Add a dirty map to address spaces

In a NUMA system it is helpful to know where the dirty pages of a mapping
are located. That way we will be able to implement writeout for applications
that are constrained to a portion of the memory of the system as required by
cpusets.

Two function are introduces to manage the dirty node map:
cpuset_clear_dirty_nodes() and cpuset_update_nodes(). Both are
defined using macros since the definition of an address space may
not be available in cpuset.h.

Signed-off-by; Christoph Lameter <clameter@sgi.com>

Index: linux-2.6.20-rc3/fs/fs-writeback.c
===================================================================
--- linux-2.6.20-rc3.orig/fs/fs-writeback.c	2007-01-05 19:42:08.292668587 -0600
+++ linux-2.6.20-rc3/fs/fs-writeback.c	2007-01-05 20:07:30.485448546 -0600
@@ -22,6 +22,7 @@
 #include <linux/blkdev.h>
 #include <linux/backing-dev.h>
 #include <linux/buffer_head.h>
+#include <linux/cpuset.h>
 #include "internal.h"
 
 /**
@@ -223,11 +224,13 @@ __sync_single_inode(struct inode *inode,
 			/*
 			 * The inode is clean, inuse
 			 */
+			cpuset_clear_dirty_nodes(inode->i_mapping);
 			list_move(&inode->i_list, &inode_in_use);
 		} else {
 			/*
 			 * The inode is clean, unused
 			 */
+			cpuset_clear_dirty_nodes(inode->i_mapping);
 			list_move(&inode->i_list, &inode_unused);
 		}
 	}
Index: linux-2.6.20-rc3/fs/inode.c
===================================================================
--- linux-2.6.20-rc3.orig/fs/inode.c	2007-01-05 19:42:08.301458414 -0600
+++ linux-2.6.20-rc3/fs/inode.c	2007-01-05 19:42:10.301632448 -0600
@@ -22,6 +22,7 @@
 #include <linux/bootmem.h>
 #include <linux/inotify.h>
 #include <linux/mount.h>
+#include <linux/cpuset.h>
 
 /*
  * This is needed for the following functions:
@@ -147,6 +148,7 @@ static struct inode *alloc_inode(struct 
 		mapping->flags = 0;
 		mapping_set_gfp_mask(mapping, GFP_HIGHUSER);
 		mapping->assoc_mapping = NULL;
+		cpuset_clear_dirty_nodes(mapping);
 		mapping->backing_dev_info = &default_backing_dev_info;
 
 		/*
Index: linux-2.6.20-rc3/include/linux/fs.h
===================================================================
--- linux-2.6.20-rc3.orig/include/linux/fs.h	2007-01-05 19:42:08.320991364 -0600
+++ linux-2.6.20-rc3/include/linux/fs.h	2007-01-05 19:42:10.326048635 -0600
@@ -446,6 +446,9 @@ struct address_space {
 	spinlock_t		private_lock;	/* for use by the address_space */
 	struct list_head	private_list;	/* ditto */
 	struct address_space	*assoc_mapping;	/* ditto */
+#ifdef CONFIG_CPUSETS
+	nodemask_t		dirty_nodes;	/* Map of nodes with dirty pages */
+#endif
 } __attribute__((aligned(sizeof(long))));
 	/*
 	 * On most architectures that alignment is already the case; but
Index: linux-2.6.20-rc3/mm/page-writeback.c
===================================================================
--- linux-2.6.20-rc3.orig/mm/page-writeback.c	2007-01-05 19:42:08.344430903 -0600
+++ linux-2.6.20-rc3/mm/page-writeback.c	2007-01-05 20:07:30.421966402 -0600
@@ -33,6 +33,7 @@
 #include <linux/syscalls.h>
 #include <linux/buffer_head.h>
 #include <linux/pagevec.h>
+#include <linux/cpuset.h>
 
 /*
  * The maximum number of pages to writeout in a single bdflush/kupdate
@@ -776,6 +777,7 @@ int __set_page_dirty_nobuffers(struct pa
 			radix_tree_tag_set(&mapping->page_tree,
 				page_index(page), PAGECACHE_TAG_DIRTY);
 		}
+		cpuset_update_dirty_nodes(mapping, page);
 		write_unlock_irq(&mapping->tree_lock);
 		if (mapping->host) {
 			/* !PageAnon && !swapper_space */
@@ -813,6 +815,7 @@ int fastcall set_page_dirty(struct page 
 		if (!spd)
 			spd = __set_page_dirty_buffers;
 #endif
+		cpuset_update_dirty_nodes(mapping, page);
 		return (*spd)(page);
 	}
 	if (!PageDirty(page)) {
Index: linux-2.6.20-rc3/fs/buffer.c
===================================================================
--- linux-2.6.20-rc3.orig/fs/buffer.c	2007-01-05 19:42:08.311224889 -0600
+++ linux-2.6.20-rc3/fs/buffer.c	2007-01-05 19:47:47.827132793 -0600
@@ -42,6 +42,7 @@
 #include <linux/bitops.h>
 #include <linux/mpage.h>
 #include <linux/bit_spinlock.h>
+#include <linux/cpuset.h>
 
 static int fsync_buffers_list(spinlock_t *lock, struct list_head *list);
 static void invalidate_bh_lrus(void);
@@ -736,6 +737,7 @@ int __set_page_dirty_buffers(struct page
 		}
 		radix_tree_tag_set(&mapping->page_tree,
 				page_index(page), PAGECACHE_TAG_DIRTY);
+		cpuset_update_dirty_nodes(mapping, page);
 	}
 	write_unlock_irq(&mapping->tree_lock);
 	__mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
Index: linux-2.6.20-rc3/include/linux/cpuset.h
===================================================================
--- linux-2.6.20-rc3.orig/include/linux/cpuset.h	2007-01-05 19:42:08.331734486 -0600
+++ linux-2.6.20-rc3/include/linux/cpuset.h	2007-01-05 20:08:11.444127529 -0600
@@ -75,6 +75,15 @@ static inline int cpuset_do_slab_mem_spr
 
 extern void cpuset_track_online_nodes(void);
 
+/*
+ * We need macros since struct address_space is not defined yet
+ */
+#define cpuset_update_dirty_nodes(__mapping, __page) \
+	node_set(page_to_nid(__page), (__mapping)->dirty_nodes)
+
+#define cpuset_clear_dirty_nodes(__mapping) \
+		(__mapping)->dirty_nodes = NODE_MASK_NONE
+
 #else /* !CONFIG_CPUSETS */
 
 static inline int cpuset_init_early(void) { return 0; }
@@ -146,6 +155,11 @@ static inline int cpuset_do_slab_mem_spr
 
 static inline void cpuset_track_online_nodes(void) {}
 
+static inline void cpuset_update_dirty_nodes(struct address_space *mapping,
+		struct page *page) {}
+
+static inline void cpuset_clear_dirty_nodes(struct address_space *mapping) {}
+
 #endif /* !CONFIG_CPUSETS */
 
 #endif /* _LINUX_CPUSET_H */