Skip to content

Commit

Permalink
mm: setup pageblock_order before it's used by sparsemem
Browse files Browse the repository at this point in the history
On architectures with CONFIG_HUGETLB_PAGE_SIZE_VARIABLE set, such as
Itanium, pageblock_order is a variable with default value of 0.  It's set
to the right value by set_pageblock_order() in function
free_area_init_core().

But pageblock_order may be used by sparse_init() before free_area_init_core()
is called along path:
sparse_init()
    ->sparse_early_usemaps_alloc_node()
	->usemap_size()
	    ->SECTION_BLOCKFLAGS_BITS
		->((1UL << (PFN_SECTION_SHIFT - pageblock_order)) *
NR_PAGEBLOCK_BITS)

The uninitialized pageblock_size will cause memory wasting because
usemap_size() returns a much bigger value then it's really needed.

For example, on an Itanium platform,
sparse_init() pageblock_order=0 usemap_size=24576
free_area_init_core() before pageblock_order=0, usemap_size=24576
free_area_init_core() after pageblock_order=12, usemap_size=8

That means 24K memory has been wasted for each section, so fix it by calling
set_pageblock_order() from sparse_init().

Signed-off-by: Xishi Qiu <qiuxishi@huawei.com>
Signed-off-by: Jiang Liu <liuj97@gmail.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Keping Chen <chenkeping@huawei.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
  • Loading branch information
Xishi Qiu authored and Linus Torvalds committed Aug 1, 2012
1 parent 51a07e5 commit ca57df7
Show file tree
Hide file tree
Showing 3 changed files with 7 additions and 2 deletions.
2 changes: 2 additions & 0 deletions mm/internal.h
Original file line number Diff line number Diff line change
Expand Up @@ -353,3 +353,5 @@ extern u32 hwpoison_filter_enable;
extern unsigned long vm_mmap_pgoff(struct file *, unsigned long,
unsigned long, unsigned long,
unsigned long, unsigned long);

extern void set_pageblock_order(void);
4 changes: 2 additions & 2 deletions mm/page_alloc.c
Original file line number Diff line number Diff line change
Expand Up @@ -4304,7 +4304,7 @@ static inline void setup_usemap(struct pglist_data *pgdat,
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE

/* Initialise the number of pages represented by NR_PAGEBLOCK_BITS */
static inline void __init set_pageblock_order(void)
void __init set_pageblock_order(void)
{
unsigned int order;

Expand Down Expand Up @@ -4332,7 +4332,7 @@ static inline void __init set_pageblock_order(void)
* include/linux/pageblock-flags.h for the values of pageblock_order based on
* the kernel config
*/
static inline void set_pageblock_order(void)
void __init set_pageblock_order(void)
{
}

Expand Down
3 changes: 3 additions & 0 deletions mm/sparse.c
Original file line number Diff line number Diff line change
Expand Up @@ -493,6 +493,9 @@ void __init sparse_init(void)
struct page **map_map;
#endif

/* Setup pageblock_order for HUGETLB_PAGE_SIZE_VARIABLE */
set_pageblock_order();

/*
* map is using big page (aka 2M in x86 64 bit)
* usemap is less one page (aka 24 bytes)
Expand Down

0 comments on commit ca57df7

Please sign in to comment.