On 06/11/2012 06:52 PM, Adamson, Andy wrote: <> >> >> Yeah, right now pnfs_return_layout always returns the whole layout >> so this patch works. Boaz's point (I think) was that marking the whole layout >> as returned in the generic layer will prevent returning of layout segments >> by the objects layout. > > Yep. I agree that _pnfs_return_layout needs to operate on an input range, and > should wait for all lseg references - the "normal" behavior. > You mean the normal "error handling behavior" right? > Then, after that is done, > I will write a patch to allow the file layout driver to call _pnfs_return_layout in > "fence mode" e.g returning the whole layout, and not waiting for the last lseg reference. > This one I do not understand. And am really trying. please be slow with me? 1. What does it means "whole layout" Is that just a short hand for "all segments" ? 2. What is "fence mode" LAYOUTRETURN. where did you take this one from? 3. How does "fence mode" permits a client to use a layout segment after the send of LAYOUTRETURN. Since that is what we do (to date) because of race. Thanks Boaz > -->Andy -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
