Bug 70790 - Login manager loses my keystrokes while inputting password
Summary: Login manager loses my keystrokes while inputting password
Status: RESOLVED WORKSFORME
Alias: None
Product: kdm
Classification: Miscellaneous
Component: general (show other bugs)
Version: unspecified
Platform: RedHat Enterprise Linux Linux
: NOR normal
Target Milestone: ---
Assignee: kdm bugs tracker
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2003-12-19 00:48 UTC by Charles
Modified: 2011-12-04 12:44 UTC (History)
1 user (show)

See Also:
Latest Commit:
Version Fixed In:


Attachments
lock kdm greeter in memory (568 bytes, patch)
2004-01-09 14:22 UTC, Oswald Buddenhagen
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Charles 2003-12-19 00:48:58 UTC
Version:            (using KDE KDE 3.1.4)
Installed from:    RedHat RPMs
Compiler:          Fedora Core 1 compiler I guess 
OS:          Linux

Running Fedora Core 1 (put Redhat in distribution selection since it's basically Redhat 10), running on spiffy and responsive IBM Thinkpad T41, I have problems logging in if I let the machine sit for a bit before logging in.  It drops some of my keystrokes.  Once it wakes up and realizes I'm trying to get in, and denies me the first time, it lets me in the next time just fine.

I am using a USB keyboard, but when I have to switch over to Windows (God forbid), its login screen doesn't drop any keystrokes ... ever.  Is there any way to make the login screen preempt whatever else the kernel is doing so I get my password typed in right the first time, every time?  I've noticed this problem with Redhat bundled KDM since Redhat 8.

I also noticed the same problem with breaking screenlock, too.  Perhaps the authentication code is shared between the two.  I thought that it would go away with my new 1.4 GHZ Pentium M laptop, but it hasn't.

(my compliments to the KDE team for a great UI)
Comment 1 Charles 2003-12-19 00:50:59 UTC
Note: my system is set to auto-focus on my password and pre-select my username.
Comment 2 Oswald Buddenhagen 2003-12-19 01:08:49 UTC
i don't think this is a kde problem, probably not even a qt one. try another Xserver (be it only an earlier or later Xfree release), another kernel (2.6 was just released), and try using a ps/2 keyboard (thereby enforcing the use of a different driver by the kernel (and possibly by the Xserver as well, i don't know)).
Comment 3 Charles 2004-01-07 21:18:36 UTC
I'm bound to the packages of my chosen operating system.  Your suggestions sound useful to an enthusiast with more time than I have.  This appears to be an issue once login manager and screensaver memory segments are swapped in virtual memory, as can happen if you step away from the system and non-interactive cache-hungry processes (like slocate ) wake up and run in your absence.  I can hear the disk madly pulling VM segments off the disk when this problem happens.  Could you send me a patch to try running kdm with realtime (RT) priority?  I know xmms has that feature, if you need a code reference.
Comment 4 Oswald Buddenhagen 2004-01-09 14:22:05 UTC
Created attachment 4067 [details]
lock kdm greeter in memory

real-time does not protect from being swapped out, only locking in memory does.
if this patch actually helps, the problem is in qt, kdelibs, or for some weird
reason even in kdm itself. if it does not help, the next step would be applying
a similar patch to the X server. if that still does not help, disable swap
alltogether.
fwiw, mlock is definitely not the correct fix - i provide this patch only to
help you identify where the problem actually lies.
Comment 5 Oswald Buddenhagen 2004-03-16 15:49:15 UTC
btw, i expect some feedback ...
Comment 6 Charles 2004-03-16 18:30:23 UTC
Sorry for the delay.  Your initial assessment was correct.  I just checked this and it appears that after the latest Fedora XFree86 update that it catches my keystrokes correctly the first time.

Running Fedora Core 1 with XFree86-4.3.0-55 rpm.  Sorry for taking up your cycles on this.
Comment 7 Charles 2004-03-16 18:30:43 UTC
.
Comment 8 Oswald Buddenhagen 2011-12-04 12:44:47 UTC
*** Bug 287231 has been marked as a duplicate of this bug. ***