In ll_fault0, the 'fault' struct is mostly cleared before the call to
cl_io_loop, but ft_flags is not reset. It is ordinarily set by
the call to filemap_fault in vvp_io_kernel_fault, but if Lustre
returns before calling filemap_fault, it still has the old value of
ft_flags.
ll_fault0 will then consume the ft_flags field. If it has the
VM_FAULT_RETRY bit set, it will be used as ll_fault0() and
ll_fault()'s return value.
This is a problem when VM_FAULT_RETRY is in ft_flags:
When fault/filemap_fault return with that flag set, they have already
released the mmap semaphore, and do_page_fault does not need to
release it.
Incorrectly returning this flag from ll_fault means mmap_sem
is not upped in the kernel's do_page_fault().
In addition to clearing ft_flags, this patch does not use it unless
it is valid. It's potentially misleading to return ft_flags in
"fault_ret" if ft_flags has not been set by filemap_fault.
This adds clarity, but does not change the current behavior:
When not valid, ft_flags is replaced by fault_ret, which is zero,
as is ft_flags when not set by filemap_fault.