|
|
#1 | |
|
Registered User
Join Date: Aug 2005
Posts: 8
|
Just fixed a memory leak in my project where XPixmaps were not deleted. What worries me about this is that the memory wasn't freed by the Xserver after the application was ended.
Since the recreations of XPixmap was on every resize event and having the size of the window, the Xserver's memory grew quite fast. Now that this is fixed, the memory is constant. Is this expected behaviour, or else is this an nvidia issue? |
|
|
|
|
|
|
#2 | |
|
NVIDIA Corporation
Join Date: Mar 2005
Posts: 2,487
|
Pixmaps are destroyed by the X server. Please see the forum sticky posts about how to report a problem, and attach a test case that reproduces this issue.
|
|
|
|
|
|
|
#3 |
|
Registered User
Join Date: Aug 2005
Posts: 8
|
Thanks for your answer. Added the log, xorg.conf and a test cpp file. See preceding comment for a compile example, you might need to adjust includes though.
I'm running debian/testing on amd64 (Intel core2) Wasn't that easy to create a small test case that has this issue. And maybe indeed I'm actually wrong about this. This case shows a XPixmap leak once per second. My observation with this test case is that terminating the application after 10s or so with ^C does not reduce memory size seen in 'top' (VIRT and RES). However at a second run, the increase takes a while, indicating a reuse, but terminating this one, bring the memory back to where it was before the first run. Further investigation shows same behavior on the bug fix mentions, show same behaviour. The alarm bells rang because the memory footprint grew to 1.7G and wasn't freed. I probably had to run the same thing twice. |
|
|
|
![]() |
| Thread Tools | |
|
|