Bug 334676

I am setting this as major because it is getting harder and harder to work with this bug.

Just as a reminder:
* current app does not matter for sure
* cause could be something with:
  tab key
  win key
  _really_ rapidly pressing a key (when you press tab several times fast and say "come on..." :-D)

Do you use compiz? Does this also happen when you run another window manager? Does it also happen under GNOME?

No compiz, no xgl, no nothing special. Just basic stuff (KDE). I don't use any other window manager (including Gnome).

Btw. it didn't happen ever before OS 10.3.

Stefan is really good at tracing down such problems ;-)...

I'm afraid Stephan overestimates me here. Anyway, please attach /etc/X11/xorg.conf and /var/log/Xorg.0.log* after right after this issue happens again. Maybe you can also find some lines in /var/log/messages and/or /var/log/warn (check for the timestamps in these files).

What should I look for in those two files you mentioned at the end?

Created attachment 180194 [details]
/etc/X11/xorg.conf

Honestly I can't say. You could just attach the lines with the timestamp, that matches the time at that this issue occurs.

Isn't this that "sticky keys" bug?
open kcontrol > "Regional & Accessibility > Accessibility. Then click in the "modifiers key" tab and UNcheck "use sticky keys". Maybe this is the problem...

So you mean this is a duplicate of Bug #106886? Maciej, did you hold the Shift key for a longer time?

Actually I'm not sure it's needed to hold shift. Maybe other keys may also cause the bug.. Let's wait for Maciej to test.

Oh man... thanks for my IQ estimation :-DDD I am not _that_ overworked :-)))

I had never used sticky keys I don't use them now (unless there is some hidden option).

I believe I upgraded to OS 10.3 a week later after its release, so far it happened three times. So I have to wait for it to happen again.

And it didn't happened ever before 10.3, so I would notice any sticky keys or similar feature (I guess).

oh, sorry. I didn't even know "sticky keys" exist. Anyway, I had a similar problem that was solved reinstalling Xorg and disabling "Plug and Play OS" in the BIOS. And I suggest not installing Xorg 7.3...

Maciej, can you please also add which file-system you're using?

Stephan, Reiserfs from the beginning (from Suse 10.0).

My question from comment #5 has still not been fully answered. :-(

Sorry, with answering I resetted the NEEDINFO. I answer all questions on next keyboard freeze.

Any news on this one? Did keyboard freeze happen again?

Stefan, it didn't happen. Maybe kernel upgrade helped?

So I am closing this as worksforme, and if it happen again I will reopen, ok?

Ok.

Created attachment 182746 [details]
/var/log/Xorg.0.log

Created attachment 182747 [details]
/var/log/messages.bz2

Created attachment 182748 [details]
/var/log/warn.bz2

warn.bz2

So. It happened again? It would have been enough to send me the lines of messages/warn at the time the issue happened. This also would have saved me the time to ask when it happened ... 

> It would have been enough to send me the lines of
> messages/warn at the time the issue happened. This also would have saved me
> the time to ask when it happened ...

Stefan, try compress files, copy them, and attach them using only mouse :-)) -- it took me almost an hour to do so, so I wanted to send those as fast as possible.  For time sake and (it didn't happen yet) I could not be sure that I don't loose mouse also.

This time it happened when I was switching desktop -- I was pressing win+1, win+2, win+3, win+4 back and forth, and at some point there was no switch.

Well, but without knowing for which time stamp I need to look the logfiles are completely useless for me. :-( 

BTW, I'm talking about system time, so this matches the time stamps in the logfiles.

What is the meaning of win+1, win+2, ...?

> Well, but without knowing for which time stamp I need to look the logfiles
> are completely useless for me. :-( 

I don't understand this -- I sent log files _right away_ when keyboard was dead. So those are the last entries -- in warn file ~20 lines in messages more. But I didn't cut anything just to provide full information.
For simplicity you can just look for the last entries from _today_. Keyboard went dead shortly after booting.

> What is the meaning of win+1, win+2, ...?

It is a keyboard shortcut for desktop switch: to desktop 1, desktop 2, and so on. Since the last time it happened for win+tab, I suspect it has something to do with win key -- but I am not sure, it happened only four times.

> I don't understand this -- I sent log files _right away_ when keyboard was
> dead. So those are the last entries -- in warn file ~20 lines in messages
> more.
Could you do this for me, please? Still it makes sense to only look at the lines of the minute in which the issue occured. And only you know this minute.

Well, I don't know the exact time -- I was busy typing "bzip2 -9 /var/log/messages' using mouse, so I focused solely on that and only that. I tried to to it as fast as possible because who know -- maybe the whole system could freeze and I would lost the opportunity to send logs.

Anyway, I could cut what I _think_ it is around that time. But I doubt it would help.

Well, it would have been more helpful to look at your clock, then relax, reboot and cut & paste the appropriate lines of your messages/warn files.

Hmm, no data would be overwritten?

Ok, so -- what should I do now? Wait for another freeze or try to separate lines from current logs?

No, data is not overwritten WRT to warn/messages files. There was nothing obviously wrong in Xorg.0.log. Easiest would be to wait for the next freeze.

So this no longer happened since 3 weeks? I suggest to reopen if it occurs again and attach the appropriate lines of messages/warn files. See comment #33. Closing as WORKSFORME for now.

It happened again, time around 7.42-7.44.

Created attachment 186110 [details]
messages

Created attachment 186111 [details]
warn

Created attachment 186112 [details]
Xorg.0.log

Only network related stuff in messages/warn. Only suspicious thing is in Xorg.0.log:

BOGUS LENGTH in write keyboard desc, expected 5188, got 5192

Not sure if this warning is related to the keyboard breakage at all. Xorg.0.log sounds like the new Xserver logfile after restarting the Xsession.
Xorg.0.log.old would be the old one. Can you attach this?

Created attachment 186129 [details]
/var/log/Xorg.0.log.old

(In reply to comment #42 from Stefan Dirsch)
> Not sure if this warning is related to the keyboard breakage at all. Xorg.0.log
> sounds like the new Xserver logfile after restarting the Xsession.
> Xorg.0.log.old would be the old one. Can you attach this?
> 

No problem, but I think that now those files are the same -- because I turned off and on computer meanwhile.

And yes, there was new Xsession because I followed your steps :-) I noted time, I closed session with dead keyboard and opened a new one with working keyboard, and then I edited those files.

Yes, these are about the same. So this time the might be interesting one has been overwritten. :-( At least now it's proven that the suspicious warning is not the culprit. I currently have no better ideas than

a) waiting for the next keyboard freeze and attach Xorg.0.log.old right after
   freeze
b) checking for keyboard BIOS settings. BTW, is this a USB or PS/2
   keyboard? ("hwinfo --keyboard" should tell us)
c) update BIOS

> a) waiting for the next keyboard freeze and attach Xorg.0.log.old right after
>    freeze

Ok.

> b) checking for keyboard BIOS settings. BTW, is this a USB or PS/2
>    keyboard? ("hwinfo --keyboard" should tell us)

Recognized as PS2. It is laptop keyboard (Dell Latitude D610).

> c) update BIOS

Ok, it is always a good idea, but it worked in 10.0, 10.1 (alpha), 10.2 and windows without any problem. It is not a proof but rather strong indication that something is wrong with 10.3 not with keyboard.

Ok. PS/2 shouldn't be a problem. I expected this, since I didn't see any USB reconnect messages. Then I'm afraid we need to go back to a) and hope that
we see something useful in the Xserver log the next time the issue occurs.

Thanks Stefan, btw. BIOS is up to date.

Created attachment 187768 [details]
Xorg.0.log

Xorg.0.log from running session (session with dead keyboard).

Thanks. Still the only suspicious warning is

  BOGUS LENGTH in write keyboard desc, expected 5296, got 5300

and we learned that this message also occurs in the logfile when the Xserver still works, so this can't be the culprit.

I'm afraid I can't help here (investigating). I propose to update Xserver and keyboard driver from time to time and/or switch to evdev driver as keyboard driver. Possibly this will help. I apologize.

Stefan, thank you for the answer.

Do you know what can be th cause of this warning? Or possible ways to fix it?

Not exactly. It's deep in Xserver code. But as already said it's not the culprit.

I though of keyboard definitions side.

> But as already said it's not the
> culprit.

True, but add one weak point to another and you get a bug as in such case.

openSUSE 11.0 Beta1 is planned for this week. In case you're considering to test again with this Beta release, whether this issue still occurs, please give me feedback about your results. Otherwise please let me know, so I will close it as LATER again and reopen it once openSUSE 11.0 final is available for testing.

I'd rather wait for final version, especially the keyboard bug needs time to occur.

openSUSE 11.0 has been released today. Please give it a try and let me know, if the issue is fixed now. Thanks.

needinfo

I close it for now, because I rather wait for KDE4.1 and then I upgrade both opensuse and KDE, so I'll be not able to send any new info for some time. If problem will occur again, I simply reopen this one, ok?

Ok. Sounds like a plan. :-)

So, back to this issue.

Now it is Dell Latitude E6500, openSUSE 11.1 64-bit, KDE 3.5.10 (because KDE4 is not ready enough for me) and the problem _exactly_ as before with workaround (close all program using mouse, and logout, login) as before.

Please let me know how I can help solving this issue.

(I am leaving the old values, because I don't remember opensuse policy -- if those values should reflect the oldest data with the problem or the newest).

Reassigning to Matthias. He's currently working on keyboard issues anyway.

It is very likely it was caused by some glitch in kernel alsa driver. I reinstalled a newer version and so far, it didn't happen (previously it happened three times already). Closing and I hope for good :-)