[RFC PATCH v2 03/13] arm/tegra: board-dt: Add AUXDATA for tegra-gpio and tegra-pinmux
Stephen Warren
swarren at nvidia.com
Wed Aug 17 06:24:00 EST 2011
Shawn Guo wrote at Monday, August 15, 2011 9:31 PM:
> On Mon, Aug 15, 2011 at 02:28:10PM -0600, Stephen Warren wrote:
> > Add AUXDATA so that Tegra GPIO and pinmux devices that are instantiated
> > from device-tree are named consistently with the static platform devices
> > in mach-tegra/devices.c. Note that the device-tree file tegra20.dtsi
> > already includes a node for the Tegra GPIO controller, whereas the node
> > for the Tegra pinmux controller will be added in a subsequent commit.
> >
> > Signed-off-by: Stephen Warren <swarren at nvidia.com>
> > ---
> > arch/arm/mach-tegra/board-dt.c | 2 ++
> > 1 files changed, 2 insertions(+), 0 deletions(-)
> >
> > diff --git a/arch/arm/mach-tegra/board-dt.c b/arch/arm/mach-tegra/board-dt.c
> > index 9f47e04..1995b79 100644
> > --- a/arch/arm/mach-tegra/board-dt.c
> > +++ b/arch/arm/mach-tegra/board-dt.c
> > @@ -50,6 +50,8 @@ void seaboard_pinmux_init(void);
> >
> >
> > struct of_dev_auxdata tegra20_auxdata_lookup[] __initdata = {
> > + OF_DEV_AUXDATA("nvidia,tegra20-gpio", TEGRA_GPIO_BASE, "tegra-gpio", NULL),
> > + OF_DEV_AUXDATA("nvidia,tegra20-pinmux", TEGRA_APB_MISC_BASE, "tegra-pinmux", NULL),
> > OF_DEV_AUXDATA("nvidia,tegra20-sdhci", TEGRA_SDMMC1_BASE, "sdhci-tegra.0", NULL),
> > OF_DEV_AUXDATA("nvidia,tegra20-sdhci", TEGRA_SDMMC2_BASE, "sdhci-tegra.1", NULL),
> > OF_DEV_AUXDATA("nvidia,tegra20-sdhci", TEGRA_SDMMC3_BASE, "sdhci-tegra.2", NULL),
> >
> The whole auxdata table was primarily added to keep clk API working
> with the device name looking up. Your gpio and pinmux can still work
> without this patch, right? If that's case, I do not think we need to
> have this patch just for consistent device name between dt and non-dt.
> They are not and they do not have to be.
Indeed, everything does work fine without the AUXDATA; there's no
dependence on any clock lookups or similar in the new drivers.
Initially, I thought consistency in device naming would be important,
since Tegra works with and without device-tree, but I guess we'll be able
to turn off the hard-coded board files soon, so device name consistency
probably isn't too relevant.
I'll drop this patch.
--
nvpublic
More information about the devicetree-discuss
mailing list