[arch-general] Keyboard lag issue

[Manuel Groß]

On Mon, 20 Jun 2016 23:15:02 +0200
Bruno Pagani <bruno.pagani at ens-lyon.org> wrote:

> Le 20/06/2016 à 22:55, Gaëtan Harter via arch-general a écrit :
> 
> > On Mon, Jun 20, 2016 at 7:17 PM, Manuel Groß <mgr at nordkrater.de>
> > wrote: 
> >> Hello there,
> >>
> >> I currently experience a weird keyboard lag issue which I yet
> >> failed to resolve. I had some help looking into it a bit (and
> >> obviously searched the web), but we couldn't identify the speficic
> >> problem. I hope someone could provide me pointers or hints on what
> >> to test out.
> >>
> >> I have a Thinkpad x250 running Arch Linux with Openbox and X11 on
> >> it. I use systemd-logind to log in, so X is started with the 'exec
> >> openbox-session' from my xinitrc. Not from some fancy lightdm or
> >> so. I do not have a certain date since when the issue occurs, but
> >> it has probably been a little over a month. I didn't make any
> >> major changes to my system then, and I do update regularly. This
> >> didn't occur when I set up the system.
> >>
> >> The issue is, that on some, certain programs, the keyboard input
> >> 'arrives' too late, with delays up to 1-2s. In other programs, this
> >> never happens at all. And in rare occasions, it is even fine for
> >> programs, in which I experience the problem.
> >> I checked logs at obvious locations, but neither the systemd
> >> journal, dmesg or the Xorg log were screaming 'I have an issue
> >> here'. The trackpoint always works as excpected and the touchpad
> >> is disabled.
> >>
> >> Programs where the delay occurs:
> >>  * sakura
> >>  * sublime
> >>  * wireshark (qt5 version)
> >>
> >> Programs which are fine:
> >>  * claws-mail
> >>  * geany
> >>  * xterm
> >>  * chromium
> >>
> >> The delay also never occurs in the system shell (Alt+F2) and
> >> xterm, so it's not the shell or oh-my-zsh.
> >> The program lists suggests, that it is not a problem due to a
> >> specific version of GTK or QT, since programs using either library
> >> (or none of them) are affected.
> >>
> >> I tried using 'libinput' instead of 'evdev' as a keyboard driver,
> >> but the change had no effect (I double checked the Xorg log which
> >> driver was actually used).
> >> In 'xev', all keystrokes arrive immediately.
> >>
> >> So seems not to be the kernel or the driver, nor a specific gtk or
> >> qt library.
> >> The next thing to test would probably be the xlib and the libxcb
> >> themselves. So I used 'downgrader' to downgrade each a few
> >> revisions (also their lib32-libraries) and checked for a change,
> >> without success.
> >>
> >> I don't know what to try next. Does anyone of you have an idea or
> >> already solved a similar issue? I'd appreciate any help on this.
> >> --
> >> Grüße/Regards
> >> Manuel Groß
> >>  
> > Hi Manuel,
> >
> > I have a Thinkpad T440p and I have the same issue with kernel 4.6,
> > with both 4.6.2-1 and 4.6.1-XX versions.
> > There is no issue with lts kernel 4.4.13-1.
> >
> > From what I experienced, gnome-terminal is impacted, and I think
> > also firefox.
> >
> >
> > Regards,
> > Gaëtan Harter  
> 
> Did you have a look at these:
> 
> https://bugs.archlinux.org/task/49431
> https://bugs.archlinux.org/task/49623
> 
> ? 

No, I did not. I completely forgot to check the bugtracker. I'll do
that firs next time I have a problem.

> If not, try disabling fbc. Eventually report upstream if it still
> hasn’t been done (I’m not affected myself, so…).

Disabling fbc solved the problem for me, I'm now back again to 4.6.2-1.
What's strange is, that not only QT applications are affected. sakura
uses GTK, and sublime uses a custom UI toolkit. All of those are fine
now.
I'll let them know in the comments, so they can check if it's the same
bug.

Thanks for your help!
-- 
Grüße/Regards
Manuel Groß
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <https://lists.archlinux.org/pipermail/arch-general/attachments/20160621/e77d0d8e/attachment-0001.asc>