Bug 146261

Created attachment 65484 [details]
xorg.conf

xorg.conf

Created attachment 65485 [details]
xorg.logs

xorg.logs

Nothing obivous I can see in the config and logfile. I'll try to reproduce with my GF 6600.

It seems that it is a x86_64 related problem with the nvidia driver shipped with OSS.

Adding Option "HWCursor" "false" in xorg.conf made the cursor visible..

Ok. Is this a PCI-X or AGP board?

It's an AGP board.

Motherboard: DFI LanParty UT nF3 250Gb (Socket 754)
CPU: AMD 64 3400+

http://www.lanparty.com.tw/Product/xx_product_spec_details_r_us.jsp?PRODUCT_ID=2840&CATEGORY_TYPE=MB&SITE=US

Unfortunately I can't reproduce sine my GF 6600 AGP has disappeared. :-(
One known problem is that the mouse cursor is broken with the nv driver after the nvidia has been in use. You need to reboot to fix this issue. Can you check this, please? Thanks.

No feedback. --> WONTFIX.

The problem is still there in beta6, ie:
- I can only use 1280*1024 in 16 bits mode (24 bits gives a scrambled screen)
- the cursor does not appear. To make it apperar, I have to add the follwing in xorg.conf:
   Options "HWCursor" "false"

This after a standard install of OSS 10.2 b6 without any tweaking or installation of anything after installation and many reboots (see comment #7)

The hardware is still:
 - Motherboard: DFI LanParty UT nF3 250Gb (Socket 754)
 - CPU: AMD 64 3400+
 - GForce 6600GT on AGP port

Thanks. I'll likely order a GF 6600 AGP board soon.

Could you attach the output of "hwinfo --gfxcard"? In case I won't manage to get a board in time I'll add the workaround for your board. Thanks.

Created attachment 71879 [details]
hwinfo --gfxcard

hwinfo --gfxcard loaded

Created attachment 71880 [details]
xorg.conf with workaround

xorg.conf with workaround

Created attachment 71881 [details]
XFree86.0.log

XFree86.0.log

Thanks. Since it's unlikey that I will get such a board in time I enabled the workaround for now. Fixed for Beta8.

Sorry, I think I have to reopen this bug for the second time.
If I the workaround you<re talking about is to add Options "HWCursor" "false" in xorgg.conf, then the bug is not solved.
This makes the cursor visible, but I stil cant use resolution depth higher than 16bits or resolution higher than 1280*1024... (see the initial description of the bug)

Oops. I'll try to reproduce this on my 6600 PCIe.

Could you try in Section "Device"

  Option "noaccel"

and if this helps these options instead (only one at the same time for now)

Option "XaaNoCPUToScreenColorExpandFill"
Option "XaaNoColor8x8PatternFillRect"
Option "XaaNoColor8x8PatternFillTrap"
Option "XaaNoDashedBresenhamLine"
Option "XaaNoDashedTwoPointLine"
Option "XaaNoImageWriteRect"
Option "XaaNoMono8x8PatternFillRect"
Option "XaaNoMono8x8PatternFillTrap"
Option "XaaNoOffscreenPixmaps"
Option "XaaNoPixmapCache"
Option "XaaNoScanlineImageWriteRect"
Option "XaaNoScreenToScreenColorExpandFill"
Option "XaaNoScreenToScreenCopy"
Option "XaaNoSolidBresenhamLine"
Option "XaaNoSolidFillRect"
Option "XaaNoSolidFillTrap"
Option "XaaNoSolidHorVertLine"
Option "XaaNoSolidTwoPointLine"
Option "XaaNoWriteBitmap"
Option "XaaNoWritePixmap"

> Oops. I'll try to reproduce this on my 6600 PCIe.
I can't. Works for me. I've tested 1280x1024 and 1600x1200, in 16 and 24bpp
on a 6600GT by MSI (PCIe).

are you testing the OSS 10.1 b6 x86_64 code?

Yes.

Denis, any results WRT comment #18?

I'll try to find some time to test it in the coming days.
It seems Fedora FC5 Test 3 has the same problem for some cards with the nv driver (this is where i found the workaround for the cursor), even if FC5t3 is using xorg v7.0 and oss 10.1, xorg v6.9 (which are supposed to be equivalent)...

In the meantime, you can have a look at:
https://www.redhat.com/archives/fedora-test-list/2006-February/msg01824.html

or here:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=182517

I have also seen there, posts of people not able to go higher than 1280*1024 or higher than 16bits with this driver on x86_64 architecture

According to these bugreports the mouse cursor issue occurs on 6100/6150 NVIDIA chipsets, i.e. Device ID 0240/0242.

I've tested all the option listed in comment #18, one by one.
- none of them made the cursor visible
- none of them allowed me to see the desktop in 1280*1024*24 (ok with 1280*1024*16)

Even not ' Option "noaccel" '? BTW, my 6600 GT AGP has reappeared meanwhile. So I can check tomorrow with mine as well.

My AGP 6600 board:

# hwinfo --gfxcard
22: PCI(AGP) 100.0: 0300 VGA compatible controller (VGA)        
  [Created at pci.295]
  UDI: /org/freedesktop/Hal/devices/pci_10de_f1
  Unique ID: VCu0.c8A7scn_Hn2
  Parent ID: vSkL.u+FWPoZGa71
  SysFS ID: /devices/pci0000:00/0000:00:01.0/0000:01:00.0
  SysFS BusID: 0000:01:00.0
  Hardware Class: graphics card
  Model: "nVidia GeForce 6600/GeForce 6600 GT"
  Vendor: pci 0x10de "nVidia Corporation"
  Device: pci 0x00f1 "GeForce 6600/GeForce 6600 GT"
  Revision: 0xa2

I can't reproduce your problems. I've tested 1280x1024 in both 16bpp and 24bpp color depth. I'll therefore close this one as WORKSFORME.

In case you find a workaround for your board feel free to reopen with a detailed description how to apply this workaround. Thanks.