[PATCH] Save some memory in mem_map on x86-64

From: Andi Kleen <ak@suse.de>

This patch saves 2MB of memory on a 1GB x86-64 machine, 20MB on a 10GB
machine.  It does this by eliminating 8 bytes of useless padding in struct
page.

This resurrects an older patch in a hopefully cleaner form.

include/asm-x86_64/bitops.h

@@ -503,6 +503,8 @@ static __inline__ int ffs(int x)
 /* find last set bit */
 #define fls(x) generic_fls(x)
 
+#define ARCH_HAS_ATOMIC_UNSIGNED 1
+
 #endif /* __KERNEL__ */
 
 #endif /* _X86_64_BITOPS_H */

include/linux/mm.h

@@ -152,6 +152,12 @@ struct pte_chain;
 struct mmu_gather;
 struct inode;
 
+#ifdef ARCH_HAS_ATOMIC_UNSIGNED
+typedef unsigned page_flags_t;
+#else
+typedef unsigned long page_flags_t;
+#endif
+
 /*
  * Each physical page in the system has a struct page associated with
  * it to keep track of whatever it is we are using the page for at the
@@ -168,7 +174,7 @@ struct inode;
  * TODO: make this structure smaller, it could be as small as 32 bytes.
  */
 struct page {
-	unsigned long flags;		/* atomic flags, some possibly
+	page_flags_t flags;		/* atomic flags, some possibly
 					   updated asynchronously */
 	atomic_t count;			/* Usage count, see below. */
 	struct list_head list;		/* ->mapping has some page lists. */
@@ -333,7 +339,7 @@ static inline void put_page(struct page *page)
  * We'll have up to (MAX_NUMNODES * MAX_NR_ZONES) zones total,
  * so we use (MAX_NODES_SHIFT + MAX_ZONES_SHIFT) here to get enough bits.
  */
-#define NODEZONE_SHIFT (BITS_PER_LONG - MAX_NODES_SHIFT - MAX_ZONES_SHIFT)
+#define NODEZONE_SHIFT (sizeof(page_flags_t)*8 - MAX_NODES_SHIFT - MAX_ZONES_SHIFT)
 #define NODEZONE(node, zone)	((node << ZONES_SHIFT) | zone)
 
 static inline unsigned long page_zonenum(struct page *page)

include/linux/mmzone.h

@@ -316,7 +316,7 @@ extern struct pglist_data contig_page_data;
 
 #include <asm/mmzone.h>
 
-#if BITS_PER_LONG == 32
+#if BITS_PER_LONG == 32 || defined(ARCH_HAS_ATOMIC_UNSIGNED)
 /*
  * with 32 bit page->flags field, we reserve 8 bits for node/zone info.
  * there are 3 zones (2 bits) and this leaves 8-2=6 bits for nodes.

include/linux/rmap-locking.h

@@ -10,8 +10,8 @@
 struct pte_chain;
 extern kmem_cache_t *pte_chain_cache;
 
-#define pte_chain_lock(page)	bit_spin_lock(PG_chainlock, &page->flags)
-#define pte_chain_unlock(page)	bit_spin_unlock(PG_chainlock, &page->flags)
+#define pte_chain_lock(page)	bit_spin_lock(PG_chainlock, (unsigned long *)&page->flags)
+#define pte_chain_unlock(page)	bit_spin_unlock(PG_chainlock, (unsigned long *)&page->flags)
 
 struct pte_chain *pte_chain_alloc(int gfp_flags);
 void __pte_chain_free(struct pte_chain *pte_chain);