Bug 6103 - Screen corruption in KDE4 with Nvidia GPU - Black dots in top left corner
Summary: Screen corruption in KDE4 with Nvidia GPU - Black dots in top left corner
Status: RESOLVED FIXED
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: Cauldron
Hardware: x86_64 Linux
Priority: Normal normal
Target Milestone: ---
Assignee: Mageia Bug Squad
QA Contact:
URL:
Whiteboard: mga2too 4beta1
Keywords: NEEDINFO
Depends on:
Blocks: 12079
  Show dependency treegraph
 
Reported: 2012-05-26 17:30 CEST by John L. ten Wolde
Modified: 2014-10-27 22:12 CET (History)
15 users (show)

See Also:
Source RPM: nvidia-current
CVE:
Status comment:


Attachments
Screenshot with the black dots at the top left of the screen (32.42 KB, image/png)
2012-05-27 07:18 CEST, Dimitrios Glentadakis
Details
Output of 'ps faux' (23.89 KB, text/plain)
2013-05-09 07:45 CEST, Dimitrios Glentadakis
Details
Output of 'systemctl' (11.38 KB, text/plain)
2013-05-09 07:46 CEST, Dimitrios Glentadakis
Details
Output of 'ps faux' (18.00 KB, text/plain)
2013-05-10 06:00 CEST, Dimitrios Glentadakis
Details
Output of 'systemctl' (11.45 KB, text/plain)
2013-05-10 06:00 CEST, Dimitrios Glentadakis
Details
Output of 'journalctl -a > journalctl.txt' (39.11 KB, text/plain)
2013-05-10 06:02 CEST, Dimitrios Glentadakis
Details
Running process list. (22.09 KB, text/plain)
2013-05-10 23:32 CEST, John L. ten Wolde
Details
Output of 'systemctl' (13.13 KB, text/plain)
2013-05-10 23:34 CEST, John L. ten Wolde
Details
Output of 'journalctl' (42.10 KB, text/plain)
2013-05-10 23:36 CEST, John L. ten Wolde
Details
Screenshot of my dots. (28.35 KB, image/png)
2013-05-11 00:12 CEST, John L. ten Wolde
Details
output of ps -faux (13.37 KB, text/plain)
2013-05-20 23:41 CEST, Florian Hubold
Details
output of systemctl (8.63 KB, text/plain)
2013-05-20 23:41 CEST, Florian Hubold
Details
output of journactl -b (127.97 KB, text/plain)
2013-05-20 23:42 CEST, Florian Hubold
Details
output of 'ps faux' (22.51 KB, text/plain)
2013-05-21 06:33 CEST, John L. ten Wolde
Details
output of 'systemctl' (10.42 KB, text/plain)
2013-05-21 06:34 CEST, John L. ten Wolde
Details
output of 'journalctl -b' (150.57 KB, text/plain)
2013-05-21 06:34 CEST, John L. ten Wolde
Details
4beta1 screenshot (553.25 KB, image/png)
2013-11-04 15:09 CET, claire robinson
Details

Description John L. ten Wolde 2012-05-26 17:30:52 CEST
I'm very impressed with MGA2 so far (superb job guys and gals!), except that on my HP Pavilion dv7 laptop (x86_64 with GeForce 9600M GT running the proprietary "NVIDIA GeForce 6100 to GeForce 360" driver) I was seeing screen corruption in KDE4.  Specifically, clusters of black pixels spattering the top left corner of the screen.  These shifted around (or disappeared momentarily) when I moved the mouse or most commonly touched window controls like Minimise or Restore with the cursor.

I got the feeling I'd dealt with this same issue on this same machine years ago, but that back then it had eventually gone away with an update.  Sure enough, from the Mandriva 2009.0 Errata and the corresponding bug report:

http://wiki.mandriva.com/en/2009.0_Errata#Graphical_corruption_with_NVIDIA_GeForce_7xxx_series_graphics_cards

https://qa.mandriva.com/show_bug.cgi?id=43716


I applied the recommendations from the bug report:

nvidia-settings -a InitialPixmapPlacement=2 -a GlyphCache=1

...and to my xorg.conf added:

Option "PixmapCacheSize" "1000000"
Option "AllowSHMPixmaps" "0"
Option "OnDemandVBlankInterrupts" "True"

On X restart not only was the corruption gone but KDE4 was noticeably more responsive (and I didn't think it was slow to begin with).

So, yeah, I'll just leave this here for the powers that be to work their magic. ;-)



Version-Release number of selected component (if applicable):

Mageia 2 (x86_64) with nvidia-current-kernel-desktop-latest-295.49-4.mga2.nonfree


How reproducible:

always (presumably)


Steps to Reproduce:
1. Install Mageia 2
2. Apply the "NVIDIA GeForce 6100 to GeForce 360" proprietary driver
3. Note black pixels wandering around screen in KDE4 while moving the mouse
Comment 1 John L. ten Wolde 2012-05-26 18:29:56 CEST
ADDENDUM: As confirmed back in 2008 it is the line "nvidia-settings -a InitialPixmapPlacement=2 -a GlyphCache=1" that corrects the issue, and for the fix to survive reboot, I've reproduced the /etc/X11/xinit.d/01nvidia-performance script on my system (originally thanks to Moreno and steletch):

<script>
#!/bin/bash

if[ -x nvidia-settings ]; then
    nvidia-settings -a InitialPixmapPlacement=2 GlyphCache=1
fi
</script>


Mageia 2's existing script /etc/nvidia-current/nvidia-settings.xinit refers to another file /etc/sysconfig/nvidia-settings but I have no such file on my system.  Could this be the problem?
Comment 2 John L. ten Wolde 2012-05-26 19:06:38 CEST
EXTRA NOTE: I forgot to mention, running the nvidia-settings command from the terminal will most likely throw out an error that the canberra-gtk-module couldn't be found.  I've gotten this message on both systems that I've installed MGA2 on so far, and from various apps not just nvidia-settings.  To fix it, install lib(64)canberra-gtk0.
Comment 3 John L. ten Wolde 2012-05-27 06:34:46 CEST
Great.  Whatever the problem is, it is NOT fixed by my efforts above.  After a cold boot, and despite the old script, the screen corruption was back.  I noticed it right at KDM.  So after logging in I removed the script, logged out, and when X restarted I arrived at KDM without the corruption.

Curious to see what InitialPixmapPlacement and GlyphCache were set to without me or the script interfering, I queried nvidia-settings and it returned:

  Attribute 'InitialPixmapPlacement' (purgatory.cosmos.net:0.0): 2.
  Attribute 'GlyphCache' (purgatory.cosmos.net:0.0): 1.


So I tried a couple of reboots and a few more log-outs (plus a CTRL+ALT+BACKSPACE for good measure) and got no corruption.  Weird.  When it appears again, I'll query nvidia-settings again and report some more.  :(
Comment 4 Dimitrios Glentadakis 2012-05-27 07:17:11 CEST
I have the same problem but i cannot runn the above command:

[root@localhost bin]# nvidia-settings -a InitialPixmapPlacement=2 GlyphCache=1
nvidia-settings: invalid option: "GlyphCache=1"

ERROR: Invalid commandline, please run `nvidia-settings --help` for usage information.

nvidia GeForce 8400 GS
x11-driver-video-nvidia-current-295.49-2.mga2.nonfree

CC: (none) => dglent

Comment 5 Dimitrios Glentadakis 2012-05-27 07:18:02 CEST
Created attachment 2381 [details]
Screenshot with the black dots at the top left of the screen
Comment 6 John L. ten Wolde 2012-05-27 07:49:28 CEST
You're missing the second "-a" (--assign).  Command line should read:

nvidia-settings -a InitialPixmapPlacement=2 -a GlyphCache=1


It may not matter however, check out my Comment 3.  I still got the corruption, even after applying that old fix.  It's not consistent.

If you've got the corruption on your current login session try running this command (you don't need to be root):

    $ nvidia-settings -q all | grep '\(Initial\|Glyph\)

I'm curious what your system reports back.

Also, try logging out and back in again.  Is the corruption still there after X restarts?  So far, I've only got the problem on about 1 in 6 restarts.
Comment 7 Dimitrios Glentadakis 2012-05-27 08:59:22 CEST
I run the fixed command now but it did nt work, i have always the black dots

Here is the output idia-settings -q all | grep '\(Initial\|Glyph\)
:
http://pastebin.com/XD8cCwft

to have this i clicked several times enter because when i enter the command i have only an input prompt:

-----------
[dglent@localhost ~]$ nvidia-settings -q all | grep '\(Initial\|Glyph\)
> 
-----------
Comment 8 John L. ten Wolde 2012-05-27 20:25:23 CEST
Whoops, now it was my turn to screw up a command line and I missed grep's closing quote.  Sorry about that.  Command should have read:

$ nvidia-settings -q all | grep '\(Initial\|Glyph\)'

The weird thing is that your pastebin is only about the last half of the output (does pastebin have a size limit?  I've never used it).  The two attributes I was asking about would have been nearer the top.  The very first line of your output should have read:

  Attributes queryable via localhost:0.0:


Anyway, after cold booting into my current session I have the dots again too, I'll log out and in again after posting this comment to report if they persist.

Running my own grep filtered nvidia-settings query, I get:

  Attribute 'InitialPixmapPlacement' (purgatory.cosmos.net:0.0): 2.
  Attribute 'GlyphCache' (purgatory.cosmos.net:0.0): 1.


Meaning that those two settings are correct whether I'm seeing the corruption or not, and the root of the problem probably lies elsewhere.  >:(

I'll log out and back in, Google some stuff and report back shortly.
Comment 9 John L. ten Wolde 2012-05-27 20:54:48 CEST
Dimitrios, are you sure that logging out and back in doesn't correct the problem?  It does for me.

After a bit of searching I found this:

http://forum.mandriva.com/en/viewtopic.php?t=93274&start=100

And from about halfway down the page:

<quote>
aapgorilla » 09 Sep 2008 23:46

I see some weird display corruption with 2.6.27-desktop-0.rc5.7.1mnb but not with 2.6.27-desktop586-0.rc5.7.1mnb in the top left corner. The corruption consists of about 10 2x2 square pixel flickering black dots. I can't make a screenshot of it to show you though (ksnapshot grabs a normal screen) :(

I removed and installed the kernel but the corruption remains :(

(driver: nvidia 177.70 ; card: 9500m gs; the corruption displays with and without the nvidia tweaks)

edit: when logging in/out the corruption will be gone, only to show at next boot...
</boot>

Note that aapgorilla also started the Mandriva bug report I linked above.  I KNEW I'd seen this before.  I can't remember if it was in MDV2008.1 or 2009.0, but from MDV2009.1 all the way to MGA1 it never bothered me again, only to regress now.

aap implies he got the dots after every boot, I've only gotten them after cold ones.  I can warm boot all I want without problem.  So far I've only cold booted this box 3 times since installing MGA2 last Tuesday.
Florian Hubold 2012-05-27 23:57:34 CEST

CC: (none) => doktor5000

Comment 10 Dimitrios Glentadakis 2012-05-28 07:38:54 CEST
I have booted before 30 minutes now, and i dont see the black dots (good!)
This is the output:


[dglent@localhost ~]$ nvidia-settings -q all | grep '\(Initial\|Glyph\)'
  Attribute 'InitialPixmapPlacement' (localhost.localdomain:0.0): 2.
    The valid values for 'InitialPixmapPlacement' are in the range 0 - 4 (inclusive).
    'InitialPixmapPlacement' can use the following target types: X Screen.
  Attribute 'GlyphCache' (localhost.localdomain:0.0): 1.
    'GlyphCache' is a boolean attribute; valid values are: 1 (on/true) and 0 (off/false).
    'GlyphCache' can use the following target types: X Screen.



I remember in the past in Mandriva, a similar problem with the same black dots but in all long of the title of the window, and the workarround was to disable spalshy. It was fixed with an update of nvidia.
Comment 11 John L. ten Wolde 2012-05-28 18:56:23 CEST
Ah.  That could be it, but I don't remember how it was corrected exactly, as I didn't actively participate in finding a fix that time, I just remember having the problem... and then having it go away.

As for your nvidia-settings output, it confirms that both ours are already InitialPixmapPlacement=2 and GlyphCache=1, so there's no need for us to fiddle with them after all.  I mistakenly thought the old bug stuff fixed it when when I logged out/in, but that would have "fixed" it anyhow.  My apologies.  Also, the session I'm reporting this from started as a cold boot without the dots.  Weird.

But it just gets weirder.  Last night I upgraded a relative's laptop from MGA1 to 2 with a clean install.  This laptop has an ATI Radeon GPU.  You're probably aware of the MGA2 Errata's warnings (§5.1) about the Free DVD and CD installations breaking if you try to apply any proprietary drivers (due to the non-free repos being inaccessible), so I did as required and applied the VESA drivers instead.  Once the system was up, I installed the flgrx drivers and radeon-firmware.  After a reboot, I tweaked some more stuff (unrelated to video) and along the way reactivated the user's composite effects.  After a (warm) boot what did I see?  Not black dots in the top-left corner, but 5 or 6 black blocks (perhaps 4 pixels square) evenly-spaced at the top of the screen that blinked when I moved the mouse.  After logging out/in, they were gone, just like on my nVidia system.  If I see it again, I'll open a new bug for Radeon, but I now wonder if the composite extensions are at fault.  :-/

I should add that my main desktop system also has a Radeon card, but it uses the FOSS drivers.  No issue(s) at all with those so far.

Do you have compositing (desktop effects) enabled?
Comment 12 Dimitrios Glentadakis 2012-05-28 19:50:49 CEST
For information i dont have the problem any more since i run the command above.
Comment 13 Dimitrios Glentadakis 2012-05-29 08:16:47 CEST
Now, i have again the black dots.

[dglent@localhost ~]$ nvidia-settings -q all | grep '\(Initial\|Glyph\)'
  Attribute 'InitialPixmapPlacement' (localhost.localdomain:0.0): 2.
    The valid values for 'InitialPixmapPlacement' are in the range 0 - 4 (inclusive).
    'InitialPixmapPlacement' can use the following target types: X Screen.
  Attribute 'GlyphCache' (localhost.localdomain:0.0): 1.
    'GlyphCache' is a boolean attribute; valid values are: 1 (on/true) and 0 (off/false).
    'GlyphCache' can use the following target types: X Screen.
Comment 14 Dimitrios Glentadakis 2012-05-29 08:18:01 CEST
I have not done any modifications to the xorg.conf file. May be it is needed ?
Comment 15 John L. ten Wolde 2012-05-29 17:54:19 CEST
No, unfortunately the xorg Options won't fix the problem at all, but adding them should't hurt either.  I think they gave my system a small, yet noticeable, performance boost.  Of course, if see your performance degrade, remove them again.

Don't bother about that nvidia-settings command any more.  I was wrong about it, and no longer believe it has anything to do with the problem. :(

Incidentally, my current session started from a another cold boot.  No black dots again this time.  The problem seems intermittent.  I think we're going to have to wait until someone with more time and understanding of these issues can look into it (or switch to the nouveau drivers).  I was so hoping I had an easy fix.
Comment 16 John L. ten Wolde 2012-05-29 20:10:48 CEST
New discoveries.  Cold booted again and this time got the dots.  Was curious if dmesg might have caught anything pertinent so I hit CTRL+ALT+F12.  Nothing odd looking that I could see, but when I tried to find the xterm on tty1 with CTRL+ALT+F1 it wasn't there, just a blank, black screen.  That's weird, because ever since installing MGA2, the xterm has started running on tty1 as opposed to 7 or 8 where it used to always be (as long as I can remember).  But this boot it was back on 7.  Once I'd found my KDE session I immediately noticed: NO BLACK DOTS!!!  So you don't even have to log out/in for the problem to go away.  Now I'm thinking it might reset ("fix" itself, whatever) when the screen resolution switches.  Memory addressing error?  Frame buffer error?

I checked Xorg's log (duh!  I probably done that days ago):

$ cat /var/log/Xorg.0.log | grep '\((EE)\|(WW)\)'
        (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
[    26.494] (WW) Falling back to old probe method for v4l
[    30.327] (WW) NVIDIA(0): No TwinView "MetaModes" specified; will fall back to Display
[    30.327] (WW) NVIDIA(0):     SubSection modes.
[    30.327] (WW) NVIDIA(0): TwinView requested, but only 1 display devices found.
[    32.099] (WW) NVIDIA(0): Option "PixmapCacheSize" is not used
[    33.016] (WW) evdev: Logitech USB Receiver: ignoring absolute axes.


Nothing Earth shattering, but at least we know not to bother with "Option PixmapCacheSize"
John L. ten Wolde 2012-05-29 20:15:48 CEST

Summary: Blast from Mandriva past: Screen corruption in KDE4 with Nvidia GPU => Screen corruption in KDE4 with Nvidia GPU - Black dots in top left corner

Comment 17 John L. ten Wolde 2012-05-29 20:17:05 CEST
I fixed up the bug's title.  Obviously this is nothing to do with 2008's bug.  :-/
Bert Aerts 2012-06-04 18:59:17 CEST

CC: (none) => bert.ram.aerts

Comment 18 Bert Aerts 2012-06-05 12:03:59 CEST
I have nVIDIA GeForce 8600M GT in my laptop Dell Inspiron 1720.
Sometimes I also have the black dots in the upper left corner of my screen.
Sometimes meaning when PC is booted, 
OR the dots are present and remain present 
OR they are not present and do not show up at all.
Manuel Hiebel 2012-06-12 12:17:34 CEST

Depends on: (none) => 3775

Manuel Hiebel 2012-06-12 12:18:05 CEST

Depends on: 3775 => (none)

Comment 19 Bert Aerts 2012-06-14 21:52:34 CEST
Still reproduced with latest nVIDIA 295.59
Comment 20 John L. ten Wolde 2012-06-15 02:43:15 CEST
@Bert: do the dots go away if you try switching to a different tty and then back to the xterm?

What I've noticed is if, on boot, the xterm starts on tty7 (and tty1 is a solid black screen, no cursor or anything) I get the dots, if the xterm starts on tty1, no dots.  This is completely consistent each time.

I've also noticed that on my Radeon system (running the FOSS driver) the xterm jumps from tty1 or tty7 from boot to boot.
Comment 21 doro2k 2012-06-15 09:18:30 CEST
Same problem here. When I switch to a different tty and then back, the dots are gone. Also, tty1 is a black screen.

CC: (none) => doro2k

Comment 22 Bert Aerts 2012-06-16 12:09:46 CEST
Same experience on my Dell Inspiron 1720 with nVIDIA GeForce 8600M GT.

No dots:
Graphical interface on tty1 (CTRL-ALT-F1)

Dots:
Graphical interface on tty7 (CTRL-ALT-F7)
tty1 is empty black screen
After switching to tty2, dots disappear.
Comment 23 Tolhildan Karker 2012-06-18 20:05:07 CEST
Can you guys/girls try this. I do not have this problem, but I may found solution for it. Lot of searching on the internet for it. 

Login as root and then try:
kwrite /var/lib/mageia/kde4-profiles/Default/share/config/kdm/kdmrc
kwrite /usr/share/config/kdm/kdmrc

Change now the line:
ServerVTs=-7

to:
ServerVTs=7

Save it and then reboot/restart. 

I have also read something about changing kernel parameter. It is about vga. Try the tips above. If that does not work maybe changing vga value maybe will fix it.

CC: (none) => tolhildan_123

Comment 24 Julien Gouesse 2012-07-10 22:20:32 CEST
I have the same problem with the proprietary driver and my Nvidia 7600 GT. However, it is not reproducible with Nouveau driver.

CC: (none) => gouessej

Comment 25 Julien Gouesse 2012-07-10 22:35:00 CEST
(In reply to comment #23)
> Can you guys/girls try this. I do not have this problem, but I may found
> solution for it. Lot of searching on the internet for it. 
> 
> Login as root and then try:
> kwrite /var/lib/mageia/kde4-profiles/Default/share/config/kdm/kdmrc
> kwrite /usr/share/config/kdm/kdmrc
> 
> Change now the line:
> ServerVTs=-7
> 
> to:
> ServerVTs=7
> 
> Save it and then reboot/restart. 
> 
> I have also read something about changing kernel parameter. It is about vga.
> Try the tips above. If that does not work maybe changing vga value maybe will
> fix it.

It works, thank you. Modifying /usr/share/config/kdm/kdmrc is enough to make it work.
Comment 26 Tolhildan Karker 2012-07-11 10:39:48 CEST
Nice it did help. I have also black dots that appearer top of the screen. Not all the time and not on every boot. For me, my, tips about changing to "ServerVTs=7" did not help. Instead I have changed in powermizer setting. The black dots appears to come because of scaling in powermizer. I disabled the scaling and have it on the same level now instead. So now the black dots do not appear at all for me. So I think this problem with black dots is about powermizer more then any thing else. What I did was:

1. Login as root
2. kwrite /etc/X11/xorg.conf
3. Add the following line under the section "Device";
Option "RegistryDwords" "PerfLevelSrc=0x2222"
4. Reboot

Now everything works perfect.
Comment 27 Julien Gouesse 2012-07-11 10:54:08 CEST
Powermizer sometimes causes freezes when I stay a long time in front of my screen without pressing a key, for example when watching a film. I will look at a way of disabling it until I switch back to Nouveau.
Comment 28 Julien Gouesse 2012-07-11 11:32:13 CEST
I will try:
Option  "RegistryDwords" "PowerMizerEnable=0x0; PerfLevelSrc=0x2222; PowerMizerLevel=0x3; PowerMizerDefault=0x3; PowerMizerDefaultAC=0x3"
Comment 29 Julien Gouesse 2012-07-11 20:02:18 CEST
Disabling Powermizer does not solve this problem or I did something wrong.
Comment 30 Julien Gouesse 2012-07-13 18:58:00 CEST
I put this into /etc/X11/xorg.conf:
Option  "RegistryDwords" "PowerMizerEnable=0x0"

and it works fine :)
Comment 31 Julien Gouesse 2012-07-14 10:42:07 CEST
After rebooting twice, it doesn't work any more.
Comment 32 Tolhildan Karker 2012-07-17 10:56:04 CEST
It seems modify xorg.conf with different Powermizer option does not help. It appeares again for me. I did never have this issue with Ubuntu/Kubuntu, but I found two links from Ubuntu forum maybe can help us solve this thing. Or we need to send mail to mailto:linux-bugs@nvidia.com. 

Here is the links from Ubuntu forum:
1.
http://ubuntuforums.org/showthread.php?t=1274648&page=3

"I seem to have solved the black dots issue. Apparently, the closed source nvidia drivers and natty's splash screen do not like each other, so I did two things:

1) in /boot/grub/grub.cfg , removed "splash" and "quiet" from the kernel command line
2) I deinstalled plymouth-theme-kubuntu-logo"

2.
http://ubuntuforums.org/showthread.php?t=784045&page=2

Read "Addie MacGruer" and "listdata" answers. 

Maybe one of this two links will solve it for us, if not it is best to contact nvidia itself. My temporary solution for this issue is to use Ctrl+Alt+F1 and then Ctrl+Alt+F7 when the black dots appeares.
Comment 33 Julien Gouesse 2012-07-17 11:47:58 CEST
Actually, I reproduce this bug even with an ATI Radeon 9250 Pro with the Xorg driver (my Nvidia 7600 GT does not work anymore). Your last tip works. I will try the first one at home.
Comment 34 Tolhildan Karker 2012-07-17 15:47:49 CEST
I tested the suggestion on link nr 2. I changed vga=788 default to vga=773 and for now I cannot see any black dots. I have made cold boot, system reboot and suspend etc without it appearing. I will test further and will come back. Hopefully changing vga value will fix this issue ones for all.
Comment 35 Florian Hubold 2012-07-17 22:11:42 CEST
Are you sure you correctly applied the workaround given by John? Which is:

nvidia-settings -a InitialPixmapPlacement=2 -a GlyphCache=1

Surely works for me and some others, also surviving a reboot and never came back.
Comment 36 Tolhildan Karker 2012-07-18 01:12:50 CEST
Floarian how do I make this to work? I mean:
nvidia-settings -a InitialPixmapPlacement=2 -a GlyphCache=1

I will apply this if changing value of vga does not work for me.
Comment 37 John L. ten Wolde 2012-07-18 23:32:06 CEST
@Florian:
OP here.  Nope, ultimately that command line did squat for me.  The settings survive reboot because they are now the default, and by now, probably have been for years.  I think you may just be lucky and win the "no-dots lottery" repeatedly on your machine.  Since my last comment, I've only been getting the dots 1:3 to 1:5 boots and "fixing" it with the tty1 to tty7 trick.

@Tolhildan:
That second Ubuntu link you supplied looks very promising.  I've edited my "vga=" line and have my fingers crossed (though I did have to install hwinfo).  I find it curious that that forum thread also dates back to 2008.  Maybe this is a side effect of the efforts to get DKMS working in the proprietary drivers?  Regressions are fun.... :-(
Comment 38 Tolhildan Karker 2012-07-18 23:59:43 CEST
@John for me it still works. After plenty system reboot, cold boot up, resume for suspend etc for testing purpose and still no black dots. In Ubuntu I never had this issue and I think it is because they do not use vga option in kernel or they handle this issue different. I will also update you all how it goes. It looks promising for now.
Comment 39 John L. ten Wolde 2012-07-21 23:05:19 CEST
Regrettably, for me the vga option does not work.  Got the dots again this boot.
Comment 40 Tolhildan Karker 2012-07-22 00:25:48 CEST
@John

For me it still works. I think you have choosen wrong colour depth for the resolution you choose. You can read here:
https://wiki.ubuntu.com/FrameBuffer

vga=773, with resolution 1024x768 have Colour depth of 8. So please try with lower colour depth for the resolution you choose. My normal screen resolution is higher then 1024x768, but that one was closest.
Comment 41 Bill K 2012-07-22 06:44:53 CEST
I'd like to add my commments, I have MGA2 on two PC's and have had the problem dots with both, strangely it seems to have fixed itself on the first one I loaded but is persisting on the second one.  (Also I am using the 32 bit install, so the problem is not restricted to 64 bit).

Both systems have nVidia graphics...

01:00.0 VGA compatible controller: nVidia Corporation NV43 [GeForce Go 6600] (rev a2)

01:00.0 VGA compatible controller: nVidia Corporation G96 [GeForce 9400 GT] (rev a1)

Both are using the "NVIDIA GeForce 6100 to GeForce 360" settings in xorg.conf

CC: (none) => forums

Comment 42 John L. ten Wolde 2012-07-22 19:52:25 CEST
@Tolhildan:
Whoops.  Disregard my last comment.  I had an error on my grub line.
Comment 43 Tolhildan Karker 2012-07-22 20:23:25 CEST
(In reply to comment #42)
> @Tolhildan:
> Whoops.  Disregard my last comment.  I had an error on my grub line.

Does it work now? When you have sorted out the error on your grub line. Because I really believe this will solve the black dots on top of the screen for Mageia users. For me it have gone days now and every thing works.
Comment 44 Bill K 2012-07-23 03:48:15 CEST
@Tolhildan

It seems changing the grub kernel parameters works.  Now from cold boot I have no dots, we'll see how it goes.

The only modification I made to /boot/grub/menu.lst was to change "vga=788" to "vga=773".
Comment 45 Tolhildan Karker 2012-07-23 12:02:16 CEST
@Bill K

Nice. I hope more will report.
Malo Deniélou 2012-07-23 12:22:32 CEST

CC: (none) => malo

Comment 46 John L. ten Wolde 2012-07-23 20:01:59 CEST
@Tolhildan:
It's only been a few boots now, but yeah, looks good so far.  System also boots faster and a peculiar lag (of about 5-10 seconds) that was holding back KDE once I'd logged in seems to be gone too.  This workaround definitely averts some severely broken stuff (possibly more than just the dots, I mean -- unless it's a related issue).

I'm going to try this on my Radeon system (using the FOSS drivers) as well because it also suffers from the KDE lag at login.

Lastly, since switching to 8bit disables the boot splash, I've started wondering if it might not be enough just to pass "nosplash" to grub instead.  Up in Comment 10, Dimitrios suggested that back when Mandriva 2009 suffered a similar issue the workaround was to disable Splashy.  We're using Plymouth now, aren't we?  I really haven't been keeping up.
Comment 47 Tolhildan Karker 2012-07-23 21:22:46 CEST
@John
Nice. I have also read about "nosplash" but I have not tried it myself. Changing vga value worked and it seems to work for others also. Performance improvement with the workaround is interesting. What exactly is broken I do not know. Maybe nothing is broken? I am not sure. 

We should have this in errata mention workaround for this. I think this issue is more spread the comment we can see here. Best will be update magic that will fix this. I am not a programmer so I cannot write patches for this. I am only pretty good bug-hunter and sometime also find a solution.
Comment 48 Julien Gouesse 2012-07-31 20:51:59 CEST
Hi

I have just reproduced this bugs twice with my ATI Radeon X1950 Pro by rebooting with nomodeset (the first time) and radeon.modeset=0 (the second time).
Comment 49 John L. ten Wolde 2012-08-26 07:18:13 CEST
Just a small addendum that the recent updates to:

kernel-desktop-3.3.8-2.mga2-1-1.mga2
nvidia-current-kernel-3.3.8-desktop-2.mga2-295.71-1.mga2.nonfree

have not resolved this issue.  I'll continue to use Tolhildan's work-around (Comment 34).
Manuel Hiebel 2013-02-24 12:56:24 CET

Source RPM: nvidia-current-kernel-desktop-latest-295.49-4.mga2.nonfree => nvidia-current

Comment 50 Samuel Verschelde 2013-04-21 14:55:26 CEST
I've got this problem in cauldron too. Installed the (future) RC from LiveDVD, everything was fine, rebooted a couple times, no pb. Today I've got the black dots. I don't know if it's related to the fact that I installed updates from cauldron yesterday or not.

unknown         : JMicron Technology Corp.|xD Host Controller [SYSTEM_OTHER] (vendor:197b device:2384 subv:103c subd:361b)
jmb38x_ms       : JMicron Technology Corp.|MS Host Controller [SYSTEM_OTHER] (vendor:197b device:2383 subv:103c subd:361b)
sdhci_pci       : JMicron Technology Corp.|Standard SD Host Controller (vendor:197b device:2381 subv:103c subd:361b)
sdhci_pci       : JMicron Technology Corp.|SD/MMC Host Controller [SYSTEM_OTHER] (vendor:197b device:2382 subv:103c subd:361b)
firewire_ohci   : JMicron Technology Corp.|IEEE 1394 Host Controller [SERIAL_FIREWIRE] (vendor:197b device:2380 subv:103c subd:361b)
r8169           : Realtek Semiconductor Co., Ltd.|RTL8111/8168 PCI Express Gigabit Ethernet controller [NETWORK_ETHERNET] (vendor:10ec device:8168 subv:103c subd:361b) (rev: 02)
iwlwifi         : Intel Corporation|PRO/Wireless 5100 AGN [Shiloh] Network Connection [NETWORK_OTHER] (vendor:8086 device:4237 subv:8086 subd:1211)
Card:NVIDIA GeForce 8100 to GeForce 360: NVIDIA Corporation|G96M [GeForce 9600M GT] [DISPLAY_VGA] (vendor:10de device:0649 subv:103c subd:361b) (rev: a1)
unknown         : Intel Corporation|82801I (ICH9 Family) Thermal Subsystem [SIGNAL_OTHER] (vendor:8086 device:2932 subv:103c subd:361b) (rev: 03)
i2c_i801        : Intel Corporation|82801I (ICH9 Family) SMBus Controller [SERIAL_SMBUS] (vendor:8086 device:2930 subv:103c subd:361b) (rev: 03)
unknown         : Intel Corporation|82801IBM/IEM (ICH9M/ICH9M-E) 4 port SATA Controller [AHCI mode] [STORAGE_SATA] (vendor:8086 device:2929 subv:103c subd:361b) (rev: 03)
lpc_ich         : Intel Corporation|ICH9M LPC Interface Controller [BRIDGE_ISA] (vendor:8086 device:2919 subv:103c subd:361b) (rev: 03)
unknown         : Intel Corporation|82801 Mobile PCI Bridge [BRIDGE_PCI] (vendor:8086 device:2448) (rev: 93)
ehci_pci        : Intel Corporation|82801I (ICH9 Family) USB2 EHCI Controller #1 [SERIAL_USB] (vendor:8086 device:293a subv:103c subd:361b) (rev: 03)
uhci_hcd        : Intel Corporation|82801I (ICH9 Family) USB UHCI Controller #6 [SERIAL_USB] (vendor:8086 device:2939 subv:103c subd:361b) (rev: 03)
uhci_hcd        : Intel Corporation|82801I (ICH9 Family) USB UHCI Controller #3 [SERIAL_USB] (vendor:8086 device:2936 subv:103c subd:361b) (rev: 03)
uhci_hcd        : Intel Corporation|82801I (ICH9 Family) USB UHCI Controller #2 [SERIAL_USB] (vendor:8086 device:2935 subv:103c subd:361b) (rev: 03)
uhci_hcd        : Intel Corporation|82801I (ICH9 Family) USB UHCI Controller #1 [SERIAL_USB] (vendor:8086 device:2934 subv:103c subd:361b) (rev: 03)
shpchp          : Intel Corporation|82801I (ICH9 Family) PCI Express Port 6 [BRIDGE_PCI] (vendor:8086 device:294a) (rev: 03)
shpchp          : Intel Corporation|82801I (ICH9 Family) PCI Express Port 5 [BRIDGE_PCI] (vendor:8086 device:2948) (rev: 03)
shpchp          : Intel Corporation|82801I (ICH9 Family) PCI Express Port 4 [BRIDGE_PCI] (vendor:8086 device:2946) (rev: 03)
shpchp          : Intel Corporation|82801I (ICH9 Family) PCI Express Port 2 [BRIDGE_PCI] (vendor:8086 device:2942) (rev: 03)
shpchp          : Intel Corporation|82801I (ICH9 Family) PCI Express Port 1 [BRIDGE_PCI] (vendor:8086 device:2940) (rev: 03)
snd_hda_intel   : Intel Corporation|82801I (ICH9 Family) HD Audio Controller [MULTIMEDIA_AUDIO_DEV] (vendor:8086 device:293e subv:103c subd:361b) (rev: 03)
ehci_pci        : Intel Corporation|82801I (ICH9 Family) USB2 EHCI Controller #2 [SERIAL_USB] (vendor:8086 device:293c subv:103c subd:361b) (rev: 03)
uhci_hcd        : Intel Corporation|82801I (ICH9 Family) USB UHCI Controller #5 [SERIAL_USB] (vendor:8086 device:2938 subv:103c subd:361b) (rev: 03)
uhci_hcd        : Intel Corporation|82801I (ICH9 Family) USB UHCI Controller #4 [SERIAL_USB] (vendor:8086 device:2937 subv:103c subd:361b) (rev: 03)
shpchp          : Intel Corporation|Mobile 4 Series Chipset PCI Express Graphics Port [BRIDGE_PCI] (vendor:8086 device:2a41) (rev: 07)
unknown         : Intel Corporation|Mobile 4 Series Chipset Memory Controller Hub [BRIDGE_HOST] (vendor:8086 device:2a40 subv:103c subd:361b) (rev: 07)
hub             : Linux 3.8.8-desktop-2.mga3 ehci_hcd|EHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0002)
uvcvideo        : QCM|HP Webcam [Video|Video Control] (vendor:0408 device:03ba)
dvb_usb_af9015  : AVerMedia|A309 (vendor:07ca device:a309)
hub             : Linux 3.8.8-desktop-2.mga3 ehci_hcd|EHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0002)
hub             : Linux 3.8.8-desktop-2.mga3 uhci_hcd|UHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0001)
hub             : Linux 3.8.8-desktop-2.mga3 uhci_hcd|UHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0001)
btusb           : Broadcom Corp|HP Integrated Module [Wireless|Radio Frequency|Bluetooth] (vendor:03f0 device:171d)
hub             : Linux 3.8.8-desktop-2.mga3 uhci_hcd|UHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0001)
hub             : Linux 3.8.8-desktop-2.mga3 uhci_hcd|UHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0001)
hub             : Linux 3.8.8-desktop-2.mga3 uhci_hcd|UHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0001)
usbhid          : MLK|RAPOO [Human Interface Device|Boot Interface Subclass|Mouse] (vendor:1bcf device:0501)
hub             : Linux 3.8.8-desktop-2.mga3 uhci_hcd|UHCI Host Controller [Hub|Unused|Full speed (or root) hub] (vendor:1d6b device:0001)
hid_generic     : MLK RAPOO

Version: 2 => Cauldron
Whiteboard: (none) => mga2too

Comment 51 Anssi Hannula 2013-05-04 12:06:43 CEST
If someone can reproduce this, please run as root:
ps faux > psfaux.txt
systemctl > systemctl.txt
journalctl -b > journalctl.txt

And then attach the tree created files. You can edit psfaux.txt to remove your own processes if you like.

Thanks.

Keywords: (none) => NEEDINFO
CC: (none) => anssi.hannula

Comment 52 Dimitrios Glentadakis 2013-05-08 07:07:22 CEST
I have this message:
[dglent@localhost ~]$ journalctl -b > journalctl.txt
journalctl: invalid option -- 'b'
Comment 53 claire robinson 2013-05-08 09:48:01 CEST
Try as root user. You can also add --no-pager to prevent it chopping the ends off long log lines.
Comment 54 claire robinson 2013-05-08 09:51:18 CEST
Sorry, it is -a for the long lines, --no-pager prevents it being piped into 'less'.
Comment 55 Dimitrios Glentadakis 2013-05-09 07:45:23 CEST
Created attachment 3908 [details]
Output of 'ps faux'
Comment 56 Dimitrios Glentadakis 2013-05-09 07:46:09 CEST
Created attachment 3909 [details]
Output of 'systemctl'
Comment 57 Dimitrios Glentadakis 2013-05-10 06:00:10 CEST
Created attachment 3917 [details]
Output of 'ps faux'

Attachment 3908 is obsolete: 0 => 1

Comment 58 Dimitrios Glentadakis 2013-05-10 06:00:56 CEST
Created attachment 3918 [details]
Output of 'systemctl'

Attachment 3909 is obsolete: 0 => 1

Comment 59 Dimitrios Glentadakis 2013-05-10 06:02:11 CEST
Created attachment 3919 [details]
Output of 'journalctl -a > journalctl.txt'
Comment 60 Dimitrios Glentadakis 2013-05-10 06:06:44 CEST
I uploaded again the files because i was nt root the first time

(In reply to claire robinson from comment #54)
> Sorry, it is -a for the long lines, --no-pager prevents it being piped into
> 'less'.

It worked with -a thanks 

[root@localhost dglent]# journalctl -b > journalctl.txt
journalctl: invalid option -- 'b'
Comment 61 John L. ten Wolde 2013-05-10 23:27:15 CEST
OP here.  Funny, it took nearly a week for the dots to reappear, but this boot I got 'em.  Requested attachments follow.  Still running MGA2.
Comment 62 John L. ten Wolde 2013-05-10 23:32:03 CEST
Created attachment 3930 [details]
Running process list.

Processes running on HP Pavilion dv7 x86_64 notebook with NVIDIA GeForce 9600M GT.
Comment 63 John L. ten Wolde 2013-05-10 23:34:38 CEST
Created attachment 3931 [details]
Output of 'systemctl'

System report on HP Pavilion dv7 x86_64 notebook with NVIDIA GeForce 9600M GT.
Comment 64 John L. ten Wolde 2013-05-10 23:36:52 CEST
Created attachment 3932 [details]
Output of 'journalctl'

systemd journal on HP Pavilion dv7 x86_64 notebook with NVIDIA GeForce 9600M GT.
Comment 65 John L. ten Wolde 2013-05-11 00:12:52 CEST
Created attachment 3934 [details]
Screenshot of my dots.

Don't know how I never noticed this before, but looking back at the screenshot Dimitrios provided last May of his dots, I realize they are virtually identical to mine, 16 in total seemingly matching down to the placement of the pixels (kind of a "Big Dipper" constellation).

In case there's some significance to this, here's a screenshot from my machine matching Dimitrios's as closely as possible for comparison.
Comment 66 John L. ten Wolde 2013-05-11 00:14:07 CEST
Whoops, forgot: Dimitrios's screenshot was back in Comment 5.
Comment 67 John L. ten Wolde 2013-05-18 20:11:44 CEST
Another addendum to announce that yesterday's (17 May my time) MGA2 updates to:

kernel-desktop-3.4.45-1.mga2-1-1.mga2
nvidia-current-kernel-3.4.45-desktop-1.mga2-295.71-16.mga2.nonfree

still have not resolved the issue.

It is now 18 May and I am eagerly awaiting the release of MGA3 final.  :-)
Comment 68 Florian Hubold 2013-05-20 23:41:19 CEST
Created attachment 4005 [details]
output of ps -faux
Comment 69 Florian Hubold 2013-05-20 23:41:43 CEST
Created attachment 4006 [details]
output of systemctl
Comment 70 Florian Hubold 2013-05-20 23:42:06 CEST
Created attachment 4007 [details]
output of journactl -b
Comment 71 Florian Hubold 2013-05-20 23:42:24 CEST
Still valid for Mageia 3.
Comment 72 John L. ten Wolde 2013-05-21 06:30:56 CEST
I can also confirm.  Attachments for comparison follow.  :-(
Comment 73 John L. ten Wolde 2013-05-21 06:33:01 CEST
Created attachment 4009 [details]
output of 'ps faux'
Comment 74 John L. ten Wolde 2013-05-21 06:34:06 CEST
Created attachment 4010 [details]
output of 'systemctl'
Comment 75 John L. ten Wolde 2013-05-21 06:34:55 CEST
Created attachment 4011 [details]
output of 'journalctl -b'
Comment 76 John L. ten Wolde 2013-05-21 06:38:30 CEST
To be clear, attachments 4009, 4010, and 4011 are for Mageia 3 on an HP Pavilion dv7 x86_64 notebook with NVIDIA GeForce 9600M GT.

This is a clean install not an upgrade.
Comment 77 Florian Hubold 2013-06-01 18:55:41 CEST
Problem seems to get worse, additional to the black dots at the top left, in many GTK applications inverted dots are spread across the whole screen.
Comment 78 Dimitrios Glentadakis 2013-10-07 06:27:11 CEST
Since Friday i have the drivers 331.13 and i belive very strongly that it solves this bug. 
I wait some more time to confirm it.
Comment 79 Malo Deniélou 2013-10-07 10:18:01 CEST
Same as Dimitrios. I can't see the black dots any more.
Comment 80 Dimitrios Glentadakis 2013-10-15 16:51:11 CEST
I had it again today.
Comment 81 claire robinson 2013-11-04 15:03:33 CET
Valid 4beta1 DVD 64 (1st build) with default kde installation.

CC: (none) => ennael1, nicolas.lecureuil, tmb
Whiteboard: mga2too => mga2too 4beta1

Comment 82 claire robinson 2013-11-04 15:09:00 CET
Created attachment 4471 [details]
4beta1 screenshot
claire robinson 2013-11-04 15:09:45 CET

CC: (none) => eeeemail

Comment 83 lectus lectus 2013-12-19 14:56:11 CET
I also got this problem.

I'm running Mageia 3 KDE x86_64 with NVIDIA Geforce 9400GT.

So far nothing solved this for me.

Researching the internet it seems to be something with plymouth: https://devtalk.nvidia.com/default/topic/527808/black-dots-in-top-left-screen-corner/

CC: (none) => fwunix

Manuel Hiebel 2013-12-22 20:21:53 CET

Blocks: (none) => 12079

Comment 84 Dimitrios Glentadakis 2014-05-04 11:33:05 CEST
I don't have the symptome since the 29 April 2014, when i installed the nvidia driver 331.67 in cauldron
Comment 85 Morgan Leijström 2014-10-27 22:12:03 CET
I too have not seen any dots on any of three machines for a long time.
Closing - feel free to reopen if anyone see it again.

Status: NEW => RESOLVED
CC: (none) => fri
Resolution: (none) => FIXED


Note You need to log in before you can comment on or make changes to this bug.