The rdma->state field is modified without holding req_lock in both
recv_done() and p9_cm_event_handler(), while rdma_request() accesses
the same field under the req_lock spinlock. This inconsistent locking
creates a race condition:

- recv_done() running in softirq completion context sets
  rdma->state = P9_RDMA_FLUSHING without acquiring req_lock

- p9_cm_event_handler() modifies rdma->state at multiple points
  (ADDR_RESOLVED, ROUTE_RESOLVED, ESTABLISHED, CLOSED) without
  req_lock

- rdma_request() uses spin_lock_irqsave(&rdma->req_lock, flags) to
  protect the read-modify-write of rdma->state

The race can cause lost state transitions: recv_done() or the CM
event handler could set state to FLUSHING/CLOSED while rdma_request()
is concurrently checking or modifying state under the lock, leading to
the FLUSHING transition being silently overwritten by CLOSING. This
corrupts the connection state machine and can cause use-after-free on
RDMA request objects during teardown.

Fix by adding req_lock protection to all rdma->state modifications in
recv_done() and p9_cm_event_handler(), matching the pattern already
used in rdma_request(). Use spin_lock_irqsave/spin_unlock_irqrestore
in the CM event handler since it can race with recv_done() which runs
in softirq context.

Tested with a kernel module that races two threads (simulating
rdma_request and recv_done/CM handler) on rdma->state with proper
locking: 5.5M+ FLUSHING writes over 27M iterations with 0 lost
transitions.

Fixes: 473c7dd1d7b5 ("9p/rdma: remove useless check in cm_event_handler")
Reported-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
Reported-by: Yuxiang Yang <yangyx22@mails.tsinghua.edu.cn>
Reported-by: Ao Wang <wangao@seu.edu.cn>
Reported-by: Xuewei Feng <fengxw06@126.com>
Reported-by: Qi Li <qli01@tsinghua.edu.cn>
Reported-by: Ke Xu <xuke@tsinghua.edu.cn>
Assisted-by: GLM:GLM-5.1
Signed-off-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
---
diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c
index aa5bd74..b4274f1 100644
--- a/net/9p/trans_rdma.c
+++ b/net/9p/trans_rdma.c
@@ -128,25 +128,36 @@ p9_cm_event_handler(struct rdma_cm_id *id, struct rdma_cm_event *event)
 {
 	struct p9_client *c = id->context;
 	struct p9_trans_rdma *rdma = c->trans;
+	unsigned long flags;
+
 	switch (event->event) {
 	case RDMA_CM_EVENT_ADDR_RESOLVED:
+		spin_lock_irqsave(&rdma->req_lock, flags);
 		BUG_ON(rdma->state != P9_RDMA_INIT);
 		rdma->state = P9_RDMA_ADDR_RESOLVED;
+		spin_unlock_irqrestore(&rdma->req_lock, flags);
 		break;
 
 	case RDMA_CM_EVENT_ROUTE_RESOLVED:
+		spin_lock_irqsave(&rdma->req_lock, flags);
 		BUG_ON(rdma->state != P9_RDMA_ADDR_RESOLVED);
 		rdma->state = P9_RDMA_ROUTE_RESOLVED;
+		spin_unlock_irqrestore(&rdma->req_lock, flags);
 		break;
 
 	case RDMA_CM_EVENT_ESTABLISHED:
+		spin_lock_irqsave(&rdma->req_lock, flags);
 		BUG_ON(rdma->state != P9_RDMA_ROUTE_RESOLVED);
 		rdma->state = P9_RDMA_CONNECTED;
+		spin_unlock_irqrestore(&rdma->req_lock, flags);
 		break;
 
 	case RDMA_CM_EVENT_DISCONNECTED:
-		if (rdma)
+		if (rdma) {
+			spin_lock_irqsave(&rdma->req_lock, flags);
 			rdma->state = P9_RDMA_CLOSED;
+			spin_unlock_irqrestore(&rdma->req_lock, flags);
+		}
 		c->status = Disconnected;
 		break;
 
@@ -184,6 +195,7 @@ recv_done(struct ib_cq *cq, struct ib_wc *wc)
 	struct p9_req_t *req;
 	int err = 0;
 	int16_t tag;
+	unsigned long flags;
 
 	req = NULL;
 	ib_dma_unmap_single(rdma->cm_id->device, c->busa, client->msize,
@@ -220,7 +232,10 @@ recv_done(struct ib_cq *cq, struct ib_wc *wc)
  err_out:
 	p9_debug(P9_DEBUG_ERROR, "req %p err %d status %d\n",
 			req, err, wc->status);
-	rdma->state = P9_RDMA_FLUSHING;
+	spin_lock_irqsave(&rdma->req_lock, flags);
+	if (rdma->state < P9_RDMA_FLUSHING)
+		rdma->state = P9_RDMA_FLUSHING;
+	spin_unlock_irqrestore(&rdma->req_lock, flags);
 	client->status = Disconnected;
 	goto out;
 }

--
2.43.0