https://bugs.winehq.org/show_bug.cgi?id=37750
--- Comment #7 from Béla Gyebrószki gyebro69@gmail.com --- (In reply to Henri Verbeet from comment #6)
If it's not too much effort, could you try bisecting the CSMT patches to see which patch fixes it? Either way, could you please attach a +tid,+d3d9,+d3d,+d3d_surface,+d3d_texture log?
The generated log is almost 700 MB in uncompressed form even though I removed the intro videos and ran the game until it reached the point in the menus where the problem is clearly visible. Download link to the compressed log: https://drive.google.com/open?id=0B-tTbLKBl-tOOE1FZHhXS3g4U2M
The commit that fixes the problem should be
44de8bdf554f05c0b416a1e356515579422174ff is the first bad commit commit 44de8bdf554f05c0b416a1e356515579422174ff Author: Stefan Dösinger stefan@codeweavers.com Date: Sat Jul 6 22:52:06 2013 +0200
wined3d: Ignore buffer->resource.map_count in the CS
This fixes sporadic test failures caused by preload refusing to load data into the GL buffer.
It's fine to preload or create a VBO if a DISCARD or NOOVERWRITE map is happening at the same time. A non-dynamic map will finish the CS before incrementing map_count.
During bisecting there was a skipped commit when the game crashed on start, there were 2 commits when so many artefacts were rendered that I could barely decide whether the issue was present or not.
'git checkout 44de8bdf554f05c0b416a1e356515579422174ff' --> the reported problem doesn't exist but there are many other graphical glitches in the game 'git reset --hard HEAD^' --> the problem is present (along with other glitches)
I used Stefan's git repository to do the bisect.