| Message ID | 1343393162-11938-14-git-send-email-lee.jones@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Fri, Jul 27, 2012 at 01:45:57PM +0100, Lee Jones wrote: > Nothing special here. We're only providing a compatible string > to ensure the driver is probed using a Device Tree boot. > + pcm: ux500-pcm { > + compatible = "stericsson,ux500-pcm"; > + }; > + One of two things is wrong here. Either you've not provided any information about the hardware so the driver won't be able to work out what to talk to or you've not provided any data so you're registering a purely virtual Linux-internal device via the device tree. Once more, the idea here is to describe the hardware not to dump Linux's internal data structures into the device tree.
On 29/07/12 21:50, Mark Brown wrote: > On Fri, Jul 27, 2012 at 01:45:57PM +0100, Lee Jones wrote: >> Nothing special here. We're only providing a compatible string >> to ensure the driver is probed using a Device Tree boot. > >> + pcm: ux500-pcm { >> + compatible = "stericsson,ux500-pcm"; >> + }; >> + > > One of two things is wrong here. Either you've not provided any > information about the hardware so the driver won't be able to work out > what to talk to or you've not provided any data so you're registering a > purely virtual Linux-internal device via the device tree. > > Once more, the idea here is to describe the hardware not to dump Linux's > internal data structures into the device tree. We use it register the driver for probe()ing. In the same way we do for the PMU and Regulators. All three of which actually belong to a different hardware block. Do you know of a better way to register those devices?
On Mon, Jul 30, 2012 at 08:32:43AM +0100, Lee Jones wrote: > On 29/07/12 21:50, Mark Brown wrote: > >On Fri, Jul 27, 2012 at 01:45:57PM +0100, Lee Jones wrote: > >>+ pcm: ux500-pcm { > >>+ compatible = "stericsson,ux500-pcm"; > >>+ }; > >>+ > >One of two things is wrong here. Either you've not provided any > >information about the hardware so the driver won't be able to work out > >what to talk to or you've not provided any data so you're registering a > >purely virtual Linux-internal device via the device tree. > >Once more, the idea here is to describe the hardware not to dump Linux's > >internal data structures into the device tree. > We use it register the driver for probe()ing. In the same way we do > for the PMU and Regulators. All three of which actually belong to a > different hardware block. Do you know of a better way to register > those devices? You've not identified which of the two cases above we're in... assuming it's a virtal device look at how the DT bindings for the other platforms are doing things.
On 30/07/12 14:40, Mark Brown wrote: > On Mon, Jul 30, 2012 at 08:32:43AM +0100, Lee Jones wrote: >> On 29/07/12 21:50, Mark Brown wrote: >>> On Fri, Jul 27, 2012 at 01:45:57PM +0100, Lee Jones wrote: > >>>> + pcm: ux500-pcm { >>>> + compatible = "stericsson,ux500-pcm"; >>>> + }; >>>> + > >>> One of two things is wrong here. Either you've not provided any >>> information about the hardware so the driver won't be able to work out >>> what to talk to or you've not provided any data so you're registering a >>> purely virtual Linux-internal device via the device tree. > >>> Once more, the idea here is to describe the hardware not to dump Linux's >>> internal data structures into the device tree. > >> We use it register the driver for probe()ing. In the same way we do >> for the PMU and Regulators. All three of which actually belong to a >> different hardware block. Do you know of a better way to register >> those devices? > > You've not identified which of the two cases above we're in... assuming > it's a virtal device look at how the DT bindings for the other platforms > are doing things. I thought it would have been obvious by my response. The PMU seems to be a virtual device with no registers or IRQs. I've found a DT which seems to do it in the same way as I do 'arch/powerpc/boot/dts/p1022ds.dts', but some others call PCM registration from the CPU-side DAI probe(). Would that be your preference?
On Mon, Jul 30, 2012 at 03:26:58PM +0100, Lee Jones wrote: > On 30/07/12 14:40, Mark Brown wrote: > >>We use it register the driver for probe()ing. In the same way we do > >>for the PMU and Regulators. All three of which actually belong to a > >>different hardware block. Do you know of a better way to register > >>those devices? > >You've not identified which of the two cases above we're in... assuming > >it's a virtal device look at how the DT bindings for the other platforms > >are doing things. > I thought it would have been obvious by my response. The PMU seems > to be a virtual device with no registers or IRQs. That's not at all obvious, no. > I've found a DT which seems to do it in the same way as I do > 'arch/powerpc/boot/dts/p1022ds.dts', but some others call PCM > registration from the CPU-side DAI probe(). Would that be your > preference? Yes, do what the other drivers are doing
diff --git a/arch/arm/boot/dts/db8500.dtsi b/arch/arm/boot/dts/db8500.dtsi index a3bee0a..a3fb05f 100644 --- a/arch/arm/boot/dts/db8500.dtsi +++ b/arch/arm/boot/dts/db8500.dtsi @@ -591,6 +591,10 @@ status = "disabled"; }; + pcm: ux500-pcm { + compatible = "stericsson,ux500-pcm"; + }; + external-bus@50000000 { compatible = "simple-bus"; reg = <0x50000000 0x4000000>;
Nothing special here. We're only providing a compatible string to ensure the driver is probed using a Device Tree boot. Signed-off-by: Lee Jones <lee.jones@linaro.org> --- arch/arm/boot/dts/db8500.dtsi | 4 ++++ 1 file changed, 4 insertions(+)