[PATCH v2 0/8] start cleanup of rte_flow_item_*

[David Marchand]

Hi Ferruh, Thomas,

On Fri, Jan 20, 2023 at 6:19 PM Ferruh Yigit <ferruh.yigit at amd.com> wrote:
>
> There was a plan to have structures from lib/net/ at the beginning
> of corresponding flow item structures.
> Unfortunately this plan has not been followed up so far.
> This series is a step to make the most used items,
> compliant with the inheritance design explained above.
> The old API is kept in anonymous union for compatibility,
> but the code in drivers and apps is updated to use the new API.
>
>
> v2: (by Ferruh)
>  * Rebased on latest next-net for v23.03
>  * 'struct rte_gre_hdr' endianness annotation added to protocol field
>  * more driver code updated for rte_flow_item_eth & rte_flow_item_vlan
>  * 'struct rte_gre_hdr' updated to have a combined "rte_be16_t c_rsvd0_ver"
>    field and updated drivers accordingly
>  * more driver code updated for rte_flow_item_gre
>  * more driver code updated for rte_flow_item_gtp
>
>
> Cc: David Marchand <david.marchand at redhat.com>

Note: it is relatively easy to run OVS checks, you only need a github
fork of ovs with a dpdk-latest branch + some github yml update to
point at a dpdk repo + branch of yours (see the last commit in my repo
below).

I ran this series in my dpdk-latest (rebased) OVS branch
https://github.com/david-marchand/ovs/commits/dpdk-latest, through
GHA.

Sparse spotted an issue on rte_flow.h header, following HIGIG2 update.
https://github.com/david-marchand/ovs/actions/runs/3979243283/jobs/6821543439#step:12:2592

2023-01-22T10:31:37.5911785Z ../../lib/ofp-packet.c: note: in included
file (through ../../lib/netdev-dpdk.h, ../../lib/dp-packet.h):
2023-01-22T10:31:37.5918848Z
/home/runner/work/ovs/ovs/dpdk-dir/build/include/rte_flow.h:645:43:
error: incorrect type in initializer (different base types)
2023-01-22T10:31:37.5919574Z
/home/runner/work/ovs/ovs/dpdk-dir/build/include/rte_flow.h:645:43:
expected restricted ovs_be16 [usertype] classification
2023-01-22T10:31:37.5920131Z
/home/runner/work/ovs/ovs/dpdk-dir/build/include/rte_flow.h:645:43:
got int
2023-01-22T10:31:37.5920720Z
/home/runner/work/ovs/ovs/dpdk-dir/build/include/rte_flow.h:646:32:
error: incorrect type in initializer (different base types)
2023-01-22T10:31:37.5921341Z
/home/runner/work/ovs/ovs/dpdk-dir/build/include/rte_flow.h:646:32:
expected restricted ovs_be16 [usertype] vid
2023-01-22T10:31:37.5921866Z
/home/runner/work/ovs/ovs/dpdk-dir/build/include/rte_flow.h:646:32:
got int
2023-01-22T10:31:37.6042168Z make[2]: *** [Makefile:4681:
lib/ofp-packet.lo] Error 1
2023-01-22T10:31:37.6042717Z make[2]: *** Waiting for unfinished jobs....

This should be fixed with:

diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h
index a215daa836..99f8340f82 100644
--- a/lib/ethdev/rte_flow.h
+++ b/lib/ethdev/rte_flow.h
@@ -642,8 +642,8 @@ struct rte_flow_item_higig2_hdr {
 static const struct rte_flow_item_higig2_hdr rte_flow_item_higig2_hdr_mask = {
        .hdr = {
                .ppt1 = {
-                       .classification = 0xffff,
-                       .vid = 0xfff,
+                       .classification = RTE_BE16(0xffff),
+                       .vid = RTE_BE16(0xfff),
                },
        },
 };

However, looking at existing code, and though I don't know HIGIG2, it
is a bit strange to use a 12 bits large mask for vid.


With this fix, OVS sparse check passes:
https://github.com/david-marchand/ovs/actions/runs/3979288868


-- 
David Marchand