Re: [tracecompass-dev] tracecompass Resources view doesn't make sense

[Joel Fernandes <joelaf@xxxxxxxxxx>]

Hi Mathieu,

On Fri, Feb 3, 2017 at 3:43 AM, Mathieu Desnoyers
<mathieu.desnoyers@xxxxxxxxxxxx> wrote:
> The per-IRQ state seems a bit odd. Unless you have very long
> threaded interrupt handlers running, I would not expect to
> see that much overlap between interrupt handlers on a single
> CPU.
>
> Can you provide the trace you are using, and info about this
> current time/time range you are looking at ? Correlating
> the text output (event-by-event) with the visual rendering should help
> us learn more about what is going on here.

Actually, any time range you pick should be fine (even just the
beginning will show the length IRQ times). We don't have any threaded
handlers that take that long, we've measure latency and it is
definitely not as high as what the tool shows.

I tarred up the CTF folder with the collected data below:
https://drive.google.com/file/d/0B6qmwBx51MQ1Rkd3UnlzS1RIWkE/view?usp=sharing

Thanks!