[PATCH] staging:brcm80211:brcmfmac:add dongle power on/off functions

[Grant Grundler]

On Wed, Oct 20, 2010 at 9:04 PM, Greg KH <greg at kroah.com> wrote:
> On Wed, Oct 20, 2010 at 06:22:57PM -0700, nohee ko wrote:
>> Add dongle power on/off functions.
>> Power save feature is essential in wi-fi
>> fullmac model. And this feature was missing.
>
> Nice feature.
>
> Unfortunately this patch (and your second one) goes against the "no more
> features" rule I said, and you all agreed to, a few days ago.

Rules are meant to be broken. :P
Seriously, I'm only seeing downside to this rule and am missing the upside.

Broadcom knows staging-next isn't where they can "compete" with other
wireless device mfg's. Beating them more with a stick isn't going to help.

While it's annoying to see Broadcom not submit more cleanup stuff,
it would be more pragmatic to let Broadcom add features that require
specialized knowledge *whenever they can* and continue to accept
"cleanup" patches from everyone (*including* Broadcom). I haven't seen
any shortage of cleanup patches for this driver.

Also, four times now I've pulled staging-next-2.6 and tested brcm drivers
using autotest suite (available on git.chromium.org) created by Sam Leffler
and Paul Stewart.  NOT including new features just means it will be that
much longer before I (or anyone else) can regression test with those features.


> Also, the merge window for new stuff for .37 is now over.

staging-next-2.6 is subject to merge windows? Why?
Sorry - this must be a FAQ but google is failing to find a
"staging-next-2.6 FAQ".

> So, care to start work on fixing other stuff in the driver first?  Then
> I will be glad to take these two patches, but until then, I will not.

Is there a list of "other stuff" someplace?
If these are in previously posted code reviews, I can look for those.

Here is my own list:
o remove get_fs/get_ds() usage - I believe this is 2.4 left-overs.
o consistent use of "ndev" vs "dev" (ie struct netdevice)
o use linux/xxx.h, not asm/xxx.h (kernel janitor work)
o rework wl_debug_level setting (it's a mess of pointless accessor functions)
o delete include/pcicfg.h

I'm not working on any of these things at the moment - other firedrills. :(

thanks,
grant