Bugzilla – Bug 1158190
missing keyboard & mouse events ...
Last modified: 2021-12-31 13:05:34 UTC
I've been suffering terrible keyboard (and some mouse) input with 15.1. My machine is ~constantly under some sort of load, and the K/B response iss terrible. I routinely get omitted key-presses, very often I lose modifiers - so editing in emacs I get 'f's and 'b's as the 'Alt' is dropped: these are often whn I'm trying to jump words to edit the missing / duplicated characters that didn't get entered. Switching virtual desktops is even wose - <ctrl><alt><arrow> keys frequenly fail to do much. I type rather quickly, but even so - hving to go back and edit amot everything I type [ ok, so I gave up here ;-], is hyper-annoying. Worse - it is a the threshold that it could jut be a c omeetence problem [ perhaps I did't press the kys ? ] - but certainly I didn't type 'ee' twice adjacently in competence but a 'p' then an 'e' just above etc. etc. Somtimes I get doubled or even trippled letters arriving too at random. Anyhow - it's slowly driving me up the wall ;-) I've tried the very latest suse kernel snapshots and no joy. I've used 'evdev' to trace the events and am missing events for the keystrokes. The problem recurs on the console as well - so, X / Wayland seem to be unrelated. I'm buying a new H/W keyboard for my Dell 7559 - but since this used to work reasonably nicely in the past, I'm fairly sure it's a S/W problem. Advice much appreciated - is there a good/easy way of testing a old eg. openSUSE 42.X kernel on this 15.1 machine somehow ? =) What else can I do to debug what the kernel is up to ? =) Thanks.
Created attachment 825274 [details] hwinfo in case it helps
For clearing ambiguous points: - This is about the laptop built-in keyboard, right? - Does this happen with the external USB keyboard as well? And, suppose that it's specific to built-in keyboard. It's i8042, and the code has been always there. So the question is whether it's a kernel regression or not. For checking that, first off, prepare the minimal test setup, e.g. boot without X (runlevel 3), and test on Linux console with evtest. Once after we get the test environment, install older kernels (Leap 15.0, Leap 42.x, etc) on the system and test with it. You can simply install it via zypper in --oldpackage kernel-default.rpm and ignore the warning about signature, then choose the old kernel from GRUB menu. But it'd be safer to increase the number of installable kernels before doing that; edit /etc/zypp/zypp.conf and add more entries for multiversion.kernels line, e.g. multiversion.kernels = latest,latest-1,latest-2,latest-3,running After testing, you can uninstall the old kernel again, and that's done.
Your intention is correct! My previous SuseLinux version was Leap 42.3. I remember me, an upgrade(!) from this version to Leap 15.1 runs Leap 15.1 perfect. Seeming runs this upgrade version 15.1 with the kernel of Leap 42.3. You should test it!
Sounds good - I re-boot rather infrequently, but let me try with the openSUSE 43.2 kernel if I can. I tested like this: BOOT_IMAGE=/boot/vmlinuz-5.3.11-1.g0a195a8-default root=UUID=d7f1d717-d6ec-45da-a33a-442675474609 quiet splash showopts nopt acpi_backlight=vendor acpi_osi= i8042.debug i8042.nomux i8042.reset=Y and beyond spamming syslog amazingly, there was nothing better about life =)
Interesting the latest (?) 42.3 kernel: Linux dell.home 4.4.180-102-default #1 SMP Mon Jun 17 13:11:23 UTC 2019 (7cfa20a) x86_64 x86_64 x86_64 GNU/Linux is also showing the problem; hmm.
Then try downgrading further to Leap 42.3 GM kernel -- I suppose it had worked certainly for you in the past? I somehow suspect it might be something in a lower level. Can you take a look at BIOS setup, e.g. disabling the Intel MEI (or whatever called the control console thingy)?
I remember me, after upgrading of Leap 42.3 to version 15.1 all runs perfect. Then I had done a system safe of the root partition (sda2) from this upgrade, but not safed the boot partition (sda1). After some time I installed Leap 15.1. new complete. Then I got this described problems with keyboard and mouse inputs laming and delaying. Now I recovered my upgraded (Leap 42.3 to 15.1) system safing (sda2), start the system new and all blocks in the boot sector by started GRUB, starting was not possible! So I came to the result, that in Leap 15.1. anything was changed in the bootsector in GRUB. I recommend, first try installing the previous boot sector GRUB from Leap 42.3 .
You need the bootsector that corresponds to the grub installed on your system.
Then it might it be some difference of boot option that is gone by some reason? e.g. any reason of passing acpi_osi option with the empty string?
I suppose, that there are some difference of boot option between Leap 42.3 and 15.1 . More I can't say at this moment, I got not more information, since the booting process stopped by my experiment.
I have done following experiments: First I updated my second Linux-OS UBUNTU in my computer with her specific bootsystem GRUB. That runs without problems and it runs OpenSuse Leap, but the started OpenSuse Leap 15.1. have the same described problems. Next I downloaded the 'OpenSuse Leap 15.1.Live' from openSuse and installed it on an USB-stick. Now I start this Live-version and was astonished, that Leap 15.1.Live runs without any problems! Here we must inspect the difference between Leap 15.1.Live and the full installed version Leap 15.1. Any package or element troubles here. If nothing goes and all fails, Leap 15.1. must rebuild new complete, element for element and package for package and test it. It's large work, I know, but it is the last possibility...
Did you try to boot the system with the very same kernel boot options as done with the (good working) live system? Of course, some additional package might influence something, but about PS2 keyboard, it's rather a direct kernel stuff, so the influence must be limited. So, at most, either the power-saving stuff or any additional kernel tuning. But the first suspect is always the kernel boot option, and module configs.
The booting of the good working live system is undepended from the Linux installed kernel boot. You start it with the BIOS before GRUB, choosing the boot medium like bootable(!) hard disks, CD's or DVD's, several external disks or USB-stick, which starts automatically. I installed and started successfully Leap 15.1 Live by an USB-stick.
however, you can look what kernel version and what kernel parameters are used in the live system and try to replicate that on your installed system.
I see in Leap 15.1 two kernels in /boot (I hope it's correct): 1) vmlinuz-,initrd-,config-,sysctl.conf-,System.map- 4.12.14-lp151.28.32 and 2) vmlinuz ....4.12.14-lp151.28.36 . In Leap Live I see only one kernel: config-,sysctl.conf-,system.map-4.12.14-lp151.28.16 and 4.12.14-lp151.28.16-default . Why? By the way: You can download Leap Live by https://software.opensuse.org
I installed kernel 4.12.14-lp151.28.16 of Leap Live, but is has no effect, the problems are the same.
Michael (Meeks), have you managed to test 42.3 GM kernel as Takashi asked? Also, if I understand comment 4 correctly, some of those debug options solved the problem. Correct? Have you tried to narrow it down to one of them?
No response, closing.
Closing.