[PATCH] (9/15) big struct block_device * push (first series)

 - this one is interesting and will play in the next series as well;
   affected place is fs/block_dev.c::do_open().  We check if bdev is
   a partition (same way it is done in generic_make_request()) and
   if it is - open entire disk and put pointer to its bdev into a
   new field bdev->bd_contains.  Otherwise (non-partition) we
   set bdev->bd_contains to bdev.  Corresponding cleanup done in
   blkdev_put() (and failure path in do_open()) - when the last opener
   goes away we close bdev->bd_contains if bdev is a partition (i.e.
   not equal to its ->bd_contains) and set it to NULL.
   Immediate effect is that we can get from bdev of partition to
   bdev of disk when submitting a bio, but it also opens a way
   to handle partition-parsing in a sane way.  That will be done
   in the next series.

3 files changed, 59 insertions, 17 deletions

diff --git a/fs/block_dev.c b/fs/block_dev.c
index ba5842f6981d..f7460a3283b9 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -353,6 +353,7 @@ struct block_device *bdget(dev_t dev)
 			atomic_set(&new_bdev->bd_count,1);
 			new_bdev->bd_dev = dev;
 			new_bdev->bd_op = NULL;
+			new_bdev->bd_contains = NULL;
 			new_bdev->bd_inode = inode;
 			inode->i_mode = S_IFBLK;
 			inode->i_rdev = kdev;
@@ -590,28 +591,62 @@ static int do_open(struct block_device *bdev, struct inode *inode, struct file *
 {
 	int ret = -ENXIO;
 	kdev_t dev = to_kdev_t(bdev->bd_dev);
+	struct module *owner = NULL;
 
 	down(&bdev->bd_sem);
 	lock_kernel();
-	if (!bdev->bd_op)
+	if (!bdev->bd_op) {
 		bdev->bd_op = get_blkfops(major(dev));
-	if (bdev->bd_op) {
-		ret = 0;
-		if (bdev->bd_op->owner)
-			__MOD_INC_USE_COUNT(bdev->bd_op->owner);
-		if (bdev->bd_op->open)
-			ret = bdev->bd_op->open(inode, file);
-		if (!ret) {
-			bdev->bd_openers++;
-			bdev->bd_inode->i_size = blkdev_size(dev);
-			bdev->bd_inode->i_blkbits = blksize_bits(block_size(dev));
-		} else {
-			if (bdev->bd_op->owner)
-				__MOD_DEC_USE_COUNT(bdev->bd_op->owner);
-			if (!bdev->bd_openers)
-				bdev->bd_op = NULL;
+		if (!bdev->bd_op)
+			goto out;
+		owner = bdev->bd_op->owner;
+		if (owner)
+			__MOD_INC_USE_COUNT(owner);
+	}
+	if (!bdev->bd_contains) {
+		unsigned minor = minor(dev);
+		struct gendisk *g = get_gendisk(dev);
+		bdev->bd_contains = bdev;
+		if (g) {
+			int shift = g->minor_shift;
+			unsigned minor0 = (minor >> shift) << shift;
+			if (minor != minor0) {
+				struct block_device *disk;
+				disk = bdget(MKDEV(major(dev), minor0));
+				ret = -ENOMEM;
+				if (!disk)
+					goto out1;
+				ret = blkdev_get(disk, file->f_mode, file->f_flags, BDEV_RAW);
+				if (ret)
+					goto out1;
+				bdev->bd_contains = disk;
+			}
+		}
+	}
+	if (bdev->bd_op->open) {
+		ret = bdev->bd_op->open(inode, file);
+		if (ret)
+			goto out2;
+	}
+	bdev->bd_openers++;
+	bdev->bd_inode->i_size = blkdev_size(dev);
+	bdev->bd_inode->i_blkbits = blksize_bits(block_size(dev));
+	unlock_kernel();
+	up(&bdev->bd_sem);
+	return 0;
+
+out2:
+	if (!bdev->bd_openers) {
+		bdev->bd_op = NULL;
+		if (bdev != bdev->bd_contains) {
+			blkdev_put(bdev->bd_contains, BDEV_RAW);
+			bdev->bd_contains = NULL;
 		}
 	}
+out1:
+	if (owner)
+		__MOD_DEC_USE_COUNT(owner);
+out:
 	unlock_kernel();
 	up(&bdev->bd_sem);
 	if (ret)
@@ -676,8 +711,13 @@ int blkdev_put(struct block_device *bdev, int kind)
 		ret = bdev->bd_op->release(bd_inode, NULL);
 	if (bdev->bd_op->owner)
 		__MOD_DEC_USE_COUNT(bdev->bd_op->owner);
-	if (!bdev->bd_openers)
+	if (!bdev->bd_openers) {
 		bdev->bd_op = NULL;
+		if (bdev != bdev->bd_contains) {
+			blkdev_put(bdev->bd_contains, BDEV_RAW);
+			bdev->bd_contains = NULL;
+		}
+	}
 	unlock_kernel();
 	up(&bdev->bd_sem);
 	bdput(bdev);
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index e7e8579dfbbd..003f996f6a2f 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -251,6 +251,7 @@ static void check_partition(struct gendisk *hd, kdev_t dev, int first_part_minor
 	else
 		printk(KERN_INFO " %s:", disk_name(hd, minor(dev), buf));
 	bdev = bdget(kdev_t_to_nr(dev));
+	bdev->bd_contains = bdev;
 	bdev->bd_inode->i_size = (loff_t)hd->part[minor(dev)].nr_sects << 9;
 	bdev->bd_inode->i_blkbits = blksize_bits(block_size(dev));
 	for (i = 0; check_part[i]; i++) {
diff --git a/include/linux/fs.h b/include/linux/fs.h
index a97a764dcc53..7a05fbe4b64c 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -420,6 +420,7 @@ struct block_device {
 	struct list_head	bd_inodes;
 	void *			bd_holder;
 	int			bd_holders;
+	struct block_device *	bd_contains;
 };
 
 struct inode {