On Wed, 8 Mar 2017, Michal Hocko wrote: > JFTR the previous version was posted here: https://lwn.net/Articles/371892/ > and Dave had some concerns https://lkml.org/lkml/2010/2/8/329 which led > to a different approach and design of the slab shrinking > https://lkml.org/lkml/2010/2/8/329. > > I haven't looked at this series yet but has those concerns been > addressed/considered? Well yes this has been discussed for a couple of years. The basic approach is not only needed for the file systems (like what Chinner was focusing on) but in general for slab caches. The objection was regarding the integration into the slab reclaim logic in vmscan.c and the filesystem reclaim in general. Dave and Matthew were at linux.conf.au and we agreed to first try it with the radix tree and then generalize from there. The reclaim logic was a bit hacky and we will have to find some better way to integrate this. There is a video on youtube capturing the discussion (My talk on movable kernel objects). -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
