View Single Post
Old 12-31-06, 09:57 PM   #3
Registered User
Join Date: Dec 2006
Posts: 3
Default Re: Problems with GeForce FX Go5200, Xorg, and DPMS

Originally Posted by netllama
The DPMS power saving display-resume issue is a known bug.
Hi Lonni, and thanks for your reply. Do you know if there are plans to fix this any time soon? This has been a problem with the driver on my hardware for the last two years or so. The first two versions of the driver I used worked fine, and then all of the subsequent updates I tried (probably around 15 other versions) did not, up to present. I'm not sure the exact version which started having the problem, but it was somewhere in the 6000's or 7000's. I got my laptop in the summer of 2004, and installed whatever release of Fedora was current at the time, probably FC2. So, it did work once, though it's not a new bug by any means, but it only started being a major problem (i.e. impossible to work around) as of FC6 (see below)... Note also that it works fine with the Free driver that comes with Xorg. Unfortunately the Free driver's performance and functionality are quite lacking.

I'm not aware of any bugs where DPMS cannot be disabled (and I've successfully done so with 1.0-9631 & 1.0-9746 on several occasions). This sounds more like an X server problem.
Recent versions of Xorg's X server enable DPMS by default, and the options to disable it no longer work. Clearly that aspect of the problem is a problem in the X server (and I've filed a bug with the Xorg team), but it wouldn't be a problem at all if the nVidia driver's PM worked properly... It's unfortunate that disabling DPMS doesn't work, but in the end that's only a workaround, and ideally the PM in the driver should be fixed...

You should be able to disable DPMS using the xset command. Specifically how are you trying to disable DPMS?
Note that turning it off is not the same as disabling it... You can turn DPMS off with xset, but you can no longer disable it as was possible until recently; i.e. anyone (e.g. malicious users) connected to the X server which is being displayed currently can turn it back on, and then wait for DPMS to shut the monitor off causing the display to lock up. This can potentially be a big problem in, say, a college environment, or a work environment where people share computers. Also, a user's desktop environment may re-enable DPMS via the screen saver or other similar irritation, where inexperienced / casual users may not be able to figure out what is going on. These features are normally enabled by default these days, so this could make for a very irritating system administration problem anywhere that affected hardware is in use... This problem is one that I've normally seen only on laptops, but I don't know that it doesn't exist for other nVidia hardware.

Formerly, I was able to disable DPMS in the X server using:

Option "DPMS" "off"

Or something very similar... It's not something I needed to do very often so usually I had to look up the exact syntax to make sure it was right, but I believe that's it. This no longer works in recent Xorg servers -- instead any mention of that option enables DPMS (which it already is by default). I have also tried these:

Option "NoPM" "True"
Option "BlankTime" "0"
Option "StandbyTime" "0"
Option "OffTime" "0"
Option "SuspendTime" "0"

And when the server starts up, I get these warning messages:

(WW) NVIDIA(0): Option "NoPM" is not used
(WW) NVIDIA(0): Option "BlankTime" is not used
(WW) NVIDIA(0): Option "StandbyTime" is not used
(WW) NVIDIA(0): Option "OffTime" is not used
(WW) NVIDIA(0): Option "SuspendTime" is not used

Obviously, none of this works.

But the truth is I'd really like DPMS to work properly so I can avoid having to work around this problem, and so I can preserve the life of my display...

FWIW, with current versions of software I am always able to get the display to restart by restarting X from a remote login (but this is disruptive enough that it isn't much better than a reboot), and usually also by closing the lid on my laptop and re-opening it. However in the past I've had much worse problems, including complete lock-up of the machine or corruption of the display that can only be fixed by rebooting.

In addition to running this on my personal laptop, I also work in IT at a medium-sized company with a fairly high percentage of Linux users. A lot of people have been itching to install FC6 or the latest versions of their favorite distros, so this is potentially going to be a problem for them... To be candid, until I can be reasonably sure this is fixed, I'm going to have to recommend that we don't buy nVidia hardware, and that our users choose personal laptops with Intel or ATI graphics hardware if they want to run recent Linux distros on them. That really pains me, having been a die-hard nVidia fan since the days of the Riva 128. But with this kind of problem, it just doesn't make sense to use nVidia right now. I'm not the sole decision maker, but my opinion does carry weight. I don't want this to sound like some kind of threat -- I'm only trying to convey how important it is to me, and potentially quite a few other people I associate with, that this get fixed. You can take that for what it's worth... [And believe me, I'm pursuing the X server angle as well... but that's a somewhat harder case to make. They'll rightly be able to argue that your driver is broken, and there would be nothing for them to fix if it worked properly. And this angle also doesn't help much with the life of the display...]

Thanks again for your reply.
captain_pizza is offline   Reply With Quote