On Fre, 2003-02-28 at 20:21, Branden Robinson wrote: > On Fri, Feb 28, 2003 at 06:55:55PM +0100, Michel D?nzer wrote: > > > > Weird. Anyway, the test certainly isn't adequate because Option > > > > "UseFBDev" doesn't work with generic framebuffer devices like vesafb, > > > > OFfb, ... > > > > > > Could you propose a better test? What will /proc/fb say when someone is > > > using a generif framebuffer device like vesafb, OFfb, and so forth? > > > > > > I can write the string parsing; just tell me what values I need to > > > ignore. > > > > Unfortunately, I don't know the output of all framebuffer devices by > > heart either. :) Collecting some user data might be a good idea, but I'm > > not sure how to best go about it. > > > > > Also, I assume that anyone using a generic framebuffer needs to be using > > > the "fbdev" XFree86 *driver*, right? > > > > No, he can use a specific driver for his chipset or another generic > > driver like vesafb if applicable. > > Uh, so it sounds like you're telling me that /proc/fb is essentially > useless for autoconfiguring the X server. > > 1) /proc/fb reporting something doesn't mean you can just switch on > "UseFBDev", because that won't work with "generic [kernel] drivers"
I do think parsing /proc/fb to determine this is feasible, we just need to gather enough data. > 2) people using "generic [kernel] drivers" could use any X server video > driver, potentially, so their case works the same as people who > aren't using a kernel framebuffer driver at all Sure, the same is even true for people using a non-generic framebuffer device. > This is extremely disappointing. It makes successful X server > autoconfiguration essentially a crap shoot. Things like the PCI ID are certainly better suited for autoconfiguration. -- Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer XFree86 and DRI project member / CS student, Free Software enthusiast
