Pinstripe installs kernel-headers-2.4.0-0.16.i386.rpm kernel-source-2.2.16-17.i386.rpm kernel-2.2.16-17.i686.rpm Why do we carry 2.4.0 header? When compiling kernel modules on such a system the modules arent usable because they use wrong headers and version information. Also (maybe rpm bug) [root@rhbeta RPMS]# rpm -i kernel-source-2.2.16-17.i386.rpm var/tmp/rpm-tmp.65670: line 14: syntax error: unexpected end of file execution of kernel-source-2.2.16-17 script failed, exit status 2 Bug we got [root@rhbeta RPMS]# df -a Filesystem 1k-blocks Used Available Use% Mounted on /dev/hda7 1344016 774200 501544 61% / none 0 0 0 - /proc usbdevfs 0 0 0 - /proc/bus/usb none 0 0 0 - /dev/pts /dev/hdc 666250 666250 0 100% /mnt/cdrom [root@rhbeta RPMS]# ls -l kernel-source-2.2.16-17.i386.rpm -rw-r--r-- 3 root root 17832250 Jul 27 01:01 kernel-source-2.2.16-17.i386.rpm But the worst thing is BADBAD: [root@rhbeta RPMS]# rpm -i kernel-source-2.2.16-17.i386.rpm package kernel-source-2.2.16-17 is already installed After a failed install is should _not_ be installed. Also:In file included from /usr/src/linux/include/linux/module.h:203, from toshiba.c:44: /usr/src/linux/include/linux/version.h:13:26: /boot/kernel.h: Datei oder Verzeichnis nicht gefunden [olaf@rhbeta src]$ ls /boot/ boot.0300 message System.map-2.2.16-17 boot.b module-info vmlinux-2.2.16-17 chain.b module-info-2.2.16-17 vmlinuz kernel.h-2.2.16 os2_d.b vmlinuz-2.2.16-17 map System.map we need a kernel.h not a kernel.h-2.2.16
The 2.4 headers are for glibc and link to /usr/include/linux which is correct This is fixed: Also (maybe rpm bug) [root@rhbeta RPMS]# rpm -i kernel-source-2.2.16-17.i386.rpm var/tmp/rpm-tmp.65670: line 14: syntax error: unexpected end of file execution of kernel-source-2.2.16-17 script failed, exit status 2 The other one may be a bug yes..
I believe the header file problem is fixed in our current source tree. post script failures have never caused rpm to roll back installs. That failure will show up once more when you upgrade to a more recent kernel, then disappear, it's just a typo caused by changing syntax from rpm to shell. So, as far as I can tell, everything here (with the exception of one thing that isn't a bug) is fixed in our current source tree.