Re: [zd1211-devs] Mainline, firmware

[Guennadi L. <g.l...@gm...>]

On Sun, 4 Dec 2005, Mayne wrote:

> On Sun, 4 Dec 2005, Guennadi Liakhovetski wrote:
> 
> >>> 1. Is anybody considering / aiming at including the driver in the
> >>> mainline?
> >> Well, I don't. ;-)
> > Ehm, is it because you think it would be too much work or you don't think
> > it's worth it?
> 
> I think only a rewritten driver should/could make it into the kernel.
> Currently, the zd1211 driver's much too bloated and comes with dozens of
> strange hacks.

Well, you, certainly, know better. 

> I'm no kernel hacker but AFAICS most device drivers come as
> one (quite small) .c file.

Not necessarily 1, but mostly, and far not many manage it "quite small". 
What about zd1201, which is already in the kernel, are there any 
similarities? Any chance of code sharing?

> The zd1211 driver is far from that. However, if
> anyone wants to work on the driver's inclusion in the standard kernel,
> just go ahead and write patches :-).

Yeah, would be very nice:-) I am, certainly, not in a position to start 
this work / carry a significant part of it, but I would try to help as 
much as I can.

Unfortunately, I am no specialist in wireless networking, just heard there 
has recently been some work done on a generic 802.11 layer in the kernel, 
I think, it has even been included in the mainline. Could, probably, use 
that one too? However, as long as its GPL (thanks Zydas), the driver 
anyway can use everything, just cannot benefit from "automatic" updates 
of in-tree code.

> [Firmware loading patch not included in zd1211.ath.cx driver version]
> > Why not? do you see any technical problems / disadvantages, or, again,
> > just don't consider that important enough?
> 
> I see the disadvantage that that way we would require the users not only
> to download, compile, and install the driver, but also to setup hotplug
> and download/install the stick's firmware.
> 
> What would be the benefit of that? If we'd aim for including the patch in
> the kernel, sure, this patch would be necessary. Otherwise I don't see the
> point in it right now. Just provide this as a separate patch should be
> enough for distribution package maintainers.

Same as for other drivers: removing the code from the kernel that doesn't 
belong there, unification with other drivers.

Thanks
Guennadi
---
Guennadi Liakhovetski