From mboxrd@z Thu Jan 1 00:00:00 1970 From: Vivien Didelot Subject: Re: [PATCH net-next v3 3/4] bridge: push bridge setting ageing_time down to switchdev Date: Sat, 10 Oct 2015 18:41:22 -0400 Message-ID: <20151010224121.GA27091@ketchup.lan> References: <1444357400-37078-1-git-send-email-sfeldma@gmail.com> <1444357400-37078-4-git-send-email-sfeldma@gmail.com> <77EF4405DD4BB54AACCE7DB593DF6A9A9F6596@SJEXCHMB14.corp.ad.broadcom.com> <20151010070434.GB1990@nanopsycho.orion> <20151010155619.GA14572@ketchup> <20151010210919.GA2217@nanopsycho.orion> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Scott Feldman , Premkumar Jonnala , "netdev@vger.kernel.org" , "davem@davemloft.net" , "siva.mannem.lnx@gmail.com" , "stephen@networkplumber.org" , "roopa@cumulusnetworks.com" , "andrew@lunn.ch" , "f.fainelli@gmail.com" To: Jiri Pirko Return-path: Received: from mail.savoirfairelinux.com ([208.88.110.44]:60804 "EHLO mail.savoirfairelinux.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750964AbbJJWld (ORCPT ); Sat, 10 Oct 2015 18:41:33 -0400 Content-Disposition: inline In-Reply-To: <20151010210919.GA2217@nanopsycho.orion> Sender: netdev-owner@vger.kernel.org List-ID: On Oct. Saturday 10 (41) 11:09 PM, Jiri Pirko wrote: > Sat, Oct 10, 2015 at 05:56:19PM CEST, vivien.didelot@savoirfairelinux.com wrote: > >On Oct. Saturday 10 (41) 09:04 AM, Jiri Pirko wrote: > >> Sat, Oct 10, 2015 at 04:53:52AM CEST, sfeldma@gmail.com wrote: > >> >On Thu, Oct 8, 2015 at 9:38 PM, Premkumar Jonnala wrote: > >> >> > >> >> > >> >>> -----Original Message----- > >> >>> From: sfeldma@gmail.com [mailto:sfeldma@gmail.com] > >> >>> Sent: Friday, October 09, 2015 7:53 AM > >> >>> To: netdev@vger.kernel.org > >> >>> Cc: davem@davemloft.net; jiri@resnulli.us; siva.mannem.lnx@gmail.com; > >> >>> Premkumar Jonnala; stephen@networkplumber.org; > >> >>> roopa@cumulusnetworks.com; andrew@lunn.ch; f.fainelli@gmail.com; > >> >>> vivien.didelot@savoirfairelinux.com > >> >>> Subject: [PATCH net-next v3 3/4] bridge: push bridge setting ageing_time down > >> >>> to switchdev > >> >>> > >> >>> From: Scott Feldman > >> >>> > >> >>> Use SWITCHDEV_F_SKIP_EOPNOTSUPP to skip over ports in bridge that don't > >> >>> support setting ageing_time (or setting bridge attrs in general). > >> >>> > >> >>> If push fails, don't update ageing_time in bridge and return err to user. > >> >>> > >> >>> If push succeeds, update ageing_time in bridge and run gc_timer now to > >> >>> recalabrate when to run gc_timer next, based on new ageing_time. > >> >>> > >> >>> Signed-off-by: Scott Feldman > >> >>> Signed-off-by: Jiri Pirko > >> > > >> > > >> > > >> >>> +int br_set_ageing_time(struct net_bridge *br, u32 ageing_time) > >> >>> +{ > >> >>> + struct switchdev_attr attr = { > >> >>> + .id = SWITCHDEV_ATTR_ID_BRIDGE_AGEING_TIME, > >> >>> + .flags = SWITCHDEV_F_SKIP_EOPNOTSUPP, > >> >>> + .u.ageing_time = ageing_time, > >> >>> + }; > >> >>> + unsigned long t = clock_t_to_jiffies(ageing_time); > >> >>> + int err; > >> >>> + > >> >>> + if (t < BR_MIN_AGEING_TIME || t > BR_MAX_AGEING_TIME) > >> >>> + return -ERANGE; > >> >>> + > >> >>> + err = switchdev_port_attr_set(br->dev, &attr); > >> >> > >> >> A thought - given that the ageing time is not a per-bridge-port attr, why are we using a "port based api" > >> >> to pass the attribute down? May be I'm missing something here? > >> > > >> >I think Florian raised the same point earlier. Sigh, I think this > >> >should be addressed....v4 coming soon...thanks guys for keeping the > >> >standard high. > >> > >> Scott, can you tell us how do you want to address this? I like the > >> current implementation. > > > >Scott, didn't you have a plan to add a struct device for the parent of > >switchdev ports? > > > >I think it would be good to introduce such device with an helper to > >retrieve this upper parent, and move the switchdev ops to this guy. > > > >So switchdev drivers may implement such API calls: > > > > .obj_add(struct device *swdev, struct switchdev_obj *obj); > > > > .port_obj_add(struct device *swdev, struct net_device *port, > > struct switchdev_obj *obj); > > > >Then switchdev code may have a parent API and the current port API may > >look like this: > > > > int switchdev_port_obj_add(struct net_device *dev, > > struct switchdev_obj *obj) > > { > > struct device *swdev = switchdev_get_parent(dev); > > int err = -EOPNOTSUPP; > > > > if (swdev && swdev->switchdev_ops && > > swdev->switchdev_ops->port_obj_add) > > err = swdev->switchdev_ops->port_obj_add(swdev, dev, obj); > > > > return err; > > } > > Fro the record, I don't see any reason for this "device". It would just > introduce unnecessary complexicity. So far, we are fine without it. I wouldn't say that. I beleive that an Ethernet switch deserves its struct device in the tree, since it is a physical chip, like any other. Configuring it through one of its port (net_device) is fine, since you want to change the port behavior, and Linux config is on per-port basis. But the complexity is already introduced in the struct net_device with the switchdev_ops. These ops really belong to the parent device, not to all of its ports. Ideally a switch device would be registered with this set of operations, creates its net_devices, and will be accessible from a port net_device through a netdev helper function. Thanks, -v