[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: multiple KDE segfault issues

On Sun, Nov 13, 2011 at 01:07:13AM -0600, Bob Tracy wrote:
> (...)
>     "kdm" currently segfaults during kdm-greeting.  To reproduce, enter
>     your login name and tab to the password entry field: "kdm" will
>     immediately segfault and reinitialize.  Using "gdm3" to access the
>     KDE4 environment works a bit better :-).  Once logged in,
>     "plasma-desktop" segfaults after doing almost anything, including
>     clicking on the "K" menu and then any submenu such as "System".
>     The common items showing up in the stack backtraces for both "kdm"
>     and "plasma-desktop" are QT4 functions: this smells like the same
>     g++ optimizer bug that kills programs based on "libxul.so", e.g.,
>     "iceweasel".

I finished the non-optimized build of all the qt4-x11 packages, and all
things KDE4 now seem to work!!  "kdm" works: got logged-in with a KDE4
session.  "plasma-desktop" no longer segfaults, BUT...

I had just fired-up an instance of "Konsole", and was resizing the
terminal window when the machine locked-up solid.  Had to hit the reset
switch to recover.  I don't think the machine crashed: couldn't find any
"oops" messages in syslog.  The ~/.xsession-errors file had quite a bit
of diagnostic information in it, but nothing that seemed relevant to the
system freeze.  Similarly, the "/var/log/Xorg.0.log*" files contained
only expected information.

At this point, I conclude we definitely have a g++ 4.4 optimizer bug.
It affects at least the binary packages produced from the "qt4-x11" 
source package, and "iceweasel".  The common symptom is application
instability manifesting as segfaults.

Separate from the optimizer issue, KMS bugs apparently *not* specific to
Alpha are being seen by others, and they are seeing black screens, system
freezes, etc., same as I am.

Bottom line: I have high hopes of having a working KDE4 environment on
Alpha before much longer, because the KMS issues are being actively
worked at high priority (due to them showing up on supported
architectures :-)).


Reply to: