At Wed, 04 Jun 2008 17:16:12 +0200, I wrote:
At Thu, 05 Jun 2008 01:12:28 +1000, Ben Stanley wrote:
So, if config.h is not needed in 'newer' kernels (and causes an error?), and is needed in 'older' kernels, it is then an error to universally include it in a configure test program where it causes an error if used inappropriately...
Well, apparently it's only RH kernels that cause the problem. So I haven't cared about it at all.
On the second thought, we may force to define the own linux/config.h to avoid this for the case > 2.6.14. A totally untested patch below.
Takashi
--- diff --git a/configure.in b/configure.in index 4c44567..e3dada2 100644 --- a/configure.in +++ b/configure.in @@ -615,6 +615,19 @@ CHECK_KERNEL_HEADER(linux/config.h, [#ifndef _LINUX_CONFIG_H #endif ])
+if test ! -f include/linux/config.h; then + if test "$kversion.$kpatchlevel" = "2.6" -a "$ksublevel" -gt 14; then + echo "Force to create linux/config.h" + mkdir -p include/linux + echo "#ifndef _LINUX_CONFIG_H +#define _LINUX_CONFIG_H +#ifndef AUTOCONF_INCLUDED +#include <linux/autoconf.h> +#endif +#endif" > include/linux/config.h + fi +fi + dnl Check kernel headers for 2.2 MODIFY_KERNEL_HEADER(linux/kmod.h, __LINUX_KMOD_H__) CHECK_KERNEL_HEADER(linux/utsrelease.h)