On Mon, Jul 06, 2026 at 10:55:07AM -0700, Stanislav Kinsburskii wrote: > +The caller must not hold ``mmap_read_lock`` before the call. > +``hmm_range_fault_unlocked()`` takes the mmap read lock internally and allows > +``handle_mm_fault()`` to drop it during fault handling. If the mmap lock is > +dropped, the function returns ``-EBUSY``. The caller must then restart the > +walk from ``range->start`` with a fresh notifier sequence. ``-EINTR`` is > +returned if a fatal signal is pending during retry handling.
It doesn't make sense to me to expose this to the caller, the caller isn't holding the lock and there is no reason to redo the mmu_interval at this point, so why not just loop inside hmm_range_fault_unlocked ? Jason
