Thanks for the bug report.  How about the attached (untested) patch 
instead?  It checks for tcgetpgrp failure.  Also, the glibc manual 
suggests that emacsclient should be killing its process group with 
SIGTTIN, not itself with SIGSTOP.  See:

http://www.gnu.org/software/libc/manual/html_node/Foreground-and-Background.html