The same kind of corruption I noticed in every game with R600g+color tiling.
I does work properly when color tiling is off.
mesa git, libdrm git, xf86-video-ati git, xorg-server-1.9, drm-radeon-testing
Please attach your xorg log and dmesg output after running the game with tiling enabled.
Created attachment 38507 [details]
Created attachment 38508 [details]
Here it is.
*** Bug 30076 has been marked as a duplicate of this bug. ***
Currently alink is working on porting wesnoth to opengl. The start was a crude hack somehow coding opengl equivalents of sdl commands. This branch does lead to screen corruptions as shown in the screenshot attached to Bug 30076 which is marked a duplicate of this one.
alink added some tests for commonly used functions in the wesnoth-gl port. It seems as if this (at least regarding the output of the function later on) does not work as expected:
20101007 15:55:17 error display: Can't read or write on back buffer.
20101007 15:55:17 error display: Can't read or write on front buffer.
20101007 15:55:17 error display: Can't copy from back buffer to front buffer.
20101007 15:55:17 error display: Can't copy from back buffer to back buffer.
Adding glFlush commands does lead the the same prob. Using r600c this does work over here, r600g does show the corruption when color tiling is active. Testbase:
rv670 512MB (HD3850), kernel 2.6.36-rc[6,7] (amd64), xorg-server-1.9.x, libdrm git master, xf86-video-ati git master, mesa git master
The function creating this output is "static bool test_buffers(bool flush)" and it is part of the following source file: http://svn.gna.org/viewcvs/wesnoth/branches/ogl/src/video.cpp?revision=46911&view=markup
Corruption is solved, but it's slow as hell. Just unplayable.