On Wed, May 15, 2024 at 04:32:36PM +0100, Matthew Wilcox wrote:
> On Fri, May 03, 2024 at 02:53:47AM -0700, Luis Chamberlain wrote:
> > +int split_folio_to_list(struct folio *folio, struct list_head *list);
>
> ...
>
> > +static inline int split_folio_to_list(struct page *page, struct list_head *list)
> > +{
>
> Type mismatch. Surprised the build bots didn't whine yet.
Good catch. As we always enabled CONFIG_THP, we didn't detect this
issue.
>
> >
> > + min_order = mapping_min_folio_order(folio->mapping);
> > + if (new_order < min_order) {
> > + VM_WARN_ONCE(1, "Cannot split mapped folio below min-order: %u",
> > + min_order);
> > + ret = -EINVAL;
> > + goto out;
> > + }
>
> Wouldn't we prefer this as:
>
> if (VM_WARN_ONCE(new_order < min_order,
> "Cannot split mapped folio below min-order: %u",
> min_order) {
> ret = -EINVAL;
> goto out;
> }
>
I don't think so:
#define VM_WARN_ONCE(cond, format...) (void)WARN_ONCE(cond, format)
So we get a build error as follows:
In file included from ./include/linux/mm.h:6,
from mm/huge_memory.c:8:
mm/huge_memory.c: In function ‘split_huge_page_to_list_to_order’:
./include/linux/mmdebug.h:93:39: error: void value not ignored as it ought to be
93 | #define VM_WARN_ONCE(cond, format...) (void)WARN_ONCE(cond, format)
| ^
mm/huge_memory.c:3158:21: note: in expansion of macro ‘VM_WARN_ONCE’
3158 | if (VM_WARN_ONCE(new_order < min_order,
