> > I wonder if, for x86, we are just missing an initialization: > > > > memset(&fl, 0, sizeof(struct file_lock)); > > > > in the beginning of locks_remove_flock(). > > Yes we are, though I think locks_init_lock() is the better > initialization. I'm not sure how I missed that when writing the patch. > > Thanks for testing! > > NeilBrown Tks for the feedback. No problem at all, that's why we are here for. Cheers o/ -- Rafael D. Tinoco Linaro Kernel Validation Team
