bluegene system.

Bug 3331

and erroneous error later on.

The old logic would result in test16.4 failing some of the time.
  The failure was caused by the sattach command attaching to a
  job step before the original srun command received a
  RESPONSE_LAUNCH_TASKS message. That messsage  would then be sent
  to the salloc command. Since srun never got the message, it
  would hang. This change does not mark the job step as RUNNING
  until after the original srun gets sent the RESPONSE_LAUNCH_TASKS
  message and sattach requests are blocked until that time.

Identify the function where an error is generated.

This bug was introduced in commit 155a65a4
  A bad array index value was used.

This isn't a new bug, but logic in this area changed and Coverity
  reported it as a new problem (duplicate logic). The logic also
  incorrectly used a strstr() rather than strcmp() to find maching
  process names.

Fixing a resource leak reported by Coverity in the SPANK logic
  resulted in unmounting the spank object and losing locally
  stored information. Regression test7.11 broke with the new
  code. We'll need to live with the error reported by Coverity
  and keep SPANK running. This is a partial reversion of commit
  6fb3fea3

Regression tests 1.63 was sometimes failing due to a race condition.
  Specifically, the srun command was signalled prior to completion
  of the task launch process, so the "tasks_started" bitmap did not
  yet have it's bits set, resulting in the signal send request being
  silently ignored.

This sets a memory pointer to NULL after releasing its memory to
  avoid a duplicate xfree call. Bug introduced in commit
  07ae3e72

This is a fix to commit d470d0a9

No longer needed (as noted by comment) since 14.11 protocol support
is already removed.

Missed in previous AIX removal.

All unused within their scope; no functional change.

3 places where a string pointer referenced an out-of-scope memory location
1 memory leak and
1 unterminated string

Accidentally commited to Slurm version 16.05 instead of master

3 places where a string pointer referenced an out-of-scope memory location
1 memory leak and
1 unterminated string

4 possible memory leaks
2 failures to check for errors returned by functions and
4 possibly unterminated strings

backfill scheduler: Stop trying to determine expected start time for a job
    after 2 seconds of wall time. This can happen if there are many running jobs
    and a pending job can not be started soon.
byg 3373

mobile devices didn't behave as expected.

This reverts commit 17549a03.

This reverts commit 7495090d.

# Conflicts:
#	src/sbcast/opts.c

some of the tables (prolog_epilog) were hard to read.