Re: [RFC][PATCH] proc: export more page flags in /proc/kpageflags (take 3)

[Wu Fengguang <fengguang.wu@intel.com>]

On Thu, Apr 23, 2009 at 04:54:06PM +0800, Andi Kleen wrote:
> > Good idea.
> > 
> > Would you recommend a good way to identify huge pages?
> > Test by page order, or by (dtor == free_huge_page)?      
> 
> That doesn't work for GB pages. The best way would be to make
> slub stop using it and then check for compound, but I don't know what 
> this implies. Otherwise would need some other way, perhaps a new
> flag?

Or play the following trick? :-)

This helps hide the internal compound page consumers(SLAB/SLUB/... and
loads of device drivers) to user space. However there are still huge
pages of different orders(IA64?). Does this make a good reason for
exporting both HEAD/TAIL flags, instead of a combined COMPOUND flag?

Thanks,
Fengguang
---

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index c722aa6..7d0bd0d 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -298,6 +298,14 @@ void prep_compound_page(struct page *page, unsigned long order)
 	}
 }
 
+/*
+ * This function helps distinguish gigantic pages from normal compound pages.
+ */
+static void free_gigantic_page(struct page *page)
+{
+	__free_pages_ok(page, compound_order(page));
+}
+
 #ifdef CONFIG_HUGETLBFS
 void prep_compound_gigantic_page(struct page *page, unsigned long order)
 {
@@ -305,7 +313,7 @@ void prep_compound_gigantic_page(struct page *page, unsigned long order)
 	int nr_pages = 1 << order;
 	struct page *p = page + 1;
 
-	set_compound_page_dtor(page, free_compound_page);
+	set_compound_page_dtor(page, free_gigantic_page);
 	set_compound_order(page, order);
 	__SetPageHead(page);
 	for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) {