report of possible data race at jffs2_rename()

[홍신 shin hong]

Hi. It seems that jffs2_rename() in jffs/dir.c may result
race condition for the following reason.

jffs2_rename() in jffs2/dir.c accesses "victim_f->inocache"
while it does not hold "mutex_lock(&victim_f->sem)".

However, it seems that it may result race conditions
since there are functions such as jffs2_do_read_inode()
which may change the value of "inocache" as "null".

Shin Hong