, and bash used shared libs from base which had been removed during the major upgrade.

This is all as expected. And I knew the problem. But forgot and rebooted before I upgraded bash to the new version. My bad! I did it to myself!

wait

i use pkg

If this happens to you then reboot and use the Boot Environments (you are using zfs surely) and select the previous system. Boot it. Log in. Then use it to fix the default system. Then reboot again.

with pkg it shouldn't happen, right?

I was using pkg too. It can most definitely happen with pkg.

but it is mayor upgrade error O.o

well you right, freebsd uppgrades are messy

If you follow the upgrade instructions then everything will work correctly no problem. I slipped and did not.

last time I lost sound in firefox after upgrade

ok i leave csh

FreeBSD upgrades are not messy as such but they are LAYERED. This is GOOD. There is BASE that is one cohesive thing that upgrades as one cohesive thing. Then there are PORTS which are upgraded rather individually.

i will manualy change to bash

There is another user "toor" (root backwards). By default it is not enabled. It is good to enable it by setting a password for it if you have moved the root shell to a port. Then you can use it to log in if you need it. Though many people do not like having any root login possibilities and will advice the opposite of what I am suggesting.

well about year ago intel driver package changed its name for example

two times

i lost driver and couln't findout why :D

my freebsd is desktop workstation not a server :)

If one does not want to enable toor which uses /bin/sh by default (still not bash) and is also not liking csh then after becoming root then run "exec bash" to switch from csh to bash and continue on that way. It's fine.

i can log as root and it is not a problem

My desktop is also FreeBSD. And I also have FreeBSD servers. It's a universal operating system.

i switch all my mashines to freebsd :)

Welcome to the community! We have cookies! :-)

:)

i use freebsd about a year

sometimes a log in here

once I helped somebody :)

i could because he has the same problem as I found solition day before ;)

rwp, thank you so much for you talk, time and conversation

it is night here. Its time to sleep. Thanks one more time. See you later :) Bye

hello

@dch, hey man

what could lead login shell to not appear in multiuser mode, but appear in single user mode?

Please clarify what you mean by "not appear".

I can read the boot log in console, and then it stops just before showing the login prompt

and it seems to not accept any input

serial or vt console?

the two most likely reasons are: not enabled for login in /etc/ttys, or some rc.d script did not complete

serial console

one other option for serial consoles is a bad choice of gettytab entry causing it to wait for DSR/DCD before showing a prompt

what hardware is this?

in /etc/ttys I see lines: "console none unknown off secure" and "ttyu0 "/usr/libexec/getty 3wire.115200" vt100 off secure"

and also "dcons "/usr/libexec/getty std.9600" vt100 off secure"

I'm omitting all the ttyvX

notice that ttyu0 is marked "off"

which means, it won't spawn a getty for it

typically that would be set to "onifconsole"

so, did that work?

I'm trying right now, thanks for the hint

it's taking some time as I'm getting mangled boot console now

and I'm trying to figure out why

could not depend from last edit to /etc/ttys

no change, still hangs in same place

rebooting in single mode

note that the 115200 is the baud rate, if you're expecting it to be 9600 or whatever then change it to that