iio7, same occured to me with signal-desktop and vscode

iio7: upgrade to 13.2 then pkg update / upgrade + reinstall signal-destop / vscode and aal is good now

s/aal/all/

and yes, sadly, there is NO warning about the removal of signal-desktop or vscode while pkg update / upgrade

I get an error every hour from gnome-keyring-daemon that looks something like this: asked to register item /org/freedesktop/secrets/collection/login/1, but it's already registered

I've found mention of it going back years in Linux, including the suggestion the suggestion that it was a Gnome bug that got fixed a few years ago. Yet here I am. I have no problems with functionality that I know of, but I would like to reduce the log noise. Any idea what it is?

Checking my open ports (udp) by; sockstat -4 | grep "udp", I see ntpd listens on my external public IP address too, alongside "*:123" and "127.0.0.1:123", the question is, do I really need ntpd to listen externally as well, when the purpose is just having a correct time?

Related to the topic, my rc.conf has; ntpd_enable="YES" - ntp_leapfile_fetch_verbose="YES" - ntpd_sync_on_start="YES" - ntpd_flags="-4"

ntpd needs to get responses back from the queries it sends

Why isn't there an rc.d script to put CARP interfaces into BACKUP state when shutting down? It seems weird to me that the server would keep sending CARP MASTER messages while shutting down. The netif script runs late enough in the shutdown process that any services have already stopped working by then.

In my mind it would make a lot of sense to have CARP MASTER interfaces switch themselves to BACKUP as one of the first things when initiating the shutdown procedure. Thoughts?

I'm a noob, I'm doing a spring clean. Checking old drive for Data, about 7 years ago I started using PC-Bsd for a while, I have added an old (Lol) 160Gb sata drive on a usb caddy, I've forgotten how to mount the drive, da1 partition 3. zpool is not showing the drve

fdisk -s da1 3: 7311360 305268736 0xa5 0x00

ahh it's a ufs disk not zfs :|

I get the error operation not permitted with "mount -t ufs -o rw /dev/da1s3 /mnt/160"

ah no superblock

da1s3 is probably not the partition

what does gpart show da3 show, and gpart show da3s1

er

da1 and da1s3

its data drive /dev/da1s3: data

btw thanks, RhodiumToad: :)

shoes its data drive "/dev/da1s3: data"

tercaL: no i don't believe it does

_Random: that is not the output I asked for

just figured out that disabling TPM on BIOS makes suspend/resume work on a lenovo laptop and all looks smooth so far including network/video

Title: What is it with TPM (Trusted Platform Module) preventing resume from suspend-to-RAM? | The FreeBSD Forums

ly2en: is there an answer to the question?

i don't think so, if you refer to the device hints, no

i'm now onto other things, like figuring out why the webcam doesn't show up in applications

rc.firewall is meant to be edited, right?

hey all, i'm having an issues with window frames around Firefox. I'm currently running GhostBSD (FreeBSD 13.1), I have three window managers installed, XFCE, NsCDE(FVWM), Lumina. I installed Lumina just to test my theory. Basically, Firefox has NO window borders in any of the window managers/desktop environments. Other applications and web browsers seem fine. Any ideas that can help?

ixmpp: apparently??

well that sounds promising

ixmpp: I use pf, so i don't know much about that side of things

ah ok

cheese is still unable to read the video from webcamd :|

also the lenovo fingerprint reader is not recognized under usbconfig

so still a way to go to get the workstation do actual work

I saw some guy complaining on twitter that for a freebsd desktop survey, only 2% of participants actually used fbsd as an os, based on browser fingerprints

ly2en: lots of people who use esoteric browsers or esoteric platforms have to fake they are chrome, so websites will work

esoteric browsers such as Firefox

aside from humanwhocodes.com/blog/2010/01/12/history-of-the-user-agent-string

Title: History of the user-agent string - Human Who Codes

meena: anyway but the bottom line was like, how can you help improving freebsd desktop by not using it ?

how does improving desktop correlate with "browser fingerprints"?

having 100x users opening that page using browser with FreeBSD in user-agent will NOT improve anything

no and that is not the point

the point is that you cannot improve X if you are not using X

and User-Agent: implies people are not using X

that's all

not using X to open some random page/vote some random poll

the random page was a poll by a freebsd developer

it's still random page

1) it is not the point I made but he made it

2) i can relate to it

you don't have to

I don't understand your stance behind somebody calling for more freebsd desktop usage

even on a "random page"

my stance is that user-agent string does not improve anything, and making assumptions based on it doesn't help

"esoteric" freebsd foundation members don't use freebsd either

(ehm, correction, directors)

maybe the guy expected at least 10% of the user-agents show freebsd and he got 2% so he went emotional/disappointed

today I was identified by Mastodon as a FirefoxOS user :)

so clearly freebsd has better odds identifying as a discontinued project with a familiar name

PAM ;-)

the infamous lenovo fingerprint reader!

you're identified as a PAM user

(ab)

Title: BSD Authentication - Wikipedia

epony: I forgot how unreadable the old MediaWiki was on mobile

thanks for reminding me

;-) there was a theme for mobile specifically

minerva or something (instead of vector)

Title: BSD Authentication - Wikipedia

epony: you can replace Minerva with any random string and you'll get the fallback skin

Title: Simonas Kareiva LY2EN: "Inspired by the post at d-s.sh/2023/deskt…" - chaos.social

so I have an apology - the guy complaining about 2% freebsd usage is a zendesk developer, not freebsd

epony: I think you're looking for #OpenBSD.

ly2en: I think that thing with wiping the disk before checking if the connection actually works is worth reporting as bug

meena: i need to check if it actually formats the media at that stage, at least it looks like it does

in fbsd 13, what is the default vt keyrate? I tried looking into etc defaults rc.conf for it but it's set to "NO" and I have not overwritten it in etc rc.conf. Is it set to fast, aka 250.34? (i'm trying to replicate my vt rate in X)

no, I think $you are looking for BSD ;-)

while getting #linux

epony: was that for me? (i just joined, have no history over what happened above me)

well, if you think it's BSD air you're breathing..

?

epony: ?

epony: are you in fbsd channel trolling again

says BSD on the box

s10g: i believe it was directed to debdrup 's comment

epony: dude =)

drobban: ty

epony: whats all the negative vibes about =)

Apple / PAM / Linux.. fake choices as usual

it's like doas and sudo, but you don't get asked / don't get a choice (at all)

?

epony: you lost me... what are we talking about?

you see, with the protocols, you have choices, with the programming interfaces, you don't get to choose, the system gets to choose instead of you, options to pick from in this short list en.wikipedia.org/wiki/Template:Authentication?useskin=vector

Title: Template:Authentication - Wikipedia

epony: I see.

Any idea why I get an error every hour from gnome-keyring-daemon that looks something like this: asked to register item /org/freedesktop/secrets/collection/login/1, but it's already registered ?

Soo far Google has told me I'm not alone, but I haven't found an answer I can use.

epony: enough.

ghoti: you using nextcloud?

i am looking for debugger that shows asm and c code of compiled application

can you suggest me something?

can't find anything with octopkg

i remember ddd was great but can't find it too

wikan: you mean c code when it's not available?

otherwise i'm pretty sure gdb can show you both :)

no, I compile my code with -g -O0

i wanna see both - asm and c on split

use gdb

isnt it terminal output program only?

there seems to be packages for at least two graphical gdb frontends: KDbg and Scope

haven't tried

and it can split a plain terminal screen, too

maybe i should learn gdb

can it be like tui?

Please use a search engine.

Do you also use zsh?

And OhMyZsh?

I use zsh in mostly ksh emulation mode

gives nice completions and other useful features without having to cope with zsh's native brokenness

the trick is emulate zsh -c '. $HOME/.zsh-completions' where .zsh-completions contains the autoload -Uz compinit; compinit; and any other completion-related tweaks you wish

got some awesome completion and suggestions out-of-box without any hassle

that means all the completion stuff is still run in zsh mode, despite emulate -R ksh being in effect elsewhere

All the ksh users I've known have been male, at least 20 years older than me, and grumpy AF.

probably we date from the days when ksh was ubiquitous on commercial unix

I use ksh93 and I'm probably not that much older.

Though I also just use the default shell often, on most OSes.

See! ksh93 user and younger.

Love Korn shell. Typed variables are useful in so many cases.

Actually, I was on a consulting engagement with AT&T once, and they were all *FORCED* to use ksh, 'cause the company had this irrational fear of all things bourne.

the fact that the base sh has no arrays leads to so, so many bits of bad shell code

for i in `cat /some/text/file`

well, back in the day, many systems had a /bin/sh that was not a POSIX shell but just traditional Bourne shell, bugs and all

I used to write a lot of scripts targeting sh for compatability between OSes.

They had to work on FreeBSD, Linux, DG/UX, Solaris, HP/UX, AIX, IRIX

they finessed the POSIX spec requirements by having a /usr/xpg4/bin/sh link to ksh, and putting that directory first on the value of confstr(SC_PATH)

maybe some more, I forget

IRIX and, um, what was the later DEC one? are the main ones i never used

Tru64?

yeah, that's it

I think I touched exactly *ONE* Tru64 box.

I've used a little less than RhodiumToad I think (minus UMAX, plus Tru64), most of them in anger.

I think the only DEC Alpha box I ever met was running the Alpha port of VMS.

The only Alpha boxen I used were running FreeBSD.

UMAX was the first Unix I met, at university

running on a 16-cpu machine, in 1988 or so

That's a lot of CPUs in 1988.

(maybe they upgraded it to 32 CPUs at some point, not sure)

Hell, it's a lot of CPUs in 2023; the maximum you can scale Intel systems to is 8 sockets.

that was an Encore Multimax

(I know we have SMP via multi-core now, but that's cheating since there's still only one system processor which does APIC and a bunch of other stuff)

NS 32032 family CPUs

"The original Multimax could support from 1 to 10 pairs of CPUs" - per wikipedia

Oh, that's the first real 32bit CPU, isn't it?

(As opposed to the Motorola 68000, which had a 16-bit ALU)

Motorola 88k series? Anyone?

Pretty sure that was slightly later.

RhodiumToad: yeah, since POSIX specs were basically written off of ksh, with only a little input from SysV sh

so making ksh your sh is basically automatic posix compliance

DataGeneral built a bunch of servers and workstations with 88k CPUs.

its not

Some were still in use when I got hired at EMC.

the 32016 (aka 16032) was the first full 32-bit microprocessor, yes, but it was buggy

"Hell, it's a lot of CPUs in 2023; the maximum you can scale Intel systems to is 8 sockets." ⬅️ wat

This is my surprised face:

meena: supermicro.com/en/products/system/mp/6u/sys-681e-tr with 8 motherboards connected to backboard

Title: SYS-681E-TR | 6U | SuperServer | Products | Supermicro

Well, four - the new motherboards are dual-CPU apparently.

xtile: yeah, the issue is that the shell is almost never invoked via PATH search :-)

:B

do you count bit slice cpus like the amd 2900?

Or CPUs that aren't single-chip?

ooookay, so this time my 13.2 upgrade didn't get me into weird boot errors. but /etc/os-release still says 13.1 after the upgrade

what might be the reason for that?

how did you upgrade?

freebsd-update -r 13.2-RELEASE upgrade

freebsd-update install

reboot

freebsd-update install

uname says 13.2

uname -U and uname -K both?

both at 1302001

I don't know then.

Title: root@vhost-alfa2 /u/h/avh# uname -U1302001root@vhost-alfa2 /u/h/avh# uname - - Pastebin.com

Hello, with FreeBSD 13.2, my laptop (Lenovo TP p14S Gen1 (Intel) + TP Thunderbolt 4 dockstation) reboots with no warning after Xorg session unlocking (i3wm + i3lock).

I does not reboot immediately after unlocking the session but within 10 minutes after. There is no trace or whatsoever in Xorg logs nor in /var/log/messages. Just a plain brutal reboot for no apparent reason.

I am without a clue to investigate more. It seems very strange that the computer reboots, which would imply some very low level error, while only interacting a user level...

I'm trying to understand an issue I just saw where `clang` on Ubuntu seemed to have different default visibility than `clang` on FreeBSD. The resolution of the issue was passing -fvisibility to the linker, but I'm wondering how the different default visibility came to be. Does FreeBSD perhaps have a defined opinion about the default visibility?

pretty sure freebsd doesn't change that from the upstream default

what visibility were you seeing on each system?

I believe it was 'default' on Ubuntu and 'hidden' on FreeBSD (I was helping uskerine with their issue, so I didn't have the FreeBSD system in front of me)

I am quite certain that visibility=hidden is not the default on freebsd.

Any other guesses as to why the same clang invocation would result in a nearly-empty shared library on FreeBSD and not on Linux, unless -fvisibility='default' was explicitly included?

I should add that I was not able to use *exactly* the same clang version in my testing, but I can't find any information about a massive change in clang's behavior between 12 (what I had) and 13 (what they had)

given that freebsd has dispensed with block devices (see docs.freebsd.org/en/books/arch-hand…ok/driverbasics/#driverbasics-block), wouldn't programs suffer performance drops due to them not using a buffer cache anymore? is there an expected way to compensate for that drop? or is the drop just not that large to worry about? (in which case, why was the buffer cache there in the first place?)

Title: Chapter 9. Writing FreeBSD Device Drivers | FreeBSD Documentation Portal

rsjw: programs that access devices directly are rare to begin with; if you want to do unaligned access to raw disks you can use a library such as libublio to do that (this is used by some ports of FUSE filesystems from linux)

rsjw: doing raw device i/o efficiently generally requires some form of AIO anyway

RhodiumToad: wouldn't regular file i/o suffer a performance drop due to the lack of buffer cache?

no

regular file i/o hasn't gone through the old-style buffer cache since the introduction of unified VM way back in the day

by the time block devices were removed in 4.x or whenever it was, the old-style buffer cache was purely vestigial

I'm not clear what the difference is between the terms "buffer cache" and "unified VM"

unified VM treats the whole of available memory as a single resource for caching both file data and program memory

(not necessarily at the same priority, but deliberately not making a hard distinction between the two types of data)

this means for example that when you mmap a file, the mapped memory pages are the same ones used to cache read/write calls, thus guaranteeing coherency

under the hood, every active vnode is associated with a vm_object that holds the memory pages with whatever part of the file data is currently cached in ram

(you can see this in action by using vmstat -o to display vm objects)

I see. the mmap reason does make sense

note that ZFS somewhat breaks this model by doing its own caching below the vm_object level

so the unified VM unified the buffer cache and "program memory"... how did this program memory cache work? is that basically the swap system, where it swaps RAM to disk?

hm, well, I can now rule out a change in clang at least. No idea what quirk the user stumbled into there, unless CPython is doing something very strange with visibility on FreeBSD which I find hard to believe

rsjw: yes. the paging system scans all of the non-wired memory to find what's not being used, and if needed, schedules pages to be written out to their backing store, which might be swap space or might be a file (vnode)

rsjw: (again, if you look at the vmstat -o output, you'll see 'sw' and 'vn' object types reflecting this)

(you'll also see 'df' (default) objects, which are allocated memory that hasn't been assigned to a pager yet, typically program memory that hasn't needed to be considered for swapping out yet)

the disk requires data to be written in blocks but the memory does not. how does the system know when to write an updated mmapped region to the disk? memory accesses no longer have information attached to them that say how large the transfer is goin to be, unlike file i/o.

memory paging is done in units of pages, or sometimes aligned clusters of pages

when you write to memory (whether allocated program memory or mmap'd memory) the dirtyness of the memory is tracked only at the page level