Pavel Machek wrote:

>Hi!
>
>
>>Anyway, our rebasing from ieee80211 up to the latest development tip is
>>done across 29 commits, with a series size of 225k.
>>
>
>Are there any plans to fix "compiled into kernel" case? It works okay
>when modular, but when compiled into kernel, it fails to load
>firmware....
>
I assume you are referring to ipw2{1,2}00 and not the ieee80211 subsystem?

When last we had moved the firmware loader to outside of the netdev init
callback, we had problems with distros not bringing up the interface
correctly (it wasn't being opened until after carrier was detected...
which meant it couldn't be configured because it hadn't been brought up,
and so hadn't had the firmware loaded)

We can put together a patch for ipw2200-1.0.7 / ipw2100-1.1.3 that
changes this behavior in order to determine if there are still problems
in user space.   I would be hesitant to incorporate into upstream or
anywhere that masses may use it until we have agreed on what the
configuration flow for user space is for network devices.  Once we have
agreed what user space should do, we can break compatibility.

The order required of user space is:

   kernel                         hotplug               hotplug script
   -----------------------        --------------        -----------
1. module load
2. netdev device registered
3.                                new device
4.                                                      ifconfig up
5. open: load firmware,
6. init device, etc.
7.                                                      configure wireless
8. scan
9. associate
A.
link                                                                            
        

B.                                                      carrier detected
C.                                                      configure link
(dhcp)

When last we looked at this, step 4 was being handled by ifplugd only
after IFF_RUNNING or carrier_detect indicated link, which with wireless
is potentially not possible until after step 7.

Changing the driver requires moving the firmware load path back to the
open handler, and extracting the MAC detection code into the probe
function (independent of fw loading) so that the MAC is available during
the hotplug scripts for use with ifrename.

Although I believe some hardware can't detect MAC until after the fw is
loaded.  So, we might want to require that hotplug scripts not attempt
to rename the interface until after it has been opened.  The rub there
being that (IIRC) ifrename requires the interface to be down in order to
rename it...

James



-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to