On Mon, 25 Sep 2023, Matthew Wilcox wrote:
> On Mon, Sep 25, 2023 at 01:32:02AM -0700, Hugh Dickins wrote:
> > {
> > struct page *page = __alloc_pages(gfp | __GFP_COMP, order,
> > - preferred_nid, nodemask);
> > + preferred_nid, nodemask);
>
> I really prefer not to do this "align arguments with opening bracket"
> style. As long as they're indented enough to make them visually distinct
> from indentation-for-if-blocks, I find it annoying when functions get
> renamed to something with a different length and somebody then wastes
> time reindenting all the arguments to match.
Okay, I don't care much about inserting spaces to align with the bracket,
but didn't like those continuation args leftward of function name above.
I'll adjust in v2, and eventually we reach a compromise.
>
> > + return page_rmappable_folio(page);
>
> I don't particularly object to the main thrust of this patch. I'm not
> sure I like it in huge_mm.h though. Maybe in mm/internal.h? I
> wouldn't want anyone outside mm/ calling it.
I was expecting more resistance :) Right, I put it in huge_mm.h to be
next to your folio_prep_large_rmappable() declaration, but it doesn't
have to be there. Ooh, there's a folio_undo_large_rmappable() in
mm/internal.h, I'll move page_rmappable_folio() next to that.
Hugh
