well

i tried enabling 8n2 mode with one usb to serial adapter and it refused to

so i tried with another serial adapter (the one that has been connected to the UPS for the last 10 years) and it crashes as soon as getty shows the login prompt

and now im trying with a usb adapter that literally costed me $2 and was giving me problems with an external modem, and it seems with 8n2 mode the terminal is very happy

tldr it was the fucking usb adapter

i hope the pcie-serial adapter will be less trouble

(famous last words)

so here i am now at 19.2kbaud no flow control yolo

Yay!

i'm gonna have to do a writeup about this on my blog

rwp: should i go back and rescue the other two that were left at the warehouse?

there is that and an extremely heavy C.Itoh printer

That is a question only you can answer.

I might say that it takes at least two terminals to play a terminal game between two people. :-)

i might go. could make a video about restoring them

this one still needs retrobright

im waiting for the summer to start retrobrighting things around here

have a commodore 128 that could use some retrobright

It is the end of summer for us on this side of the planet. Winter is coming.

i need to map colors to make better use of this terminal but im not sure how its done

my own nick is underlined here but that's about it

i see that *text* like this appears also underlined

any hints on updating an extracted base.txz distribution with freebsd-update? neither -b nor -d seem to do the trick...

so i have a weird issue with the serial terminal. i can't run sudo

it does nothing, util i ctrl-C

then it prints the usage instructions

it works fine over ssh

Do you have the same problem if you try to change your password?

I am guessing it is asking you for a password and changing the tty to noecho for the entering of it.

i have it nopasswd

Hmm... That is weird then.

even truss sudo shows nothing either

Are you in the wheel group? Also try "su - $USER" just for additional data points.

"su - $USER" will stack a shell. Exit from there to return to the previous shell.

haro.. I found my way here from 20 yrs ago.

nice

we been waiting for you... now we can move on finally

I only remember asmodai from the user list.

Does saffron and blackend come by anymore?

Welcome PawTeeGal! I do not recognize those nicks but who knows?

People from probably the late 90's used to be efnet freebsd group. I know blackend is still on the official FreeBSD project. saffron used to be regular too, I think she used to work for Xi Systems.

Though there has been the big migration to Libera.Chat the others still also exist. Except for that one which was taken from us and we were forced to walk away from which I do not speak of now. Perhaps they are still using EFnet? (Shrug.) You might find them there.

Until then you have many new friends here to meet.

Sure I'll hang around. I looked around on efnet looks like it got turned into the internet ghettos.. shame..

And hi Rep, Good meeting you. I figure going down memory lanes a bit and found me way here.

We have been enjoying hjf's adventure down museum serial terminal lane. From hjf: i.imgur.com/eZqPwuw.jpeg

now thats vintage.. alas I got rid of all my vintage stuff. back from the fbsd 2.2.8 days and my sparc 20. (good riddance honestly)

It's an obscure TERM=viewpoint3a+ type and how shall I say, there has been some termcap issues with it.

dont think I've ever seen one..

I did started out on a Hearthkit H19 tho

before i forget all my learnings

how does the package repository work. Why am I missing almost every py39- package that py311 has?

im running with latest right now, the same thing applies for quarterly but for other packages.

is it just a temporary glitch or am I doing something wrong.

drobban: ports only supports a single python version, which is currently 3.11 by default, so that's what the packager builders use. it's possible to configure a package to build for multiple versions, but the vast majority of python packages don't do that

to get a different python version you would have to build your own packages from ports with python set to 3.9 (or whatever)

ivy: thanks for the clarification!

i phrased that a little poorly, ports supports multiple python versions, but you have to choose one version as the default version, which is what the py*-* packages will build against

and the default default version is 3.11

ivy: =) think that was how I understood you the first time as well.

ports named tcl9.0 and tcl9 why two?

dandyn: without checking, i guess tcl9 is 9.latest and tcl9.0 is fixed at 9.0. check the makefiles to be sure

okok, thnx

dandyn maybe you can ask in #tcl, they are pretty helpful and as per they conversations they are now fully engaged with tcl9, I think most people use 8.6

Hi, where can I find a howto or tutorial to report a bug in the base system and potentially submit a correction? forums.freebsd.org/threads/rcp-issu…commands-package.95211/#post-674568

uskerine: if you found a bug, you can report it at bugs.freebsd.org. however if you have a code fix, you should not file a bug but instead send a pull request to the FreeBSD github repository, github.com/freebsd/freebsd-src

uskerine: this uses the normal git workflow, if you're familiar with that

can someone explain how wireless works on freebsd

i have this network card: iwm0: <Intel(R) Dual Band Wireless AC 7260> mem 0xf0400000-0xf0401fff at device 0.0 on pci2

but i get this: ssid LF channel 6 (2437 MHz 11g) bssid 74:4d:28:8e:7a:8e

surely this card should do better than .11g?

according to the Intel spec sheet, it should do .11ac

According to docs.freebsd.org/en/books/handbook/network/#network-wireless “802.11ac support on FreeBSD is currently under development.”—that might mean, it does not work yet? My hardware is far too old to run into such a problem :-)

But it says there: “FreeBSD supports networks that operate using 802.11a, 802.11b, 802.11g and 802.11n.”

even so, surely it should work as .11n

Which version of FreeBSD are you on? `uname -v`

i'm not at that device right now but it's 14.1 from about a week ago

(had some weird issues with 15.0 on it, so i had to downgrade to 14.1)

iwm(4) says 802.11ac hmmmm... docs.freebsd.org/en/books/handbook/network/#network-wireless

wrong link: man.freebsd.org/cgi/man.cgi?query=i…path=FreeBSD+14.1-RELEASE+and+Ports

You have tried to put `iwm7260fw_load="YES"` into your `/boot/loader.conf`?

Sometimes loading it after that stage results in some strange problems.

i will try that next, thanks

I hope, it will solve your problem.

Kalten: unfortunately it did not

i tried loading all the iwl-related modules in loader.conf

iwm7260fw_load=YES

if_iwm_load=YES

if_iwlwifi_load=YES

but i'm still stuck on .11g: ssid LF channel 6 (2437 MHz 11g) bssid 74:4d:28:8e:7a:8e

i really hope i don't have to install wifibox to make this work, this ancient laptop only has 4GB RAM

ivy well I ahve not tested, the fix is described there, but there might be more things to do

so shall I just submit the bug?

uskerine: if you don't have a direct code fix then yes you can just submit the bug. i didn't read your link though so not sure what the specific issue is

brb

when they removed the r commands from base, there is a header that points rsh to /usr/bin but it is no longer there, it is now at /usr/local/bin, as a result of that, rcp fails

apparently the issue shall be fixed just by changing a header which has hardcoded /usr/bin/rsh

and that file is in base

trying to rule out a power issue with a usb-serial adapter by feeding it 5V directly instead of bus power

ok, so wifibox sort of worked but it's put me behind some weird kind of NAT and IPv6 doesn't work

i assume this isn't how it's meant to work

maybe i configured something wrong

the suggestion is the networking depends on what the guest exposes

I have horrendous trouble with Linux wireless, let alone Linux wireless in a VM on my laptop!

ok seems like wifibox does work with ipv6, had to rtfm a bit

traceroute6 to hemlock.eden.le-fay.org (fd12:8247:3:1::5) from fd00::1, 64 hops max, 28 byte packets

1 fd00::ffff 0.623 ms 1.204 ms 0.921 ms

2 2001:8b0:aab5:c103::1 2.471 ms 2.967 ms 2.177 ms

3 hemlock.eden.le-fay.org 2.060 ms 2.036 ms 2.134 ms

i guess fd00::ffff is the NAT router

let's see if it's any faster

[ ID] Interval Transfer Bitrate Retr

[ 5] 0.00-10.00 sec 108 MBytes 91.0 Mbits/sec 422 sender

well that's pretty bad for a 2x2 .11ac connection, but i was only getting 20Mbps with FreeBSD iwlwifi, so i suppose it's an improvement

bit sad that i get 550Mbps on my iphone with zero effort

from the same AP

ivy: my solution to wireless issues was an M.2 ath9k card. I get 120Mbps on 5Ghz on 15.0-CURRENT (this is A b/g/n card). Is AC really significant?

duncan: if you only get 120Mbps i'd say ac is significant, yes

i don't really care enough to investigate what's going on here (at least i'm getting more than the 20Mbps i had before) but...

maybe this laptop only does 20MHz or something

I guess what I was getting at is, 108Mb might be the best the card can do

yeah i just rebooted with the wifibox console enabled so i can see what the card is actually negotiating

how on earth do you do this in linux

It's an Alpine Linux VM right? I'd boot that up (it's pretty well thought out) and do a speed test, preferably one run by your ISP

The nominal bitrate negotiated is rarely the speed you get in practice

i already did a speedtest, i'm trying to work out what specifically it negotiated with the AP. e.g. if it's actually using the full 80MHz channel

apparently iw(1) is the right command but usage is a bit cryptic

The interesting question would be whether the VM differs from bare metal with the same PCI device?

channel 6 (2437 MHz), width: 20 MHz, center1: 2437 MHz

why on earth is it using 2.4GHz

welcome to the world of Linux/UNIX wireless

my response to these crazy scenes was ath9k but there mut be something in between 'workhorse decade and a half old original OSS driver' and '2020s broken WiFi 7 chipsets'

must*

do linux users just not use wifi?

surely there must be a way to make it prefer 5GHz

most users use NetworkManager - it provides a GUI (this is quite good), sugar over low level config options like power saving, preferred channels

you can do similar with iwconfig but it's horrible

but the general malaise is in terms of what drivers expose and how stable they are when interacting with e.g. IOMMU, power management

ivy: maybe more luck (just for testing) to start (as root) bsdconfig, select “Networking Management” and than “Wireless Networks”. It should write to /etc/rc.conf in the end

Maybe your 2.4 and 5.0 can be chosen there?

Kalten: i doubt it since ifconfig wouldn't even accept 'mode 11n' on freebsd

i'm not too bothered, 100Mbps is fine for what i use this laptop for, i just wish i could run iperf here instead of running it on my phone

ivy: promotion of 5GHz band is usualy done on AP side (if supported by AP)

I have nothing to trie it out, but ifconfig(8) talks about something: chanlist (could be `chanlist n`) and channel (could be `channel n` or `channel nt/40` or `channel ht/40`???)

mzar: what do you mean by "promotion"? band steering?

yes

mzar: but shouldn't the client naturally choose 5GHz if the signal is good? that's what all our other clients do

I have not heard of relevant standardisation with this regard

Well: 2.4G has wider range (good for moving around with your laptop)

but 2.4GHz is much more congested, bad if you're anywhere with your laptop

also much slower...

(at least in our case as the 5GHz SSID is 80MHz wide)

so your AP has to support band steering

and it must be turned on

as freebsd is the only client so far that doesn't do this properly i will probably not bother with this, tbh

but i will bear it in mind if it comes up in future, thanks

ah, our vendor doesn't support it anyway

interesting story, so Linux clients prefer 5 over 2.4 without band steering enabled on the AP ?

but it does do .11k and .11r so i would think the client could find a better frequency

mzar: is that a statement or a question? my client (wifibox, which is linux) seems to strongly prefer 2.4GHz

OK

maybe i'll install linux on this laptop and see if it works any better natively

Linux has better WiFi support, that's unquestionable

i agree, but i thought wifibox would make this work as well as it does under native linux

probably yes

maybe signal on 2.4 is stronger

but in fact no, it sticks to 2.4GHz

if i get 600MBps in iperf3 on my iphone and 120Mbps in iperf3 on my laptop... something is wrong somewhere, i think

s/MBps/Mbps

(600MBps on wifi would be nice, heh)

what OS is on that laptop ?

FreeBSD 14.1 amd64

i have a user with Steam Deck (a wifi device that runs Linux) and that does manage about 500Mbps over wifi

hmm... i wonder if the 5GHz SSID is configured for 80MHz, and the laptop's wifi card doesn't support that for some reason, maybe it would fall back to 2.4GHz rather than using a narrower channel?

ha... AFAIR FreeBSD supports only 802.11n, perhaps your phone supports 802.11ac or ax ?

right, i'm not using freebsd wifi stack, i'm using wifibox

no ax here, the AP, phone, and laptop are all .11ac

(well i think the phone does support .11ax, but the AP doesn't)

I would try and produce a minimum working example: Alpine or Debian, try and get 5G.

cool, but AFAIK with FreeBSD you can't use ac yet

no, freebsd doesn't support .11ac, but wifibox does? isn't that the whole point of using it?

does ifconfig show ac for this WLAN NIC ?

ifconfig just shows the wifibox tap interface

i'd have to go turn on the laptop again to see what the wifibox VM says

interesting

mzar: this is unhelpful - this is wifibox, Linux handles wireless, FreeBSD just talks to it

you have put a lot of work to make it working ivy, congrats

i really didn't, i just installed wifibox and set up /etc/rc.conf like the manpage says

s/work/effort

tap interface could be also a bottleneck

and it does work and tbh it works well enough that i don't really mind if i only get 100Mbps... the only reason i care about this is i was having an issue with our APs (not related to FreeBSD) where with ~500Mbps wireless traffic they would sometimes crash

and it felt silly that i had to use my iphone to generate such traffic with iperf instead of a laptop

that's sad story

although honestly, this laptop is old enough that my iphone probably has a faster CPU

You need an MWE. MWEs are good both for articulating the bug report and for (not infrequently) nailing down the problem in the process of composing it

try to trigger this AP failure with multiple UDP streams from your FreeBSD-driven laptop, it could work

i think i fixed the AP issue with a stronger PSU, but we shall see

If Linux proper can't do faster than 100Mbit, you'll know it's probably hardware

duncan: what's an MWE?

minimum working example

oh, right

i mean like i said i don't care enough to actually file a bug

at least i've improved it from 20Mbps under iwlwifi(4) :-d

that's not a bug, only missing feature

omg stop being so pedantic

well, you don't have to report it, but it could be helpful to you later

later e.g. it becomes clear how to got 5Ghz

not worries, let's hope WiFi performance will get improved a bit freebsdfoundation.org/blog/why-lapt…ategic-move-toward-broader-adoption

if wifi under freebsd is really this bad i do hope the foundation spends some $$$ on improving it

duncan: what do you mean by "becomes clear how to got 5Ghz" ?

Wifi will improve over time. Thunderbolt/USB4 is a bigger issue

but tbh if i bought another laptop it's going to be a macbook and not run freebsd

(i love freebsd on servers/vms though, and it's also pretty good on desktop)

ivy: it is easier to work out why a simpler Alpine/Debian Linux isn't getting 5Ghz than that in a VM. Fewer variables, easier for others to reproduce. that's why its an MWE.

ivy: cool, same here, it works fine in all scenarios

mzar: well...

i did have to submit a patch to alc(4) to make it work on my desktop by default

(some motherboard have broken MSI-X for this card)

IMHO it's just a waste of time to try to improve this setup working thanks to ivy's effort and tap-driven duct tape

not offending ivy

but let's say "most scenarios" :-)

mzar: why is it a waste of time?

because ivy's time is important

i don't know what you mean by "ivy's efforts", wifibox is nothing to do with me, i did not contribute anything to that project

wifibox did improve my wireless performance from 20Mbps to 100MBps, so i'm certainly not complaining about that

cool

s/MBps/Mbps

(ffs)

no, I don't get it. wireless should perform similarly in a VM to bare metal given same PCIe device access. the interesting question is why not

wifibox is neat but belies real disinterest from manufacturers compared to Linux/embedded

duncan: because there are plenty of bottlenecks, and the most relevant in this case is probably tap(4)

very unlikely, i get 10Gbps from tap(4) in my bhyve VMs

so it's fast laptop

I agree, but I don't know enough about how these things interact to say. Narrowing it down is time-consuming, but it's still worthwhile to do so.

never mind

am I right that the freebsd iscsi initiator does not *verify* the header and/or data digest? I saw in the man-page that they are generated, but are they als verified for incoming PDUs?

I'm asking as my toy-iscsi-target generated invalid digest-values and yet freebsd (and linux) seem to ignore that

f0lkert: you have to read that code and tell us how it works

f0lkert: i think someone is more likely to know on the mailing list, though mzar's playful comment about reading the code might ultimately be true

jmnbtsls1E: I'm considering doing so :)

ok so i found yet another issue. the CH340 driver for FreeBSD is incomplete unfortunately

this is why it was causing issues. It doesn't support any modes other than 8N1

trying to compile an application with FreeBSD 14.1-RELEASE after an upgrade from 14.0, but getting 'undefined reference' on a whole heap of symbols that I know are in SOs in /usr/lib & /usr/local/lib, both with GCC and CLang. File permissions are correct too

GCC's /usr/local/bin/ld cannot find symbols in libPoco*.so files, which are in /usr/local/lib, installed with pkg

same code compiles fine under Linux (GCC), MSYS2, etc.

hackkitten: I'm not familiar with libPoco, but chances are you're just missing some linker flags (LDFLAGS)

probably -L/usr/local/lib

what does the compiler say ? can you post the output somewhere (pastebin, whatever)

compiler AND linker, just post the whole thing

I did add that library path, yes, but it doesn't make a difference. It does find the libs at least (no errors), but cannot seem to use them. Ran a pkg upgrade earlier, seeing whether that helped

oprs: here's the pastebin link command: pastebin.com/Md3YJ18R

pkg upgrade didn't change matters, sadly

well there's the obious undefined reference to Poco::SystemException::SystemException

but I also see references to FreeBSD 14.0 in the object paths (obj/x86_64-portbld-freebsd14.0/chronotrigger.o, etc…)

did you properly clean your build tree before compiling on 14.1 ? (i.e. make clean)

yes, I did clean it beforehand, this is also the first time that I'm building this application on this FreeBSD system

cloned it off GitHub earlier

hm

though the 14.0 reference is interesting, yes

that's the output from g++ -dumpmachine

I use those strings to separate different build outputs

with uname -a I get: FreeBSD freebsd-nya 14.1-RELEASE-p5 FreeBSD 14.1-RELEASE-p5 GENERIC amd64

I can try to reinstall the system, do a clean install of 14.1, maybe

I just realized there's a multimedia/nymphcast port, you could just install that

that doesn't explain why it doesn't compile though

didn't know there was a pkg online

I'm the NymphCast developer, so that's why I want to compile it myself :)

lol, I guess that makes sense )

will be setting up a quick new VM with 14.1, and also GhostBSD alongside it, see whether that works better

thanks for the assistance :)

well, there seems to be a few FreeBSD specific patches: cgit.freebsd.org/ports/tree/multimedia/nymphcast/files

ah, was wondering about that

you may want to have a look

indeed, I'd want to integrate those into the main tree

much obliged :)

yw :)

just learn that, in the very same way that I can install python packages from the binary pkg system, I can install R packages. Who creates them?

s/learn/learnt/

Do you mean who are all the port maintainers? Or who are the upstream devs?