[PATCH v2 01/10] nfsd: fix cpntf publish race in nfs4_init_cp_state

Jeff Layton <[email protected]>
Newsgroups gmane.linux.kernel,gmane.linux.nfs
Message-ID <[email protected]>
From: Chris Mason <[email protected]>

nfs4_alloc_init_cpntf_state() published the new cpntf entry into the
s2s_cp_stateids IDR (with cs_type set) in one s2s_cp_lock section, then
took the lock again to list_add() it onto p_stid->sc_cp_list. In the gap
the entry is reachable by so_id but cp_list is still {NULL,NULL} from
kzalloc. A racing OFFLOAD_CANCEL (so_id is echoed to the client as
cnr_stateid, so any NFSv4.2 client can drive it) reaches
manage_cpntf_state() -> _free_cpntf_state_locked() and does list_del() on
the zeroed list_head, oopsing the server.

Fold the cs_type assignment and the list_add() into the same critical
section as idr_alloc_cyclic(), so a concurrent lookup either misses the
entry or sees a fully linked cp_list. INIT_LIST_HEAD() the entry after
allocation and switch _free_cpntf_state_locked() to list_del_init() so a
stale unlink is a no-op. nfs4_init_copy_state() passes NULL p_stid and
skips the list_add, preserving NFS4_COPY_STID semantics.

Fixes: 624322f1adc5 ("NFSD add COPY_NOTIFY operation")
Assisted-by: kres:claude-opus-4-7
Signed-off-by: Chris Mason <[email protected]>
---
 fs/nfsd/nfs4state.c | 42 ++++++++++++++++++++++++++++++++----------
 1 file changed, 32 insertions(+), 10 deletions(-)

diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index a4398dc861a5..b8946db3ebaa 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -944,7 +944,7 @@ struct nfs4_stid *nfs4_alloc_stid(struct nfs4_client *cl, struct kmem_cache *sla
  * Create a unique stateid_t to represent each COPY.
  */
 static int nfs4_init_cp_state(struct nfsd_net *nn, copy_stateid_t *stid,
-			      unsigned char cs_type)
+			      unsigned char cs_type, struct nfs4_stid *p_stid)
 {
 	int new_id;
 
@@ -954,19 +954,37 @@ static int nfs4_init_cp_state(struct nfsd_net *nn, copy_stateid_t *stid,
 	idr_preload(GFP_KERNEL);
 	spin_lock(&nn->s2s_cp_lock);
 	new_id = idr_alloc_cyclic(&nn->s2s_cp_stateids, stid, 0, 0, GFP_NOWAIT);
-	stid->cs_stid.si_opaque.so_id = new_id;
-	stid->cs_stid.si_generation = 1;
+	if (new_id >= 0) {
+		stid->cs_stid.si_opaque.so_id = new_id;
+		stid->cs_stid.si_generation = 1;
+		/*
+		 * Publish cs_type and link onto the parent stid's
+		 * sc_cp_list inside the same critical section that
+		 * installed the entry into nn->s2s_cp_stateids. A
+		 * concurrent manage_cpntf_state() either fails the
+		 * idr_find() (entry not yet visible) or observes a
+		 * fully linked cp_list, so list_del_init() in
+		 * _free_cpntf_state_locked() is always well-defined.
+		 */
+		stid->cs_type = cs_type;
+		if (p_stid) {
+			struct nfs4_cpntf_state *cps =
+				container_of(stid, struct nfs4_cpntf_state,
+					     cp_stateid);
+
+			list_add(&cps->cp_list, &p_stid->sc_cp_list);
+		}
+	}
 	spin_unlock(&nn->s2s_cp_lock);
 	idr_preload_end();
 	if (new_id < 0)
 		return 0;
-	stid->cs_type = cs_type;
 	return 1;
 }
 
 int nfs4_init_copy_state(struct nfsd_net *nn, struct nfsd4_copy *copy)
 {
-	return nfs4_init_cp_state(nn, &copy->cp_stateid, NFS4_COPY_STID);
+	return nfs4_init_cp_state(nn, &copy->cp_stateid, NFS4_COPY_STID, NULL);
 }
 
 struct nfs4_cpntf_state *nfs4_alloc_init_cpntf_state(struct nfsd_net *nn,
@@ -977,13 +995,17 @@ struct nfs4_cpntf_state *nfs4_alloc_init_cpntf_state(struct nfsd_net *nn,
 	cps = kzalloc_obj(struct nfs4_cpntf_state);
 	if (!cps)
 		return NULL;
+	/*
+	 * Initialize cp_list so any stale unlink (e.g. on an
+	 * entry that never reached its parent's sc_cp_list)
+	 * degrades to a benign self-unlink via list_del_init().
+	 */
+	INIT_LIST_HEAD(&cps->cp_list);
 	cps->cpntf_time = ktime_get_boottime_seconds();
 	refcount_set(&cps->cp_stateid.cs_count, 1);
-	if (!nfs4_init_cp_state(nn, &cps->cp_stateid, NFS4_COPYNOTIFY_STID))
+	if (!nfs4_init_cp_state(nn, &cps->cp_stateid, NFS4_COPYNOTIFY_STID,
+				p_stid))
 		goto out_free;
-	spin_lock(&nn->s2s_cp_lock);
-	list_add(&cps->cp_list, &p_stid->sc_cp_list);
-	spin_unlock(&nn->s2s_cp_lock);
 	return cps;
 out_free:
 	kfree(cps);
@@ -7854,7 +7876,7 @@ _free_cpntf_state_locked(struct nfsd_net *nn, struct nfs4_cpntf_state *cps)
 	WARN_ON_ONCE(cps->cp_stateid.cs_type != NFS4_COPYNOTIFY_STID);
 	if (!refcount_dec_and_test(&cps->cp_stateid.cs_count))
 		return;
-	list_del(&cps->cp_list);
+	list_del_init(&cps->cp_list);
 	idr_remove(&nn->s2s_cp_stateids,
 		   cps->cp_stateid.cs_stid.si_opaque.so_id);
 	kfree(cps);

-- 
2.55.0
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.