iw coalesce show - disabled at init?

iw coalesce show - disabled at init?

[Kobi Cohen-Arazi]

Hi,

When loading a wifi driver that supports set_coalesce and running "iw
show" command:
$ iw ... coalesce show
It would always show coalesce as disabled.
I think the reason for this is that there are no coalesce rules cached
in cfg/netlink layers yet. (empty set of rules).

Is my understanding correct? if yes, Is this by design?
Ideally, at startup, 80211 should query the driver for the current
coalesce rules and use it as the initial state.

Thank you,
- Kobi

Re: iw coalesce show - disabled at init?

[Johannes Berg]

On Wed, 2015-08-12 at 11:36 -0700, Kobi Cohen-Arazi wrote:
> Hi,
> 
> When loading a wifi driver that supports set_coalesce and running "iw
> show" command:
> $ iw ... coalesce show
> It would always show coalesce as disabled.
> I think the reason for this is that there are no coalesce rules 
> cached
> in cfg/netlink layers yet. (empty set of rules).
> 
> Is my understanding correct? if yes, Is this by design?
> Ideally, at startup, 80211 should query the driver for the current
> coalesce rules and use it as the initial state.
> 

Typically devices/drivers shouldn't *have* any such rules at startup,
so the point is moot. If the driver/device you're using behaves
differently, arguably that's a bug. Marvell?

johannes

Re: iw coalesce show - disabled at init?

[Kobi Cohen-Arazi]

Hi Johannes,


On Wed, Aug 12, 2015 at 2:08 PM, Johannes Berg
<johannes@sipsolutions.net> wrote:
> On Wed, 2015-08-12 at 11:36 -0700, Kobi Cohen-Arazi wrote:
>> Hi,
>>
>> When loading a wifi driver that supports set_coalesce and running "iw
>> show" command:
>> $ iw ... coalesce show
>> It would always show coalesce as disabled.
>> I think the reason for this is that there are no coalesce rules
>> cached
>> in cfg/netlink layers yet. (empty set of rules).
>>
>> Is my understanding correct? if yes, Is this by design?
>> Ideally, at startup, 80211 should query the driver for the current
>> coalesce rules and use it as the initial state.
>>
>
> Typically devices/drivers shouldn't *have* any such rules at startup,
> so the point is moot. If the driver/device you're using behaves
> differently, arguably that's a bug. Marvell?

Nope :) not Marvell.

So what you're saying is that at startup, drivers should have coalesce
_disabled_ (i.e. no rules for coalesce) ?

- Kobi

Re: iw coalesce show - disabled at init?

[Johannes Berg]

On Wed, 2015-08-12 at 14:13 -0700, Kobi Cohen-Arazi wrote:
> 
> So what you're saying is that at startup, drivers should have 
> coalesce _disabled_ (i.e. no rules for coalesce) ?
> 

I think so, yes. It's a service that should be actively enabled, with
user-configured rules. Mechanism vs. policy, and all that ...

johannes

Re: iw coalesce show - disabled at init?

[Kobi Cohen-Arazi]

On Wed, Aug 12, 2015 at 2:28 PM, Johannes Berg
<johannes@sipsolutions.net> wrote:
> On Wed, 2015-08-12 at 14:13 -0700, Kobi Cohen-Arazi wrote:
>>
>> So what you're saying is that at startup, drivers should have
>> coalesce _disabled_ (i.e. no rules for coalesce) ?
>>
>
> I think so, yes. It's a service that should be actively enabled, with
> user-configured rules. Mechanism vs. policy, and all that ...

I think that in most (if not all) NICs, this specific service is
enabled by default.

Kobi