Re: Opreport Output details

[Maynard J. <may...@us...>]

Ashok . wrote:
> Hi,
> 
> 1. I am newbie to Oprofile & Performance measurement tools. So I am not able
> to understand the output of Opreport. I didn't find any document which
> explains each column in detail. Can you please provide me some explanation?
> 
> 2. how the percentage is calculated?.
> 
> 3. what exactly is presented in each block?
> 
> # # opreport --merge=all --long-filenames --callgraph
 From the looks of the report below, I'm guessing you did not run the profiler 
with callgraph enabled, so the callgraph data in the report is meaningless.  See 
the "OProfile manual", available at http://oprofile.sourceforge.net/docs/ for 
details about how to run the profiler and use the post-processing tools 
(opreport, opannotate).  And if you need/want to see callgraph output, be aware 
that only a few architectures currently have the kernel support needed (x86, 
ARM, PowerPC).  If you do collect a callgraph profile, then see chapter 4, 
section 2.3 for tips on how to read a callgraph report.

> CPU: CPU with timer interrupt, speed 0 MHz (estimated)
> Profiling through timer interrupt
> samples  %        app name                 symbol name
> -------------------------------------------------------------------------------
> 41213    95.6817  /mnt/vmlinux_a1b23c      default_idle
>   41213    100.000  /mnt/vmlinux_a1b23c      default_idle [self]
> -------------------------------------------------------------------------------
> 141       0.3274  /mnt/vmlinux_a1b23c      sched_setscheduler
>   141      100.000  /mnt/vmlinux_a1b23c      sched_setscheduler [self]
> -------------------------------------------------------------------------------
> 113       0.2623  /mnt/vmlinux_a1b23c      smp_call_function_on_cpu
>   113      100.000  /mnt/vmlinux_a1b23c      smp_call_function_on_cpu [self]
> -------------------------------------------------------------------------------
> 76        0.1764  /mnt/vmlinux_a1b23c      schedule
>   76       100.000  /mnt/vmlinux_a1b23c      schedule [self]
> -------------------------------------------------------------------------------
> 61        0.1416  /lib/libpthread-0.9.29.so (no symbols)
>   61       100.000  /lib/libpthread-0.9.29.so (no symbols) [self]
> -------------------------------------------------------------------------------
> 53        0.1230  /lib/libuClibc-0.9.29.so (no symbols)
>   53       100.000  /lib/libuClibc-0.9.29.so (no symbols) [self]
> -------------------------------------------------------------------------------
> 34        0.0789  /bin/busybox             (no symbols)
>   34       100.000  /bin/busybox             (no symbols) [self]
> 
> 
> 4.  is it possible to dump system time or jiffies value with this output?
No, the sample data that comes from the oprofile kernel driver does not contain 
such info.

-Maynard
> 
> 
> thanks
> Ashok
> 
> 
> 
> ------------------------------------------------------------------------
> 
> ------------------------------------------------------------------------------
> Crystal Reports - New Free Runtime and 30 Day Trial
> Check out the new simplified licensing option that enables unlimited
> royalty-free distribution of the report engine for externally facing 
> server and web deployment.
> http://p.sf.net/sfu/businessobjects
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> oprofile-list mailing list
> opr...@li...
> https://lists.sourceforge.net/lists/listinfo/oprofile-list