Re: [kvm-ppc-devel] First detailed look at instruction emulation using kvm_stat instruction extension

[Christian E. <ehr...@li...>]

Christian Ehrhardt wrote:
> We all saw in prior kvm_stat counters that listed the different exit 
> reasons the high count of instruction emulation. The recent patch to 
> kvm-ppc-devel splits up those into the different instructions. I 
> attached a log of a simple guest boot&halt FYI. E.g. see the high number 
> of m[ft]msr instructions.
[...]
sorry for that type - I meants m[ft]spr activity.
The full instruction tracing allows us to check e.g. which sprn's are accessed, in which sequences they are accessed, with which valued and from which pc of the guest kernel.

I still don't fully trust these tracing or the kvm_stat extension because I get dfferent numbers of instructions (but the ratio between them look the same, so I think it's valid to look at the results). I expect that I increment by more than 1 in the kvm_stat instr patches atm - comments welcome (see the 1/3 in the patch series just posted to kvm-ppc-devel).

I attached only a 2000 line partial log of the later part of the guest execution, because the full log is very big (3.9M).
I can send a full log to everyone interested - please send me a mail (hmmm... maybe I upload that somewhere).

A short check of the ratio between the different sprn's used gave that listing:
grep sprn boot_halt_v4.instr.log | awk '{ printf("%s %s\n",$11,$12) }' | sort | uniq -c| sort -rn
   2461 0x01b SRR1
   2256 0x110 SPRG0
   2245 0x111 SPRG1
   2004 0x01a SRR0
   1538 0x113 SPRG3
   1127 0x114 SPRG4
   1034 0x3b2 MMUCR
    979 0x016 DEC
    597 0x030 PID
    509 0x115 SPRG5
    499 0x117 SPRG7
    475 0x150 TSR
    358 0x03d DEAR
    106 0x03e ESR
      3 0x11f PVR
      1 0x19f IVOR15
      1 0x19e IVOR14
      1 0x19d IVOR13
      1 0x19c IVOR12
      1 0x19b IVOR11
      1 0x19a IVOR10
      1 0x199 IVOR9
      1 0x198 IVOR8
      1 0x197 IVOR7
      1 0x196 IVOR6
      1 0x195 IVOR5
      1 0x194 IVOR4
      1 0x193 IVOR3
      1 0x192 IVOR2
      1 0x191 IVOR1
      1 0x190 IVOR0
      1 0x03f IVPR

While the instruction statistics for all instructions in that log is:
$awk '{ printf("%s\n",$8) }' boot_halt_v4.instr.log | sort | uniq -c | sort -rn
   9157 mtspr
   7494 wrteei
   7051 mfspr
   6073 mfmsr
   4532 wrtee
   1693 rfi
   1647 tlbwe
    943 mtmsr
     89 lbz
     84 stb
     81 tlbsx
     22 mfdcr
     21 lbzx
     20 mtdcr
      7 stbx
      3 sthbrx
      3 iccci


P.S. FYI - atm I think of adding such summary reports as optional part to the post processing script either packing most of the function into awk commands or by extending the python code.

-- 

Grüsse / regards, 
Christian Ehrhardt
IBM Linux Technology Center, Open Virtualization