On Tue, 2024-10-08 at 08:42 +0200, Greg Kroah-Hartman wrote: > On Mon, Oct 07, 2024 at 11:32:25PM +0200, Philipp Hortmann wrote: > > On 10/7/24 20:26, Johannes Berg wrote: > > > From: Johannes Berg<johannes.berg@xxxxxxxxx> > > > > > > This driver is using lib80211 and any driver that plans to ever > > > leave staging should never have done that, so remove the driver > > > to enable cleaning up lib80211 into libipw inside the old Intel > > > drivers. > > > > > > Signed-off-by: Johannes Berg<johannes.berg@xxxxxxxxx> > > > --- > > > > Hi Johannes, > > > > this driver is not only working, it is also supporting a form factor that is > > usable and is available on the used hardware market. > > > > Is it possible to migrate the driver from lib80211 to libipw? > > I don't think a "migration" is simple or easy, Indeed. In this case, as you can see in the driver, it already "vendors" much of what was the ieee80211 softmac stack years ago, and partially became libipw/lib80211 as "rtllib". I don't know why it even depends on lib80211 (which does crypto) if it has its own rtllib_crypt_* code, but I'm not sure I want to know either. I'm not very inclined to bother finding out, because the thing is, there's basically no way it will ever "graduate" from staging in this way. It's been around for just over 15 years (!), and nobody cared, and in all that time it has already been the case that it should be using mac80211, not ieee80211/lib80211. > I only know of one driver > that ever made that move and it was not easy. It's simpler to just take > the existing driver, hack it to pieces out-of-tree and then submit a new > driver using the correct apis. Indeed, or integrate with the existing realtek driver frameworks. > > It does not have to be you who is doing this. How urgent is it to remove > > this driver? > > No staging driver should ever get in the way of api changes elsewhere in > the kernel, that's one of the rules of this part of the tree. So from > my opinion, it's fine to delete it now. It can always come back in a > new way later on. Right. It's not per se urgent, we just found that lib80211.h didn't exist in MAINTAINERS, and then I realized that it didn't need to exist at all (since we threw out some ancient wifi drivers.) So no, it doesn't need to happen now, but also as Greg said staging shouldn't be in the way, and as I wrote above, nothing happened with this driver in 15 years. I'm not going to wait that long again with lib80211, and I see no reason why there should suddenly be activity on it now. And if suddenly and unexpectedly there is now any development interest in it, it could even be restored without the lib80211 dependency, as Greg said, though I doubt that'd be useful - more useful to integrate with existing code instead. johannes
