Re: [RFC PATCH] mm, meminit: remove init_reserved_page()

[Wei Yang <richard.weiyang@gmail.com>]

On Mon, Nov 19, 2018 at 11:48:45AM +0800, Wei Yang wrote:
>Function init_reserved_page() is introduced in commit 7e18adb4f80b ("mm:
>meminit: initialize remaining struct pages in parallel with kswapd").
>While I am confused why it uses for_each_mem_pfn_range() in
>deferred_init_memmap() to initialize deferred pages structure.
>
>After commit 2f47a91f4dab ("mm: deferred_init_memmap improvements"),
>deferred_init_memmap() uses for_each_free_mem_range() to initialize
>page structure. This means the reserved memory is not touched.
>
>The original context before commit 7e18adb4f80b ("mm: meminit: initialize
>remaining struct pages in parallel with kswapd"), reserved memory's page
>structure is just SetPageReserved, which means they are not necessary to be
>initialized.
>
>This patch removes init_reserved_page() to restore the original context.

Oops, my understanding is not correct :-(

>
>Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
>
>---
>
>I did bootup and kernel build test with the patched kernel, it looks good.
>
>One of my confusion is the commit 7e18adb4f80b ("mm: meminit: initialize
>remaining struct pages in parallel with kswapd") works fine. Does it eat some
>reserved pages? Either I don't see the reason in commit 2f47a91f4dab ("mm:
>deferred_init_memmap improvements") of changing pfn iteration from
>for_each_mem_pfn_range() to for_each_free_mem_range().
>
>Another question is in function reserve_bootmem_region(), we add a
>INIT_LIST_HEAD() in commit 1d798ca3f164 ("mm: make compound_head() robust").
>While the reserved page is never visible in page allocator. Do we still need
>to do this step?
>
>---
> mm/page_alloc.c | 28 ----------------------------
> 1 file changed, 28 deletions(-)
>
>diff --git a/mm/page_alloc.c b/mm/page_alloc.c
>index 2d3c54201255..48cf24766343 100644
>--- a/mm/page_alloc.c
>+++ b/mm/page_alloc.c
>@@ -1192,32 +1192,6 @@ static void __meminit __init_single_page(struct page *page, unsigned long pfn,
> #endif
> }
> 
>-#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
>-static void __meminit init_reserved_page(unsigned long pfn)
>-{
>-	pg_data_t *pgdat;
>-	int nid, zid;
>-
>-	if (!early_page_uninitialised(pfn))
>-		return;
>-
>-	nid = early_pfn_to_nid(pfn);
>-	pgdat = NODE_DATA(nid);
>-
>-	for (zid = 0; zid < MAX_NR_ZONES; zid++) {
>-		struct zone *zone = &pgdat->node_zones[zid];
>-
>-		if (pfn >= zone->zone_start_pfn && pfn < zone_end_pfn(zone))
>-			break;
>-	}
>-	__init_single_page(pfn_to_page(pfn), pfn, zid, nid);
>-}
>-#else
>-static inline void init_reserved_page(unsigned long pfn)
>-{
>-}
>-#endif /* CONFIG_DEFERRED_STRUCT_PAGE_INIT */
>-
> /*
>  * Initialised pages do not have PageReserved set. This function is
>  * called for each range allocated by the bootmem allocator and
>@@ -1233,8 +1207,6 @@ void __meminit reserve_bootmem_region(phys_addr_t start, phys_addr_t end)
> 		if (pfn_valid(start_pfn)) {
> 			struct page *page = pfn_to_page(start_pfn);
> 
>-			init_reserved_page(start_pfn);
>-
> 			/* Avoid false-positive PageTail() */
> 			INIT_LIST_HEAD(&page->lru);
> 
>-- 
>2.15.1

-- 
Wei Yang
Help you, Help me