I've just fixed that.  Turned out there were two bugs playing along:
r/g/b positions within pixel where not passed correctly to rasops, and
the endianness of the pixel accesses was not set correctly either.
That resulted in bits messed up in such a way, that the "white" color
was actually "#381830" - i.e. something in the very dark gray20 area.

Committed to current.  I'll request pull-up to 3.0 after people have
chance to test it on other hpc* ports.

SY, Uwe
-- 
uwe@ptc.spbu.ru                         |       Zu Grunde kommen
http://www.ptc.spbu.ru/~uwe/            |       Ist zu Grunde gehen