fs: fix UAF/GPF bug in nilfs_mdt_destroy

jira VULN-155283
cve CVE-2022-50367
commit-author Dongliang Mu <[email protected]>
commit 2e488f1
upstream-diff |
	Adjusted context in inode_init_always to due rh specific variable
	rh_reserved2 being initialized to 0 added in commit:
	dbb05b7 ("Rebuild centos8 with kernel-4.18.0-80.el8")

In alloc_inode, inode_init_always() could return -ENOMEM if
security_inode_alloc() fails, which causes inode->i_private
uninitialized. Then nilfs_is_metadata_file_inode() returns
true and nilfs_free_inode() wrongly calls nilfs_mdt_destroy(),
which frees the uninitialized inode->i_private
and leads to crashes(e.g., UAF/GPF).

Fix this by moving security_inode_alloc just prior to
this_cpu_inc(nr_inodes)

Link: https://lkml.kernel.org/r/CAFcO6XOcf1Jj2SeGt=jJV59wmhESeSKpfR0omdFRq+J9nD1vfQ@mail.gmail.com
	Reported-by: butt3rflyh4ck <[email protected]>
	Reported-by: Hao Sun <[email protected]>
	Reported-by: Jiacheng Xu <[email protected]>
	Reviewed-by: Christian Brauner (Microsoft) <[email protected]>
	Signed-off-by: Dongliang Mu <[email protected]>
	Cc: Al Viro <[email protected]>
	Cc: [email protected]
	Signed-off-by: Al Viro <[email protected]>
(cherry picked from commit 2e488f1)
	Signed-off-by: Roxana Nicolescu <[email protected]>

Author: roxanan1996

fs/inode.c

@@ -168,8 +168,6 @@ int inode_init_always(struct super_block *sb, struct inode *inode)
 
 	inode->rh_reserved2 = 0;
 
-	if (security_inode_alloc(inode))
-		goto out;
 	spin_lock_init(&inode->i_lock);
 	lockdep_set_class(&inode->i_lock, &sb->s_type->i_lock_key);
 
@@ -200,11 +198,12 @@ int inode_init_always(struct super_block *sb, struct inode *inode)
 	inode->i_fsnotify_mask = 0;
 #endif
 	inode->i_flctx = NULL;
+
+	if (unlikely(security_inode_alloc(inode)))
+		return -ENOMEM;
 	this_cpu_inc(nr_inodes);
 
 	return 0;
-out:
-	return -ENOMEM;
 }
 EXPORT_SYMBOL(inode_init_always);