Comment 1 Berthold Daum 2010-08-03 06:37:36 EDT

Created attachment 175761 [details]
Test program to reproduce the error

Comment 2 Berthold Daum 2010-08-03 06:41:08 EDT

Created attachment 175762 [details]
Screenshot of system monitor after 1800 passes

Comment 3 Berthold Daum 2010-08-03 06:41:37 EDT

Created attachment 175763 [details]
System configuration

Comment 4 Felipe Heidrich 2010-08-04 14:34:17 EDT

Works for me. I ran your snippet on my Linux machine (Fedora Core 11) and the memory did not go up like you said.

Comment 5 Berthold Daum 2010-08-06 09:09:11 EDT

Can we do something to sort this out?

Comment 6 Felipe Heidrich 2010-08-06 10:40:13 EDT

can you tell me exactly which calls is causing the leak ?
If you remove this line:
gc.drawImage(image, 0, 0, bounds.width, bounds.height, 0, 0, width,height);
does the leak stop ?

Comment 7 Berthold Daum 2010-08-09 04:40:55 EDT

Yes, it is gc.drawImage() that causes the leak.

Comment 8 Felipe Heidrich 2010-08-10 11:50:48 EDT

Works for me, try this snippet:

public static void main(String[] args) {
	
		int count = 10000;
		Display display = new Display();
		System.out.println("Write down the memory used by this process and hit enter");
		try {
			System.in.read();
		} catch (IOException e) {
		}
		Image image = new Image(display, 3200, 2400);
		Image thumbnail = new Image(image.getDevice(), 320, 240);
		GC gc = new GC(thumbnail);
		long time = System.currentTimeMillis();
		for (int i = 0; i < count; i++) {
			gc.drawImage(image, 0, 0, 3200, 2400, 0, 0, 320, 240);
		}
		System.out.println(System.currentTimeMillis() - time);

		gc.dispose();
		image.dispose();
		thumbnail.dispose();
		System.gc();
		System.out.println("Compare the memory used by this process now to the initial mark, does it go up if count goes up ?");
		try {
			System.in.read();
		} catch (IOException e) {
		}
		display.dispose();
	}


For me it starts with 8.1mb and ends with 9mb, it doesn't matter the value for count (I tested with 1 and 10000). I also check all the code, no leaks anywhere.

Comment 9 Berthold Daum 2010-08-10 17:05:47 EDT

Here is the reading from the system monitor, before and after:

Before:

Virtual 514.1 MB
Resident 25.4 MB
Writable 10.9 MB
Shared 14.6 MB
XServer 1.0 kB


After:

Virtual 1.0 GB
Resident 411.8 MB
Writable 399.7 MB
Shared 10.5 MB
XServer 30 MB

Comment 10 Felipe Heidrich 2010-08-12 16:04:04 EDT

What is the version of X server are you running ?
My guess this is a bug in the X server.

Silenio, do you have ubuntu 10.4 ? can you try the snippet ?

Comment 11 Silenio Quarti 2010-08-13 10:28:34 EDT

Tried the snippet on Ubuntu 10.04. Let it run around 3000 iterations. Memory usage is stable.

Comment 12 Berthold Daum 2010-08-13 17:49:31 EDT

XServer version is 7.5+5ubuntu1

In the meantime I found the following: https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/565981

Looks somehow similar.

Comment 13 Eric Williams 2016-10-04 12:00:25 EDT

I can't reproduce this issue, and comment 12 points to an issue outside of SWT.