Skip to contentSkip to conversation

cifs: fix GlobalMid_Lock bug in cifs_reconnect

PATCH In review Ronnie Sahlberg 3 participants 3 messages Jun 14, 2019 – Jun 17, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 cifs: fix GlobalMid_Lock bug in cifs_reconnect

We can not hold the GlobalMid_Lock spinlock during the
dfs processing in cifs_reconnect since it invokes things that may sleep and thus trigger :

BUG: sleeping function called from invalid context at kernel/locking/rwsem.c:23

Thus we need to drop the spinlock during this code block.

RHBZ: 1716743

---
 fs/cifs/connect.c | 2 ++
 1 file changed, 2 insertions(+)
fs/cifs/connect.c +2 -0
index 8c4121da624e..8dd6637a3cbb 100644
@@ -476,6 +476,7 @@ cifs_reconnect(struct TCP_Server_Info *server)
 	spin_lock(&GlobalMid_Lock);
 	server->nr_targets = 1;
 #ifdef CONFIG_CIFS_DFS_UPCALL
+	spin_unlock(&GlobalMid_Lock);
 	cifs_sb = find_super_by_tcp(server);
 	if (IS_ERR(cifs_sb)) {
 		rc = PTR_ERR(cifs_sb);
@@ -493,6 +494,7 @@ cifs_reconnect(struct TCP_Server_Info *server)
 	}
 	cifs_dbg(FYI, "%s: will retry %d target(s)\n", __func__,
 		 server->nr_targets);
+	spin_lock(&GlobalMid_Lock);
 #endif
 	if (server->tcpStatus == CifsExiting) {
 		/* the demux thread will exit normally
Signature · 1 line
2.13.6
#1 lore raw

tentatively merged into cifs-2.6.git for-next and to the github tree

Quoted 40 lines · Ronnie Sahlberg

We can not hold the GlobalMid_Lock spinlock during the
dfs processing in cifs_reconnect since it invokes things that may sleep and thus trigger :

BUG: sleeping function called from invalid context at kernel/locking/rwsem.c:23

Thus we need to drop the spinlock during this code block.

RHBZ: 1716743

---
 fs/cifs/connect.c | 2 ++
 1 file changed, 2 insertions(+)
fs/cifs/connect.c +2 -1
index 8c4121da624e..8dd6637a3cbb 100644
@@ -476,6 +476,7 @@ cifs_reconnect(struct TCP_Server_Info *server)
        spin_lock(&GlobalMid_Lock);
        server->nr_targets = 1;
 #ifdef CONFIG_CIFS_DFS_UPCALL
+       spin_unlock(&GlobalMid_Lock);
        cifs_sb = find_super_by_tcp(server);
        if (IS_ERR(cifs_sb)) {
                rc = PTR_ERR(cifs_sb);
@@ -493,6 +494,7 @@ cifs_reconnect(struct TCP_Server_Info *server)
        }
        cifs_dbg(FYI, "%s: will retry %d target(s)\n", __func__,
                 server->nr_targets);
+       spin_lock(&GlobalMid_Lock);
 #endif
        if (server->tcpStatus == CifsExiting) {
                /* the demux thread will exit normally
--

2.13.6

Signature · 3 lines
Thanks,

Steve
#2 lore raw

--
Best regards,
Pavel Shilovsky

чт, 13 июн. 2019 г. в 20:57, Steve French <[email protected]>:

Quoted 50 lines

tentatively merged into cifs-2.6.git for-next and to the github tree

Quoted 38 lines · Ronnie Sahlberg

We can not hold the GlobalMid_Lock spinlock during the
dfs processing in cifs_reconnect since it invokes things that may sleep and thus trigger :

BUG: sleeping function called from invalid context at kernel/locking/rwsem.c:23

Thus we need to drop the spinlock during this code block.

RHBZ: 1716743

---
 fs/cifs/connect.c | 2 ++
 1 file changed, 2 insertions(+)
fs/cifs/connect.c +2 -1
index 8c4121da624e..8dd6637a3cbb 100644
@@ -476,6 +476,7 @@ cifs_reconnect(struct TCP_Server_Info *server)
        spin_lock(&GlobalMid_Lock);
        server->nr_targets = 1;
 #ifdef CONFIG_CIFS_DFS_UPCALL
+       spin_unlock(&GlobalMid_Lock);
        cifs_sb = find_super_by_tcp(server);
        if (IS_ERR(cifs_sb)) {
                rc = PTR_ERR(cifs_sb);
@@ -493,6 +494,7 @@ cifs_reconnect(struct TCP_Server_Info *server)
        }
        cifs_dbg(FYI, "%s: will retry %d target(s)\n", __func__,
                 server->nr_targets);
+       spin_lock(&GlobalMid_Lock);
 #endif
        if (server->tcpStatus == CifsExiting) {
                /* the demux thread will exit normally
--
Quoted 2 lines

2.13.6

--
Thanks,

Steve