On Thu, Dec 08, 2022 at 04:24:19PM -0800, John Hubbard wrote: > It does! And now for the comment, I'll think you'll find that this suffices: > > /* > * All callers that get here will already hold the i_mmap_rwsem. > * Therefore, no additional locks need to be taken before > * calling hugetlb_walk(). > */ > > ...which, considering all the data above, is probably the mother of > all summaries. :) But really, it's all that people need to know here, and > it's readily understandable without wondering what KSM has to do with this, > for example. I'm okay with the change. :) I think what I'll do is I'll move part of the original one into commit message, and take the new version in the code. Thanks, -- Peter Xu
