Description of problem:

Attempting to specify a kdump_pre hook in /etc/kdump.conf produces the
following error message from mkdumprd:

	/etc/kdump.conf: Unsupported type kdump_pre

This is because, although an entry for kdump_pre has been added to
the case statement starting around line 879 in /sbin/mkdumprd, the
corresponding entry was not added to the case statement near line
1587.  The result is that the latter case statement falls through
to its default, which is to print the above error message and exit.

In addition, having corrected for the above problem, the resulting
init in the dump initrd invokes the wrong kdump_pre script.  They
way it's coded, it always invokes "/bin/<scriptname>", even if
kdump.conf specifies some other executable for the hook.

This latter problem can be fixed by using $KDUMP_PRE instead of
$KDUMP_PRE_INTERNAL in several places in the mkdumprd script.
This makes the pre- code behave just as does the (working) post-
code.

Version-Release number of selected component (if applicable):

# rpm -qa | grep kexec
kexec-tools-1.102pre-10.el5

How reproducible:

Happens every time.

Steps to Reproduce:

1. add a kdump_pre hook to /etc/kdump.conf
2. invoke 'service kdump restart' to cause the dump initrd to be rebuilt
  
Actual results:

	/etc/kdump.conf: Unsupported type kdump_pre

Expected results:

	Successful restart of kdump service.


Additional info:

I've attached a patch to mkdumprd that appears to correct both
problems.