Re: [RFC V2] cfg80211: introduce critical protocol indication from user-space

[Dan Williams <dcbw@redhat.com>]

On Thu, 2013-03-28 at 16:30 -0700, Ben Greear wrote:
> On 03/28/2013 04:01 PM, Dan Williams wrote:
> > On Thu, 2013-03-28 at 23:44 +0100, Johannes Berg wrote:
> >> On Thu, 2013-03-28 at 17:42 -0500, Dan Williams wrote:
> >>
> >>>> Well, you can do DHCP a second or so, I'd think? And EAPOL much quicker,
> >>>> of course. I don't really see any reasonable minimum time? We might want
> >>>> to enforce a max though, maybe.
> >>>
> >>> Not quite.  A lot is dependent on the server itself, and I've had users
> >>> on university and corporate networks report it sometimes takes 30 to 60
> >>> seconds for the whole DHCP transaction to complete (DISCOVER, REQUEST,
> >>> OFFER, ACK).  Sometimes there's a NAK in there if the server doesn't
> >>> like your lease, which means you need another round-trip.  So in many
> >>> cases, it's a couple round-trips and each of these packets may or may
> >>> not get lost in noisy environments.
> >>
> >> Oh, yes, of course. However, we're talking about optimising the good
> >> cases, not the bad ones. Think of it this way: if it goes fast, we
> >> shouldn't make it slow by putting things like powersave or similar in
> >> the way. If it's slow, then it'll still work, just slower. But when
> >> "slower" only means a few hundred milliseconds, it doesn't matter if
> >> everything takes forever (30-60 secs)
> >
> > True, but at least 4 or 5 seconds is the minimum time I'd recommend here
> > for DHCP.
> 
> Couldn't dhcp just turn off the critical protection as soon as it is done?
> 
> Then, you only need to worry about the max time allowed.

Yes, that's really what I meant.  4 - 5 seconds is the "best worst-case
scenario", clearly when a lease is acquired the critical protection
would be turned off by the connection manager.

But if something doesn't turn it off, and the 802.11 stack needs a
timeout value, I would suggest 4 or 5 seconds for that.

Dan

> Also, you would probably need to enforce in the kernel that only
> x out of y time in any given period can be locked, otherwise lots
> of different dhclient processes (perhaps erroneously spawned..or
> running on lots of different VIFs) could basically disable scanning
> or channel changes...
> 
> Thanks,
> Ben
>