Hi Raphael!
Many thanks for your 2007 work and page
http://raphael.g-system.at/tech/netbsd/epfb/ based on Hamajima's epfb
driver for the Cirrus Logic EP93xx range of ARM chips. Without those
extra bits it would have taken me much longer to get it working, if at
all.
The first results are promising: the display is stable at bit depths
at which the Cirrus FB driver for Linux is already jumping about,
though I see what you mean about the precise video timing issues.
Like you, I am now looking at getting the X server to run on it, but
when I naively build the X sets for evbarm, the X server (and a wscons
driver for it) are not built and xserver.tgz just contains the fake
drivers Xprt, Xvfb etc.
I regret I'm new to NetBSD's build system, if old to BSD itself, and
am still figuring out how the contents of each architecture's .tgzs is
determined.
Before I carry on banging my head against this, may I ask whether
you (or anyone else) has made any further progress on epfb or in
getting NetBSD/X running on this chip, and should I be using a CVS
stable branch or head rather than plain 4.0 + patches?