Re: [PATCH] mmc: add SDHCI driver for STM platforms.

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 4 Aug 2010 07:35:20 +0200 Peppe CAVALLARO <peppe.cavallaro@st.com> wrote:

> Hi Andrew,
> 
> > -----Original Message-----
> > From: Andrew Morton [mailto:akpm@linux-foundation.org]
> > Sent: Wednesday, August 04, 2010 12:21 AM
> > To: Matt Fleming
> > Cc: Peppe CAVALLARO; linux-mmc@vger.kernel.org
> > Subject: Re: [PATCH] mmc: add SDHCI driver for STM platforms.
> > 
> > On Mon, 2 Aug 2010 15:36:31 +0100
> > Matt Fleming <matt@console-pimps.org> wrote:
> > 
> > > > > What is devm_stm_pad_claim()? I can't find this function in linus'
> > > > > tree, or in -next, or in Andrew's -mm series. Where is this function
> > > > > from?
> > > >
> > > > This comes from our STMicroelectronics platform pad management
> > > > infrastructure; code available from:
> > > > http://git.stlinux.com/?p=stm/linux-sh4-2.6.32.y.git;a=summary It's
> > > > used to claim/manage various resources i.e. PIO and clock lines.
> > >
> > > That's interesting. So, this driver won't build with linus' tree? I
> > > didn't think that the kernel catered for external interfaces like this
> > > - it causes all sorts of issues, not least of which is the fact that
> > > the driver can't be built or tested with the mainline kernel.
> > >
> > > Andrew, what do you make of this?
> > 
> > It definitely needs to be compileable.  Otherwise it's dead code to
> > everyone else and nobody can maintain it as they make tree-wide or
> > subsystem-wide changes.
> > 
> > Also I do think the code in mainline should be able to be excecutable
> > (and testable!) without external code dependencies.  This gets into the
> > realm of licensing concerns: is the out-of-tree code GPL?  If not then
> > people will make a big fuss.  If it _is_ GPL then get it merged up!
> 
> It is under GPL license and available from git.stlinux.com at this time.
> Please also consider that I'm going to maintain and update the sdhci-stm
> driver, currently on high priority in our side.
> 

So why not merge the sdhci-stm driver into mainline at the same time as
"mmc: add SDHCI driver for STM platforms"?  Then all of this won't be an
issue at all.