Comment 8 for bug 326445

I think there may be probability of hardware bug(?) or timing problem of memory access.
But we should consider such problem.

I will implement some safety code.

(*) page_hash_latch is not locked, but x_lock is still waiting.

lock_word = 1048576 ( unlocked ), reader_count = 0, writer = 351 ( RW_LOCK_EXCLUSIVE ),
writer_thread = 1204177232 ( 47c64950 ) , writer_count = 0, pass = 0, s_waiters = 1,
x_waiters = 0, wait_ex_waiters = 1,
last_s_file_name = 0x8c413a "btr/btr0sea.c",
last_x_file_name = 0x8c5f16 "buf/buf0lru.c", writer_is_wait_ex = 1,
last_s_line = 821, last_x_line = 1450,