[PATCH] pdflush exclusion infrastructure

Collision avoidance for pdflush threads.

Turns the request_queue-based `unsigned long ra_pages' into a structure
which contains ra_pages as well as a longword.

That longword is used to record the fact that a pdflush thread is
currently writing something back against this request_queue.

Avoids the situation where several pdflush threads are sleeping on the
same request_queue.

This patch provides only the infrastructure for the pdflush exclusion.
This infrastructure gets used in pdflush-single.patch

4 files changed, 35 insertions, 8 deletions

diff --git a/include/linux/backing-dev.h b/include/linux/backing-dev.h
new file mode 100644
index 000000000000..075cacc389e1
--- /dev/null
+++ b/include/linux/backing-dev.h
@@ -0,0 +1,30 @@
+/*
+ * include/linux/backing-dev.h
+ *
+ * low-level device information and state which is propagated up through
+ * to high-level code.
+ */
+
+#ifndef _LINUX_BACKING_DEV_H
+#define _LINUX_BACKING_DEV_H
+
+/*
+ * Bits in backing_dev_info.state
+ */
+enum bdi_state {
+	BDI_pdflush,		/* A pdflush thread is working this device */
+	BDI_unused,		/* Available bits start here */
+};
+
+struct backing_dev_info {
+	unsigned long ra_pages;	/* max readahead in PAGE_CACHE_SIZE units */
+	unsigned long state;	/* Always use atomic bitops on this */
+};
+
+extern struct backing_dev_info default_backing_dev_info;
+
+int writeback_acquire(struct backing_dev_info *bdi);
+int writeback_in_progress(struct backing_dev_info *bdi);
+void writeback_release(struct backing_dev_info *bdi);
+
+#endif		/* _LINUX_BACKING_DEV_H */
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index d8175ccc104c..ac373e6a2454 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -7,6 +7,7 @@
 #include <linux/tqueue.h>
 #include <linux/list.h>
 #include <linux/pagemap.h>
+#include <linux/backing-dev.h>
 
 #include <asm/scatterlist.h>
 
@@ -162,11 +163,7 @@ struct request_queue
 	make_request_fn		*make_request_fn;
 	prep_rq_fn		*prep_rq_fn;
 
-	/*
-	 * The VM-level readahead tunable for this device.  In
-	 * units of PAGE_CACHE_SIZE pages.
-	 */
-	unsigned long ra_pages;
+	struct backing_dev_info	backing_dev_info;
 
 	/*
 	 * The queue owner gets to use this for whatever they like.
@@ -328,7 +325,7 @@ extern void blk_queue_hardsect_size(request_queue_t *q, unsigned short);
 extern void blk_queue_segment_boundary(request_queue_t *q, unsigned long);
 extern void blk_queue_assign_lock(request_queue_t *q, spinlock_t *);
 extern void blk_queue_prep_rq(request_queue_t *q, prep_rq_fn *pfn);
-extern unsigned long *blk_get_ra_pages(struct block_device *bdev);
+extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
 
 extern int blk_rq_map_sg(request_queue_t *, struct request *, struct scatterlist *);
 extern void blk_dump_rq_flags(struct request *, char *);
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 25578c7a5e62..374045884cb8 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -305,6 +305,7 @@ struct address_space_operations {
 	int (*direct_IO)(int, struct inode *, struct kiobuf *, unsigned long, int);
 };
 
+struct backing_dev_info;
 struct address_space {
 	struct inode		*host;		/* owner: inode, block_device */
 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
@@ -320,7 +321,7 @@ struct address_space {
 	spinlock_t		i_shared_lock;  /* and spinlock protecting it */
 	unsigned long		dirtied_when;	/* jiffies of first page dirtying */
 	int			gfp_mask;	/* how to allocate the pages */
-	unsigned long 		*ra_pages;	/* device readahead */
+	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
 	spinlock_t		private_lock;	/* for use by the address_space */
 	struct list_head	private_list;	/* ditto */
 	struct address_space	*assoc_mapping;	/* ditto */
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 2f0b56f0183b..451cdff1ec16 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -454,7 +454,6 @@ void do_page_cache_readahead(struct file *file,
 void page_cache_readahead(struct file *file, unsigned long offset);
 void page_cache_readaround(struct file *file, unsigned long offset);
 void handle_ra_thrashing(struct file *file);
-extern unsigned long default_ra_pages;
 
 /* vma is the first one with  address < vma->vm_end,
  * and even  address < vma->vm_start. Have to extend vma. */