Red Hat Bugzilla – Bug 448351
directfb application segfaults randomly
Last modified: 2008-10-27 14:45:05 EDT
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:184.108.40.206) Gecko/20080404 Firefox/220.127.116.11
Description of problem:
I wrote a program for directfb that worked fine under i386 architecture. However, under x86_64, it apears to randomly segfault. It ALWAYS segfaults at dfb_config_init, conf.c:1274.
gdb shows that all of the variables being passed in are being "optimized out". This is probably why the segfault is occurring, argv, argc, *and* i are being optimized out. I am not running gcc with any optimization flags.
a related problem that I may need to open another bug for, is that gdb refuses to read the symbols in the program I compiled - it seems to only use the ones from the debuginfo. If I try to load the object files manually, it says "the address where <file> has been loaded is missing".
The really befuddling part is - it only doesn't work *sometimes*. A few times, the program will start.
There are no hardware or memory issues that I'm aware of, I just built the machine three days ago.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Write a program using directfb, and pass argc and argv in to the init routine.
2. ???? *
* south park reference, there is no step 2.
stock fedora install.
FYI directfb has been updated to 1.2.6 in Rawhide.
You might want to test against this newer version.
Actually it should be possible to downgrade the release field when building directfb. So it will have the same SONAME as the current F-9 1.0.0 version.
I'm closing this bug to rawhide as there is low chances to have the package updated because of the ABI bump.