Am 30.04.10 15:15 schrieb(en) Michel Dänzer:
As you booted with modeset=0, there's no driver usable for the console. Either 
try without modeset=0 (you may need radeon.agpmode=1 instead for AGP to work 
reliably with KMS) or re-enable radeonfb.


> The strange effect in Xine is *completely* gone.

So as I suspected probably bitrot in the non-DRI big endian paths.

Does EXA also work without workarounds now?

I ran a bunch of further tests, now with the latest stable kernel (2.6.33.3).

A kernel built without CONFIG_FB_RADEON (neither 'm' nor 'y') fails to start X.  The X11 
log file (interestingly, Xorg.1.log) ejects with "(EE) Screen(s) found, but none 
have a usable configuration".  Thus, I continued with 'CONFIG_FB_RADEON=y' and the 
'radeon.modeset=0' as kernel arg.

I now compared the following 'Option' settings:
1. "AccelMethod" "XAA"
2. "AccelMethod" "EXA"
3. "AccelMethod" "EXA" with "RenderAccel" "off"

All three setups show more or less the same Glx performance, between 4150 and 
4200 glxgears frames.  The xine bitrot issue is now resolved in all three modes.

Mode #2 (EXA w/o RenderAccel off) shows several rendering errors:
- font rendering on the Gnome desktop is partly broken, typically showing black 
horizontal artefacts in the glyphs;
- in Gnome themes where buttons have a gradient (e.g. in gdm's default theme), 
the gradient shows horizontal artefacts;
- bright flickering pixels in the Xine splash screen;
- in the OpenOffice spreadsheet, single flickering pixels are visible in cells 
where cells in preceding columns are not empty.

In Mode #3 (EXA w/ RenderAccel off), I notices two issues:
- the flickering pixels in OpenOffice as above;
- all OpenOffice toolbar icons are completely broken.

Thus, I returned to mode #1 (XAA), which does also show the OpenOffice 
flickering spreadsheet pixels issue as the EXA modes, but the rest looks quite 
ok afaict.  I still have no access to the text consoles.

So the setup boils down to the following:

- kernel config (only activated options):
<snip>
CONFIG_AGP=m
CONFIG_AGP_UNINORTH=m
CONFIG_DRM=m
CONFIG_DRM_KMS_HELPER=m
CONFIG_DRM_TTM=m
CONFIG_DRM_RADEON=m
CONFIG_DRM_RADEON_KMS=y
CONFIG_FB_RADEON=y
CONFIG_FB_RADEON_I2C=y
CONFIG_FB_RADEON_BACKLIGHT=y
</snip>

- xorg.conf (the 108 MHz pixel clock mode was found by x11, but I have to 
explicitly select a mode, as I otherwise get the mac mini resolution...):
<snip>
Section "Device"
        Identifier      "Configured Video Device"
        Driver          "radeon"
        Option          "AccelMethod"   "XAA"
        Option          "MacModel"      "mini-internal"
EndSection
Section "Monitor"
        Identifier      "Formac TFT"
        Modeline        "1280x1024"     108.00  1280 1328 1440 1688  1024 1025 
1028 1066 +hsync +vsync
EndSection
Section "Screen"
        Identifier      "Default Screen"
        Monitor         "Formac TFT"
        Device          "Configured Video Device"
        DefaultDepth    24
        SubSection      "Display"
                Depth   24
                Modes   "1280x1024"
        EndSubSection
EndSection
Section "ServerLayout"
        Identifier      "Default Layout"
        Screen          "Default Screen"
EndSection
</snip>

Needless to say that this is not really intuitive...  I know, though, that I 
use somewhat exotic hardware, and I have no idea if it would be possible to bet 
some of the missing information e.g from OpenFirmware.

So, thanks a lot for your help!  Of course, if you have any ideas how a setup 
with less tweaks would be possible, please let me know!

Best, Albrecht.

Attachment: pgp8mMQJwtvO4.pgp
Description: PGP signature

_______________________________________________
xorg-driver-ati mailing list
[email protected]
http://lists.x.org/mailman/listinfo/xorg-driver-ati

Reply via email to