rate_bytes_per_sec might be 0, check for it.

Signed-off-by: David Yang <mmyangfl@gmail.com>
---
 drivers/net/dsa/ocelot/felix.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/net/dsa/ocelot/felix.c b/drivers/net/dsa/ocelot/felix.c
index 9e5ede932b42..fb2d02ff0fe7 100644
--- a/drivers/net/dsa/ocelot/felix.c
+++ b/drivers/net/dsa/ocelot/felix.c
@@ -2011,6 +2011,10 @@ static int felix_port_policer_add(struct dsa_switch *ds, int port,
 		.burst = policer->burst,
 	};
 
+	if (dsa_mall_policer_tc_entry_type(policer) !=
+	    DSA_MALL_POLICER_TC_KNOWN)
+		return -EOPNOTSUPP;
+
 	return ocelot_port_policer_add(ocelot, port, &pol);
 }
 
-- 
2.51.0

On 1/26/26 7:13 AM, David Yang wrote:
> rate_bytes_per_sec might be 0, check for it.
> 
> Signed-off-by: David Yang <mmyangfl@gmail.com>
> ---
>  drivers/net/dsa/ocelot/felix.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/net/dsa/ocelot/felix.c b/drivers/net/dsa/ocelot/felix.c
> index 9e5ede932b42..fb2d02ff0fe7 100644
> --- a/drivers/net/dsa/ocelot/felix.c
> +++ b/drivers/net/dsa/ocelot/felix.c
> @@ -2011,6 +2011,10 @@ static int felix_port_policer_add(struct dsa_switch *ds, int port,
>  		.burst = policer->burst,
>  	};
>  
> +	if (dsa_mall_policer_tc_entry_type(policer) !=
> +	    DSA_MALL_POLICER_TC_KNOWN)
> +		return -EOPNOTSUPP;

I'm not sure this strict mode is what we want or even we can apply it
now: it could cause functional regression with setup previously
completing successfully in presence of 'new' params and working to some
degree and now failing to start.

Generally speaking some/most offloads are free to ignore
tunable/parameter not available in the H/W.

I think a way to opt-in the new behavior could help.

/P