Bug 1044820

Summary: xf86-input-libinput-0.25.1-1.1 makes non-GTK applications unusable
Product: [openSUSE] openSUSE Distribution Reporter: Forgotten User QFh0D9oMPg <forgotten_QFh0D9oMPg>
Component: X.OrgAssignee: E-mail List <xorg-maintainer-bugs>
Status: RESOLVED WONTFIX QA Contact: E-mail List <xorg-maintainer-bugs>
Severity: Major    
Priority: P2 - High CC: cfeck, forgotten_BUvhHT2XhU, forgotten_QFh0D9oMPg, msrb
Version: Leap 42.3Flags: msrb: needinfo? (forgotten_BUvhHT2XhU)
Target Milestone: ---   
Hardware: x86-64   
OS: SUSE Other   
Whiteboard:
Found By: --- Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: output of "strace firefox"

Description Forgotten User QFh0D9oMPg 2017-06-18 18:15:32 UTC
I use an Microsoft Surface Pro 3 (Intel graphics), which I updated from 42.2 to 42.3. This meant also an upgrade to xf86-input-libinput-0.25.1-1.1 (currently).

Using Gnome and IceWM, only GTK apps are starting. Other apps (i.e.: Qt-based, Firefox, Chrome) simply crash at startup.

Reverting the single library given above to the version provided in 42.2 (currently xf86-input-libinput-0.19.1-1.1) solves the problem.

Example of the coredumps:
https://lists.opensuse.org/archive/opensuse-factory/2017-06/msg00430.html

More info on the hw:
https://lists.opensuse.org/archive/opensuse-factory/2017-06/msg00466.html

I tried 42.3 stock kernel and HEAD - no difference.
I removed the Xorg Intel driver in order to use generic modesetting - no difference.
Comment 1 Stefan Dirsch 2017-06-18 20:12:31 UTC
Sounds weird to me, but if it's really the xf86-input-libinput driver version it would be rather easy to git bisect ...
Comment 2 Forgotten User QFh0D9oMPg 2017-06-19 03:52:11 UTC
I did not find any useful information in the usual logs. If I can provide any better input, just tell me.
Comment 3 Stefan Dirsch 2017-06-19 14:19:28 UTC
That's why it would be easiest to git bisect the issue. If anybody else can reproduce the issue. Otherwise you would need to do it yourself. I'll try to reproduce the issue next week. I'm not in the office this week. If you're familiar with git bisecting just go ahead. ;-)
Comment 4 Forgotten User QFh0D9oMPg 2017-06-19 17:44:12 UTC
Created attachment 729426 [details]
output of "strace firefox"

Hum, I don't even know what git bisect is, I'm afraid. Maybe the strace output helps as an alternative?
Comment 5 Stefan Dirsch 2017-06-26 10:27:04 UTC
Hmm. I cannot reproduce this issue at all. Tested in a VM with the current Leap 42.3 packages. I made sure that only libinput driver is installed and being used.

firefox, chromium and digkam can be started without any problems.

I'm afraid you need to bisect this issue yourself. :-(

Closing as WORKSFORME.
Comment 6 Forgotten User BUvhHT2XhU 2017-07-01 07:20:22 UTC
I'm also affected by this bug on a Surface Pro 3.

I have git bisect the bug and found out, that the problem has something to do with the N-Trig digitizer for the pen and libinput.
If I tell the x-server, to use the evdev driver instead of the libinput driver, everything works fine.
On openSUSE Tumbleweed, everything works fine for me.
When I have time, I'll try to figure out why everything is working in Tumbleweed but not in Leap 42.3.

@Larx Noname:
Can you please append the following lines at the end of the file 
"/etc/X11/xorg.conf.d/40-libinput.conf" and tell me, if this fixes your problem?

    Section "InputClass"
         Identifier "N-Trig pen"
         MatchProduct "NTRG0001:01 1B96:1B05 Pen"
         MatchIsTablet "true"
         MatchDevicePath "/dev/input/event*"
         Driver "evdev"
     EndSection
Comment 7 Forgotten User QFh0D9oMPg 2017-07-01 09:54:30 UTC
I can confirm that the proposed change to X config does the trick. So I'll keep this line in there until a fix is found. Thanks a lot!
Comment 8 Michal Srb 2017-07-03 12:21:19 UTC
(In reply to Raphael Geissinger from comment #6)
> I have git bisect the bug and found out, that the problem has something to
> do with the N-Trig digitizer for the pen and libinput.

Do you remember which was the first broken commit?
Comment 9 Stefan Dirsch 2017-08-07 13:34:54 UTC
Raphael, still available? Since you've git bisected that issue, it should be fairly easy to tell us, what the first broken commit is, right?
Comment 10 Stefan Dirsch 2017-08-22 14:59:23 UTC
Raphael?
Comment 11 Stefan Dirsch 2017-10-19 14:35:05 UTC
I'm afraid without the help of Raphael we can't do much here. :-(