Description of problem: Linux sludge.devel.redhat.com 2.4.20-2.49 #1 Mon Feb 17 11:29:55 EST 2003 i686 i686 i386 GNU/Linux [root@sludge test]# rpm -q procps procps-2.0.11-6 [root@sludge test]# rpm -q kernel kernel-2.4.20-2.14 kernel-2.4.20-2.24 kernel-2.4.20-2.47.1 kernel-2.4.20-2.49 kernel-2.4.20-8 Not exactly sure whats going, but it appears top is showing no processes as using anything more than 0.0% cpu, and only a few processes as taking more than 0% mem... top -b -n 1 output attached, but a sample chunk... 1 root 15 0 92 56 40 S 0.0 0.0 0:00 0 init 2 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 keventd 3 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kapmd 4 root RT 0 0 0 0 SW 0.0 0.0 0:00 0 kdog/0 5 root 34 19 0 0 0 SWN 0.0 0.0 0:00 0 ksoftirqd_CPU 8 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 bdflush 6 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kswapd 7 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kscand 9 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kupdated 10 root 25 0 0 0 0 SW 0.0 0.0 0:00 0 mdrecoveryd 14 root 15 0 0 0 0 DW 0.0 0.0 0:00 0 kjournald 72 root 25 0 0 0 0 SW 0.0 0.0 0:00 0 khubd 1725 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kjournald Thats supposedly sorted by CPU useage, on a machine with a couple of cpu using things running... This seems to have started happening recently, perhaps after my last reboot. Offhand I'm not sure if I changed kernels at the time, but thats definately possible...
Created attachment 91154 [details] top -b -n1 output
same thing has happened to me, kernel 2.4.20-2.48smp and glibc-2.3.2-27.9
Its very strange that all TIME field are zero. That would seem to indicate that both utime and stime are zero. Can you paste in /proc/$pid/stat for some process when this happens? Also, are you using LD_ASSUME_KERNEL or anything strange thread-related?
/proc/pid/stat works: 11278 (slurpcpu) R 11257 11278 11102 34834 11281 0 20 0 60 0 1180 0 0 0 25 0 0 0 84160728 1380352 58 4294967295 134512640 134513592 3221218944 3221218864 134513412 0 0 0 0 0 0 0 17 0 1180 0 0 0 11278 (slurpcpu) R 11257 11278 11102 34834 11283 0 20 0 60 0 1538 0 0 0 25 0 0 0 84160728 1380352 58 4294967295 134512640 134513592 3221218944 3221218864 134513412 0 0 0 0 0 0 0 17 0 1538 0 0 0 11278 (slurpcpu) R 11257 11278 11102 34834 11285 0 20 0 60 0 2047 0 0 0 25 0 0 0 84160728 1380352 58 4294967295 134512640 134513592 3221218944 3221218864 134513412 0 0 0 0 0 0 0 17 0 2047 0 0 0 I do use LD_ASSUME_KERNEL on threaded applications which need it (Mozilla, Python etc.), but not on the system as a whole.
Ah, I see what's happening. Thats not the final RHL9 kernel. It doesn't have the &P->rtprio, &P->policy, fields in /proc/$pid/stat, so the fields added to stat in that kernel version are incompatible. You need to upgrade the kernel.