* Peng Zhang <zhangpeng.00@xxxxxxxxxxxxx> [230731 07:45]: > > > 在 2023/7/26 23:08, Matthew Wilcox 写道: > > On Wed, Jul 26, 2023 at 11:02:52AM -0400, Liam R. Howlett wrote: > > > * Peng Zhang <zhangpeng.00@xxxxxxxxxxxxx> [230726 04:10]: > > > > static inline > > > > -enum maple_type mas_parent_type(struct ma_state *mas, struct maple_enode *enode) > > > > +enum maple_type ma_parent_type(struct maple_tree *mt, struct maple_node *node) > > > > > > I was trying to keep ma_* prefix to mean the first argument is > > > maple_node and mt_* to mean maple_tree. I wasn't entirely successful > > > with this and I do see why you want to use ma_, but maybe reverse the > > > arguments here? > > > > I think your first idea is better. Usually we prefer to order the > > arguments by "containing thing" to "contained thing". So always use > > (struct address_space *, struct folio *), for example. Or (struct > > mm_struct *, struct vm_area_struct *). > There are disagreements here, so how to decide? But I don't know if the > new version still has this helper. Please keep the maple tree as the first argument and use: mt_parent_type as the name.. if you still need it.
