HFSPlus: simplify inode mode settting logic

From: Wyatt Banks <[email protected]>

HFSPlus: Fix broken inode mode setting logic. Fix broken umask mount option.

---
Patched against 2.6.21.5

hfsplus_cat_read_inode() decides if the selected inode is a directory or not. It then sets the 3rd parameter of hfsplus_get_perms() accordingly. Inside hfsplus_get_perms() it again decides based on its 3rd parameter if the inode is a directory or not. Besides this redundancy, mode is the BSD file type and mode bits (see Apple TechNote 1150 for details) and is never 0. This makes a large portion of the branching logic unused. What actually does happen in the branching is that the UGO mode bits are combined with the directory or regular file bit after the 2nd decision and this resultant mode is assigned to the inode. This also loses suid, sgid, sticky, pipe, fifo, etc.

This patch also fixes the umask mount option as a result. I figured instead of filing a bug report, I'd just go ahead and fix it. :-)

diff -uprN -X linux-2.6.21.5/Documentation/dontdiff linux-2.6.21.5/fs/hfsplus/inode.c linux-2.6.21.5-devel/fs/hfsplus/inode.c
@@ -173,7 +173,7 @@ out:

	return NULL;

}

-static void hfsplus_get_perms(struct inode *inode, struct hfsplus_perm *perms, int dir) +static void hfsplus_get_perms(struct inode *inode, struct hfsplus_perm *perms)

 {
	struct super_block *sb = inode->i_sb;
	u16 mode;
@@ -188,14 +188,7 @@ static void hfsplus_get_perms(struct ino
	if (!inode->i_gid && !mode)
		inode->i_gid = HFSPLUS_SB(sb).gid;

- if (dir) {
- mode = mode ? (mode & S_IALLUGO) :
- (S_IRWXUGO & ~(HFSPLUS_SB(sb).umask));
- mode |= S_IFDIR;
- } else if (!mode)
- mode = S_IFREG | ((S_IRUGO|S_IWUGO) &
- ~(HFSPLUS_SB(sb).umask));
- inode->i_mode = mode;
+ inode->i_mode = mode & ~(HFSPLUS_SB(sb).umask);

	HFSPLUS_I(inode).rootflags = perms->rootflags;
	HFSPLUS_I(inode).userflags = perms->userflags;
@@ -415,7 +408,7 @@ int hfsplus_cat_read_inode(struct inode
			/* panic? */;
		hfs_bnode_read(fd->bnode, &entry, fd->entryoffset,
					sizeof(struct hfsplus_cat_folder));
-		hfsplus_get_perms(inode, &folder->permissions, 1);
+		hfsplus_get_perms(inode, &folder->permissions);
		inode->i_nlink = 1;
		inode->i_size = 2 + be32_to_cpu(folder->valence);
		inode->i_atime = hfsp_mt2ut(folder->access_date);

@@ -435,7 +428,7 @@ int hfsplus_cat_read_inode(struct inode

		hfsplus_inode_read_fork(inode, HFSPLUS_IS_DATA(inode) ?
					&file->data_fork : &file->rsrc_fork);
-		hfsplus_get_perms(inode, &file->permissions, 0);
+		hfsplus_get_perms(inode, &file->permissions);
		inode->i_nlink = 1;
		if (S_ISREG(inode->i_mode)) {
			if (file->permissions.dev)

Hi,

This is wrong, there is an extra note:

"If the S_IFMT field (upper 4 bits) of the fileMode field is zero, then Mac OS X assumes that the permissions structure is uninitialized"

bye, Roman