Posts by metala

metala · Packaging for Devuan

On my test machine (AMD Athlon 64 X2 TK-57 w/ 512MB, Debian 13.3, XFCE4) NM and wicd run in parallel.

After boot:

$ sudo ./ps_mem.py -p "`pgrep -f -d, 'wicd|NetworkManager|nm-applet'`"
 Private  +   Shared  =  RAM used	Program
 16.0 KiB + 843.5 KiB = 859.5 KiB	wicd-client
984.0 KiB + 513.5 KiB =   1.5 MiB	python3.13
  1.9 MiB + 642.5 KiB =   2.5 MiB	nm-applet
  2.8 MiB + 661.5 KiB =   3.5 MiB	wicd
  4.0 MiB + 762.5 KiB =   4.7 MiB	NetworkManager

After some time:

984.0 KiB + 429.5 KiB =   1.4 MiB	python3.13
  1.3 MiB + 843.5 KiB =   2.1 MiB	wicd-client
  2.0 MiB + 583.5 KiB =   2.5 MiB	nm-applet
  2.8 MiB + 576.5 KiB =   3.3 MiB	wicd
  3.3 MiB + 307.5 KiB =   3.6 MiB	NetworkManager

With Wicd GUI window open and Wifi networks shown:

  1.6 MiB + 815.5 KiB =   2.4 MiB	python3.13
  2.2 MiB + 597.5 KiB =   2.8 MiB	nm-applet
  2.5 MiB + 346.5 KiB =   2.9 MiB	NetworkManager
  5.9 MiB +   1.0 MiB =   6.9 MiB	wicd
 37.3 MiB +   5.9 MiB =  43.2 MiB	wicd-client

Wicd GUI closed (hidden), nm-applet showing Available networks->

  1.6 MiB + 815.5 KiB =   2.4 MiB	python3.13
  2.8 MiB + 346.5 KiB =   3.2 MiB	NetworkManager
  5.9 MiB +   1.0 MiB =   6.9 MiB	wicd
  5.7 MiB +   2.7 MiB =   8.5 MiB	nm-applet
 36.7 MiB +   4.9 MiB =  41.6 MiB	wicd-client

(After a while...) Wicd GUI closed (hidden), nm-applet showing Available networks->

  1.6 MiB + 814.5 KiB =   2.4 MiB	python3.13
  2.8 MiB + 346.5 KiB =   3.2 MiB	NetworkManager
  6.0 MiB +   1.0 MiB =   7.0 MiB	wicd
  5.9 MiB +   2.7 MiB =   8.7 MiB	nm-applet
 36.7 MiB +   4.9 MiB =  41.6 MiB	wicd-client

Running wicd-curses:

  1.5 MiB + 673.5 KiB =   2.2 MiB	python3.13
  2.8 MiB + 337.5 KiB =   3.1 MiB	NetworkManager
  7.1 MiB + 881.5 KiB =   7.9 MiB	wicd
  5.9 MiB +   2.7 MiB =   8.6 MiB	nm-applet
 27.5 MiB +   2.4 MiB =  29.9 MiB	wicd-curses
 34.8 MiB +   4.6 MiB =  39.4 MiB	wicd-client

(30 minutes later...) wicd-curses is running:

  1.5 MiB + 437.5 KiB =   1.9 MiB	python3.13
 44.0 KiB +   2.0 MiB =   2.0 MiB	wicd-client
  2.4 MiB + 905.5 KiB =   3.3 MiB	nm-applet
  4.0 MiB + 201.5 KiB =   4.2 MiB	NetworkManager
  6.0 MiB + 859.5 KiB =   6.8 MiB	wicd
  5.6 MiB +   1.8 MiB =   7.4 MiB	wicd-curses

This might be irrelevant, but it is part of the NM-suite.

400.0 KiB + 115.5 KiB = 515.5 KiB    ModemManager

PS. I don't know how to memory test ephemeral process like wicd-cli.

metala · Packaging for Devuan

As far as I remember, that's the original behaviour of wicd, it disconnects the current connection and connects to the other.

I really don't know if we can change the behaviour. If someone starts a bug I could have a local patch in the package, that addresses it, but that's up for discussion.

Wouldn't having multiple connections mess up the routing table?

metala · Packaging for Devuan

It seems the *.deb are not directly available from the mentors. I am investigating why.

Workaround: https://pgit.metala.org/dl/wicd/1.8.0~g … 9785e9f-1/
VCS: https://git.devuan.org/metala/wicd/src/ … erimental/

From IRC:
> mentors discards the binary, it's only there to review the source package

metala · Packaging for Devuan

Hi all,

tldr;
Source at mentors: https://mentors.debian.net/package/wicd/
Version: 1.8.0~git20260213.19785e9f-1
Binaries/Sources: https://pgit.metala.org/dl/wicd/1.8.0~g … 9785e9f-1/
VCS: https://git.devuan.org/metala/wicd/src/ … 19785e9f-1
GPG KeyID = 3E0561FB254A6F2C289874DBF350FFD9E01E8F4B, keyserver=keys.openpgp.org
wicd_1.8.0~git20260213.19785e9f-1_amd64.changes is signed and contains checksums.

I have been working on packaging wicd for Debian and If it gets sponsored, we will get it as well.
However, I would like it to be well-tested and fixed, before I submit it for unstable and look for sponsors.

If you are keen on helping and you have free time, and maybe a spare computer (it is in experimental, after all), please, check it out and write down your thoughts. Patches are also welcome. You can post comments here, but also over to my dev email or the pkg-wicd-maint@ on Debian. I am reading my email every day and I would try to be active here on the forum the following weeks.

Side note. Most of the work on porting wicd to Python3 was done by the current upstream maintainer lp/~hanaguro. The praise should go to him.

Kind regards,
Marin

metala · Desktop and Multimedia

I think I've found the culprit `orca` not running well after autostart, but it seems to be a more complex issue.

I am using Awesome WM and I am starting DE session to get more useful desktop experience. I used to start mate-session manually, but I've decided to spawn mate-session on awesome WM startup.

.config/awesome/rc.lua

-- Autostart
awful.spawn.with_shell("~/.config/awesome/autorun.sh")

.config/awesome/autorun.sh

#!/bin/sh

run() {
  if ! pgrep -f "$1" ; then
    "$@" &
  fi
}

run mate-session

This works fine, but some things seem to break. I've got orca installed, at some point, which tries to autostart itself:

$ dpkg -S /etc/xdg/autostart/orca-autostart.desktop 
orca: /etc/xdg/autostart/orca-autostart.desktop

However, probably due to some race condition with pipewire, the orca starts, but does not work. If I start the session manually, the speech dispatcher is working correctly. Running a `killall -9 orca` fixes the mate-terminal freezing issue.

I guess I will switch to manual startup of the mate-session, because something gets broken, when starting it up like that.

metala · Desktop and Multimedia

[    15.421] (II) LoadModule: "amdgpu"
[    15.422] (II) Loading /usr/lib/xorg/modules/drivers/amdgpu_drv.so
[    15.426] (II) Module amdgpu: vendor="X.Org Foundation"
[    15.426]    compiled for 1.21.1.7, module version = 23.0.0
[    15.426]    Module class: X.Org Video Driver
[    15.426]    ABI class: X.Org Video Driver, version 25.2
[    15.426] (II) AMDGPU: Driver for AMD Radeon:
        All GPUs supported by the amdgpu kernel driver
... <truncated for brevity> ...
[    15.694] (II) LoadModule: "glamoregl"
[    15.694] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[    15.699] (II) Module glamoregl: vendor="X.Org Foundation"
[    15.699]    compiled for 1.21.1.7, module version = 1.0.1
[    15.699]    ABI class: X.Org ANSI C Emulation, version 0.4
[    15.719] (II) AMDGPU(0): glamor X acceleration enabled on AMD Radeon Graphics (renoir, LLVM 15.0.6, DRM 3.57, 6.9.7+bpo-amd64)
[    15.719] (II) AMDGPU(0): glamor detected, initialising EGL layer.
[    15.719] (**) AMDGPU(0): TearFree property default: on

Linux 6.9.7+bpo-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.9.7-1~bpo12+1 (2024-07-03) x86_64 GNU/Linux

metala · Desktop and Multimedia

I am using the amdgpu driver for Xorg w/ Vega 8 APU + TearFree option enabled.

However, I doubt that this causes the issue. If I kill the mate-session it gets fixed and if I restart the mate-session, the issue does not reappear. It is also limited to mate-terminal and xfce4-terminal, the browser is playing videos while the terminal is hanging/freezing.

I've been given hints to look into gvfs and gio. I will try to strace it using uxterm or even debug it, if I can get the gdb to follow the detached child of /usr/bin/mate-terminal.

Do you know if Devuan has debug symbols of the compiled packages?

metala · Desktop and Multimedia

Software:
Devuan 5
xorg 1:7.7+23
awesome 4.3-7
mate-session 1.26.0-1+deb12u1
pipewire 0.3.65-3+deb12u1
orca 43.1-1

The Problem:
After some updates or something, the mate-terminal started. I've also tried the XFCE4's terminal emulator and it had the same issue. However, the uxterm works just fine without any hanging.

Solution:
Killing and restarting the mate-session fixes the problem.

Thoughts:
I have the same configuration on a different machine, but I don't face the issue. The difference is that there I start the mate-session manually, after awesome WM is loaded. Here on this machine I have it spawned in the awesome WM configuration on startup. I don't see anything suspicious in .xsession-errors.

Question:
Any ideas how to troubleshoot this?

UPDATE:
The on-screen reader (orca) is causing the freezing. If kill it the terminal emulator no longer freezes. Orca is autostarted by mate-session, but it seems to fail if mate-session is spawned by awesome WM on startup. If I run mate-session manually, the screen reader works fine and synthesises speech. However if it's started by the awesome WM, the orca does not seem to work. I assume that there is a startup race between pipewire and mate-session + orca.

The simplest solution is to run mate-sessoin manually, remove the orca package or remove it from /etc/xdg/autostart.

metala · Hardware & System Configuration

The Problem:
I have installed fresh Devuan 5 without X server on an 2011-ish laptop. However, when I closed the lid of the "soon to be server" laptop, the device went to sleep.

The Solution:
1. Edit /etc/elogind/logind.conf

2. Uncomment and set the following parameters:

HandleLidSwitch=ignore
HandleLidSwitchExternalPower=ignore
HandleLidSwitchDocked=ignore

3. Reload elogind:
# service elogind reload

Note. I did it few months ago and I am not sure if the elogind reload helped or I needed a restart. service elogind reload works fine.

The officially official Devuan Forum!

#1 Re: Packaging for Devuan » wicd 1.8.0 package » 2026-02-19 20:27:48

#2 Re: Packaging for Devuan » wicd 1.8.0 package » 2026-02-19 11:17:05

#3 Re: Packaging for Devuan » wicd 1.8.0 package » 2026-02-19 00:58:19

#4 Packaging for Devuan » wicd 1.8.0 package » 2026-02-19 00:24:58

#5 Re: Desktop and Multimedia » [SOLVED] mate-terminal freezes briefly every X seconds due to orca » 2024-09-20 06:59:04

#6 Re: Desktop and Multimedia » [SOLVED] mate-terminal freezes briefly every X seconds due to orca » 2024-09-19 11:13:22

#7 Re: Desktop and Multimedia » [SOLVED] mate-terminal freezes briefly every X seconds due to orca » 2024-09-18 19:07:42

#8 Desktop and Multimedia » [SOLVED] mate-terminal freezes briefly every X seconds due to orca » 2024-09-18 16:45:22

#9 Hardware & System Configuration » [HOW-TO] Prevent laptop suspend / sleep on lid close » 2024-09-12 21:14:02

Board footer