Patch for resolving loop devices

Commit Message

Michael McMaster Dec. 11, 2012, 11:16 a.m. UTC

I experienced an error where mkfs.btrfs failed when an unrelated
cryptoloop device was mounted.

$ sudo /tmp/btrfs-progs/mkfs.btrfs  /dev/mapper/mydisk

WARNING! - Btrfs v0.20-rc1-37-g91d9eec-dirty IS EXPERIMENTAL
WARNING! - see http://btrfs.wiki.kernel.org before using

error checking /dev/mapper/mydisk mount status

$ sudo losetup -a
/dev/loop0: [0014]:71314 (/home/michael/loopfile), encryption aes (type 18)

I found that resolve_loop_device() method in utils.c was returning "aes"
as the associated file, instead of /home/michael/loopfile.  This
eventually causes the realpath() call in is_same_blk_file() to fail. The
attached patch fixes this problem.

Regards,
Michael McMaster

Patch

--- utils.c.orig	2012-12-11 21:00:07.000000000 +1000
+++ utils.c	2012-12-11 21:00:26.000000000 +1000
@@ -653,16 +653,16 @@ 
 {
 	int loop_fd;
 	int ret_ioctl;
-	struct loop_info loopinfo;
+	struct loop_info64 loopinfo;
 
 	if ((loop_fd = open(loop_dev, O_RDONLY)) < 0)
 		return -errno;
 
-	ret_ioctl = ioctl(loop_fd, LOOP_GET_STATUS, &loopinfo);
+	ret_ioctl = ioctl(loop_fd, LOOP_GET_STATUS64, &loopinfo);
 	close(loop_fd);
 
 	if (ret_ioctl == 0) {
-		strncpy(loop_file, loopinfo.lo_name, max_len);
+		strncpy(loop_file, loopinfo.lo_file_name, max_len);
 		if (max_len > 0)
 			loop_file[max_len-1] = 0;
 	} else