Still problem with red preempted boxes for tasks

Still problem with red preempted boxes for tasks

[Steven Rostedt]

Hi Yordan,

While looking at my data, I found that the red empty boxes to represent
preemption for task plots is still unreliable.

Down load this: http://rostedt.org/private/trace-views.tar.bz2

Then in the /tmp directory:

 /tmp$ tar xvf trace-views.tar.bz2

Start up KernelShark, and import view1.json. Notice the red preemption
box and what tasks they are for. Then load view2.json and do the same,
the view3.json, and that as well. You'll notice that each one gives
different preemption boxes in the same location.

-- Steve

P.S. The session files really help with passing you debugging info like
this :-)

Re: Still problem with red preempted boxes for tasks

[Yordan Karadzhov (VMware)]

Hi Steven,

Thanks a lot for this feedback!

I will try to fix this before the end of the week. Not sure I manage to 
understand what has to be done for the colors of the task plots but I 
guess you can show me at our next 1:1.

My talk tomorrow starts at 5 PM Sofia time (10 AM for you). I still 
don't have the link for the streaming but I hope to have it tomorrow.

cheers,
Yordan


On  3.10.2018 21:48, Steven Rostedt wrote:
> Hi Yordan,
> 
> While looking at my data, I found that the red empty boxes to represent
> preemption for task plots is still unreliable.
> 
> Down load this: http://rostedt.org/private/trace-views.tar.bz2
> 
> Then in the /tmp directory:
> 
>   /tmp$ tar xvf trace-views.tar.bz2
> 
> Start up KernelShark, and import view1.json. Notice the red preemption
> box and what tasks they are for. Then load view2.json and do the same,
> the view3.json, and that as well. You'll notice that each one gives
> different preemption boxes in the same location.
> 
> -- Steve
> 
> P.S. The session files really help with passing you debugging info like
> this :-)
> 

Re: Still problem with red preempted boxes for tasks

[Steven Rostedt]

On Wed, 3 Oct 2018 22:10:30 +0300
"Yordan Karadzhov (VMware)" <y.karadz@gmail.com> wrote:

> Hi Steven,
> 
> Thanks a lot for this feedback!
> 
> I will try to fix this before the end of the week. Not sure I manage to 
> understand what has to be done for the colors of the task plots but I 
> guess you can show me at our next 1:1.

If you look at what the old kernelshark does, you should see what I'm
talking about.

For CPU plots, the colors match for what task is running on the CPU.

For task plots, the colors should match for what CPU the task is
running on. Thus, if a task migrates, it should have a different color.

> 
> My talk tomorrow starts at 5 PM Sofia time (10 AM for you). I still 
> don't have the link for the streaming but I hope to have it tomorrow.

Ah, so I shouldn't have canceled our tracing meeting, but instead made
your talk what we do for it ;-)

-- Steve