Bug 203284

Summary: arm64: bogus CPU % accounting in top
Product: Base System Reporter: Ed Maste <emaste>
Component: kernAssignee: freebsd-bugs (Nobody) <bugs>
Status: Closed FIXED    
Severity: Affects Only Me    
Priority: ---    
Version: CURRENT   
Hardware: arm64   
OS: Any   
Bug Depends on:    
Bug Blocks: 203349    

Description Ed Maste freebsd_committer freebsd_triage 2015-09-23 13:43:58 UTC
a.out is a CPU-bound process that spins at 100% reading and writing memory in a loop, with 32 copies running in parallel on the 48-core system, so I'd expect to see all of them reporting 100% CPU.

A snapshot of top looks like:

  PID USERNAME    THR PRI NICE   SIZE    RES STATE   C   TIME    WCPU COMMAND
  822 root          1 103    0  1030M  1024M CPU2    2  34:38 1779.83% a.out
  819 root          1 103    0  1030M  1026M CPU17  17  34:41 458.17% a.out
  817 root          1 103    0  1030M  1026M CPU44  44  34:41 444.14% a.out
  815 root          1 103    0  1030M  1026M CPU42  42  34:41 444.14% a.out
  806 root          1 103    0  1030M  1012M CPU38  38  34:44 356.67% a.out
  807 root          1 103    0  1030M  1013M CPU30  30  34:44 356.24% a.out
  812 root          1 103    0  1030M  1021M CPU5    5  34:41  31.81% a.out
  837 root          1  20    0  7660K  1896K CPU25  25   3:16   1.63% top
  809 root          1 103    0  1030M  1020M CPU32  32  34:41   0.00% a.out
  814 root          1 103    0  1030M  1021M CPU26  26  34:40   0.00% a.out
  811 root          1 103    0  1030M  1020M CPU45  45  34:40   0.00% a.out
  810 root          1 103    0  1030M  1016M CPU24  24  34:40   0.00% a.out
  805 root          1 103    0  1030M   995M CPU29  29  34:38   0.00% a.out
  820 root          1 103    0  1030M  1016M CPU33  33  34:38   0.00% a.out
  821 root          1 103    0  1030M  1020M CPU8    8  34:37   0.00% a.out
  818 root          1 103    0  1030M  1005M CPU19  19  34:37   0.00% a.out
  808 root          1 103    0  1030M  1005M CPU36  36  34:36   0.00% a.out

Sometimes many process report 400% or so, sometimes only a few.
Comment 1 Ed Maste freebsd_committer freebsd_triage 2015-09-25 13:47:17 UTC
See r288216
https://svnweb.freebsd.org/changeset/base/288216
Comment 2 Ed Maste freebsd_committer freebsd_triage 2015-11-27 22:06:58 UTC
Addressed by r288216