On Fri, 2011-08-19 at 11:51 -0600, James Tappin (AFRL SMEI Scientist) wrote: > Package: linux-2.6 > Version: 2.6.32-35 > Severity: normal > > > When a process exceeds 2^31-1 (probably) milliseconds of CPU, the CPU > time counter becomes corrupt. [...] This wouldn't entirely surprise me on a 32-bit kernel with 1000 Hz timer, but we don't build any kernel packages with that configuration. Which kernel did you actually find this on? Please send contents of /proc/<pid>/stat for a process in this state. Ben.
Attachment:
signature.asc
Description: This is a digitally signed message part