I set RC_DEVICE_TARBALL="no" in /etc/conf.d/rc and then 'modprobe nvidia' doesn't create device /dev/nvidia0 (this device is created after xorg startup) , so it would be nice, if you little modify nvclock init script. I added this to start(): create_some_shit() { modprobe nvidia if [ ! -c /dev/nvidia0 ]; then mknod /dev/nvidia0 c 195 0 fi if [ ! -c /dev/nvidiactl ]; then mknod /dev/nvidiactl c 195 255 fi }
The best solution is to change "before xdm" to "need xdm" into /etc/init.d/nvclock
Yes, I confirm this. I always needed to restart nvclock script when X starts in order to have it working. I'll test the change suggested by Stefano. Could this script change be commited if it works?
Re-assign wrt Bug 171232
(In reply to comment #1) > The best solution is to change "before xdm" to "need xdm" into > /etc/init.d/nvclock > No effect. I changed before xdm, for need xdm and I have still same bug because I have parallel startup enabled. Maybe xdm appears as started for nvclock script where X is not fully started and causes this bug.
(In reply to comment #4) > (In reply to comment #1) > > The best solution is to change "before xdm" to "need xdm" into > > /etc/init.d/nvclock > > > > No effect. I changed before xdm, for need xdm and I have still same bug because > I have parallel startup enabled. Maybe xdm appears as started for nvclock > script where X is not fully started and causes this bug. So maybe the nvclock script needs to sleep a while before it does its thing?
As stated in the Handbook[1], when a service is required (even if it's only needed to load nvidia.ko), then we use `need', not `before' (which only states that /if/ the other service is present, we load the current one first). I made the change suggested in comment #1 and that should fix the issue, even if comment #4 suggests it does not (which would make it a bug in sys-apps/baselayout, not in media-video/nvclock). [1] http://www.gentoo.org/doc/en/handbook/handbook-x86.xml?part=2&chap=4#doc_chap4