[PATCH] IB/hfi1: Drop stale TID RDMA packets

[PATCH] IB/hfi1: Drop stale TID RDMA packets

[Mike Marciniszyn]

From: Kaike Wan <kaike.wan@intel.com>

Upstream commit d58c1834bf0d218a0bc00f8fb44874551b21da84.

In a congested fabric with adaptive routing enabled, traces show that
the sender could receive stale TID RDMA NAK packets that contain newer
KDETH PSNs and older Verbs PSNs. If not dropped, these packets could
cause the incorrect rewinding of the software flows and the incorrect
completion of TID RDMA WRITE requests, and eventually leading to memory
corruption and kernel crash.

The current code drops stale TID RDMA ACK/NAK packets solely based
on KDETH PSNs, which may lead to erroneous processing. This patch
fixes the issue by also checking the Verbs PSN. Addition checks are
added before rewinding the TID RDMA WRITE DATA packets.

[ported to 5.2 from upstream accounting for fspsn replacing flpsn.]

Fixes: 9e93e967f7b4 ("IB/hfi1: Add a function to receive TID RDMA ACK packet")
Cc: <stable@vger.kernel.org>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
---
 0 files changed

diff --git a/drivers/infiniband/hw/hfi1/tid_rdma.c b/drivers/infiniband/hw/hfi1/tid_rdma.c
index fe7e709..8fcad80 100644
--- a/drivers/infiniband/hw/hfi1/tid_rdma.c
+++ b/drivers/infiniband/hw/hfi1/tid_rdma.c
@@ -4511,7 +4511,7 @@ void hfi1_rc_rcv_tid_rdma_ack(struct hfi1_packet *packet)
 	struct rvt_swqe *wqe;
 	struct tid_rdma_request *req;
 	struct tid_rdma_flow *flow;
-	u32 aeth, psn, req_psn, ack_psn, fspsn, resync_psn, ack_kpsn;
+	u32 aeth, psn, req_psn, ack_psn, flpsn, resync_psn, ack_kpsn;
 	unsigned long flags;
 	u16 fidx;
 
@@ -4540,6 +4540,9 @@ void hfi1_rc_rcv_tid_rdma_ack(struct hfi1_packet *packet)
 		ack_kpsn--;
 	}
 
+	if (unlikely(qp->s_acked == qp->s_tail))
+		goto ack_op_err;
+
 	wqe = rvt_get_swqe_ptr(qp, qp->s_acked);
 
 	if (wqe->wr.opcode != IB_WR_TID_RDMA_WRITE)
@@ -4552,7 +4555,8 @@ void hfi1_rc_rcv_tid_rdma_ack(struct hfi1_packet *packet)
 	trace_hfi1_tid_flow_rcv_tid_ack(qp, req->acked_tail, flow);
 
 	/* Drop stale ACK/NAK */
-	if (cmp_psn(psn, full_flow_psn(flow, flow->flow_state.spsn)) < 0)
+	if (cmp_psn(psn, full_flow_psn(flow, flow->flow_state.spsn)) < 0 ||
+	    cmp_psn(req_psn, flow->flow_state.resp_ib_psn) < 0)
 		goto ack_op_err;
 
 	while (cmp_psn(ack_kpsn,
@@ -4714,8 +4718,12 @@ void hfi1_rc_rcv_tid_rdma_ack(struct hfi1_packet *packet)
 		switch ((aeth >> IB_AETH_CREDIT_SHIFT) &
 			IB_AETH_CREDIT_MASK) {
 		case 0: /* PSN sequence error */
+			if (!req->flows)
+				break;
 			flow = &req->flows[req->acked_tail];
-			fspsn = full_flow_psn(flow, flow->flow_state.spsn);
+			flpsn = full_flow_psn(flow, flow->flow_state.lpsn);
+			if (cmp_psn(psn, flpsn) > 0)
+				break;
 			trace_hfi1_tid_flow_rcv_tid_ack(qp, req->acked_tail,
 							flow);
 			req->r_ack_psn = mask_psn(be32_to_cpu(ohdr->bth[2]));

Re: [PATCH] IB/hfi1: Drop stale TID RDMA packets

[Greg KH]

On Mon, Aug 26, 2019 at 12:01:50PM -0400, Mike Marciniszyn wrote:
> From: Kaike Wan <kaike.wan@intel.com>
> 
> Upstream commit d58c1834bf0d218a0bc00f8fb44874551b21da84.
> 
> In a congested fabric with adaptive routing enabled, traces show that
> the sender could receive stale TID RDMA NAK packets that contain newer
> KDETH PSNs and older Verbs PSNs. If not dropped, these packets could
> cause the incorrect rewinding of the software flows and the incorrect
> completion of TID RDMA WRITE requests, and eventually leading to memory
> corruption and kernel crash.
> 
> The current code drops stale TID RDMA ACK/NAK packets solely based
> on KDETH PSNs, which may lead to erroneous processing. This patch
> fixes the issue by also checking the Verbs PSN. Addition checks are
> added before rewinding the TID RDMA WRITE DATA packets.
> 
> [ported to 5.2 from upstream accounting for fspsn replacing flpsn.]

Now applied, thanks.

greg k-h