[RFC PATCH 01/13] ARM: davinci: move private EDMA API to arm/common

Matt Porter mporter at ti.com
Sat Sep 22 04:34:46 EST 2012


On Fri, Sep 21, 2012 at 10:42:05AM +0100, Russell King wrote:
> On Fri, Sep 21, 2012 at 09:33:42AM +0000, Hebbar, Gururaja wrote:
> > On Fri, Sep 21, 2012 at 14:59:23, Russell King - ARM Linux wrote:
> > > On Thu, Sep 20, 2012 at 10:43:34AM -0400, Matt Porter wrote:
> > > > Move mach-davinci/dma.c to common/edma.c so it can be used
> > > > by OMAP (specifically AM33xx atm) as well. This just moves
> > > > the private EDMA API but does not support OMAP.
> > > > 
> > > > Signed-off-by: Matt Porter <mporter at ti.com>
> > > > ---
> > > >  arch/arm/Kconfig                           |    1 +
> > > >  arch/arm/common/Kconfig                    |    3 +
> > > >  arch/arm/common/Makefile                   |    1 +
> > > >  arch/arm/common/edma.c                     | 1588 ++++++++++++++++++++++++++++
> > > >  arch/arm/include/asm/mach/edma.h           |  267 +++++
> > > 
> > > asm/mach should not be used as a dumping ground for platform header files.
> > > It is there to provide the interfaces between generic ARM architecture
> > > code and platform code.  (At least four files that are there at the
> > > moment need to be moved out of there - patch series to follow...)
> > 
> > Can this be moved to include/linux/platform_data/ ?
> 
> Here's the pertinant question: "is it platform data?"  Looking at the
> file, it appears to be internal data structures and register definitions
> for the driver itself.  Therefore, it isn't platform data, and it
> shouldn't be living separately from the driver.
> 
> If the driver itself only makes use of the data structures, the data
> structures should be defined either within the driver, or a header file
> co-located next to the driver itself.  The same goes for register
> definitions too.
> 
> The only structure that I can find which isn't internal to the driver
> is struct edma_soc_info, struct edma_rsv_info, and the enum dma_event_q.
> Those can go to include/linux/platform_data, but the rest should not.

Ok, but is it ok to keep the actual private EDMA API portion in
arch/arm/include/asm/mach/? It's not a problem to move the internal
portions to a local include and that pdata to the appropriate place.
We still need a place independent of mach-davinci and mach-omap2 to
keep that portion of the include. I suppose it could be put in with
the dmaengine wrapper's include/linux/edma.h but I hate to clutter
that up when the private API will go away later.

-Matt


More information about the devicetree-discuss mailing list