dtraseasm profiler

[Eugeniu Rabii]

After a day of debugging I understood that, I should have made 
"frequency" parameter lower. It is 1001 by default.

Eugene.

On 2/14/20 4:37 PM, Eugeniu Rabii wrote:
>
> Hello,
>
> I am trying to run dtraceasm on Mac with JMH, here are my settings:
>
>     JMH version is 1.24-SNAPSHOT (I build it from sources)
>
>     Mac Version is 10.15.3 (the latest one)
>
>     dtrace is at : Sun D 1.15
>
>     jdk version:  openjdk version "13.0.2" 2020-01-14
>
>     csrutil status: System Integrity Protection status: disabled.
>
>
> The first issue is that is that under these settings, 
> DTraceAsmProfier.java fails with an IllegalStateException, but the fix 
> for this is trivial:
>
> Collection<String> messages =Utils.destroy(dtraceProcess); if (!messages.isEmpty() && !messages.iterator().next().equals("")) {
>      throw new IllegalStateException(messages.toString()); }
>
> on line 90.
>
>
> No matter the @Benchmark I write (and run it with:  java -jar 
> target/benchmarks.jar -prof dtraceasm:hotThreshold=0.01 ), the result 
> is always the same, for example:
>
>
> # Processing profiler results: DTraceAsmProfiler
>
>
> Result "org.sample.MyBenchmark.left":
>   27580.157 ±(99.9%) 616.272 ns/op [Average]
>   (min, avg, max) = (27452.109, 27580.157, 27854.915), stdev = 160.044
>   CI (99.9%): [26963.885, 28196.429] (assumes normal distribution)
>
> Secondary result "org.sample.MyBenchmark.left:·asm":
> PrintAssembly processed: 35658 total address lines.
> Perf output processed (skipped 17.278 seconds):
>  Column 1: sampled_pc (0 events)
>
> WARNING: No hottest code region above the threshold (0.01%) for 
> disassembly.
> Use "hotThreshold" profiler option to lower the filter threshold.
>
> ....[Hottest 
> Regions]...............................................................................
> ....................................................................................................
>          <totals>
>
> ....[Hottest Methods (after 
> inlining)]..............................................................
> ....................................................................................................
>          <totals>
>
> ....[Distribution by 
> Source]........................................................................
> ....................................................................................................
>          <totals>
>
> WARNING: The perf event count is suspiciously low (0). The performance 
> data might be
> inaccurate or misleading. Try to do the profiling again, or tune up 
> the sampling frequency.
> With some profilers on Mac OS X, System Integrity Protection (SIP) may 
> prevent profiling.
> In such case, temporarily disabling SIP with 'csrutil disable' might help.
>
>
> Is there anything else I can try?
>
> Thank you,
>
> Eugene.
>
>