linux-kernel - [PATCH 4.9 17/83] NFSv4: Fix a potential sleep while atomic in nfs4_do

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20190904175305.430416704@linuxfoundation.org>
Date:   Wed,  4 Sep 2019 19:53:09 +0200
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     linux-kernel@...r.kernel.org
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        stable@...r.kernel.org, John Hubbard <jhubbard@...dia.com>,
        Trond Myklebust <trond.myklebust@...merspace.com>,
        Sasha Levin <sashal@...nel.org>
Subject: [PATCH 4.9 17/83] NFSv4: Fix a potential sleep while atomic in nfs4_do_reclaim()

[ Upstream commit c77e22834ae9a11891cb613bd9a551be1b94f2bc ]

John Hubbard reports seeing the following stack trace:

nfs4_do_reclaim
   rcu_read_lock /* we are now in_atomic() and must not sleep */
       nfs4_purge_state_owners
           nfs4_free_state_owner
               nfs4_destroy_seqid_counter
                   rpc_destroy_wait_queue
                       cancel_delayed_work_sync
                           __cancel_work_timer
                               __flush_work
                                   start_flush_work
                                       might_sleep:
                                        (kernel/workqueue.c:2975: BUG)

The solution is to separate out the freeing of the state owners
from nfs4_purge_state_owners(), and perform that outside the atomic
context.

Reported-by: John Hubbard <jhubbard@...dia.com>
Fixes: 0aaaf5c424c7f ("NFS: Cache state owners after files are closed")
Signed-off-by: Trond Myklebust <trond.myklebust@...merspace.com>
Signed-off-by: Sasha Levin <sashal@...nel.org>
---
 fs/nfs/nfs4_fs.h    |  3 ++-
 fs/nfs/nfs4client.c |  5 ++++-
 fs/nfs/nfs4state.c  | 27 ++++++++++++++++++++++-----
 3 files changed, 28 insertions(+), 7 deletions(-)

diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index 1452177c822db..c719389381dc4 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -434,7 +434,8 @@ static inline void nfs4_schedule_session_recovery(struct nfs4_session *session,
 
 extern struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *, struct rpc_cred *, gfp_t);
 extern void nfs4_put_state_owner(struct nfs4_state_owner *);
-extern void nfs4_purge_state_owners(struct nfs_server *);
+extern void nfs4_purge_state_owners(struct nfs_server *, struct list_head *);
+extern void nfs4_free_state_owners(struct list_head *head);
 extern struct nfs4_state * nfs4_get_open_state(struct inode *, struct nfs4_state_owner *);
 extern void nfs4_put_open_state(struct nfs4_state *);
 extern void nfs4_close_state(struct nfs4_state *, fmode_t);
diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index 43f42cc30a606..1ec6dd4f3e2e4 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -781,9 +781,12 @@ found:
 
 static void nfs4_destroy_server(struct nfs_server *server)
 {
+	LIST_HEAD(freeme);
+
 	nfs_server_return_all_delegations(server);
 	unset_pnfs_layoutdriver(server);
-	nfs4_purge_state_owners(server);
+	nfs4_purge_state_owners(server, &freeme);
+	nfs4_free_state_owners(&freeme);
 }
 
 /*
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 6f474b0670323..4e63daeef6339 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -611,24 +611,39 @@ void nfs4_put_state_owner(struct nfs4_state_owner *sp)
 /**
  * nfs4_purge_state_owners - Release all cached state owners
  * @server: nfs_server with cached state owners to release
+ * @head: resulting list of state owners
  *
  * Called at umount time.  Remaining state owners will be on
  * the LRU with ref count of zero.
+ * Note that the state owners are not freed, but are added
+ * to the list @head, which can later be used as an argument
+ * to nfs4_free_state_owners.
  */
-void nfs4_purge_state_owners(struct nfs_server *server)
+void nfs4_purge_state_owners(struct nfs_server *server, struct list_head *head)
 {
 	struct nfs_client *clp = server->nfs_client;
 	struct nfs4_state_owner *sp, *tmp;
-	LIST_HEAD(doomed);
 
 	spin_lock(&clp->cl_lock);
 	list_for_each_entry_safe(sp, tmp, &server->state_owners_lru, so_lru) {
-		list_move(&sp->so_lru, &doomed);
+		list_move(&sp->so_lru, head);
 		nfs4_remove_state_owner_locked(sp);
 	}
 	spin_unlock(&clp->cl_lock);
+}
 
-	list_for_each_entry_safe(sp, tmp, &doomed, so_lru) {
+/**
+ * nfs4_purge_state_owners - Release all cached state owners
+ * @head: resulting list of state owners
+ *
+ * Frees a list of state owners that was generated by
+ * nfs4_purge_state_owners
+ */
+void nfs4_free_state_owners(struct list_head *head)
+{
+	struct nfs4_state_owner *sp, *tmp;
+
+	list_for_each_entry_safe(sp, tmp, head, so_lru) {
 		list_del(&sp->so_lru);
 		nfs4_free_state_owner(sp);
 	}
@@ -1764,12 +1779,13 @@ static int nfs4_do_reclaim(struct nfs_client *clp, const struct nfs4_state_recov
 	struct nfs4_state_owner *sp;
 	struct nfs_server *server;
 	struct rb_node *pos;
+	LIST_HEAD(freeme);
 	int status = 0;
 
 restart:
 	rcu_read_lock();
 	list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) {
-		nfs4_purge_state_owners(server);
+		nfs4_purge_state_owners(server, &freeme);
 		spin_lock(&clp->cl_lock);
 		for (pos = rb_first(&server->state_owners);
 		     pos != NULL;
@@ -1798,6 +1814,7 @@ restart:
 		spin_unlock(&clp->cl_lock);
 	}
 	rcu_read_unlock();
+	nfs4_free_state_owners(&freeme);
 	return 0;
 }
 
-- 
2.20.1