RE: [EXT] Re: [net-next PATCH V2] octeontx2-af: Install TC filter rules in hardware based on priority

[Suman Ghosh]

>On Sun, Jul 16, 2023 at 11:54:42PM +0530, Suman Ghosh wrote:
>> As of today, hardware does not support installing tc filter rules
>> based on priority. This patch fixes the issue and install the hardware
>> rules based on priority. The final hardware rules will not be
>> dependent on rule installation order, it will be strictly priority
>> based, same as software.
>>
>> Signed-off-by: Suman Ghosh <sumang@xxxxxxxxxxx>
>> ---
>> Changes since v1:
>> - Rebased the patch on top of current 'main' branch
>>
>>  .../net/ethernet/marvell/octeontx2/af/mbox.h  |   9 +-
>>  .../marvell/octeontx2/af/rvu_npc_fs.c         |   9 +-
>>  .../marvell/octeontx2/af/rvu_switch.c         |   6 +-
>>  .../marvell/octeontx2/nic/otx2_common.h       |  11 +-
>>  .../marvell/octeontx2/nic/otx2_devlink.c      |   1 -
>>  .../marvell/octeontx2/nic/otx2_ethtool.c      |   1 +
>>  .../marvell/octeontx2/nic/otx2_flows.c        |   2 +
>>  .../ethernet/marvell/octeontx2/nic/otx2_tc.c  | 313
>> +++++++++++++-----
>>  8 files changed, 248 insertions(+), 104 deletions(-)
>
><...>
>
>> +static void otx2_tc_del_from_flow_list(struct otx2_flow_config
>*flow_cfg,
>> +				       struct otx2_tc_flow *node)
>>  {
>> +	struct otx2_tc_flow *tmp;
>> +	struct list_head *pos, *n;
>
>Please declared variables in rversed Christmas tree, in all places,
>thanks.
[Suman] Ack, will update in v3 patch.
>
>> +
>> +	list_for_each_safe(pos, n, &flow_cfg->flow_list_tc) {
>> +		tmp = list_entry(pos, struct otx2_tc_flow, list);
>> +		if (node == tmp) {
>> +			list_del(&node->list);
>> +			return;
>> +		}
>> +	}
>> +}
>> +
>> +static int otx2_tc_add_to_flow_list(struct otx2_flow_config
>*flow_cfg,
>> +				    struct otx2_tc_flow *node)
>> +{
>> +	struct otx2_tc_flow *tmp;
>> +	struct list_head *pos, *n;
>> +	int index = 0;
>
>Ditto.
[Suman] Ack, will update in v3 patch.
>
>> +
>
><...>
>
>> +static int otx2_del_mcam_flow_entry(struct otx2_nic *nic, u16 entry,
>> +u16 *cntr_val) {
>> +	struct npc_delete_flow_rsp __maybe_unused *rsp;
>
>Why __maybe_unused? This keyword is usually used when in some CONFIG_*
>option, it won't be used. It is not the case here.
[Suman] Ack, not needed her. Will update in v3 patch
>
>>  	struct npc_delete_flow_req *req;
>>  	int err;
>
><...>
>
>> +	ntuple = !!(nic->netdev->features & NETIF_F_NTUPLE);
>
>No need in !! for bool variables.
[Suman] Ack, will update in v3 patch.
>
>Thanks