Re: [PATCH 06/13] mm/munlock: maintain page->mlock_count while unevictable

[Hugh Dickins]

On Fri, 11 Feb 2022, Vlastimil Babka wrote:
> On 2/6/22 22:40, Hugh Dickins wrote:
> > @@ -115,6 +116,7 @@ void lruvec_add_folio_tail(struct lruvec *lruvec, struct folio *folio)
> >  
> >  	update_lru_size(lruvec, lru, folio_zonenum(folio),
> >  			folio_nr_pages(folio));
> > +	/* This is not expected to be used on LRU_UNEVICTABLE */
> 
> Felt uneasy about this at first because it's just a _tail version of
> lruvec_add_folio, and there's probably nothing fundamental about the users
> of _tail to not encounter unevictable pages. But if the assumption is ever
> violated, the poisoned list head should make it immediately clear, so I
> guess that's fine.

Yes, I could have made that one check against LRU_UNEVICTABLE too, but
thought we would rather see the crash on the poisoned list head: since
specifically choosing the tail of an unordered (and imaginary) list
raises questions - it might turn out to be best permitted, it might turn
out to require a rethink; but until there is a case, let's crash on it.

Hugh