Re: Status of the RTL8712 staging driver

[Christian Lamparter <chunkeey@googlemail.com>]

On Tuesday, April 02, 2013 11:02:21 PM Larry Finger wrote:
> On 04/02/2013 12:41 PM, Christian Lamparter wrote:
> > On Tuesday, April 02, 2013 02:51:19 PM Juan Carlos Garza Fernandez wrote:
> >> Now I have another question regarding the 8712 driver which has been
> >> in staging forever.
> >>
> >> In your opinion, are there any chances that it will be improved to
> >> reach the mainline level?
> 
> Yes, r8712u has been in staging for a long time; however, it is not a matter of 
> improving it. For a new wireless driver to be included in the main wireless 
> directories, it must use mac80211. Making the change from the current softmac in 
> r8712u to a mac80211-based driver is essentially a complete rewrite., 
Oh, Joshua Roys had already investigated a mac80211 driver and abandoned
it since it is not possible:

"... Some differences from rtl8192se were eliminated, although all PHY/BB/RF
reg commands are still commented out.  *When any of those regs were accessed 
directly, the device would die*. When accessed through the IOCMD regs, the
firmware on the device would seem to disable automatic control of some 
important functions causing a large drop in receive sensitivity and general
performance." <http://marc.info/?l=linux-wireless&m=133644367402433>

The driver can't initialize the PHY and while the firmware has a command
to set the phy to a specified channel, you can't define the channel mode:
whenever it is supposed to be legacy, HT20, HT40+ or HT40-...

That's the reason why the new r92su now interfaces with cfg80211 instead
of mac80211. The *su firmware really forces you to use the "high-level"
commands it has for scan, connect, disconnect and join_ibss. 
(Furthermore, the scan/connect/... cmds also do the whole MLME work: probe,
(de-)auth, (dis-/re-)assoc and action for BA...  Unfortunately, mac80211
wants to do this a as well, so a mac80211 driver for rtl8192su would have
to discard all the mgmt frames from mac80211?!)   

> As r8712u functions quite well in station mode, I have not felt inclined to 
> rework it for AP mode, particularly when a user can obtain a functioning
> router/AP for about > $30. I feel that my time is better spent converting
> the drivers for Realtek devices that are not currently in the kernel. That
> list includes the rtl8723ae driver in 3.8, the rtl8188ee driver that will
> be in 3.10, and 3 other drivers that I currently have on my todo list.
Very true! (Is the RTL8812AU (Realtek's 11ac chip) also on the list?)

Regards,
	Christian