Hi there, I'm running gentoo-sources-2.4.20-r1 on my desktop now and installed and ran openoffice-bin for the first time since upgrading the kernel. The problem is that (say, when running oowriter) the initial window appears - however it quickly becomes apparent that it is very sluggish, then OO siezes up completely when trying to open a dialog from a menu or such. The window will not refresh, and xkill is the only way out. Marked as UNCONFIRMED because I am not at all certain that this has anything to do with the kernel. I had a few problems installing OO on this occasion - see bug #14092, so maybe this bug is a non-starter. Nonetheless, I thought I'd file this just in case. I do recall OO behaving a little eratically when I was running a kernel many months ago (not yours!) which was havily patched in terms of the scheduling timer, preemptive and so on. My sources have the preemptive option enabled, and I've tried OO both with lowlatency on and off (via the proc interface). I guess the best thing I can do is compile an older kernel to rule this in or out so I'll try and do that ASAP. I've attached my .config, just in case.
Created attachment 7397 [details] My current .config for gentoo-sources-2.4.20-r1
Created attachment 7514 [details] My kernel .config
I'm experiencing the exact same problems with openoffice-bin and gentoo-sources-2.4.20-r1. I have both the low latency and kernel preemption options enabled. I experienced no problems with gentoo-sources-2.4.19-r10.
Hmmm, strange. After I re-installed OO properly (in lieu of #14092), I no longer had a problem. I've rebooted since, so low-latency will be off again (although pre-emptive is still on). In light of what George said, maybe there's something in that.
I had this problem for a long time. It only occured when the CPU usage hit 100%, i.e. all the time when a distributed computing program was running, or nearly continuously when a compile was running (tiny breaks from compiling during file i/o gave a short burst of CPU time to OOo, enough to let it refresh). The problem was corrected in later kernels. I recall seeing a changelog entry a la "Scheduler fixed: process priorities now obeyed :)" or similar when upgrading to the kernel that fixed the problem. The point of all this drivel was to say "hmm, let's close this bug, shall we?" Nothing like squashing open bugs. Whee, I feel like I'm actually contributing to Gentoo! :)
Hey, thanks... that makes sense that the fix to priorities would fix this.
