[Bug 93720] Random crashes on (EE) TouchListenerGone: couldn't allocate events

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=93720

--- Comment #11 from Franck Delache <franck.delache at grassvalley.com> ---
For the record, my Java program renders a lot of images on the screen.
It seems that every time the JVM wants to draw an image it allocates a pixmap
and links it to the Java XRSurfaceData object (class used internally by the
JVM).

Once the Java XRSurfaceData object is not used anymore (i.e. cannot be strongly
reached) AND is garbage collected then the pixmap is freed (one can look at the
sun.java2d.Disposer class for more details).

The problem is that even if the Java XRSurfaceData object is not referenced
anymore, it does not mean it is immediately freed. One must wait for the
Garbage Collector to do its job and free those objects.

Thus the XServer and kernel resources are freed only when the JVM decides to
cleanup its own heap.

My application does not keep any reference over those XRSurfaceData objects and
using a profiler, I can see that there are a lot of weakly references
XRSurfaceData objects (i.e. objects ready to be garbage collected), but
unfortunately the Garbage Collector does not run yet since the JVM still has
enough free heap.
Which goes to a point where the host resources are exhausted before the JVM
consider it is time to do some cleanup in its heap.

I am still scratching my head on how to resolve this issue, but would like some
clarification on the resources involved. Could you answer some questions?

 - Does pixmaps are allocated in user space?
 - Does buffers (backing the pixmaps) are allocated in user space or kernel
space?
 - What kind of kernel resources are used for each buffer?
 - Can we tune the amount of memory reserved by the kernel for those resources?

That would help me understand the system behavior and see what could I tune to
avoid that bad effect.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20160128/20d5ae60/attachment.html>