dkrand
03-10-08, 01:11 PM
Forwarded from freebsd-x11@freebsd.org at the request of Florent Thoumie.
I've recently replaced my old workstation running FreeBSD 6.3, xorg-server 1.4_4,1 and nvidia-driver 100.14.19 with a new one running FreeBSD 7.0, xorg-server 1.4_4,1 and nvidia-driver 162.12 and on the new workstation the X server seems to grow with out bound, causing me to have to restart the X server every few days.
On the old system here the X server:
USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
root 772 3.1 15.3 280804 238884 ?? Ss 20Jan08 1761:05.23 /usr/local/bin/X :0 -auth /usr/local/lib/X11/xdm/authdir/authfiles/A:0-QvpfCq (Xorg)
And on the new system:
USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
root 35061 3.1 9.8 1027484 203664 ?? Ss 4:47PM 6:41.43 /usr/local/bin/X :0 -auth /usr/local/lib/X11/xdm/authdir/authfiles/A:0-c2ZMy4 (Xorg)
Note that the old X server has been running for about 7 weeks, and I'm lucky to get 3 days out of my new one. And even after 7 weeks the old one has a VSZ of under 300 MB, while the new one starts out at about 1 GB, and keeps growing.
What will happen is that the VSZ on the new server will just get bigger and bigger until it eventually reaches around 2.8 GB, then X starts throwing errors like this one from "xv -max -quit -root":
insufficient memory in X server to store root pixmap
I have, from habit, this in my /boot/loader.conf which is why it gets to grow this large:
kern.maxdsiz=2048M
kern.dfldsiz=256M
I finally got the time to try downgrading the nvidia-driver from 162.12 to 100.14.19 and everything is back to normal. After 24 hours the server has only grown to 175 MB, with the 162.12 nvidia-driver it would have already been near 2 GB.
I've recently replaced my old workstation running FreeBSD 6.3, xorg-server 1.4_4,1 and nvidia-driver 100.14.19 with a new one running FreeBSD 7.0, xorg-server 1.4_4,1 and nvidia-driver 162.12 and on the new workstation the X server seems to grow with out bound, causing me to have to restart the X server every few days.
On the old system here the X server:
USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
root 772 3.1 15.3 280804 238884 ?? Ss 20Jan08 1761:05.23 /usr/local/bin/X :0 -auth /usr/local/lib/X11/xdm/authdir/authfiles/A:0-QvpfCq (Xorg)
And on the new system:
USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
root 35061 3.1 9.8 1027484 203664 ?? Ss 4:47PM 6:41.43 /usr/local/bin/X :0 -auth /usr/local/lib/X11/xdm/authdir/authfiles/A:0-c2ZMy4 (Xorg)
Note that the old X server has been running for about 7 weeks, and I'm lucky to get 3 days out of my new one. And even after 7 weeks the old one has a VSZ of under 300 MB, while the new one starts out at about 1 GB, and keeps growing.
What will happen is that the VSZ on the new server will just get bigger and bigger until it eventually reaches around 2.8 GB, then X starts throwing errors like this one from "xv -max -quit -root":
insufficient memory in X server to store root pixmap
I have, from habit, this in my /boot/loader.conf which is why it gets to grow this large:
kern.maxdsiz=2048M
kern.dfldsiz=256M
I finally got the time to try downgrading the nvidia-driver from 162.12 to 100.14.19 and everything is back to normal. After 24 hours the server has only grown to 175 MB, with the 162.12 nvidia-driver it would have already been near 2 GB.