Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   zhenwei pi <pizhenwei@bytedance.com>
To:     kbusch@kernel.org, axboe@fb.com, hch@lst.de, sagi@grimberg.me
Cc:     linux-nvme@lists.infradead.org, linux-kernel@vger.kernel.org,
        lengchao@huawei.com, pizhenwei@bytedance.com
Subject: [PATCH v3] nvme-rdma: handle nvme completion data length
Date:   Sun, 25 Oct 2020 19:51:24 +0800
Message-Id: <20201025115124.1430678-1-pizhenwei@bytedance.com>
Precedence: bulk

Hit a kernel warning:
refcount_t: underflow; use-after-free.
WARNING: CPU: 0 PID: 0 at lib/refcount.c:28

RIP: 0010:refcount_warn_saturate+0xd9/0xe0
Call Trace:
 <IRQ>
 nvme_rdma_recv_done+0xf3/0x280 [nvme_rdma]
 __ib_process_cq+0x76/0x150 [ib_core]
 ...

The reason is that a zero bytes message received from target, and the
host side continues to process without length checking, then the
previous CQE is processed twice.

Do sanity check on received data length, try to recovery for corrupted
CQE case.

Because zero bytes message in not defined in spec, using zero bytes
message to detect dead connections on transport layer is not
standard, currently still treat it as illegal.

Thanks to Chao Leng & Sagi for suggestions.

Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
---
 drivers/nvme/host/rdma.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c
index aad829a2b50d..40a0a3b6476c 100644
--- a/drivers/nvme/host/rdma.c
+++ b/drivers/nvme/host/rdma.c
@@ -1768,6 +1768,14 @@ static void nvme_rdma_recv_done(struct ib_cq *cq, struct ib_wc *wc)
 		return;
 	}
 
+	/* sanity checking for received data length */
+	if (unlikely(wc->byte_len < len)) {
+		dev_err(queue->ctrl->ctrl.device,
+			"Unexpected nvme completion length(%d)\n", wc->byte_len);
+		nvme_rdma_error_recovery(queue->ctrl);
+		return;
+	}
+
 	ib_dma_sync_single_for_cpu(ibdev, qe->dma, len, DMA_FROM_DEVICE);
 	/*
 	 * AEN requests are special as they don't time out and can
-- 
2.11.0