ceph: avoid putting the realm twice when decoding snaps fails

jira VULN-65927
cve CVE-2022-49770
commit-author Xiubo Li <[email protected]>
commit 51884d1
upstream-diff Absence of the following commit caused a merge conflict:
    2e58664 ("ceph: do not update snapshot context when there is no new snapshot")

When decoding the snaps fails it maybe leaving the 'first_realm'
and 'realm' pointing to the same snaprealm memory. And then it'll
put it twice and could cause random use-after-free, BUG_ON, etc
issues.

	Cc: [email protected]
Link: https://tracker.ceph.com/issues/57686
	Signed-off-by: Xiubo Li <[email protected]>
	Reviewed-by: Ilya Dryomov <[email protected]>
	Signed-off-by: Ilya Dryomov <[email protected]>
(cherry picked from commit 51884d1)
	Signed-off-by: Pratham Patel <[email protected]>

Author: thefossguy-ciq

fs/ceph/snap.c

@@ -665,14 +665,15 @@ int ceph_update_snap_trace(struct ceph_mds_client *mdsc,
 	struct ceph_mds_snap_realm *ri;    /* encoded */
 	__le64 *snaps;                     /* encoded */
 	__le64 *prior_parent_snaps;        /* encoded */
-	struct ceph_snap_realm *realm = NULL;
+	struct ceph_snap_realm *realm;
 	struct ceph_snap_realm *first_realm = NULL;
 	int invalidate = 0;
 	int err = -ENOMEM;
 	LIST_HEAD(dirty_realms);
 
 	dout("update_snap_trace deletion=%d\n", deletion);
 more:
+	realm = NULL;
 	ceph_decode_need(&p, e, sizeof(*ri), bad);
 	ri = p;
 	p += sizeof(*ri);