Hello, After following the ULE status thread with some interest, I decided to enable on my i386, hyperthread-enabled, UNP machine. Perhaps not the best test bed, but a test none the less. I'm running a series of make -jX buildworld to put load on the system and running X with XFCE4 to get an idea of system responsiveness. The issue I'm seeing, seconds after launching gnome-terminal, gnome-terminal terminal completely froze. I can move and resize the window, but cannot kill or close it. A quick look at top shows: 38043 djp 20 0 21040K 16456K RUN 1 0:01 0.00% 0.00% gnome-terminal Also, gkrellmd and gkrellm have also both froze. Here's that info: 39481 djp 20 0 16624K 13948K RUN 1 0:02 0.00% 0.00% gkrellm 39321 djp 20 0 4324K 2220K RUN 1 0:00 0.00% 0.00% gkrellmd In my kernel, I've added: options SCHED_ULE # ULE scheduler options SMP # Symmetric MultiProcessor Kernel options PREEMPTION Is there any info I can provide that would shed some light on this? -- Regards, Doug,
On Thu, Feb 10, 2005 at 05:10:22PM -0600, Doug Poland wrote:> > The issue I'm seeing, seconds after launching gnome-terminal, > gnome-terminal terminal completely froze. I can move and resize the > window, but cannot kill or close it. A quick look at top shows: > > 38043 djp 20 0 21040K 16456K RUN 1 0:01 0.00% 0.00% gnome-terminal > 39481 djp 20 0 16624K 13948K RUN 1 0:02 0.00% 0.00% gkrellm > 39321 djp 20 0 4324K 2220K RUN 1 0:00 0.00% 0.00% gkrellmd > > In my kernel, I've added: > > options SCHED_ULE # ULE scheduler > options SMP # Symmetric MultiProcessor Kernel > options PREEMPTION >In addition, both mozilla and firefox "hung" moments after entering the command from an aterm. They were both in the same state as above. I was pushing loads of about 15 and using Xorg quite smoothly (with Nvidia drivers), also decoding mp3 stream with no skips or interruptions. Then I launched OpenOffice.org-1.1.4. I got the splash screen and the entire interface froze, no pings, nothing. Forced to perform a software reset. Is this information useful? Is there something else I can do as a ULE "tester" to make these data more useful? -- Regards, Doug