Hi Marc,

Could you tell me how many %CPU and %Mem rt_xtt_gtk and Xorg are consuming using the "top" command?.

Are you using the Claes´s patch?

Eduardo

Hi Eduardo,

I am renting an small office at the other side of the town where I live.
I was in a hurry and shut down the PC.

What I can rember Xorg took around 40% CPU, it's an old PC and we don't have the same hardware so difficult to compare. As Av I took the scantime connected to the pwr_values.

Hoppefully I have time next week to be there.

Just like you I really want to know what the problem is.

/Marc

Hi Eduardo,

Just checked the intermedio graph at home on my laptop.

This is the original:




This is without pipes:




I don't know what the dynamical relation is, in graphs, concerning connection points and pipes etc.
Maybe you should clean your graph and slowly build it up again while keeping an eye on Xorg...

By the way I am running v5.1.0-4.

/Marc

Hi Marc,

Thanks for your info, I'm getting to same conclusion like you about pipes and conections.
I'm testing Claes´s patch at "nomal" scan time 0.5Sec, testing five GeGraph in dual monitor since 30/6 and all is Ok up to now. (2% cpu rt_xtt_gtk, 1% cpu Xorg)

Eduardo

Hi Eduardo and Marc,

I've gradually reduced the code to some simple Xlib calls and I still get the same exception, so I'm quite convinced that this is an Xlib bug. I also strongly suspect this the patch for it

http://patchwork.freedesktop.org/patch/16297

It's a sequence counter wrapping problem and after 4294967296 calls to Xlib the exception will occur. Thus sooner or later is will crash and when it's just a question of the rate of drawing an clipping requests to Xlib.

Obviously only a problem on 32-bit.

/Claes

Hi Eduardo,

I just did "quick" patch test on an Ubuntu 14.04.
Use a test machine!
I screwed up Xorg.

off topic:

Normally I never watch football but Holland is doing pretty well.
Tonight we are going to send Argentina home......