Bug 157581

Summary: xorg-x11-driver-video: nv driver module creates garbage on screen
Product: [openSUSE] SUSE Linux 10.1 Reporter: Heiko Rommel <heiko.rommel>
Component: X.OrgAssignee: Matthias Hopf <mhopf>
Status: RESOLVED DUPLICATE QA Contact: E-mail List <qa-bugs>
Severity: Normal    
Priority: P5 - None CC: aritger, dmacvicar, eich, sndirsch
Version: Beta 7   
Target Milestone: ---   
Hardware: Other   
OS: Other   
Whiteboard:
Found By: Component Test Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Bug Depends on:    
Bug Blocks: 144304    
Attachments: /var/log/Xorg*
/etc/X11/xorg.conf
output of "hwinfo --gfxcard"

Description Heiko Rommel 2006-03-13 13:44:19 UTC
When using applications like Firefox or Thunderbird to edit text I get a lot of gargabe thin cursors on the screen. As soon as the focus leaves the the text window these garbage cursors vanish.

This makes editing of text very painful.

I don't see this defect with driver i810.
Comment 1 Stefan Dirsch 2006-03-13 13:47:52 UTC
Could you attach /var/log/Xorg.0.log and /etc/X11/xorg.conf? Thanks.
Comment 2 Heiko Rommel 2006-03-13 14:43:13 UTC
Created attachment 72565 [details]
/var/log/Xorg*
Comment 3 Heiko Rommel 2006-03-13 14:43:49 UTC
Created attachment 72566 [details]
/etc/X11/xorg.conf
Comment 4 Stefan Dirsch 2006-03-13 15:57:17 UTC
Matthias, since you're currently using a GeForce 6200. Could you try to reproduce this and fix it if required in case it's a driver bug? Thanks.
Comment 5 Matthias Hopf 2006-03-17 18:15:59 UTC
Can you please attach the output of 'hwinfo --gfxcard'? There are too many different GeForce 6200 on the market, I think I have a different one.

Can you test the Card with
  Option "NoAccel"
and/or
  Option "HWcursor" "off"
in the Device section of /etc/X11/xorg.conf?

For "NoAccel" prepare to see a very sloooow Xserver...

Does this happen with the binary only driver?
Comment 6 Heiko Rommel 2006-03-20 11:55:33 UTC
Created attachment 73950 [details]
output of "hwinfo --gfxcard"
Comment 7 Heiko Rommel 2006-03-20 11:58:11 UTC
If using 
   Option "NoAccel"
the defect is gone.

Using
   Option "HWcursor" "off"
has no visiable effect.

If using the binary driver (the patched NVIDIA-Linux-x86-1.0-8178-pkg1.run one) the defect is gone, too.
Comment 8 Matthias Hopf 2006-03-22 14:03:42 UTC
I'm waiting for the 6200 Stefan is trying to get. I have a different one (PCIe instead of AGP).
Comment 10 Jiri Dluhos 2006-04-27 16:05:38 UTC
Possibly this is the same problem as with bug #161873 ?
Comment 11 Matthias Hopf 2006-05-31 15:20:11 UTC
I don't think so.
I recently had this issue myself, but the nv driver is too involved to be easily fixed w/o documentation.

Heiko, did you see this with SL10.0 as well, or is this a new installation? Adding a screenshot would be a great idea.

Also try the options XaaNoImageWriteRect, XaaNoScanlineImageWriteRect, XaaNoScreenToScreenColorExpandFill, XaaNoScreenToScreenCopy (a single one could suffice), instead of NoAccel. Also XaaNoOffscreenPixmaps and/or XaaNoPixmapCache (might need both).

Adding Andy Ritger, maybe he's got another idea.
Comment 12 Jiri Dluhos 2006-05-31 15:28:29 UTC
*** Bug 161873 has been marked as a duplicate of this bug. ***
Comment 13 Jiri Dluhos 2006-05-31 15:32:50 UTC
I have added bug 161873 as a duplicate; we have a screenshot there, and also we have a machine that safely reproduces it any time :-)

We have observed the behavior on SLES10 RC2 x86, with nVidia 6200 TurboCache (OEM AsusTek).
Comment 14 Jiri Dluhos 2006-05-31 15:35:07 UTC
Bug 168849 might possibly be related to the same problem.
Comment 15 Jiri Dluhos 2006-05-31 15:37:20 UTC
Ouch. Now I see that Matthias thinks it is not the same problem as bug 161873 (comment #11). Stupid me. Sorry.

However, it still seems to me the very same thing. Should I re-split the bugs?
Comment 16 Matthias Hopf 2006-05-31 15:51:36 UTC
No necessarily ATM. It *could* be the same bug. Let's analyze further first.
Comment 17 Stefan Dirsch 2006-06-24 09:31:16 UTC
Heiko, we still need your feedback. See comment #11.
Comment 18 Martin Mrazik 2006-06-26 08:51:15 UTC
I tried the option from comment #11:

NoAccel: solves the problem

Having these option in xorg.conf didn't help: XaaNoImageWriteRect, XaaNoScanlineImageWriteRect, XaaNoScreenToScreenColorExpandFill, XaaNoScreenToScreenCopy 

The same with XaaNoOffscreenPixmaps and/or XaaNoPixmapCache (having them in xorg.conf doesn't solve the problem).

I observed this also on SL10.0. You can find a screenshot in bug #161873
Comment 19 Matthias Hopf 2006-06-26 09:56:38 UTC
Ok, that at least gives us a hint. This is a bug that hasn't been reported and addressed before, and no side effect of fixing another bug (which I had in mind).

Can you please try (only one at a time)

XaaNoSolidFillRect, XaaNoSolidHorVertLine, XaaNoSolidTwoPointLine, XaaNoSolidBresenhamLine, XaaNoScanlineCPUToScreenColorExpandFill, XaaNoScreenToScreenColorExpandFill, XaaNoCPUToScreenColorExpandFill

(in decreasing order of probable help)?
Comment 20 Jiri Dluhos 2006-06-26 10:50:37 UTC
XaaNoSolidFillRect helps
XaaNoSolidHorVertLine does not help

Should I try the rest?
Comment 21 Matthias Hopf 2006-06-26 11:04:43 UTC
No, I don't think this is needed. Thanks!
Comment 23 Matthias Hopf 2006-06-26 16:17:45 UTC
This bug exists against plain SL 10.1 as well -> changing product.
Comment 24 Matthias Hopf 2006-06-26 17:22:55 UTC
*** Bug 183749 has been marked as a duplicate of this bug. ***
Comment 25 Matthias Hopf 2006-06-27 14:43:26 UTC
*** Bug 168849 has been marked as a duplicate of this bug. ***
Comment 26 andy ritger 2006-06-30 21:10:23 UTC
Sorry for the slow response.  Aaron Plattner just released an updated nv driver (nv-1_2_0).  I would recommend to start by testing with that.  Based on the results from that, someone from NVIDIA can help debug this.

Thanks,
- Andy
Comment 27 Holger Sickenberg 2006-08-01 13:18:03 UTC
I do have same problem here with a GeFORCE 6200
Comment 28 Stefan Dirsch 2006-08-02 08:32:53 UTC
Could you test again with openSUSE 10.2 Alpha3, which includes latest X.Org and nv driver?
Comment 29 Matthias Hopf 2006-08-17 13:27:36 UTC
Holger or Heiko, could you please retest with Alpha3?
I could never reproduce this here with my 6200...
Comment 30 Matthias Hopf 2006-09-29 12:51:22 UTC
Any news on that?
There has been no post for 6 weeks now...
Comment 31 Martin Mrazik 2006-11-14 15:46:47 UTC
I'll try to test next week
Comment 32 Stefan Dirsch 2006-11-30 21:25:28 UTC
Heiko, Holger, Martin, Jiri. Any updates?
Comment 33 Matthias Hopf 2006-12-11 17:37:39 UTC
This bug has been open without external testing for half a year now, and the last external comment is a month old.

Closing this as FIXED, as I assume the driver update did it. If the issue still pops up, feel free to reopen.
Comment 34 Martin Mrazik 2007-02-28 10:18:39 UTC
sorry for _late_ answer... But I still have this issue with both openSUSE 10.3 alpha1 and SP1 beta4
Comment 35 Stefan Dirsch 2007-02-28 10:47:14 UTC
Ok. Could you also try the options mentioned in comments #19/20?
Comment 36 Martin Mrazik 2007-02-28 11:34:14 UTC
(In reply to comment #35)
> Ok. Could you also try the options mentioned in comments #19/20?

Same results as in comment #20 on openSUSE 10.3-alpha1:

XaaNoSolidFillRect helps
XaaNoSolidHorVertLine does not help

(actually I have pretty same HW as jdluhos)

Comment 37 Stefan Dirsch 2007-02-28 11:39:02 UTC
Thanks. I can close this one finally as a duplicate.

*** This bug has been marked as a duplicate of bug 246421 ***