[dpdk-dev] [PATCH] net/mlx5: fix counter offset support discover

[dpdk-dev] [PATCH] net/mlx5: fix counter offset support discover

[Suanming Mou]

Currently, the counter offset support is discovered by creating the
rule with invalid offset counter and jump action in root table. If
the rule creation fails with EINVAL errno, that mean counter offset
is not supported in root table.

However, jump action may not be supported in some rdma-core version.
In this case, the discover code will not work properly.

This commits changes the jump action to generic drop action. That
makes the discover code to be more compatible.

Fixes: 994829e695c0 ("net/mlx5: remove single counter container")

Signed-off-by: Suanming Mou <suanmingm@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
---
 drivers/net/mlx5/mlx5_flow_dv.c | 22 +++++-----------------
 1 file changed, 5 insertions(+), 17 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c
index 78c710f..bb2898f 100644
--- a/drivers/net/mlx5/mlx5_flow_dv.c
+++ b/drivers/net/mlx5/mlx5_flow_dv.c
@@ -12294,19 +12294,15 @@ struct mlx5_cache_entry *
 		.match_mask = (void *)&mask,
 	};
 	void *actions[2] = { 0 };
-	struct mlx5_flow_tbl_resource *tbl = NULL, *dest_tbl = NULL;
+	struct mlx5_flow_tbl_resource *tbl = NULL;
 	struct mlx5_devx_obj *dcs = NULL;
 	void *matcher = NULL;
 	void *flow = NULL;
-	int i, ret = -1;
+	int ret = -1;
 
 	tbl = flow_dv_tbl_resource_get(dev, 0, 0, 0, false, NULL, 0, 0, NULL);
 	if (!tbl)
 		goto err;
-	dest_tbl = flow_dv_tbl_resource_get(dev, 1, 0, 0, false,
-					    NULL, 0, 0, NULL);
-	if (!dest_tbl)
-		goto err;
 	dcs = mlx5_devx_cmd_flow_counter_alloc(priv->sh->ctx, 0x4);
 	if (!dcs)
 		goto err;
@@ -12314,10 +12310,7 @@ struct mlx5_cache_entry *
 						    &actions[0]);
 	if (ret)
 		goto err;
-	ret = mlx5_flow_os_create_flow_action_dest_flow_tbl
-				(dest_tbl->obj, &actions[1]);
-	if (ret)
-		goto err;
+	actions[1] = priv->drop_queue.hrxq->action;
 	dv_attr.match_criteria_enable = flow_dv_matcher_enable(mask.buf);
 	ret = mlx5_flow_os_create_flow_matcher(sh->ctx, &dv_attr, tbl->obj,
 					       &matcher);
@@ -12346,17 +12339,12 @@ struct mlx5_cache_entry *
 				     "support detection");
 		ret = 0;
 	}
-	for (i = 0; i < 2; i++) {
-		if (actions[i])
-			claim_zero(mlx5_flow_os_destroy_flow_action
-				   (actions[i]));
-	}
+	if (actions[0])
+		claim_zero(mlx5_flow_os_destroy_flow_action(actions[0]));
 	if (matcher)
 		claim_zero(mlx5_flow_os_destroy_flow_matcher(matcher));
 	if (tbl)
 		flow_dv_tbl_resource_release(MLX5_SH(dev), tbl);
-	if (dest_tbl)
-		flow_dv_tbl_resource_release(MLX5_SH(dev), dest_tbl);
 	if (dcs)
 		claim_zero(mlx5_devx_cmd_destroy(dcs));
 	return ret;
-- 
1.8.3.1

Re: [dpdk-dev] [PATCH] net/mlx5: fix counter offset support discover

[Thomas Monjalon]

12/11/2020 09:51, Suanming Mou:
> Currently, the counter offset support is discovered by creating the
> rule with invalid offset counter and jump action in root table. If
> the rule creation fails with EINVAL errno, that mean counter offset
> is not supported in root table.
> 
> However, jump action may not be supported in some rdma-core version.
> In this case, the discover code will not work properly.
> 
> This commits changes the jump action to generic drop action. That
> makes the discover code to be more compatible.
> 
> Fixes: 994829e695c0 ("net/mlx5: remove single counter container")
> 
> Signed-off-by: Suanming Mou <suanmingm@nvidia.com>
> Acked-by: Matan Azrad <matan@nvidia.com>

Applied as "net/mlx5: fix detection of counter offset support"
in next-net-mlx, thanks.