9p: add more conservative locking
During the reorganization some of the multi-theaded locking assumptions were
accidently relaxed. This patch moves us back towards a more conservative
locking strategy.
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
diff --git a/net/9p/trans_fd.c b/net/9p/trans_fd.c
index e8ebe2c..be65d82 100644
--- a/net/9p/trans_fd.c
+++ b/net/9p/trans_fd.c
@@ -188,8 +188,16 @@
LIST_HEAD(cancel_list);
P9_DPRINTK(P9_DEBUG_ERROR, "mux %p err %d\n", m, err);
- m->err = err;
+
spin_lock_irqsave(&m->client->lock, flags);
+
+ if (m->err) {
+ spin_unlock_irqrestore(&m->client->lock, flags);
+ return;
+ }
+
+ m->err = err;
+
list_for_each_entry_safe(req, rtmp, &m->req_list, req_list) {
req->status = REQ_STATUS_ERROR;
if (!req->t_err)
@@ -352,8 +360,9 @@
/* not an else because some packets (like clunk) have no payload */
if ((m->req) && (m->rpos == m->rsize)) { /* packet is read in */
P9_DPRINTK(P9_DEBUG_TRANS, "got new packet\n");
-
+ spin_lock(&m->client->lock);
list_del(&m->req->req_list);
+ spin_unlock(&m->client->lock);
p9_client_cb(m->client, m->req);
m->rbuf = NULL;
@@ -651,9 +660,8 @@
if (m->err < 0)
return m->err;
- req->status = REQ_STATUS_UNSENT;
-
spin_lock(&client->lock);
+ req->status = REQ_STATUS_UNSENT;
list_add_tail(&req->req_list, &m->unsent_req_list);
spin_unlock(&client->lock);
@@ -672,19 +680,21 @@
{
struct p9_trans_fd *ts = client->trans;
struct p9_conn *m = ts->conn;
+ int ret = 1;
P9_DPRINTK(P9_DEBUG_TRANS, "mux %p req %p\n", m, req);
spin_lock(&client->lock);
list_del(&req->req_list);
- spin_unlock(&client->lock);
if (req->status == REQ_STATUS_UNSENT) {
req->status = REQ_STATUS_FLSHD;
- return 0;
+ ret = 0;
}
- return 1;
+ spin_unlock(&client->lock);
+
+ return ret;
}
/**