Radeon DPMS problems? backlight turns back on after a minute

[Lowell Alleman]

On Tue, Apr 7, 2009 at 9:31 AM, Lowell Alleman
<lowell at allemansonline.com> wrote:
> On Tue, Apr 7, 2009 at 3:05 AM, Magnus Kessler <Magnus.Kessler at gmx.net> wrote:
>> On Tuesday 07 April 2009, Lowell Alleman wrote:
>>> Are there any known issues with DPMS not working properly with the latest
>>> radeon driver?  I've found some old bugs that sound similar, but don't
>>> exactly match and the workarounds I've found seem to make a difference.
>>>  Here is what I'm experiencing:
>>>
>>> My laptop (IBM T42, r300) LCD will not going into power saving mode on
>>> its own.  If I force it go into one of these modes, it only stays in that
>>> mode for a short period of time (a few seconds to a couple of minutes)
>>> before the screen turns back on again. I'm pretty sure this isn't due to
>>> any keyboard/mouse events (since the screen saver doesn't
>>> pop up a login dialog) and I don't see any new events /var/log/acpid that
>>> correlate to the LCD being re-enabled.  However, every time the LCD turns
>>> back on, I do see the following message in the Xorg.0 log file:  "enable
>>> LVDS".  It appears that the X server still thinks that the DPMS mode is
>>> enabled, because "xset q" says that the monitor is still in "Suspend" (or
>>> whatever mode I forced it to be in) even after the LCD was turned on
>>> again.
>>>
>>>
>>> Here are some additional details about my setup, and what I've tried:
>>>
>>> Relevant output from "xset q":
>>> DPMS (Energy Star):
>>>  Standby: 600    Suspend: 660    Off: 1200
>>>  DPMS is Enabled
>>>  Monitor is On
>>>
>>> After the display has been disabled using "xset force suspend" the
>>> display shuts down.  If you check "xset q" it now reports "Monitor is in
>>> Suspend". A minute or so later the LCD is turned back on.  Now I can see
>>> screensaver (which is DPMS-aware, and therefore has been paused to save
>>> CPU cycles).  At this point, "xset q" still shows "Monitor is in
>>> Suspend".
>>>
>>> I've tried switching between EXA and XAA acceleration.  As of radeon
>>> v6.12.0 XAA says it is not supported any more, but EXA doesn't appear to
>>> be the default yet.  (Not sure what that means; but in both cases this
>>> problems still occurs).  I've also tried adding the "NoPM" option, as I
>>> found as an old workaround suggested, but that didn't make a difference
>>> either.
>>>
>>>
>>> Versions:
>>>
>>> X.Org X Server 1.5.2
>>> Radeon driver:  6.12.0  (I compiled this locally for improved
>>> performance in latest release)
>>> Linux 2.6.27-11-generic #1 SMP Thu Jan 29 19:24:39 UTC 2009 i686
>>> GNU/Linux Kubuntu 8.10
>>> KDE 4.1.4
>>>
>>> I can provide additional info (xorg.conf and Xorg.0.log) if this is an
>>> unknown situation and those would be helpful to have.
>>>
>>> Thanks,
>>>
>>> - Lowell
>>
>> Hi Lowell,
>>
>> this was a bug in the KDE 4 screensaver code [0]. It is certainly fixed in
>> the latest version of KDE. Try updating to KDE 4.2.1 or above.
>>
>> HTH,
>>
>> Magnus
>>
>> [0] http://bugs.kde.org/show_bug.cgi?id=165265
>>
>> _______________________________________________
>> xorg mailing list
>> xorg at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/xorg
>>
>
> Hey Magnus,
>
> Thanks for that link.  I do think that that bug is *part* of my
> problem, but not *all* of it.  (I plan on upgrading to Kubuntu 9.04,
> which has KDE 4.2.x once it is released last this month.)  Based on
> the info in that bug report, I ran the following test:
>
> I set my screen saver to come on after 5 minutes.  Then I ran "xset
> dpms 5 10 15".  The screen does blank after 5 seconds, but it then
> comes back on after about 2 seconds, with the screen saver visible in
> the background (which was unexpected.)  Then at the 10 second mark,
> the screen turns off again and comes back on after a few minutes.
>
> One general observation is that it seems like the first time the LCD
> enters a power saving mode, it lasts for only a few seconds.  Then
> each additional switch into powersavings mode lasts for about a minute
> or two.  (When I say "first time", I mean after an Xorg restart.)
>
> Just to rule out KDE as the whole problem, I'm going to try starting
> up in a different environment; perhaps some failsafe xterm mode.  I'll
> let you know what I find.
>
> - Lowell
>

Ok. I did some additional testing and I can confirm that this ONLY is
a problem when I'm running KDE.   And while using KDE, it seems that
If I stick with just DPMS modes and disable the sceensaver, then my
LCD will shut off as it should.  (Thanks again Magnus for pointing me
in the right direction.)

So one final question.  Why does "xset q" still report "Monitor is in
Suspend" after the display has turned back on?  It is clear that at
least at some level X knows that the Monitor has been turned on again
because I see the message "enable LVDS" in the Xorg.log file each time
the display turns back on.  Is this a bug or is there a legitimate
reason for this discrepancy?  Any thoughts?

- Lowell