[PATCH 1/6] hfsplus: fix BKL leak in hfsplus_ioctl

[Christoph Hellwig <hch@lst.de>]

Currenly the HFSPLUS_IOC_EXT2_GETFLAGS case never unlocks the BKL, which
can lead to easily reproduced lockups when doing multiple GETFLAGS ioctls.

Fix this by only taking the BKL for the HFSPLUS_IOC_EXT2_SETFLAGS case
as neither HFSPLUS_IOC_EXT2_GETFLAGS not the default error case needs it.

This behaviour was introduced by "hfsplus: Push down BKL into ioctl function"
during the 2.6.35-rc series.

Signed-off-by: Christoph Hellwig <hch@tuxera.com>

Index: linux-2.6/fs/hfsplus/ioctl.c
===================================================================
--- linux-2.6.orig/fs/hfsplus/ioctl.c	2010-09-26 01:47:23.992790058 +0900
+++ linux-2.6/fs/hfsplus/ioctl.c	2010-09-26 01:49:16.701790597 +0900
@@ -26,7 +26,6 @@ long hfsplus_ioctl(struct file *filp, un
 	struct inode *inode = filp->f_path.dentry->d_inode;
 	unsigned int flags;
 
-	lock_kernel();
 	switch (cmd) {
 	case HFSPLUS_IOC_EXT2_GETFLAGS:
 		flags = 0;
@@ -39,6 +38,8 @@ long hfsplus_ioctl(struct file *filp, un
 		return put_user(flags, (int __user *)arg);
 	case HFSPLUS_IOC_EXT2_SETFLAGS: {
 		int err = 0;
+
+		lock_kernel();
 		err = mnt_want_write(filp->f_path.mnt);
 		if (err) {
 			unlock_kernel();
@@ -93,7 +94,6 @@ setflags_out:
 		return err;
 	}
 	default:
-		unlock_kernel();
 		return -ENOTTY;
 	}
 }