Re: [PATCH bluetooth-next] nl802154: export supported commands

[Alexander Aring <alex.aring@gmail.com>]

On Wed, Jun 03, 2015 at 12:43:17PM +0530, Varka Bhadram wrote:
> On 06/03/2015 12:19 PM, Alexander Aring wrote:
> 
> >Hi,
> >
> >On Wed, Jun 03, 2015 at 09:44:01AM +0530, Varka Bhadram wrote:
> >>This patch will export the supported commands by the devices
> >>to the user. This is required because user should know supported
> >>commands by the IEEE-802.15.4 devices. Drivers that are there in
> >>the mainline are not supporting all the functionalities that the
> >>core is providing to us.
> >>
> >I think this is not the reason why this was implemented. You check here
> >cfg802154_ops which ends in mac802154 (SoftMAC) OR _possible_ HardMAC drivers.
> >
> >If you want to make sure that a driver implement it _and_ really sure
> >you need to check on ieee802154_ops.
> >
> >I suppose the real reason why wireless has this implementation, is
> >because to check if HardMAC drivers can support a specific command.
> >This is because every HardMAC driver should implement then the full
> >cfg802154_ops callback structure. It's simple wrong here to say "to
> >check if the driver support the functionalities".
> 
> I found this thread for the reason used in wireless for implementation.
> 
> http://www.spinics.net/lists/linux-wireless/msg29923.html
> 

Yea, but in this case he means really HardMAC drivers only which
implements some cfg802154_ops which others doesn't implement.

I would split our use cases in two categories: setting phy or mac cmds.


for phy setttings cmds:

You can check if the phy supports the specific CMD. E.g. tx_power,
before calling the specific cmd. The cmd will not set because you check the
phy flags.


for mac setttings cmds:

This currently makes no sense, because these are all supported by the
mac802154 layer and we have currently no HardMAC drivers. Each HardMAC
driver implements the cfg802154_ops and then it could make sense to
check if the HardMAC driver supports a specific command.

At the moment for mac settings (in case of SoftMAC) it should display
always the same.



Nevertheless, I would ack this. This would be useful if somebody wants
to implement a HardMAC driver.

- Alex