From d58c1834bf0d218a0bc00f8fb44874551b21da84 Mon Sep 17 00:00:00 2001 From: Kaike Wan Date: Thu, 15 Aug 2019 15:20:33 -0400 Subject: [PATCH] IB/hfi1: Drop stale TID RDMA packets 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. Fixes: 9e93e967f7b4 ("IB/hfi1: Add a function to receive TID RDMA ACK packet") Cc: Reviewed-by: Mike Marciniszyn Signed-off-by: Kaike Wan Signed-off-by: Dennis Dalessandro Link: https://lore.kernel.org/r/20190815192033.105923.44192.stgit@awfm-01.aw.intel.com Signed-off-by: Doug Ledford --- drivers/infiniband/hw/hfi1/tid_rdma.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/drivers/infiniband/hw/hfi1/tid_rdma.c b/drivers/infiniband/hw/hfi1/tid_rdma.c index 996fc298207e..94070144fef5 100644 --- a/drivers/infiniband/hw/hfi1/tid_rdma.c +++ b/drivers/infiniband/hw/hfi1/tid_rdma.c @@ -4509,7 +4509,7 @@ 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, resync_psn, ack_kpsn; + u32 aeth, psn, req_psn, ack_psn, flpsn, resync_psn, ack_kpsn; unsigned long flags; u16 fidx; @@ -4538,6 +4538,9 @@ 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) @@ -4550,7 +4553,8 @@ 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, @@ -4712,7 +4716,12 @@ done: 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]; + 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])); -- 2.30.2