[Bug 26119] kernel32/pipe tests show some valgrind warnings

https://bugs.winehq.org/show_bug.cgi?id=26119

--- Comment #3 from Austin English  ---
A similar issue in hlink/tests/hlink.c:
==11822== Thread 3:
==11822== Syscall param writev(vector[...]) points to uninitialised byte(s)
==11822==    at 0x4E9528CC: writev (in /usr/lib/libc-2.18.so)
==11822==    by 0x7BC7FE6F: send_request (server.c:213)
==11822==    by 0x7BC80001: wine_server_call (server.c:294)
==11822==    by 0x7BC80E6E: server_select (server.c:591)
==11822==    by 0x7BC8A6F6: NtWaitForMultipleObjects (sync.c:863)
==11822==    by 0x7BC8A746: NtWaitForSingleObject (sync.c:872)
==11822==    by 0x7BC47912: server_ioctl_file (file.c:1375)
==11822==    by 0x7BC48312: NtFsControlFile (file.c:1623)
==11822==    by 0x7B8796BF: ConnectNamedPipe (sync.c:1608)
==11822==    by 0x469EF26: listen_thread (rpc_transport.c:131)
==11822==    by 0x7BC87137: ??? (signal_i386.c:2571)
==11822==    by 0x7BC87180: call_thread_func (signal_i386.c:2630)
==11822==    by 0x7BC87115: ??? (signal_i386.c:2571)
==11822==    by 0x7BC8E560: start_thread (thread.c:428)
==11822==    by 0x4EA7BD89: start_thread (in /usr/lib/libpthread-2.18.so)
==11822==    by 0x4E95CA0D: clone (in /usr/lib/libc-2.18.so)
==11822==  Address 0x51be59c is on thread 1's stack
==11822==  Uninitialised value was created by a stack allocation
==11822==    at 0x7B8795C0: ConnectNamedPipe (sync.c:1594)
==11822==