>> From: Wang Qing <wangq...@vivo.com> >> >> sched_domain_flags_f() are statically set now, but actually, we can get a >> lot of necessary information based on the cpu_map. e.g. we can know whether >> its cache is shared. >> >> Allows custom extension without affecting current. > >This all still makes absolutely no sense. The architecture builds these >masks, the architecture is in charge of which flags function is called >on which mask. > >Passing the mask back in means it lost the plot somewhere and doens't >know wth it's doing anymore.
It is not passing the mask back, sched_domain_flags_f() doesn't use cpumask at all, it always return fixed values. sd_topology_level select different sd_flag() to config, it's too primitive. If an architecture can describe its cache topology clearly in every level, the sd only need sched_domain_flags_f(cpumask) to get its ShPR flag. Thanks, Wang > >NAK
