[freespace2] More on the text coming and going

[Taylor Richards]

On Fri, 2003-03-14 at 21:30, Steven Fuller wrote:
> On Fri, 14 Mar 2003, Bradley Bell wrote:
> 
> > It's just as Josh described.  As far as I remember, the fonts and the various
> > cockpit bitmaps change into boxes shortly after appearing, and once in a while
> > will flash back to their normal appearance for a short time.
> 
> This sounds like the internal texture state in gropengl.cpp is messed up
> (textures not being correctly bound), but anyone without a G400 has been
> unable to reproduce it and I have not been able to find any obvious errors
> when inspecting the code.  I do not believe it is something specifically
> wrong with the font.
> 
> 
> Steven Fuller
> 

I don't think it's something with the fonts either since the problem
goes away if you disable the shaded background behind the text in
mainhallmenu.cpp.  I don't know if it's related but from a valgrind log
on my G400 test system there are a butt load of mga related errors that
don't show up with NVIDIA drivers or running with software GL.  I didn't
really look carefully at it (since I don't quite understand it) but I
think the mga is using a 565 texture format instead of 1555.

==26689== 1253376 errors in context 80 of 80:
==26689== Invalid write of size 4
==26689==    at 0x4478CB11: mgaConvertTexture (mgatexcnv.c:117)
==26689==    by 0x4478D8D8: mgaUploadSubImageLocked (mgatexmem.c:384)
==26689==    by 0x4478D9A2: mgaUploadTexLevel (mgatexmem.c:422)
==26689==    by 0x4478DB44: mgaUploadTexImages (mgatexmem.c:509)
==26689==    Address 0x4932D000 is not stack'd, malloc'd or free'd

I can send you the whole log it you want.  Here are some links to those
source files if you want to look them over:

http://www.atomised.org/docs/XFree86-4.2.1/mgatexmem_8c-source.html
http://www.atomised.org/docs/XFree86-4.2.1/mgatexcnv_8c-source.html


Taylor

-- 
Taylor Richards <mtrs at bellsouth.net>