Re: [PATCH v4 net-next 4/9] net/sched: mqprio: add an extack message to mqprio_parse_opt()

From: Jamal Hadi Salim
Date: Fri Apr 07 2023 - 12:09:05 EST


On Mon, Apr 3, 2023 at 6:35 AM Vladimir Oltean <vladimir.oltean@xxxxxxx> wrote:
>
> Ferenc reports that a combination of poor iproute2 defaults and obscure
> cases where the kernel returns -EINVAL make it difficult to understand
> what is wrong with this command:
>
> $ ip link add veth0 numtxqueues 8 numrxqueues 8 type veth peer name veth1
> $ tc qdisc add dev veth0 root mqprio num_tc 8 map 0 1 2 3 4 5 6 7 \
> queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7
> RTNETLINK answers: Invalid argument
>
> Hopefully with this patch, the cause is clearer:
>
> Error: Device does not support hardware offload.
>
> The kernel was (and still is) rejecting this because iproute2 defaults
> to "hw 1" if this command line option is not specified.
>
> Link: https://lore.kernel.org/netdev/ede5e9a2f27bf83bfb86d3e8c4ca7b34093b99e2.camel@xxxxxxxxxxx/
> Signed-off-by: Vladimir Oltean <vladimir.oltean@xxxxxxx>
> Reviewed-by: Ferenc Fejes <fejes@xxxxxxxxxxx>
> Reviewed-by: Simon Horman <simon.horman@xxxxxxxxxxxx>

Acked-by: Jamal Hadi Salim <jhs@xxxxxxxxxxxx>

cheers,
jamal
> ---
> v3->v4: none
> v2->v3: change link from patchwork to lore
> v1->v2: slightly reword last paragraph of commit message
>
> net/sched/sch_mqprio.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/net/sched/sch_mqprio.c b/net/sched/sch_mqprio.c
> index 5a9261c38b95..9ee5a9a9b9e9 100644
> --- a/net/sched/sch_mqprio.c
> +++ b/net/sched/sch_mqprio.c
> @@ -133,8 +133,11 @@ static int mqprio_parse_opt(struct net_device *dev, struct tc_mqprio_qopt *qopt,
> /* If ndo_setup_tc is not present then hardware doesn't support offload
> * and we should return an error.
> */
> - if (qopt->hw && !dev->netdev_ops->ndo_setup_tc)
> + if (qopt->hw && !dev->netdev_ops->ndo_setup_tc) {
> + NL_SET_ERR_MSG(extack,
> + "Device does not support hardware offload");
> return -EINVAL;
> + }
>
> return 0;
> }
> --
> 2.34.1
>