Description of problem:
cli crashes during volume set when it is invoked from the gluster cli prompt.

[root@kd glusterfs]# gluster
gluster> volume set v3 auth.allow "write-behind"
Segmentation fault (core dumped)

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


How reproducible:


Steps to Reproduce:
1. Go to the gluster cli prompt ( gluster>)
2. do volume set on auth.allow with value equal to "write-behind" (without omitting the double quotes), which is invalid but nonetheless, shouldn't cause the cli to crash.

Actual results:


Expected results:


Additional info:
The following info is taken from valgrind log.


==10540== Invalid write of size 2
==10540==    at 0x421FED: cli_to_glusterd (cli-rpc-ops.c:6008)
==10540==    by 0x417FAF: gf_cli_set_volume (cli-rpc-ops.c:2704)
==10540==    by 0x40DBFD: cli_cmd_volume_set_cbk (cli-cmd-volume.c:908)
==10540==    by 0x40A2CB: cli_cmd_process (cli-cmd.c:140)
==10540==    by 0x40A463: cli_cmd_process_line (cli-cmd.c:202)
==10540==    by 0x40AA97: cli_rl_process_line (cli-rl.c:103)
==10540==    by 0x40B15A: cli_rl_input (cli-rl.c:368)
==10540==    by 0x3DE8A07D13: start_thread (pthread_create.c:309)
==10540==    by 0x3DE86F167C: clone (clone.S:115)
==10540==  Address 0x56145c7 is 39 bytes inside a block of size 40 alloc'd
==10540==    at 0x4A06F18: calloc (vg_replace_malloc.c:566)
==10540==    by 0x4C6E062: __gf_default_calloc (mem-pool.h:75)
==10540==    by 0x4C6E46B: __gf_calloc (mem-pool.c:113)
==10540==    by 0x421F56: cli_to_glusterd (cli-rpc-ops.c:5999)
==10540==    by 0x417FAF: gf_cli_set_volume (cli-rpc-ops.c:2704)
==10540==    by 0x40DBFD: cli_cmd_volume_set_cbk (cli-cmd-volume.c:908)
==10540==    by 0x40A2CB: cli_cmd_process (cli-cmd.c:140)
==10540==    by 0x40A463: cli_cmd_process_line (cli-cmd.c:202)
==10540==    by 0x40AA97: cli_rl_process_line (cli-rl.c:103)
==10540==    by 0x40B15A: cli_rl_input (cli-rl.c:368)
==10540==    by 0x3DE8A07D13: start_thread (pthread_create.c:309)
==10540==    by 0x3DE86F167C: clone (clone.S:115)