Re: [PATCH v2 1/3] mm: document zone device struct page field usage

[Ira Weiny]

On Sun, Jul 21, 2019 at 09:02:04AM -0700, Matthew Wilcox wrote:
> On Fri, Jul 19, 2019 at 12:29:53PM -0700, Ralph Campbell wrote:
> > Struct page for ZONE_DEVICE private pages uses the page->mapping and
> > and page->index fields while the source anonymous pages are migrated to
> > device private memory. This is so rmap_walk() can find the page when
> > migrating the ZONE_DEVICE private page back to system memory.
> > ZONE_DEVICE pmem backed fsdax pages also use the page->mapping and
> > page->index fields when files are mapped into a process address space.
> > 
> > Restructure struct page and add comments to make this more clear.
> 
> NAK.  I just got rid of this kind of foolishness from struct page,
> and you're making it harder to understand, not easier.  The comments
> could be improved, but don't lay it out like this again.

Was V1 of Ralphs patch ok?  It seemed ok to me.

Ira

> 
> > @@ -76,13 +76,35 @@ struct page {
> >  	 * avoid collision and false-positive PageTail().
> >  	 */
> >  	union {
> > -		struct {	/* Page cache and anonymous pages */
> > -			/**
> > -			 * @lru: Pageout list, eg. active_list protected by
> > -			 * pgdat->lru_lock.  Sometimes used as a generic list
> > -			 * by the page owner.
> > -			 */
> > -			struct list_head lru;
> > +		struct {	/* Page cache, anonymous, ZONE_DEVICE pages */
> > +			union {
> > +				/**
> > +				 * @lru: Pageout list, e.g., active_list
> > +				 * protected by pgdat->lru_lock. Sometimes
> > +				 * used as a generic list by the page owner.
> > +				 */
> > +				struct list_head lru;
> > +				/**
> > +				 * ZONE_DEVICE pages are never on the lru
> > +				 * list so they reuse the list space.
> > +				 * ZONE_DEVICE private pages are counted as
> > +				 * being mapped so the @mapping and @index
> > +				 * fields are used while the page is migrated
> > +				 * to device private memory.
> > +				 * ZONE_DEVICE MEMORY_DEVICE_FS_DAX pages also
> > +				 * use the @mapping and @index fields when pmem
> > +				 * backed DAX files are mapped.
> > +				 */
> > +				struct {
> > +					/**
> > +					 * @pgmap: Points to the hosting
> > +					 * device page map.
> > +					 */
> > +					struct dev_pagemap *pgmap;
> > +					/** @zone_device_data: opaque data. */
> > +					void *zone_device_data;
> > +				};
> > +			};
> >  			/* See page-flags.h for PAGE_MAPPING_FLAGS */
> >  			struct address_space *mapping;
> >  			pgoff_t index;		/* Our offset within mapping. */
> > @@ -155,12 +177,6 @@ struct page {
> >  			spinlock_t ptl;
> >  #endif
> >  		};
> > -		struct {	/* ZONE_DEVICE pages */
> > -			/** @pgmap: Points to the hosting device page map. */
> > -			struct dev_pagemap *pgmap;
> > -			void *zone_device_data;
> > -			unsigned long _zd_pad_1;	/* uses mapping */
> > -		};
> >  
> >  		/** @rcu_head: You can use this to free a page by RCU. */
> >  		struct rcu_head rcu_head;
> > -- 
> > 2.20.1
> > 
>