Re: [RFC] Move brcm80211 to mainline

["Henry Ptasinski" <henryp@broadcom.com>]

On Thu, Jul 07, 2011 at 07:58:22AM -0700, Greg KH wrote:
> On Thu, Jul 07, 2011 at 07:46:28AM -0700, Henry Ptasinski wrote:
> > On Wed, Jul 06, 2011 at 06:45:12PM -0700, Greg KH wrote:
> > > On Wed, Jul 06, 2011 at 08:58:11PM -0400, Pavel Roskin wrote:
> > > > On 07/06/2011 08:40 PM, Rafał Miłecki wrote:
> > > > >Short question, without commenting on brcm80211 code yet:
> > > > >
> > > > >Why should we want 2 mainline drivers for the same hardware?
> > > > 
> > > > Also, checkpatch.pl finds a lot of bad stuff:
> > > > 
> > > > total: 1 errors, 1505 warnings, 105454 lines checked
> > > > 
> > > > $ scripts/checkpatch.pl 0001-wireless-testing-add-brcm80211.patch >Log
> > > > $ egrep '^(WARNING|ERROR)' Log |sort |uniq -c
> > > >       1 ERROR: do not use assignment in if condition
> > > >      76 WARNING: braces {} are not necessary for any arm of this statement
> > > >     205 WARNING: braces {} are not necessary for single statement blocks
> > > >       4 WARNING: consider using a completion
> > > >      11 WARNING: consider using kstrto* in preference to simple_strtoul
> > > >      20 WARNING: do not add new typedefs
> > > >       5 WARNING: externs should be avoided in .c files
> > > >    1164 WARNING: line over 80 characters
> > > >      20 WARNING: Use of volatile is usually wrong: see
> > > 
> > > Almost all of these should be fixed, especially the volatile one, that's
> > > just horrible.
> > 
> > Fixes for these are in progress.  We'll keep posting patches to address this
> > cleanup in staging.
> 
> Ok, at the least these should be resolved before ever asking for the
> code to be moved out of staging.  Sorry for me not checking this
> previously, I wrongly assumed that you had finished this task already.

No worries.  We'll have almost all of these cleaned up shortly. The 80 char
lines will be nice and tedious, though, unless somebody can suggest a decent
tool that doesn't make the code unreadable.

> 
> > > Also, there's the whole big endian mess that needs to be fixed up before
> > > the driver can be moved out of staging.  Henry, what's the status of
> > > that?
> > 
> > We're looking at BE MIPS, and I sent some cards to Benjamin Herrenschmidt so he
> > can test on PPC platforms.
> 
> Don't forget just building the thing on other arches (SPARC for example)
> and what happens when you try to do that...

Building some cross-compiler toolchains now.  SPARC hardware to do anything
native with seems to be getting very scarce, and trying to fit a PCIe minicard
into one might be challenging ...

- Henry