LU-2432 ptlrpc: alloc_rqbd spin on vmap_area_lock

vmalloc based allocations can potentially take a very long time to
complete due to a regression in the kernel. As a result, MDS service
threads might lock up for certain periods of time while all of the
cores spin on the vmap_area_lock down in ptlrpc_alloc_rqbd.

Tihs patch only allow one thread of each CPT to enter rqbd
alloation path.

Signed-off-by: Liang Zhen <liang@whamcloud.com>
Change-Id: I4ba442801859ae58cf8e8dd8ae18af1062379639
Reviewed-on: http://review.whamcloud.com/4939
Tested-by: Hudson
Tested-by: Maloo <whamcloud.maloo@gmail.com>
Reviewed-by: Bobi Jam <bobijam@gmail.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Prakash Surya <surya1@llnl.gov>
Reviewed-by: Oleg Drokin <green@whamcloud.com>