If we fail to write the pid to the pid file, we should still close the
file before returning from write_pid_file().  The consequences of this
bug are trivial because kadmin is just going to exit regardless.
Reported by Will Fiveash <will.fiveash@oracle.com>.

Based on a patch from Xi Wang <xi@mit.edu>.

If an iprop slave tries to load a dump from the master and it fails,
reset the ulog header so we take another full dump, instead of
reporting that the slave is current when it isn't.  Reported by
Richard Basch <basch@alum.mit.edu>.

ticket: 7588
target_version: 1.11.2
tags: pullup

Commit 0780e46fc13dbafa177525164997cd204cc50b51 matched a %ld format
string with the integer 0, which is an int rather than a long.  Just
put 0 in the format string instead.  Noted by David Benjamin
<davidben@mit.edu>.

k5srvutil is a little more convenient to use for rolling keys than
kadmin is.  When migrating off 1DES, though, it may be desirable to
explicitly specify the desired keysalts.  This adds an option, -e, to
k5srvutil to specify desired keysalts.

[ghudson@mit.edu: style fix; make whitespace in keysalt list work]

ticket: 7589 (new)

Add k5-int.h static functions to duplicate byte ranges, optionally
with a trailing zero byte, and set an error code like k5alloc does.
Use them where they would shorten existing code.

Reformat and simplify dump.c code according to current coding
standards.  No functional changes except for some error messages.

When dumping, use a common iterator function to unpack the dump_args
structure, unparse and filter the principal name, and convert master
keys.  Add helper functions to dump and load the "octets or -1" format
used for optional binary fields in the current dump format.

Without changing anything (except to make a few internal functions
static), reorder dump.c to bottom-up order so that forward function
declarations aren't needed.