RDMA/hns: Fix signed-unsigned mixed comparisons
authorChengchang Tang <tangchengchang@huawei.com>
Tue, 17 Oct 2023 12:52:35 +0000 (20:52 +0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Mon, 20 Nov 2023 10:59:21 +0000 (11:59 +0100)
[ Upstream commit b5f9efff101b06fd06a5e280a2b00b1335f5f476 ]

The ib_mtu_enum_to_int() and uverbs_attr_get_len() may returns a negative
value. In this case, mixed comparisons of signed and unsigned types will
throw wrong results.

This patch adds judgement for this situation.

Fixes: 30b707886aeb ("RDMA/hns: Support inline data in extented sge space for RC")
Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://lore.kernel.org/r/20231017125239.164455-4-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
drivers/infiniband/hw/hns/hns_roce_hw_v2.c

index 2b8f648..3daa684 100644 (file)
@@ -270,7 +270,7 @@ static bool check_inl_data_len(struct hns_roce_qp *qp, unsigned int len)
        struct hns_roce_dev *hr_dev = to_hr_dev(qp->ibqp.device);
        int mtu = ib_mtu_enum_to_int(qp->path_mtu);
 
-       if (len > qp->max_inline_data || len > mtu) {
+       if (mtu < 0 || len > qp->max_inline_data || len > mtu) {
                ibdev_err(&hr_dev->ib_dev,
                          "invalid length of data, data len = %u, max inline len = %u, path mtu = %d.\n",
                          len, qp->max_inline_data, mtu);