Bugzilla – Bug 246421
nv driver has redraw problems
Last modified: 2007-03-16 07:33:38 UTC
The attached screenshot says all. This is just Firefox with the cursor moved back. Dragging window etc. which forces a redraw solves the problem. Of course, it doesn't depend on the windowmanager. Hardware information is also attached.
Created attachment 119723 [details] Screenshot Sorry, my computer crashed while 'hwinfo'. Maybe I shouldn't run two different X sessions with different drivers ...
Created attachment 119725 [details] Hardware information
Could you also attach /etc/X11/xorg.conf and /var/log/Xorg.0.log? Thanks.
Created attachment 119742 [details] xorg.conf
Created attachment 119743 [details] Xorg.0.log
Configuration looks ok to me. Nothing obvious in logfile. Probably it's one of the accelerated function which is broken. Check with Option "NoAccel" in Section "Device" of your /etc/X11/xorg.conf if this is correct. If the problem no longer occurs with this option active it would be good to figure out which XAA function is broken. Available options to disable these: XaaNoCPUToScreenColorExpandFill XaaNoColor8x8PatternFillRect XaaNoColor8x8PatternFillTrap XaaNoDashedBresenhamLine XaaNoDashedTwoPointLine XaaNoImageWriteRect XaaNoMono8x8PatternFillRect XaaNoMono8x8PatternFillTrap XaaNoOffscreenPixmaps XaaNoPixmapCache XaaNoScanlineCPUToScreenColorExpandFill XaaNoScanlineImageWriteRect XaaNoScreenToScreenColorExpandFill XaaNoScreenToScreenCopy XaaNoSolidBresenhamLine XaaNoSolidFillRect XaaNoSolidFillTrap XaaNoSolidHorVertLine XaaNoSolidTwoPointLine Begin with XaaNoOffscreenPixmaps, XaaNoPixmapCache, XaaNoScreenToScreenCopy, XaaNoSolidFillRect.
It's XaaNoSolidFillRect.
Thanks for testing! As expected, it's a function, which to disable, makes the driver pretty unusable.
Model: "Micro-Star International GeForce 6200 TurboCache(TM)" Vendor: pci 0x10de "nVidia Corporation" Device: pci 0x0161 "GeForce 6200 TurboCache(TM)" SubVendor: pci 0x1462 "Micro-Star International Co., Ltd." SubDevice: pci 0x0271
I remember we had this issue some time ago. AFAIR this couldn't be reproduced in-house at that time, and as the binary driver fixed the issue it was closed. Adding Andy Ritger. Andy, is this a known issue to you?
Adding Lonni to Cc as well.
*** Bug 157581 has been marked as a duplicate of this bug. ***
Lonni, can you reproduce this problem with any of your 6200 TC cards?
No, I cannot. If you'd like me to look into this further, I'll need you to ship me the card(s) that exhibit this behavior. thanks.
Yes, would be fair. Any of the reporters, who volunteers to give his card to NVIDIA? I'll try to find a replacement for this time. Bernhard, Martin, Jiri?
Lonni, I'll send you the card of Bernhard after testing it on another machine to verify that it doesn't depend on the machine.
I could easily reproduce this issue (using firefox type something in a Novell Buzilla comment field and try to go back with cursor left --> result is the screenshot of comment #1) on openSUSE 10.3 Alpha with xserver 1.2.99.901 and nv driver 1.2.2.
Indeed, the only XaaNo* option, which workarounds this issue - by making the driver pretty useless - is "XaaNoSolidFillRect".
Lonni, could you give me an detailed adress (private email) where to send the card? Thanks. Even more easier to reproduce this issue is xinit /usr/bin/firefox http://www.opensuse.org Type something into the search field and go back with cursor left.
Private email sent with shipping address. thanks
Thanks. I'll send the card to you ASAP.
I've received the graphics card and reproduced the problem. Bug 298152 has been opened for this issue. thanks
Sounds good. Thanks!
To: xorg-commit@lists.freedesktop.org Date: Thu, 15 Mar 2007 18:37:53 -0700 (PDT) From: Aaron Plattner <aplattner@kemper.freedesktop.org> Subject: xf86-video-nv: 2 commits - src/nv_driver.c src/nv_hw.c src/nv_driver.c | 2 ++ src/nv_hw.c | 1 + 2 files changed, 3 insertions(+) New commits: diff-tree 90c53c5322e50bce06996e3c857e6bf7531235d4 (from +9763b9dcef144f649ccbcfd302601769bc940ffb) Author: Aaron Plattner <aplattner@nvidia.com> Date: Thu Mar 15 17:07:25 2007 -0700 Fix GXxor corruption on nv4x. SUSE bug #246421. diff --git a/src/nv_hw.c b/src/nv_hw.c index 6232cd6..d7a0c19 100644 --- a/src/nv_hw.c +++ b/src/nv_hw.c @@ -1190,6 +1190,7 @@ void NVLoadStateExt ( pNv->PGRAPH[0x008C/4] = 0x60de8051; pNv->PGRAPH[0x0090/4] = 0x00008000; pNv->PGRAPH[0x0610/4] = 0x00be3c5f; + pNv->PGRAPH[0x0bc4/4] |= 0x00008000; j = pNv->REGS[0x1540/4] & 0xff; if(j) {
Fixed for STABLE/Factory. Check for xorg-x11-driver-video RPM changelog: ------------------------------------------------------------------- Fri Mar 16 08:28:10 CET 2007 - sndirsch@suse.de - xf86-video-nv-corruption-cursor.diff: * fix GXxor corruption on nv4x (Bug #246421) * Hide the cursor in CloseScreen. Otherwise, if we're restoring a graphical framebuffer mode the cursor will still be visible.