On Thu, Mar 29, 2018 at 01:11:09PM +0100, Mel Gorman wrote: >On Thu, Mar 29, 2018 at 11:36:07AM +0800, Wei Yang wrote: >> set_pageblock_order() is a standalone function which sets pageblock_order, >> while current implementation calls this function on each ZONE of each node >> in free_area_init_core(). >> >> Since free_area_init_node() is the only user of free_area_init_core(), >> this patch moves set_pageblock_order() up one level to invoke >> set_pageblock_order() only once on each node. >> >> Signed-off-by: Wei Yang <richard.weiyang@xxxxxxxxx> > >The patch looks ok but given that set_pageblock_order returns immediately >if it has already been called, I expect the benefit is marginal. Was any >improvement in boot time measured? No, I don't expect measurable improvement from this since the number of nodes and zones are limited. This is just a code refine from logic point of view. > >-- >Mel Gorman >SUSE Labs -- Wei Yang Help you, Help me
