Lock slurmstepd in memory

If slurmstepd had been swapped out before upgrade happened it could
  easily lead to SIGBUS at any time after upgrade. Prevent that by
  mlocking it.
bug 2334

NEWS

@@ -17,6 +17,8 @@ documents those changes that are of interest to users and administrators.
     that had a partition in them.
  -- Don't return the extern step from sstat by default.
  -- In sstat print 'extern' instead of 4294967295 for the extern step.
+ -- Lock slurmstepd in memory to avoid possible SIGBUS if the daemon is paged
+    out at the time of a Slurm upgrade (changing plugins).
 
 * Changes in Slurm 16.05.3
 ==========================

src/slurmd/slurmstepd/slurmstepd.c

@@ -157,6 +157,17 @@ main (int argc, char *argv[])
 	 * on STDERR_FILENO for us. */
 	dup2(STDERR_FILENO, STDOUT_FILENO);
 
+	/* slurmstepd is the only daemon that should survive upgrade. If it
+	 * had been swapped out before upgrade happened it could easily lead
+	 * to SIGBUS at any time after upgrade. Avoid that by locking it
+	 * in-memory. */
+#ifdef _POSIX_MEMLOCK
+	if (mlockall(MCL_FUTURE | MCL_CURRENT) < 0)
+		info("failed to mlock() slurmstepd pages: %m");
+#else
+	info("mlockall() system call does not appear to be available");
+#endif
+
 	/* This does most of the stdio setup, then launches all the tasks,
 	 * and blocks until the step is complete */
 	rc = job_manager(job);