| Summary: | xorg 7.2 (xorg-server-1.2.0) consumes memory | ||||||
|---|---|---|---|---|---|---|---|
| Product: | xorg | Reporter: | Joel Cunningham <uber.leoj> | ||||
| Component: | Server/General | Assignee: | Xorg Project Team <xorg-team> | ||||
| Status: | RESOLVED FIXED | QA Contact: | Xorg Project Team <xorg-team> | ||||
| Severity: | normal | ||||||
| Priority: | medium | CC: | andrewdk, aplattner, miho, x | ||||
| Version: | 7.2 (2007.02) | ||||||
| Hardware: | x86-64 (AMD64) | ||||||
| OS: | Linux (All) | ||||||
| Whiteboard: | |||||||
| i915 platform: | i915 features: | ||||||
| Attachments: |
|
||||||
|
Description
Joel Cunningham
2007-02-17 10:34:50 UTC
I can prove the issue and reverting to the old version of xorg-server-1.1.1 solves the problem. Sorry about the phenomenal bug spam, guys. Adding xorg-team@ to the QA contact so bugs don't get lost in future. System: Gentoo Linux, amd64 Xorg: 7.2 Driver: nvidia drivers 1.0.9746 Desktop Env: KDE-3.5.6 1GiB RES, 1.2GiB Virt after 8 days of uptime. No problems noticed in xrestop. Very interesting /proc/pid/maps tho: http://forum.pctuning.cz/zaloha/miho/maps.txt System: Gentoo Linux, x86 Xorg: 7.2 Driver: i810-1.7.4 Desktop Env: Gnome+Beryl Same behaviour, I can see memory usage increasing when I move mouse over kiba-dock, but the leak is there even without kiba. Same here; Xorg: 7.2 Driver: nvidia 1.0.9746 Env: Gnome+Beryl (beryl alone w/gnome panel and fluxbox also exhibit this, seems to be WM/DE independent) It's bad enough to fill up half a gb of ram *and* 500mb swap in about 2 days. i suppose it might be related with 64bit architecture. only one user wrotes, he has x86, but in fact he didn't write if he has 64bit procesor with 32bit system or not. i've got two computers with gentoo linux installed with very similar CFLAGS and environment. the main diffrence is that one is based on athlon-xp cpu and the other one is sempron. both systems are compiled with -march=athlon-xp however. i've noticed memory leaks in system with sempron, but everything is just fine on computer with athlon-xp I've got two x86 systems, one prescott, one Core Duo, both compiled with -march=prescott, both exhibit this bug. It is architecture agnostic, at least for x86+amd64... CFLAGS="-O2 -march=i686 -fomit-frame-pointer -pipe" CHOST="i686-pc-linux-gnu" Intel(R) Pentium(R) M processor 1.60GHz Created attachment 9064 [details]
Simple connect/disconnect test case
I can get runaway memory usage with the attached test case which simply opens and closes a display connection repeatedly. With git master, I don't see this problem.
Found the fix: commit 811675733e97416c990e6dc9c19271b43d96248d Author: Daniel Stone <daniel@fooishbar.org> Date: Thu Feb 15 19:09:00 2007 +0200 os: fix client privates leak Minor leak here. Oops. Someone should cherry-pick that over to server-1.2-branch and release a new server. (In reply to comment #10) > Someone should cherry-pick that over to server-1.2-branch and release a new > server. server-1.2-branch is kind of a trainwreck; if someone wants to do 1.2.1, they'll have to start again. It's already on server-1.3-branch, though. I have patched xorg-server-1.2.0 with commit 811675733e97416c990e6dc9c19271b43d96248d and leak is still there :( I will try complete git tree later... My bad... Commit 811675733e97416c990e6dc9c19271b43d96248d seems to fix the leak... Can anyone else confirm that? (patched ebuild for gentoo here: http://sklad.krtek.net/xorg-server.tar.gz) As some other people (http://forums.gentoo.org/viewtopic-t-537634-postdays-0-postorder-asc-highlight-leak-start-75.html) has successed in fixing the leak with commit 811675733e97416c990e6dc9c19271b43d96248d, suggesting closing bug (FIXED? WONTFIX?) Nothing more can be done here... FIXED |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.