Re: rt73 driver trouble
- From: Dan Williams <dcbw redhat com>
- To: Jouni Malinen <j w1 fi>
- Cc: hostap shmoo com, Lamarque Vieira Souza <lamarque syst com br>, networkmanager-list gnome org
- Subject: Re: rt73 driver trouble
- Date: Wed, 04 Apr 2007 07:51:15 -0400
On Tue, 2007-04-03 at 19:32 -0700, Jouni Malinen wrote:
> On Tue, Apr 03, 2007 at 09:35:50PM -0400, Dan Williams wrote:
> > On Tue, 2007-04-03 at 15:52 -0300, Lamarque Vieira Souza wrote:
> > > I managed to make NetworkManager to work with rt73 driver. The two
> > > patches I used are attached. The wpa_supplicant-0.4.8-ralink.diff
> > > patch is just a patch form of the files supplied by Ralink in
> > > directory RT73_Linux_STA_Drv184.108.40.206/WPA_Supplicant.
> > Custom wpa_supplicant drivers are not the answer. The answer is to fix
> > the driver to work correctly with the WEXT driver. That's why
> > NetworkManager _only_ uses the WEXT driver with wpa_supplicant, because
> > there needs to be standardization on _one_ interface so that we don't
> > need to re-write stuff 50 times for each driver.
> While I agree with the part that drivers should really use WEXT, my
> policy on driver specific interface implementations for wpa_supplicant
> is quite a bit more open than the one for NM. However..
Along the same lines, I'm quite looking forward to mac80211/cfg80211
because there will be _one_ interface to a large bloc of drivers. If
you're aggressive about pushing new features and such back upstream to
cfg80211 and the stack so that driver needs don't get so far ahead of
configuration API like with WEXT, then that just means less work for you
in the long run.
Part of the problem is that since there's no good configuration
interface on Linux quite yet, vendors have no good reason to support the
common interface that does exist, I understand that. But without a
little bit of pushback, nothing will change.
For example, the rt73 example used _obviously_ doesn't support WE-18 and
later since it uses iwpriv for WPA. That's just wrong and I hope you
wouldn't accept a driver that has duplicated the interface that's
"standard" and already exists on Linux.
> > > Do you think the wpa_supplicant-0.4.8-ralink.diff can be applied to
> > > main wpa_supplicant? There is not indication of which license the
> > > code in RT73_Linux_STA_Drv220.127.116.11/WPA_Supplicant uses, the code in
> > > RT73_Linux_STA_Drv18.104.22.168/Module is GPL.
> This is an issue. I will not apply any random patch from a vendor that
> someone else sends to me, especially so, if there is no clear license
> statement or the change is licensed only under GPL. In other words, if
> someone really wants to get this patch into wpa_supplicant, that someone
> will need to talk with the copyright holder and confirm that this patch
> is available under a compatible (both GPLv2 and BSD) license. In many
> cases, it may be easier to just get the driver converted to support
> WE-18 for WPA configuration..
] [Thread Prev