Merge kroah.com:/home/greg/linux/BK/bleeding_edge-2.5

into kroah.com:/home/greg/linux/BK/gregkh-2.5

12 files changed, 130 insertions, 104 deletions

diff --git a/include/linux/cache.h b/include/linux/cache.h
index 37d7a6fe4b6e..95faa6ec3aa7 100644
--- a/include/linux/cache.h
+++ b/include/linux/cache.h
@@ -44,4 +44,13 @@
 #endif /* CONFIG_SMP */
 #endif
 
+#if !defined(____cacheline_maxaligned_in_smp)
+#if defined(CONFIG_SMP)
+#define ____cacheline_maxaligned_in_smp \
+	__attribute__((__aligned__(1 << (L1_CACHE_SHIFT_MAX))))
+#else
+#define ____cacheline_maxaligned_in_smp
+#endif
+#endif
+
 #endif /* __LINUX_CACHE_H */
diff --git a/include/linux/dcache.h b/include/linux/dcache.h
index 897d414ac466..f99a03f17e60 100644
--- a/include/linux/dcache.h
+++ b/include/linux/dcache.h
@@ -181,8 +181,6 @@ extern void shrink_dcache_parent(struct dentry *);
 extern void shrink_dcache_anon(struct list_head *);
 extern int d_invalidate(struct dentry *);
 
-#define shrink_dcache() prune_dcache(0)
-struct zone_struct;
 /* dcache memory management */
 extern int shrink_dcache_memory(int, unsigned int);
 extern void prune_dcache(int);
diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index 317d45415250..10021357c093 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -40,7 +40,7 @@
  * virtual kernel addresses to the allocated page(s).
  */
 extern struct page * FASTCALL(_alloc_pages(unsigned int gfp_mask, unsigned int order));
-extern struct page * FASTCALL(__alloc_pages(unsigned int gfp_mask, unsigned int order, zonelist_t *zonelist));
+extern struct page * FASTCALL(__alloc_pages(unsigned int gfp_mask, unsigned int order, struct zonelist *zonelist));
 extern struct page * alloc_pages_node(int nid, unsigned int gfp_mask, unsigned int order);
 
 static inline struct page * alloc_pages(unsigned int gfp_mask, unsigned int order)
diff --git a/include/linux/ide.h b/include/linux/ide.h
index 62fd4b0fae50..aacdaad8f594 100644
--- a/include/linux/ide.h
+++ b/include/linux/ide.h
@@ -470,7 +470,6 @@ typedef struct ide_drive_s {
 	unsigned no_unmask	: 1;	/* disallow setting unmask bit */
 	unsigned no_io_32bit	: 1;	/* disallow enabling 32bit I/O */
 	unsigned nobios		: 1;	/* flag: do not probe bios for drive */
-	unsigned revalidate	: 1;	/* request revalidation */
 	unsigned atapi_overlap	: 1;	/* flag: ATAPI overlap (not supported) */
 	unsigned nice0		: 1;	/* flag: give obvious excess bandwidth */
 	unsigned nice2		: 1;	/* flag: give a share in our own bandwidth */
@@ -530,6 +529,7 @@ typedef struct ide_drive_s {
 	byte		acoustic;	/* acoustic management */
 	unsigned int	failures;	/* current failure count */
 	unsigned int	max_failures;	/* maximum allowed failure count */
+	struct list_head list;
 } ide_drive_t;
 
 /*
@@ -889,6 +889,7 @@ read_proc_t proc_ide_read_geometry;
 #define IDE_SUBDRIVER_VERSION	1
 
 typedef struct ide_driver_s {
+	struct module			*owner;
 	const char			*name;
 	const char			*version;
 	byte				media;
@@ -913,10 +914,11 @@ typedef struct ide_driver_s {
 	unsigned long	(*capacity)(ide_drive_t *);
 	ide_startstop_t	(*special)(ide_drive_t *);
 	ide_proc_entry_t        *proc;
-	int		(*init)(void);
 	int		(*reinit)(ide_drive_t *);
 	void		(*ata_prebuilder)(ide_drive_t *);
 	void		(*atapi_prebuilder)(ide_drive_t *);
+	struct list_head drives;
+	struct list_head drivers;
 } ide_driver_t;
 
 #define DRIVER(drive)		((drive)->driver)
@@ -926,7 +928,6 @@ typedef struct ide_driver_s {
  */
 #define IDE_CHIPSET_MODULE		0	/* not supported yet */
 #define IDE_PROBE_MODULE		1
-#define IDE_DRIVER_MODULE		2
 
 typedef int	(ide_module_init_proc)(void);
 
@@ -947,7 +948,6 @@ typedef struct ide_module_s {
  */
 #ifndef _IDE_C
 extern	ide_hwif_t	ide_hwifs[];		/* master data repository */
-extern	ide_module_t	*ide_modules;
 extern	ide_module_t	*ide_probe;
 #endif
 extern int noautodma;
@@ -1210,37 +1210,16 @@ extern struct block_device_operations ide_fops[];
 extern ide_proc_entry_t generic_subdriver_entries[];
 #endif
 
-int ide_reinit_drive (ide_drive_t *drive);
+int ata_attach(ide_drive_t *drive);
 
 #ifdef _IDE_C
 #ifdef CONFIG_BLK_DEV_IDE
 int ideprobe_init (void);
 #endif /* CONFIG_BLK_DEV_IDE */
-#ifdef CONFIG_BLK_DEV_IDEDISK
-int idedisk_reinit (ide_drive_t *drive);
-int idedisk_init (void);
-#endif /* CONFIG_BLK_DEV_IDEDISK */
-#ifdef CONFIG_BLK_DEV_IDECD
-int ide_cdrom_reinit (ide_drive_t *drive);
-int ide_cdrom_init (void);
-#endif /* CONFIG_BLK_DEV_IDECD */
-#ifdef CONFIG_BLK_DEV_IDETAPE
-int idetape_reinit (ide_drive_t *drive);
-int idetape_init (void);
-#endif /* CONFIG_BLK_DEV_IDETAPE */
-#ifdef CONFIG_BLK_DEV_IDEFLOPPY
-int idefloppy_reinit (ide_drive_t *drive);
-int idefloppy_init (void);
-#endif /* CONFIG_BLK_DEV_IDEFLOPPY */
-#ifdef CONFIG_BLK_DEV_IDESCSI
-int idescsi_reinit (ide_drive_t *drive);
-int idescsi_init (void);
-#endif /* CONFIG_BLK_DEV_IDESCSI */
 #endif /* _IDE_C */
 
-int ide_register_module (ide_module_t *module);
-void ide_unregister_module (ide_module_t *module);
-ide_drive_t *ide_scan_devices (byte media, const char *name, ide_driver_t *driver, int n);
+int ide_register_driver(ide_driver_t *driver);
+void ide_unregister_driver(ide_driver_t *driver);
 int ide_register_subdriver (ide_drive_t *drive, ide_driver_t *driver, int version);
 int ide_unregister_subdriver (ide_drive_t *drive);
 int ide_replace_subdriver(ide_drive_t *drive, const char *driver);
diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
index 09e6b86ee763..26588de0b514 100644
--- a/include/linux/interrupt.h
+++ b/include/linux/interrupt.h
@@ -50,9 +50,9 @@ enum {
 #if !CONFIG_SMP
 # define cli()			local_irq_disable()
 # define sti()			local_irq_enable()
-# define save_flags(x)		local_irq_save(x)
+# define save_flags(x)		local_save_flags(x)
 # define restore_flags(x)	local_irq_restore(x)
-# define save_and_cli(x)	local_irq_save_off(x)
+# define save_and_cli(x)	local_irq_save(x)
 #endif
 
 
diff --git a/include/linux/kernel.h b/include/linux/kernel.h
index 022587afdc46..95965970e49b 100644
--- a/include/linux/kernel.h
+++ b/include/linux/kernel.h
@@ -191,4 +191,10 @@ struct sysinfo {
 };
 
 #define BUG_ON(condition) do { if (unlikely((condition)!=0)) BUG(); } while(0)
+
+/* Trap pasters of __FUNCTION__ at compile-time */
+#if __GNUC__ > 2 || __GNUC_MINOR__ >= 95
+#define __FUNCTION__ (__func__)
+#endif
+
 #endif
diff --git a/include/linux/list.h b/include/linux/list.h
index 3bc81f885ce1..913c93c44286 100644
--- a/include/linux/list.h
+++ b/include/linux/list.h
@@ -211,6 +211,19 @@ static inline void list_splice_init(list_t *list, list_t *head)
 	for (pos = (head)->next, n = pos->next; pos != (head); \
 		pos = n, n = pos->next)
 
+/**
+ * list_for_each_entry	-	iterate over list of given type
+ * @pos:	the type * to use as a loop counter.
+ * @head:	the head for your list.
+ * @member:	the name of the list_struct within the struct.
+ */
+#define list_for_each_entry(pos, head, member)				\
+	for (pos = list_entry((head)->next, typeof(*pos), member),	\
+		     prefetch(pos->member.next);			\
+	     &pos->member != (head); 					\
+	     pos = list_entry(pos->member.next, typeof(*pos), member),	\
+		     prefetch(pos->member.next))
+
 #endif /* __KERNEL__ || _LVM_H_INCLUDE */
 
 #endif
diff --git a/include/linux/mm.h b/include/linux/mm.h
index d731c02741b9..045a861e4024 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -157,7 +157,7 @@ struct page {
 	struct address_space *mapping;	/* The inode (or ...) we belong to. */
 	unsigned long index;		/* Our offset within mapping. */
 	struct list_head lru;		/* Pageout list, eg. active_list;
-					   protected by pagemap_lru_lock !! */
+					   protected by zone->lru_lock !! */
 	union {
 		struct pte_chain * chain;	/* Reverse pte mapping pointer.
 					 * protected by PG_chainlock */
@@ -268,10 +268,10 @@ void FASTCALL(__free_pages_ok(struct page *page, unsigned int order));
 #define NODE_SHIFT 4
 #define ZONE_SHIFT (BITS_PER_LONG - 8)
 
-struct zone_struct;
-extern struct zone_struct *zone_table[];
+struct zone;
+extern struct zone *zone_table[];
 
-static inline zone_t *page_zone(struct page *page)
+static inline struct zone *page_zone(struct page *page)
 {
 	return zone_table[page->flags >> ZONE_SHIFT];
 }
@@ -449,7 +449,6 @@ static inline int can_vma_merge(struct vm_area_struct * vma, unsigned long vm_fl
 		return 0;
 }
 
-struct zone_t;
 /* filemap.c */
 extern unsigned long page_unuse(struct page *);
 extern void truncate_inode_pages(struct address_space *, loff_t);
diff --git a/include/linux/mm_inline.h b/include/linux/mm_inline.h
new file mode 100644
index 000000000000..47762ca695a5
--- /dev/null
+++ b/include/linux/mm_inline.h
@@ -0,0 +1,40 @@
+
+static inline void
+add_page_to_active_list(struct zone *zone, struct page *page)
+{
+	list_add(&page->lru, &zone->active_list);
+	zone->nr_active++;
+}
+
+static inline void
+add_page_to_inactive_list(struct zone *zone, struct page *page)
+{
+	list_add(&page->lru, &zone->inactive_list);
+	zone->nr_inactive++;
+}
+
+static inline void
+del_page_from_active_list(struct zone *zone, struct page *page)
+{
+	list_del(&page->lru);
+	zone->nr_active--;
+}
+
+static inline void
+del_page_from_inactive_list(struct zone *zone, struct page *page)
+{
+	list_del(&page->lru);
+	zone->nr_inactive--;
+}
+
+static inline void
+del_page_from_lru(struct zone *zone, struct page *page)
+{
+	list_del(&page->lru);
+	if (PageActive(page)) {
+		ClearPageActive(page);
+		zone->nr_active--;
+	} else {
+		zone->nr_inactive--;
+	}
+}
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index df4523bc5341..f30bd4f44863 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -8,6 +8,8 @@
 #include <linux/spinlock.h>
 #include <linux/list.h>
 #include <linux/wait.h>
+#include <linux/cache.h>
+#include <asm/atomic.h>
 
 /*
  * Free memory management - zoned buddy allocator.
@@ -27,6 +29,21 @@ typedef struct free_area_struct {
 struct pglist_data;
 
 /*
+ * zone->lock and zone->lru_lock are two of the hottest locks in the kernel.
+ * So add a wild amount of padding here to ensure that they fall into separate
+ * cachelines.  There are very few zone structures in the machine, so space
+ * consumption is not a concern here.
+ */
+#if defined(CONFIG_SMP)
+struct zone_padding {
+	int x;
+} ____cacheline_maxaligned_in_smp;
+#define ZONE_PADDING(name)	struct zone_padding name;
+#else
+#define ZONE_PADDING(name)
+#endif
+
+/*
  * On machines where it is needed (eg PCs) we divide physical memory
  * into multiple physical zones. On a PC we have 3 zones:
  *
@@ -34,7 +51,8 @@ struct pglist_data;
  * ZONE_NORMAL	16-896 MB	direct mapped by the kernel
  * ZONE_HIGHMEM	 > 896 MB	only page cache and user processes
  */
-typedef struct zone_struct {
+
+struct zone {
 	/*
 	 * Commonly accessed fields:
 	 */
@@ -43,6 +61,17 @@ typedef struct zone_struct {
 	unsigned long		pages_min, pages_low, pages_high;
 	int			need_balance;
 
+	ZONE_PADDING(_pad1_)
+
+	spinlock_t		lru_lock;	
+	struct list_head	active_list;
+	struct list_head	inactive_list;
+	atomic_t		refill_counter;
+	unsigned long		nr_active;
+	unsigned long		nr_inactive;
+
+	ZONE_PADDING(_pad2_)
+
 	/*
 	 * free areas of different sizes
 	 */
@@ -89,7 +118,7 @@ typedef struct zone_struct {
 	 */
 	char			*name;
 	unsigned long		size;
-} zone_t;
+} ____cacheline_maxaligned_in_smp;
 
 #define ZONE_DMA		0
 #define ZONE_NORMAL		1
@@ -107,16 +136,16 @@ typedef struct zone_struct {
  * so despite the zonelist table being relatively big, the cache
  * footprint of this construct is very small.
  */
-typedef struct zonelist_struct {
-	zone_t * zones [MAX_NR_ZONES+1]; // NULL delimited
-} zonelist_t;
+struct zonelist {
+	struct zone *zones[MAX_NR_ZONES+1]; // NULL delimited
+};
 
 #define GFP_ZONEMASK	0x0f
 
 /*
  * The pg_data_t structure is used in machines with CONFIG_DISCONTIGMEM
  * (mostly NUMA machines?) to denote a higher-level memory zone than the
- * zone_struct denotes.
+ * zone denotes.
  *
  * On NUMA machines, each NUMA node would have a pg_data_t to describe
  * it's memory layout.
@@ -126,8 +155,8 @@ typedef struct zonelist_struct {
  */
 struct bootmem_data;
 typedef struct pglist_data {
-	zone_t node_zones[MAX_NR_ZONES];
-	zonelist_t node_zonelists[GFP_ZONEMASK+1];
+	struct zone node_zones[MAX_NR_ZONES];
+	struct zonelist node_zonelists[GFP_ZONEMASK+1];
 	int nr_zones;
 	struct page *node_mem_map;
 	unsigned long *valid_addr_bitmap;
@@ -142,7 +171,8 @@ typedef struct pglist_data {
 extern int numnodes;
 extern pg_data_t *pgdat_list;
 
-static inline int memclass(zone_t *pgzone, zone_t *classzone)
+static inline int
+memclass(struct zone *pgzone, struct zone *classzone)
 {
 	if (pgzone->zone_pgdat != classzone->zone_pgdat)
 		return 0;
@@ -156,10 +186,10 @@ static inline int memclass(zone_t *pgzone, zone_t *classzone)
  * prototypes for the discontig memory code.
  */
 struct page;
-extern void show_free_areas_core(pg_data_t *pgdat);
-extern void free_area_init_core(int nid, pg_data_t *pgdat, struct page **gmap,
+void free_area_init_core(int nid, pg_data_t *pgdat, struct page **gmap,
   unsigned long *zones_size, unsigned long paddr, unsigned long *zholes_size,
   struct page *pmap);
+void get_zone_counts(unsigned long *active, unsigned long *inactive);
 
 extern pg_data_t contig_page_data;
 
@@ -181,7 +211,7 @@ extern pg_data_t contig_page_data;
  * next_zone - helper magic for for_each_zone()
  * Thanks to William Lee Irwin III for this piece of ingenuity.
  */
-static inline zone_t * next_zone(zone_t * zone)
+static inline struct zone *next_zone(struct zone *zone)
 {
 	pg_data_t *pgdat = zone->zone_pgdat;
 
@@ -198,7 +228,7 @@ static inline zone_t * next_zone(zone_t * zone)
 
 /**
  * for_each_zone - helper macro to iterate over all memory zones
- * @zone - pointer to zone_t variable
+ * @zone - pointer to struct zone variable
  *
  * The user only needs to declare the zone variable, for_each_zone
  * fills it in. This basically means for_each_zone() is an
@@ -206,7 +236,7 @@ static inline zone_t * next_zone(zone_t * zone)
  *
  * for (pgdat = pgdat_list; pgdat; pgdat = pgdat->node_next)
  * 	for (i = 0; i < MAX_NR_ZONES; ++i) {
- * 		zone_t * z = pgdat->node_zones + i;
+ * 		struct zone * z = pgdat->node_zones + i;
  * 		...
  * 	}
  * }
diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index 9801b15876d9..5a49020e728b 100644
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -28,7 +28,7 @@
  *
  * Note that the referenced bit, the page->lru list_head and the active,
  * inactive_dirty and inactive_clean lists are protected by the
- * pagemap_lru_lock, and *NOT* by the usual PG_locked bit!
+ * zone->lru_lock, and *NOT* by the usual PG_locked bit!
  *
  * PG_error is set to indicate that an I/O error occurred on this page.
  *
@@ -76,8 +76,6 @@ extern struct page_state {
 	unsigned long nr_dirty;
 	unsigned long nr_writeback;
 	unsigned long nr_pagecache;
-	unsigned long nr_active;	/* on active_list LRU */
-	unsigned long nr_inactive;	/* on inactive_list LRU */
 	unsigned long nr_page_table_pages;
 	unsigned long nr_reverse_maps;
 } ____cacheline_aligned_in_smp page_states[NR_CPUS];
@@ -163,6 +161,7 @@ extern void get_page_state(struct page_state *ret);
 #define SetPageActive(page)	set_bit(PG_active, &(page)->flags)
 #define ClearPageActive(page)	clear_bit(PG_active, &(page)->flags)
 #define TestClearPageActive(page) test_and_clear_bit(PG_active, &(page)->flags)
+#define TestSetPageActive(page) test_and_set_bit(PG_active, &(page)->flags)
 
 #define PageSlab(page)		test_bit(PG_slab, &(page)->flags)
 #define SetPageSlab(page)	set_bit(PG_slab, &(page)->flags)
diff --git a/include/linux/swap.h b/include/linux/swap.h
index a7f1f96ff9f1..affa89d77eb1 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -139,7 +139,7 @@ struct task_struct;
 struct vm_area_struct;
 struct sysinfo;
 struct address_space;
-struct zone_t;
+struct zone;
 
 /* linux/mm/rmap.c */
 extern int FASTCALL(page_referenced(struct page *));
@@ -163,7 +163,7 @@ extern void swap_setup(void);
 
 /* linux/mm/vmscan.c */
 extern wait_queue_head_t kswapd_wait;
-extern int FASTCALL(try_to_free_pages(zone_t *, unsigned int, unsigned int));
+extern int try_to_free_pages(struct zone *, unsigned int, unsigned int);
 
 /* linux/mm/page_io.c */
 int swap_readpage(struct file *file, struct page *page);
@@ -209,55 +209,8 @@ extern struct swap_list_t swap_list;
 asmlinkage long sys_swapoff(const char *);
 asmlinkage long sys_swapon(const char *, int);
 
-extern spinlock_t _pagemap_lru_lock;
-
 extern void FASTCALL(mark_page_accessed(struct page *));
 
-/*
- * List add/del helper macros. These must be called
- * with the pagemap_lru_lock held!
- */
-#define DEBUG_LRU_PAGE(page)			\
-do {						\
-	if (!PageLRU(page))			\
-		BUG();				\
-	if (PageActive(page))			\
-		BUG();				\
-} while (0)
-
-#define __add_page_to_active_list(page)		\
-do {						\
-	list_add(&(page)->lru, &active_list);	\
-	inc_page_state(nr_active);		\
-} while (0)
-
-#define add_page_to_active_list(page)		\
-do {						\
-	DEBUG_LRU_PAGE(page);			\
-	SetPageActive(page);			\
-	__add_page_to_active_list(page);	\
-} while (0)
-
-#define add_page_to_inactive_list(page)		\
-do {						\
-	DEBUG_LRU_PAGE(page);			\
-	list_add(&(page)->lru, &inactive_list);	\
-	inc_page_state(nr_inactive);		\
-} while (0)
-
-#define del_page_from_active_list(page)		\
-do {						\
-	list_del(&(page)->lru);			\
-	ClearPageActive(page);			\
-	dec_page_state(nr_active);		\
-} while (0)
-
-#define del_page_from_inactive_list(page)	\
-do {						\
-	list_del(&(page)->lru);			\
-	dec_page_state(nr_inactive);		\
-} while (0)
-
 extern spinlock_t swaplock;
 
 #define swap_list_lock()	spin_lock(&swaplock)