Now that I've learned something about the CONFIG_RTNETLINK option for the Linux kernel configuration (thanks to everyone who responded), I've found that there is a connection between it and a new daemon's build (not just the kernel's build), via the /usr/include/linux/autoconf.h file.
 
Initially, this .h file only contained the line "#error:  Invalid kernel header included in userspace" which caused the compilation of a file to fail.  Then I found that there were autoconf.h files under each of the kernel versions I have worked with, and they looked like they were generated from the .config file that's generated by make menuconfig/make xconfig.  So I copied the one from my /usr/src/linux-2.4.18-3/include subdirectory into /usr/include/linux, and my build went OK.  Then I inserted a "#define  CONFIG_RTNETLINK 1" in it, and my build went OK and grew by 15KB.
 
So my questions are:
    1. Assuming that my make menuconfig/make xconfig are broken, so that I have to manually edit the .config file to turn RTNETLINK on, what do I have to do to generate autoconf.h properly?  I tried just running "autoconf," and that really messed me up after coming back really fast.  It's gotta be specific to the kernel version whose .config file I modify.
    2. What utility do I have to use to get this autoconf.h "installed" at /usr/include/linux, where it's apparently expected?
    3. Are the first two questions the right questions?  (I think I also have to re-build my kernel with this option.)
Thanks again,
  Dennis Ching
 

Reply via email to