Tracking Debian testing/unstable with BunsenLabs

AndyMender · 2016-04-05 19:55:59

I don't feel comfortable with Debian Unstable, but recently I did upgrade 2 freshly set-up computers from BL Hydrogen rc2 to Debian Testing repos. Everything went perfectly fine, except for Volti breaking (dependency version mismatch). I substituted Volti with Volumeicon to fix this .

The new apt (I started using Debian during the good old aptitude times) does dependency resolution much better so less overall breakage.

Head_on_a_Stick · 2016-04-07 19:25:22

Any users of Intel graphics cards on newer machines should try removing the xserver-xorg-video-intel package (the Intel DDX driver) and rely on modesetting instead.

packages.debian.org wrote:

The use of this driver is discouraged if your hw is new enough (ca. 2007 and newer). You can try uninstalling this driver and let the server use it's builtin modesetting driver instead.

As of Xorg 1.18, this now provides DRI3 page flipping & Glamor acceleration and is less buggy than the Intel (DDX) drivers.

I've just tried it in my Arch Haswell (HD4600) system and it works a treat.

More details here:
https://www.reddit.com/r/archlinux/comm … ideointel/

Snap · 2016-04-08 05:48:10

Not my case, but this is a great piece of info to keep in mind. So many computers are delivered with stock Intel cards. Thanks, HoaS.

Eraph · 2016-04-08 06:07:23

Snap wrote:

Not my case, but this is a great piece of info to keep in mind. So many computers are delivered with stock Intel cards. Thanks, HoaS.

+1! Seems as though SID provides more stability in some areas at the expense of others.

Head_on_a_Stick · 2016-04-08 08:04:47

It appears there is some debate over this issue:
https://bugs.freedesktop.org/show_bug.cgi?id=94843

I will report back with some benchmarks...

Head_on_a_Stick · 2016-04-09 17:16:40

Head_on_a_Stick wrote:

I will report back with some benchmarks...

Here is the set for the modesetting driver:
https://openbenchmarking.org/result/160 … MPTY690222

Watch this space for the Intel DDX results...

EDIT:
https://openbenchmarking.org/result/160 … MPTY960422

Looks like the Intel driver does have an advantage.

Eraph · 2016-04-09 21:52:14

Are these all X11 benchmarks? I wonder how it compares for OpenGL performance. Or is that all down to mesa?

Head_on_a_Stick · 2016-04-09 22:49:29

Eraph wrote:

is that all down to mesa?

Yes.

The xserver-xorg-video-intel package provides the DDX driver for 2D acceleration, Mesa povides the DRI driver for 3D acceleration.

The kernel component, the i915 module, completes the Intel driver triumvirate.

The Intel DDX driver does not support Glamor acceleration (in Arch & Debian) but it does support DRI3 page flipping if it is enabled as an option; the modesetting driver uses Glamor acceleration by default but does not offer DRI3, at least according to my Xorg.0.log which is strange because modesetting(4) claims that DRI3 should be used by default

EDIT: the modesetting driver does indeed support DRI3 but it does not write this to the log; use `glxinfo` to determine the DRI status.

Last edited by Head_on_a_Stick (2016-04-27 18:24:53)

KrunchTime · 2016-04-10 05:20:09

libc-dev-bin upgrade to remove a number of packages:

The following packages will be REMOVED:
  build-essential debhelper dh-autoreconf dh-make dh-strip-nondeterminism
  equivs g++ g++-5 geany-plugin-py geany-plugins libc6-dev libexpat1-dev
  libpython2.7-dev libstdc++-5-dev libtool
The following packages will be upgraded:
  libc-dev-bin

I placed the package on hold, dist-upgraded, and then cancelled the hold.

Last edited by KrunchTime (2016-04-10 05:20:32)

KrunchTime · 2016-04-11 08:13:35

libc-dev-bin upgrade issue has been resolved. I dist-upgraded again last evening without issue.

dolly · 2016-04-26 07:08:26

Lost my wi-fi after a dist-upgrade today. Rebooted and it works again.

KrunchTime · 2016-04-27 13:46:04

dolly wrote:

Lost my wi-fi after a dist-upgrade today. Rebooted and it works again.

This usually is a temporary thing after a network-manager upgrade. If you wait awhile, your wi-fi should automatically re-connect. At least that has been my experience.

Okay, I came here for a specific reason. A new package was recently installed automatically during a dist-upgrade. The package name is xserver-xorg-input-libinput. Now I'm not able to configure tap-to-click using synclient commands on my desktop (Logitech wireless touchpad) and tap-to-click no longer works on my laptop (Focalpoint touchpad). I've done some research and this may be a temporary issue due to transitioning to using libinput as a replacement for evdev and synclient for input devices under X11.

Additionally, after a recent dist-upgrade, the box lines for the username and password fields in the logon screen are no longer visible; not a big deal, but a tad annoying. I also had similar issues for fields in some apps using the Bunsen theme, so now I'm using the Industrial theme.

Last edited by KrunchTime (2016-04-27 13:48:02)

Head_on_a_Stick · 2016-04-27 18:00:26

KrunchTime wrote:

xserver-xorg-input-libinput

You should just be able to uninstall xserver-xorg-input-libinput and your system will fall back to xserver-xorg-input-synaptics again.

If you can't remove it without upsetting your applecart (if you see what I mean) then there should be a template configuration file that you can copy over and edit:

# cp /usr/share/X11/xorg.conf.d/90-libinput.conf /etc/X11/xorg.conf.d/
# vim /etc/X11/xorg.conf.d/90-libinput.conf

See libinput(4) for all the options

I tried libinput a while ago in Arch and it's very good these days but I prefer the scrolling algorithm that synaptics uses, the libinput version doesn't "freewheel" enough for me and it's not quite as configurable (yet).

YMMV.

EDIT: https://wiki.archlinux.org/index.php/Libinput

Last edited by Head_on_a_Stick (2016-04-27 18:09:15)

Head_on_a_Stick · 2016-04-27 18:03:15

KrunchTime wrote:

Additionally, after a recent dist-upgrade, the box lines for the username and password fields in the logon screen are no longer visible; not a big deal, but a tad annoying.

I take it that testing/unstable has gone to GTK3.20 then?

See https://bbs.archlinux.org/viewtopic.php?id=210815

tl;dr: update your theme or switch to Adwaita

KrunchTime · 2016-04-28 11:08:05

@HoaS: You beat me to it. devil, over at Siduction, had a similar response. However, I'll still include the link to his response since he provided the command to turn tap-to-click on.

Yeah, I figured the GUI issues might be due to the GTK 3.20 update. There's an upgrade warning about it over in the Siduction forums.

Thank you for your input.

Head_on_a_Stick · 2016-04-28 18:00:43

If you want to keep on using synaptics with the libinput package installed, specify the driver in an xorg configuration file:

Section "InputClass"
        Identifier "touchpad catchall"
        Driver "synaptics"
        MatchIsTouchpad "on"
        MatchDevicePath "/dev/input/event*"
EndSection

KrunchTime · 2016-05-11 01:00:06

^ Thank you for the tip. I'm sticking with libinput for now. Double-tap works somewhat well, although sometimes a single-tap is registered or nothing is registered at all. Update today makes single-tap less responsive and I'm experiencing some weird anomolies, like context menus appearing without touching the touchpad.

KrunchTime · 2016-08-12 23:24:17

Just a heads up, a recent upgrade borked wireless connectivity on my Unstable instance. I'm not sure which package it was, but a dist-upgrade today seems to have resolved the issue.

dolly · 2016-08-16 06:58:14

Got this today, but how critical is it?

critical bugs of systemd (231-1 → 231-3) <Outstanding>
 b1 - #834367 - systemctl daemon-reexec (as run on systemd upgrade) causes all keystrokes to go to text console in addition to X (including passwords)
serious bugs of gnupg (1.4.20-6 → 2.1.14-5) <Outstanding>
 b2 - #834368 - gnupg: update to gnupg2 broke gpg-agent autostart
 b3 - #834399 - gnupg: gnupg2-bases gpg breaks Gajim

Head_on_a_Stick · 2016-08-16 07:41:40

dolly wrote:

how critical is it?

That would depend on how often you run the command

systemctl daemon-reexec

I've never even heard of that option so I'm thinking you should be OK

In fact, from systemctl(1):

This command is of little use except for debugging and package upgrades.

No keystrokes are passed during the upgrade anyway, the password is given before APT is invoked so it should be "safe".

If you were feeling paranoid then `sudo -i` could be used to attain a root shell first and then run `apt-get dist-upgrade` from there but it shouldn't make any difference.

Always remember to backup or snapshot before a `dist-upgrade` though!
O:)

#261 2016-04-05 19:55:59

Re: Tracking Debian testing/unstable with BunsenLabs

#262 2016-04-07 19:25:22

Re: Tracking Debian testing/unstable with BunsenLabs

#263 2016-04-08 05:48:10

Re: Tracking Debian testing/unstable with BunsenLabs

#264 2016-04-08 06:07:23

Re: Tracking Debian testing/unstable with BunsenLabs

#265 2016-04-08 08:04:47

Re: Tracking Debian testing/unstable with BunsenLabs

#266 2016-04-09 17:16:40

Re: Tracking Debian testing/unstable with BunsenLabs

#267 2016-04-09 21:52:14

Re: Tracking Debian testing/unstable with BunsenLabs

#268 2016-04-09 22:49:29

Re: Tracking Debian testing/unstable with BunsenLabs

#269 2016-04-10 05:20:09

Re: Tracking Debian testing/unstable with BunsenLabs

#270 2016-04-11 08:13:35

Re: Tracking Debian testing/unstable with BunsenLabs

#271 2016-04-26 07:08:26

Re: Tracking Debian testing/unstable with BunsenLabs

#272 2016-04-27 13:46:04

Re: Tracking Debian testing/unstable with BunsenLabs

#273 2016-04-27 18:00:26

Re: Tracking Debian testing/unstable with BunsenLabs

#274 2016-04-27 18:03:15

Re: Tracking Debian testing/unstable with BunsenLabs

#275 2016-04-28 11:08:05

Re: Tracking Debian testing/unstable with BunsenLabs

#276 2016-04-28 18:00:43

Re: Tracking Debian testing/unstable with BunsenLabs

#277 2016-05-11 01:00:06

Re: Tracking Debian testing/unstable with BunsenLabs

#278 2016-08-12 23:24:17

Re: Tracking Debian testing/unstable with BunsenLabs

#279 2016-08-16 06:58:14

Re: Tracking Debian testing/unstable with BunsenLabs

#280 2016-08-16 07:41:40

Re: Tracking Debian testing/unstable with BunsenLabs

Board footer