[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20181206070633.GI2318@nanopsycho>
Date: Thu, 6 Dec 2018 08:06:33 +0100
From: Jiri Pirko <jiri@...nulli.us>
To: Vasundhara Volam <vasundhara-v.volam@...adcom.com>
Cc: David Miller <davem@...emloft.net>,
"michael.chan@...adcom.com" <michael.chan@...adcom.com>,
Jiri Pirko <jiri@...lanox.com>, Netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH net-next RFC 1/7] devlink: Add devlink_param for port
register and unregister
Thu, Dec 06, 2018 at 07:02:59AM CET, vasundhara-v.volam@...adcom.com wrote:
>Thank you reviewing the patches.
>
>On Wed, Dec 5, 2018 at 5:24 PM Jiri Pirko <jiri@...nulli.us> wrote:
>>
>> Wed, Dec 05, 2018 at 06:56:54AM CET, vasundhara-v.volam@...adcom.com wrote:
>> >Add functions to register and unregister for the driver supported
>> >configuration parameters table per port.
>> >
>> >Cc: Jiri Pirko <jiri@...lanox.com>
>> >Signed-off-by: Vasundhara Volam <vasundhara-v.volam@...adcom.com>
>> >---
>> > include/net/devlink.h | 29 +++++++++++
>> > net/core/devlink.c | 130 ++++++++++++++++++++++++++++++++++++++++++++++----
>> > 2 files changed, 151 insertions(+), 8 deletions(-)
>> >
>> >diff --git a/include/net/devlink.h b/include/net/devlink.h
>> >index 67f4293..9b4d80b 100644
>> >--- a/include/net/devlink.h
>> >+++ b/include/net/devlink.h
>> >@@ -48,6 +48,7 @@ struct devlink_port_attrs {
>> >
>> > struct devlink_port {
>> > struct list_head list;
>> >+ struct list_head param_list;
>> > struct devlink *devlink;
>> > unsigned index;
>> > bool registered;
>> >@@ -419,6 +420,13 @@ enum devlink_param_generic_id {
>> > .validate = _validate, \
>> > }
>> >
>> >+enum devlink_port_param_generic_id {
>> >+ /* add new param generic ids above here */
>> >+ __DEVLINK_PORT_PARAM_GENERIC_ID_MAX,
>> >+ DEVLINK_PORT_PARAM_GENERIC_ID_MAX =
>> >+ __DEVLINK_PORT_PARAM_GENERIC_ID_MAX - 1,
>> >+};
>>
>> I don't see the need for enum just for per-port params. The existing
>> params enum should be enough.
>In that case there won't be any differentiation between generic device
>and port params.
Yes, you don't need that.
>Also, if enum is not needed, then separate struct
>devlink_port_param_generic is also not needed.
Yep.
>
>Based on this, entire patchset needs a change.
[...]
Powered by blists - more mailing lists