Re: rt20 scheduling latency testcase and failure data

["Sébastien Dugué" <sebastien.dugue@bull.net>]

On Thu, 2006-05-18 at 10:56 +0200, Ingo Molnar wrote:
> * Sébastien Dugué <sebastien.dugue@bull.net> wrote:
> 
> > > thanks for tracking this down. FYI, the latency of stopping the trace is 
> > > that expensive because we are copying large amounts of trace data 
> > > around, to ensure that /proc/latency_trace is always consistent and is 
> > > updated atomically, and to make sure that we can update the trace from 
> > > interrupt contexts too - without /proc/latency_trace accesses blocking 
> > > them. The latency of stopping the trace is hidden from the tracer itself 
> > > - but it cannot prevent indirect effects such as your app from missing 
> > > periods, if the periods are in the 5msec range.
> > > 
> > 
> >   Thanks for the explanation, will have to look deeper into the code 
> > to understand how it works though.
> 
> there's another complexity on SMP: if trace_all_cpus is set then the 
> per-cpu trace buffers are sorted chronologically as well while the 
> copying into the current-max-trace-buffer, to produce easier to read 
> latency_trace output.
> 
  Well, that's not the case here, but thanks for the info.

  Sébastien.