On Mon, May 22, 2017 at 01:30:15PM -0500, Gustavo A. R. Silva wrote:
> Add null check before dereferencing pointer sgid_attr.ndev
> inside function rdma_vlan_dev_vlan_id().
>
> Addresses-Coverity-ID: 1373979
> Signed-off-by: Gustavo A. R. Silva <garsilva@xxxxxxxxxxxxxx>
> ---
> drivers/infiniband/hw/qedr/qedr_cm.c | 10 ++++++----
> 1 file changed, 6 insertions(+), 4 deletions(-)
+ Ram,
Ram, IB part of qedr driver doesn't exist in MAINTAINERS file.
>
> diff --git a/drivers/infiniband/hw/qedr/qedr_cm.c b/drivers/infiniband/hw/qedr/qedr_cm.c
> index 3d7705c..d86dbe8 100644
> --- a/drivers/infiniband/hw/qedr/qedr_cm.c
> +++ b/drivers/infiniband/hw/qedr/qedr_cm.c
> @@ -270,11 +270,13 @@ static inline int qedr_gsi_build_header(struct qedr_dev *dev,
> return rc;
> }
>
> - vlan_id = rdma_vlan_dev_vlan_id(sgid_attr.ndev);
> - if (vlan_id < VLAN_CFI_MASK)
> - has_vlan = true;
> - if (sgid_attr.ndev)
> + if (sgid_attr.ndev) {
> + vlan_id = rdma_vlan_dev_vlan_id(sgid_attr.ndev);
> + if (vlan_id < VLAN_CFI_MASK)
> + has_vlan = true;
> +
> dev_put(sgid_attr.ndev);
> + }
I do agree that it is better than kernel panic trying to access NULL
pointer, but I'm not sure that retaining default vlan_id = 0 is correct
solution here.
>
> if (!memcmp(&sgid, &zgid, sizeof(sgid))) {
> DP_ERR(dev, "gsi post send: GID not found GID index %d\n",
> --
> 2.5.0
>
Attachment:
signature.asc
Description: PGP signature
