On Wed, Aug 28, 2019 at 10:01:40AM +0200, Vlastimil Babka wrote: >On 8/28/19 8:06 AM, Wei Yang wrote: >> When insert and delete a vma, it will compute and propagate related subtree >> gap. After some investigation, we can reduce subtree gap propagation a little. >> >> [1]: This one reduce the propagation by update *next* gap after itself, since >> *next* must be a parent in this case. >> [2]: This one achieve this by unlinking vma from list. >> >> After applying these two patches, test shows it reduce 0.3% function call for >> vma_compute_subtree_gap. > >BTW, what's the overall motivation of focusing so much >micro-optimization effort on the vma tree lately? This has been rather >stable code where we can be reasonably sure of all bugs being found. Now >even after some review effort, subtle bugs can be introduced. And >Matthew was warning for a while about an upcoming major rewrite of the >whole data structure, which will undo all this effort? > Hi, Vlastimil Thanks for your comment. I just found there could be some refine for the code and then I modify and test it. Hope this could help a little. You concern is valid. The benefits / cost may be not that impressive. The community have the final decision. For me, I just want to make it better if we can. -- Wei Yang Help you, Help me
