Thanks for the reply, I figured it out and you are kind of right about the rpm doing something strange. Part of the rpm post is to have a custom fglrx module built via script. The way the sh script looks for the kernel headers causes it to look in
/usr/src/linux/include for the headers. In the latest kernel releases they no longer use /usr/src/linux/include. Usually they just use /usr/src/redhat/kernel-version/... and /lib/modules/kernel-version/... , The fglrx bin that will build you custom rpm's for different distro's builds in a way for the rpm to find your kernel headers based on uname queries. The problem with doing this during installation is that one of the conditionals in the includes if-else routines comes up true, which causes the /usr/src/linux/include to be used as opposed to the real /lib/modules/kernel-version/build/include. It's either something weird that happens do to the chroot env your operating in, or maybe its something totally different.
The only work around I've found is to extract the rpm, pull the post script and have the sh script run with hard coded settings as part of the post, which then of course defeats the whole point of having an rpm...
Last edited by dizowned; 02-26-2010 at 05:47 AM.