Re: [PATCH v2 05/15] DMA: shdma: pass SoC-specific configuration to the driver via OF matching

[Guennadi Liakhovetski <g.liakhovetski@gmx.de>]

On Mon, 22 Jul 2013, Laurent Pinchart wrote:

> Hi Guennadi,
> 
> Thanks for the patch.
> 
> On Friday 19 July 2013 18:29:30 Guennadi Liakhovetski wrote:
> > Similar to the non-DT case, this patch passes SoC-specific configuration
> > to the driver via device ID matching, instead of platform data.
> > 
> > Signed-off-by: Guennadi Liakhovetski <g.liakhovetski+renesas@gmail.com>
> > ---
> > 
> > v2: adjust spacing within array definitions to keep a uniform style.
> > 
> >  Documentation/devicetree/bindings/dma/shdma.txt |    7 +++++--
> >  drivers/dma/sh/shdmac.c                         |   22 +++++++++++++-------
> >  2 files changed, 20 insertions(+), 9 deletions(-)
> > 
> > diff --git a/Documentation/devicetree/bindings/dma/shdma.txt
> > b/Documentation/devicetree/bindings/dma/shdma.txt index c15994a..7702e35
> > 100644
> > --- a/Documentation/devicetree/bindings/dma/shdma.txt
> > +++ b/Documentation/devicetree/bindings/dma/shdma.txt
> > @@ -22,7 +22,10 @@ Optional properties (currently unused):
> >  * DMA controller
> > 
> >  Required properties:
> > -- compatible:	should be "renesas,shdma"
> > +- compatible:	should be one of
> > +		"renesas,shdma-r8a73a4" for the system DMAC on r8a73a4 SoC
> > +		"renesas,shdma-r8a7740" for the DMACs (not RTDMAC) on r8a7740
> > +		"renesas,shdma" for a generic DMAC
> > 
> >  Example:
> >  	dmac: dma-mux0 {
> > @@ -36,7 +39,7 @@ Example:
> >  		ranges;
> > 
> >  		dma0: shdma@fe008020 {
> > -			compatible = "renesas,shdma";
> > +			compatible = "renesas,shdma-r8a7740";
> >  			reg = <0xfe008020 0x270>,
> >  				<0xfe009000 0xc>;
> >  			interrupt-parent = <&gic>;
> > diff --git a/drivers/dma/sh/shdmac.c b/drivers/dma/sh/shdmac.c
> > index 859ddbe..f80543c 100644
> > --- a/drivers/dma/sh/shdmac.c
> > +++ b/drivers/dma/sh/shdmac.c
> > @@ -20,6 +20,8 @@
> > 
> >  #include <linux/init.h>
> >  #include <linux/module.h>
> > +#include <linux/of.h>
> > +#include <linux/of_device.h>
> >  #include <linux/slab.h>
> >  #include <linux/interrupt.h>
> >  #include <linux/dmaengine.h>
> > @@ -663,6 +665,14 @@ static const struct shdma_ops sh_dmae_shdma_ops = {
> >  	.get_partial = sh_dmae_get_partial,
> >  };
> > 
> > +static const struct of_device_id sh_dmae_of_match[] = {
> > +	{.compatible = "renesas,shdma",},
> > +	{.compatible = "renesas,shdma-r8a73a4", .data = r8a73a4_shdma_devid,},
> > +	{.compatible = "renesas,shdma-r8a7740", .data = r8a7740_shdma_devid,},
> 
> Nit-picking here, OF device ID entries are usually ordered from most specific 
> to most generic compatible strings. It's up to you.

Yeah, in fact, I'm not even sure we need the generic line, so far I think 
we always need SoC configuration. Maybe I shall just remove it. Could do 
in an incremental patch.

> > +	{}
> > +};
> > +MODULE_DEVICE_TABLE(of, sh_dmae_of_match);
> 
> Shouldn't you guard the table with #ifdef CONFIG_OF ? If the driver can only 
> be used on ARM platforms it might not be worth it, but if it can be used on SH 
> as well that would make sense.

Well, it is a pretty common practice, I admit, but what exactly does it 
bring us apart from saving a couple of bytes? I just tried a SuperH build 
- no problem.

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/