Am Freitag 08 September 2006 21:10 schrieb Matthias Schwarzott: > On Friday 08 September 2006 19:41, Martin Wache wrote: > > Hi all, > > > > thinking again about the OSD troubles with nvidia, and that Stefan > > suggested that it might be a wrong range of YUV values, I have prepared > > some patches. Could someone who has these problems with the OSD please > > try them separately and tell me if one resolves the problems? > > > > I checked our conversion RGB->YUV, it respects the range. But there is > > still this small bug in the scaling algorithm which sometimes produces > > stray pixels at the edges. Those pixel may have any values and if the > > nVidia cards gets confused by those pixels, it may well be that the > > complete line looks strange. > > > > Thank you for any reports! > > Martin > > I tried all your patches applied against my ebuild for cvs from today. They > all do not change anything. > Then I tried using insight as debugger but that crashed :( > > Then I found a working version: > Compiling softdevice with -O0 instead of -O2 made it working :))) > (Exactly I use normally: -O2 -march=athlon-xp -fomit-frame-pointer -pipe, > and changed the -O2 to -O0 leaving the rest as they are).
Could you bisect thorugh the source files, compiling some with with O2 and O0, till you find the affected files? (probabaly needs soe Makefile modifications) Then bisect through the functions in that files by copying them to a seperate file till you find the broken ones? Then one could analyze what goes wrong. Cheers, -- (°= =°) //\ Prakash Punnoor /\\ V_/ \_V
pgpgj3SocSqUl.pgp
Description: PGP signature
_______________________________________________ Softdevice-devel mailing list [email protected] https://lists.berlios.de/mailman/listinfo/softdevice-devel
