diff mbox

[1/2] omap2+: add drm device

Message ID 1326487320-8781-1-git-send-email-rob.clark@linaro.org
State New
Headers show

Commit Message

Rob Clark Jan. 13, 2012, 8:41 p.m. UTC
From: Rob Clark <rob@ti.com>

Register OMAP DRM/KMS platform device, and reserve a CMA region for
the device to use for buffer allocation.

v1: initial patch
v2: move platform data structs into plat-omap to avoid having to
    #include headers from drivers/staging and cleanups

Signed-off-by: Rob Clark <rob@ti.com>
---
Note: after applying this patch there will be duplicate copies of the
platform data structs (until the 2nd patch is applied).  But I tested
to ensure this does not cause build breaks.  So the 2nd patch which
should go thru staging tree is safe to be held until this patch hits
Linus's master branch.

 arch/arm/plat-omap/Makefile           |    2 +-
 arch/arm/plat-omap/common.c           |    3 +-
 arch/arm/plat-omap/drm.c              |   83 +++++++++++++++++++++++++++++++++
 arch/arm/plat-omap/include/plat/drm.h |   70 +++++++++++++++++++++++++++
 4 files changed, 156 insertions(+), 2 deletions(-)
 create mode 100644 arch/arm/plat-omap/drm.c
 create mode 100644 arch/arm/plat-omap/include/plat/drm.h

Comments

Felipe Contreras Jan. 13, 2012, 8:59 p.m. UTC | #1
On Fri, Jan 13, 2012 at 10:41 PM, Rob Clark <rob.clark@linaro.org> wrote:
> diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
> index 9a58461..b86e6cb 100644
> --- a/arch/arm/plat-omap/Makefile
> +++ b/arch/arm/plat-omap/Makefile
> @@ -4,7 +4,7 @@
>
>  # Common support
>  obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
> -        usb.o fb.o counter_32k.o
> +        usb.o fb.o counter_32k.o drm.o

Should be something like this:
obj-$(CONFIG_DRM_OMAP) += drm.o

No?
Rob Clark Jan. 13, 2012, 9:04 p.m. UTC | #2
On Fri, Jan 13, 2012 at 2:59 PM, Felipe Contreras
<felipe.contreras@gmail.com> wrote:
> On Fri, Jan 13, 2012 at 10:41 PM, Rob Clark <rob.clark@linaro.org> wrote:
>> diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
>> index 9a58461..b86e6cb 100644
>> --- a/arch/arm/plat-omap/Makefile
>> +++ b/arch/arm/plat-omap/Makefile
>> @@ -4,7 +4,7 @@
>>
>>  # Common support
>>  obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
>> -        usb.o fb.o counter_32k.o
>> +        usb.o fb.o counter_32k.o drm.o
>
> Should be something like this:
> obj-$(CONFIG_DRM_OMAP) += drm.o
>
> No?

I think it would work either way.  Currently drm.c would be empty if
the driver isn't enabled in the kconfig.. if it is preferred to do
that in the Makefile, I could change it.

BR,
-R

> --
> Felipe Contreras
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/dri-devel
Rob Clark Jan. 13, 2012, 9:19 p.m. UTC | #3
On Fri, Jan 13, 2012 at 2:59 PM, Felipe Contreras
<felipe.contreras@gmail.com> wrote:
> On Fri, Jan 13, 2012 at 10:41 PM, Rob Clark <rob.clark@linaro.org> wrote:
>> diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
>> index 9a58461..b86e6cb 100644
>> --- a/arch/arm/plat-omap/Makefile
>> +++ b/arch/arm/plat-omap/Makefile
>> @@ -4,7 +4,7 @@
>>
>>  # Common support
>>  obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
>> -        usb.o fb.o counter_32k.o
>> +        usb.o fb.o counter_32k.o drm.o
>
> Should be something like this:
> obj-$(CONFIG_DRM_OMAP) += drm.o

btw, how does that work if CONFIG_DRM_OMAP=m?  Do you end up w/ a
plat-omap module?

BR,
-R

> No?
>
> --
> Felipe Contreras
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/dri-devel
Felipe Contreras Jan. 16, 2012, 2:12 p.m. UTC | #4
On Fri, Jan 13, 2012 at 11:19 PM, Rob Clark <rob.clark@linaro.org> wrote:
> On Fri, Jan 13, 2012 at 2:59 PM, Felipe Contreras
> <felipe.contreras@gmail.com> wrote:
>> On Fri, Jan 13, 2012 at 10:41 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>> diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
>>> index 9a58461..b86e6cb 100644
>>> --- a/arch/arm/plat-omap/Makefile
>>> +++ b/arch/arm/plat-omap/Makefile
>>> @@ -4,7 +4,7 @@
>>>
>>>  # Common support
>>>  obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
>>> -        usb.o fb.o counter_32k.o
>>> +        usb.o fb.o counter_32k.o drm.o
>>
>> Should be something like this:
>> obj-$(CONFIG_DRM_OMAP) += drm.o
>
> btw, how does that work if CONFIG_DRM_OMAP=m?  Do you end up w/ a
> plat-omap module?

Yes, and platform drivers are supposed to be loaded automatically at
boot-time by udev (or similar).
Rob Clark Jan. 16, 2012, 4:37 p.m. UTC | #5
On Mon, Jan 16, 2012 at 8:12 AM, Felipe Contreras
<felipe.contreras@gmail.com> wrote:
> On Fri, Jan 13, 2012 at 11:19 PM, Rob Clark <rob.clark@linaro.org> wrote:
>> On Fri, Jan 13, 2012 at 2:59 PM, Felipe Contreras
>> <felipe.contreras@gmail.com> wrote:
>>> On Fri, Jan 13, 2012 at 10:41 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>> diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
>>>> index 9a58461..b86e6cb 100644
>>>> --- a/arch/arm/plat-omap/Makefile
>>>> +++ b/arch/arm/plat-omap/Makefile
>>>> @@ -4,7 +4,7 @@
>>>>
>>>>  # Common support
>>>>  obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
>>>> -        usb.o fb.o counter_32k.o
>>>> +        usb.o fb.o counter_32k.o drm.o
>>>
>>> Should be something like this:
>>> obj-$(CONFIG_DRM_OMAP) += drm.o
>>
>> btw, how does that work if CONFIG_DRM_OMAP=m?  Do you end up w/ a
>> plat-omap module?
>
> Yes, and platform drivers are supposed to be loaded automatically at
> boot-time by udev (or similar).

oh, but this won't work, because common.c has to call
omapdrm_reserve_vram() (similar to how omapfb_reserve_sdram_memblock()
works).. so I think it has to stay the way it is in this patch.

I guess this is the reason that omapfb (fb.c is the example I copied)
device is setup in the same way.

BR,
-R

> --
> Felipe Contreras
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/dri-devel
Felipe Contreras Jan. 16, 2012, 4:59 p.m. UTC | #6
On Mon, Jan 16, 2012 at 6:37 PM, Rob Clark <rob.clark@linaro.org> wrote:
> On Mon, Jan 16, 2012 at 8:12 AM, Felipe Contreras
> <felipe.contreras@gmail.com> wrote:
>> On Fri, Jan 13, 2012 at 11:19 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>> On Fri, Jan 13, 2012 at 2:59 PM, Felipe Contreras
>>> <felipe.contreras@gmail.com> wrote:
>>>> On Fri, Jan 13, 2012 at 10:41 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>> diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
>>>>> index 9a58461..b86e6cb 100644
>>>>> --- a/arch/arm/plat-omap/Makefile
>>>>> +++ b/arch/arm/plat-omap/Makefile
>>>>> @@ -4,7 +4,7 @@
>>>>>
>>>>>  # Common support
>>>>>  obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
>>>>> -        usb.o fb.o counter_32k.o
>>>>> +        usb.o fb.o counter_32k.o drm.o
>>>>
>>>> Should be something like this:
>>>> obj-$(CONFIG_DRM_OMAP) += drm.o
>>>
>>> btw, how does that work if CONFIG_DRM_OMAP=m?  Do you end up w/ a
>>> plat-omap module?
>>
>> Yes, and platform drivers are supposed to be loaded automatically at
>> boot-time by udev (or similar).
>
> oh, but this won't work, because common.c has to call
> omapdrm_reserve_vram() (similar to how omapfb_reserve_sdram_memblock()
> works).. so I think it has to stay the way it is in this patch.

#if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
extern void omapdrm_reserve_vram(void);
#else
static inline void omapdrm_reserve_vram(void) { }
#endif

Like how it's done with DSP stuff.
Rob Clark Jan. 16, 2012, 5:01 p.m. UTC | #7
On Mon, Jan 16, 2012 at 10:59 AM, Felipe Contreras
<felipe.contreras@gmail.com> wrote:
> On Mon, Jan 16, 2012 at 6:37 PM, Rob Clark <rob.clark@linaro.org> wrote:
>> On Mon, Jan 16, 2012 at 8:12 AM, Felipe Contreras
>> <felipe.contreras@gmail.com> wrote:
>>> On Fri, Jan 13, 2012 at 11:19 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>> On Fri, Jan 13, 2012 at 2:59 PM, Felipe Contreras
>>>> <felipe.contreras@gmail.com> wrote:
>>>>> On Fri, Jan 13, 2012 at 10:41 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>>> diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
>>>>>> index 9a58461..b86e6cb 100644
>>>>>> --- a/arch/arm/plat-omap/Makefile
>>>>>> +++ b/arch/arm/plat-omap/Makefile
>>>>>> @@ -4,7 +4,7 @@
>>>>>>
>>>>>>  # Common support
>>>>>>  obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
>>>>>> -        usb.o fb.o counter_32k.o
>>>>>> +        usb.o fb.o counter_32k.o drm.o
>>>>>
>>>>> Should be something like this:
>>>>> obj-$(CONFIG_DRM_OMAP) += drm.o
>>>>
>>>> btw, how does that work if CONFIG_DRM_OMAP=m?  Do you end up w/ a
>>>> plat-omap module?
>>>
>>> Yes, and platform drivers are supposed to be loaded automatically at
>>> boot-time by udev (or similar).
>>
>> oh, but this won't work, because common.c has to call
>> omapdrm_reserve_vram() (similar to how omapfb_reserve_sdram_memblock()
>> works).. so I think it has to stay the way it is in this patch.
>
> #if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
> extern void omapdrm_reserve_vram(void);
> #else
> static inline void omapdrm_reserve_vram(void) { }
> #endif
>
> Like how it's done with DSP stuff.

right, but then you'd miss the omapdrm_reserve_vram() call at startup..

BR,
-R

> --
> Felipe Contreras
> --
> To unsubscribe from this list: send the line "unsubscribe linux-omap" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
Felipe Contreras Jan. 16, 2012, 8:37 p.m. UTC | #8
On Mon, Jan 16, 2012 at 7:01 PM, Rob Clark <rob.clark@linaro.org> wrote:
> On Mon, Jan 16, 2012 at 10:59 AM, Felipe Contreras
> <felipe.contreras@gmail.com> wrote:
>> On Mon, Jan 16, 2012 at 6:37 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>> On Mon, Jan 16, 2012 at 8:12 AM, Felipe Contreras
>>> <felipe.contreras@gmail.com> wrote:
>>>> On Fri, Jan 13, 2012 at 11:19 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>> On Fri, Jan 13, 2012 at 2:59 PM, Felipe Contreras
>>>>> <felipe.contreras@gmail.com> wrote:
>>>>>> On Fri, Jan 13, 2012 at 10:41 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>>>> diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
>>>>>>> index 9a58461..b86e6cb 100644
>>>>>>> --- a/arch/arm/plat-omap/Makefile
>>>>>>> +++ b/arch/arm/plat-omap/Makefile
>>>>>>> @@ -4,7 +4,7 @@
>>>>>>>
>>>>>>>  # Common support
>>>>>>>  obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
>>>>>>> -        usb.o fb.o counter_32k.o
>>>>>>> +        usb.o fb.o counter_32k.o drm.o
>>>>>>
>>>>>> Should be something like this:
>>>>>> obj-$(CONFIG_DRM_OMAP) += drm.o
>>>>>
>>>>> btw, how does that work if CONFIG_DRM_OMAP=m?  Do you end up w/ a
>>>>> plat-omap module?
>>>>
>>>> Yes, and platform drivers are supposed to be loaded automatically at
>>>> boot-time by udev (or similar).
>>>
>>> oh, but this won't work, because common.c has to call
>>> omapdrm_reserve_vram() (similar to how omapfb_reserve_sdram_memblock()
>>> works).. so I think it has to stay the way it is in this patch.
>>
>> #if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
>> extern void omapdrm_reserve_vram(void);
>> #else
>> static inline void omapdrm_reserve_vram(void) { }
>> #endif
>>
>> Like how it's done with DSP stuff.
>
> right, but then you'd miss the omapdrm_reserve_vram() call at startup..

Why?
Rob Clark Jan. 16, 2012, 9:24 p.m. UTC | #9
On Mon, Jan 16, 2012 at 2:37 PM, Felipe Contreras
<felipe.contreras@gmail.com> wrote:
> On Mon, Jan 16, 2012 at 7:01 PM, Rob Clark <rob.clark@linaro.org> wrote:
>> On Mon, Jan 16, 2012 at 10:59 AM, Felipe Contreras
>> <felipe.contreras@gmail.com> wrote:
>>> On Mon, Jan 16, 2012 at 6:37 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>> On Mon, Jan 16, 2012 at 8:12 AM, Felipe Contreras
>>>> <felipe.contreras@gmail.com> wrote:
>>>>> On Fri, Jan 13, 2012 at 11:19 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>>> On Fri, Jan 13, 2012 at 2:59 PM, Felipe Contreras
>>>>>> <felipe.contreras@gmail.com> wrote:
>>>>>>> On Fri, Jan 13, 2012 at 10:41 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>>>>> diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
>>>>>>>> index 9a58461..b86e6cb 100644
>>>>>>>> --- a/arch/arm/plat-omap/Makefile
>>>>>>>> +++ b/arch/arm/plat-omap/Makefile
>>>>>>>> @@ -4,7 +4,7 @@
>>>>>>>>
>>>>>>>>  # Common support
>>>>>>>>  obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
>>>>>>>> -        usb.o fb.o counter_32k.o
>>>>>>>> +        usb.o fb.o counter_32k.o drm.o
>>>>>>>
>>>>>>> Should be something like this:
>>>>>>> obj-$(CONFIG_DRM_OMAP) += drm.o
>>>>>>
>>>>>> btw, how does that work if CONFIG_DRM_OMAP=m?  Do you end up w/ a
>>>>>> plat-omap module?
>>>>>
>>>>> Yes, and platform drivers are supposed to be loaded automatically at
>>>>> boot-time by udev (or similar).
>>>>
>>>> oh, but this won't work, because common.c has to call
>>>> omapdrm_reserve_vram() (similar to how omapfb_reserve_sdram_memblock()
>>>> works).. so I think it has to stay the way it is in this patch.
>>>
>>> #if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
>>> extern void omapdrm_reserve_vram(void);
>>> #else
>>> static inline void omapdrm_reserve_vram(void) { }
>>> #endif
>>>
>>> Like how it's done with DSP stuff.
>>
>> right, but then you'd miss the omapdrm_reserve_vram() call at startup..
>
> Why?

I guess drm.o is ending up in a module, but the code that calls that
(in common.c) is not in a module, so you get an unresolved symbol at
link

BR,
-R

> --
> Felipe Contreras
> --
> To unsubscribe from this list: send the line "unsubscribe linux-omap" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
Cousson, Benoit Jan. 23, 2012, 5:24 p.m. UTC | #10
Hi Rob,

On 1/13/2012 9:41 PM, Rob Clark wrote:
> From: Rob Clark<rob@ti.com>

[...]

> +static int __init omap_init_gpu(void)

Why is the function to init drm device is named gpu?

> +{
> +	struct omap_hwmod *oh = NULL;
> +
> +	/* lookup and populate the DMM information, if present - OMAP4+ */
> +	oh = omap_hwmod_lookup("dmm");
> +
> +	if (oh) {
> +		dmm_platdata.base = omap_hwmod_get_mpu_rt_va(oh);
> +		dmm_platdata.irq = oh->mpu_irqs[0].irq;

These are internal hwmod attributes that should not be retrieved here. 
They are included in the device resources and this is up to the driver 
to get them using the platform_get_resource.

> +
> +		if (dmm_platdata.base)
> +			omapdrm_platdata.dmm_pdata =&dmm_platdata;

pdata is supposed to be used for storing board level information, and we 
are in the process of removing all of them for device tree adaptation. 
So you should not use that at all in this case if this is not strictly 
needed.

> +	}
> +
> +	return platform_device_register(&omap_drm_device);

This is not the proper way to init a device nowadays.

If you want to take advantage of the pm functionality, you should create 
an omap_device.
Please have a look at the other OMAP devices creation code (GPIO, UART...).

> +}
> +
> +arch_initcall(omap_init_gpu);
> +
> +void omapdrm_reserve_vram(void)
> +{
> +#ifdef CONFIG_CMA
> +	/*
> +	 * Create private 32MiB contiguous memory area for omapdrm.0 device
> +	 * TODO revisit size.. if uc/wc buffers are allocated from CMA pages
> +	 * then the amount of memory we need goes up..
> +	 */
> +	dma_declare_contiguous(&omap_drm_device.dev, 32 * SZ_1M, 0, 0);
> +#else
> +#  warning "CMA is not enabled, there may be limitations about scanout buffer allocations on OMAP3 and earlier"
> +#endif
> +}
> +
> +#endif
> diff --git a/arch/arm/plat-omap/include/plat/drm.h b/arch/arm/plat-omap/include/plat/drm.h
> new file mode 100644
> index 0000000..e29be29
> --- /dev/null
> +++ b/arch/arm/plat-omap/include/plat/drm.h

Ideally, platform data should not exist anymore in a device tree world, 
but meanwhile there is a directory to store them:
include/linux/platform_data

> @@ -0,0 +1,70 @@
> +/*
> + * DRM/KMS device registration for TI OMAP platforms
> + *
> + * Copyright (C) 2012 Texas Instruments
> + * Author: Rob Clark<rob.clark@linaro.org>
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License version 2 as published by
> + * the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful, but WITHOUT
> + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
> + * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
> + * more details.
> + *
> + * You should have received a copy of the GNU General Public License along with
> + * this program.  If not, see<http://www.gnu.org/licenses/>.
> + */
> +
> +#ifndef __PLAT_OMAP_DRM_H__
> +#define __PLAT_OMAP_DRM_H__
> +
> +/*
> + * Optional platform data to configure the default configuration of which
> + * pipes/overlays/CRTCs are used.. if this is not provided, then instead the
> + * first CONFIG_DRM_OMAP_NUM_CRTCS are used, and they are each connected to
> + * one manager, with priority given to managers that are connected to
> + * detected devices.  Remaining overlays are used as video planes.  This
> + * should be a good default behavior for most cases, but yet there still
> + * might be times when you wish to do something different.
> + */
> +struct omap_kms_platform_data {
> +	/* overlays to use as CRTCs: */
> +	int ovl_cnt;
> +	const int *ovl_ids;
> +
> +	/* overlays to use as video planes: */
> +	int pln_cnt;
> +	const int *pln_ids;
> +
> +	int mgr_cnt;
> +	const int *mgr_ids;
> +
> +	int dev_cnt;
> +	const char **dev_names;
> +};
> +
> +struct omap_drm_platform_data {
> +	struct omap_kms_platform_data *kms_pdata;
> +	struct omap_dmm_platform_data *dmm_pdata;
> +};

Since the dmm_platform_data should not exist in theory, it should not be 
used by this structure either.

Where is the driver that will use these devices?

Regards,
Benoit
Rob Clark Jan. 23, 2012, 5:48 p.m. UTC | #11
On Mon, Jan 23, 2012 at 11:24 AM, Cousson, Benoit <b-cousson@ti.com> wrote:
> Hi Rob,
>
> On 1/13/2012 9:41 PM, Rob Clark wrote:
>>
>> From: Rob Clark<rob@ti.com>
>
>
> [...]
>
>> +static int __init omap_init_gpu(void)
>
>
> Why is the function to init drm device is named gpu?
>

drm drivers are typically gpu drivers (although due to lack of opensrc
pvr userspace most of the actual gpu bits are stripped out)

The function can be renamed.. the name is really not too important
since it is only used within this file

>> +{
>> +       struct omap_hwmod *oh = NULL;
>> +
>> +       /* lookup and populate the DMM information, if present - OMAP4+ */
>> +       oh = omap_hwmod_lookup("dmm");
>> +
>> +       if (oh) {
>> +               dmm_platdata.base = omap_hwmod_get_mpu_rt_va(oh);
>> +               dmm_platdata.irq = oh->mpu_irqs[0].irq;
>
>
> These are internal hwmod attributes that should not be retrieved here. They
> are included in the device resources and this is up to the driver to get
> them using the platform_get_resource.
>


hmm, I don't claim to be a hwmod expert, but how is the platform
device mapped back to the needed resources then?

Andy looked more at that part so I'll let him comment.


>> +
>> +               if (dmm_platdata.base)
>> +                       omapdrm_platdata.dmm_pdata =&dmm_platdata;
>
>
> pdata is supposed to be used for storing board level information, and we are
> in the process of removing all of them for device tree adaptation. So you
> should not use that at all in this case if this is not strictly needed.
>
>
>> +       }
>> +
>> +       return platform_device_register(&omap_drm_device);
>
>
> This is not the proper way to init a device nowadays.
>
> If you want to take advantage of the pm functionality, you should create an
> omap_device.
> Please have a look at the other OMAP devices creation code (GPIO, UART...).
>


Because omapdss is a separate layer below omapdrm, there really isn't
any PM in omapdrm (yet).  We do need to add support to restore the LUT
on resume although there are some complications when it comes to
correct sequence, ie. ensure LUT is reprogrammed before DSS overlays
are enabled, or IVAHD, etc, is allowed to start accessing buffer.
(But at least it should be possible to do correctly now, compared to
old state where there was a separate standalone tiler driver.)



>> +}
>> +
>> +arch_initcall(omap_init_gpu);
>> +
>> +void omapdrm_reserve_vram(void)
>> +{
>> +#ifdef CONFIG_CMA
>> +       /*
>> +        * Create private 32MiB contiguous memory area for omapdrm.0
>> device
>> +        * TODO revisit size.. if uc/wc buffers are allocated from CMA
>> pages
>> +        * then the amount of memory we need goes up..
>> +        */
>> +       dma_declare_contiguous(&omap_drm_device.dev, 32 * SZ_1M, 0, 0);
>> +#else
>> +#  warning "CMA is not enabled, there may be limitations about scanout
>> buffer allocations on OMAP3 and earlier"
>> +#endif
>> +}
>> +
>> +#endif
>> diff --git a/arch/arm/plat-omap/include/plat/drm.h
>> b/arch/arm/plat-omap/include/plat/drm.h
>> new file mode 100644
>> index 0000000..e29be29
>> --- /dev/null
>> +++ b/arch/arm/plat-omap/include/plat/drm.h
>
>
> Ideally, platform data should not exist anymore in a device tree world, but
> meanwhile there is a directory to store them:
> include/linux/platform_data
>
>
>> @@ -0,0 +1,70 @@
>> +/*
>> + * DRM/KMS device registration for TI OMAP platforms
>> + *
>> + * Copyright (C) 2012 Texas Instruments
>> + * Author: Rob Clark<rob.clark@linaro.org>
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> it
>> + * under the terms of the GNU General Public License version 2 as
>> published by
>> + * the Free Software Foundation.
>> + *
>> + * This program is distributed in the hope that it will be useful, but
>> WITHOUT
>> + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
>> + * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
>> for
>> + * more details.
>> + *
>> + * You should have received a copy of the GNU General Public License
>> along with
>> + * this program.  If not, see<http://www.gnu.org/licenses/>.
>> + */
>> +
>> +#ifndef __PLAT_OMAP_DRM_H__
>> +#define __PLAT_OMAP_DRM_H__
>> +
>> +/*
>> + * Optional platform data to configure the default configuration of which
>> + * pipes/overlays/CRTCs are used.. if this is not provided, then instead
>> the
>> + * first CONFIG_DRM_OMAP_NUM_CRTCS are used, and they are each connected
>> to
>> + * one manager, with priority given to managers that are connected to
>> + * detected devices.  Remaining overlays are used as video planes.  This
>> + * should be a good default behavior for most cases, but yet there still
>> + * might be times when you wish to do something different.
>> + */
>> +struct omap_kms_platform_data {
>> +       /* overlays to use as CRTCs: */
>> +       int ovl_cnt;
>> +       const int *ovl_ids;
>> +
>> +       /* overlays to use as video planes: */
>> +       int pln_cnt;
>> +       const int *pln_ids;
>> +
>> +       int mgr_cnt;
>> +       const int *mgr_ids;
>> +
>> +       int dev_cnt;
>> +       const char **dev_names;
>> +};
>> +
>> +struct omap_drm_platform_data {
>> +       struct omap_kms_platform_data *kms_pdata;
>> +       struct omap_dmm_platform_data *dmm_pdata;
>> +};
>
>
> Since the dmm_platform_data should not exist in theory, it should not be
> used by this structure either.
>
> Where is the driver that will use these devices?

drivers/staging/omapdrm

BR,
-R

> Regards,
> Benoit
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-omap" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
Felipe Contreras Jan. 24, 2012, 3:33 p.m. UTC | #12
On Mon, Jan 16, 2012 at 11:24 PM, Rob Clark <rob.clark@linaro.org> wrote:
> On Mon, Jan 16, 2012 at 2:37 PM, Felipe Contreras
> <felipe.contreras@gmail.com> wrote:
>> On Mon, Jan 16, 2012 at 7:01 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>> On Mon, Jan 16, 2012 at 10:59 AM, Felipe Contreras
>>> <felipe.contreras@gmail.com> wrote:
>>>> #if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
>>>> extern void omapdrm_reserve_vram(void);
>>>> #else
>>>> static inline void omapdrm_reserve_vram(void) { }
>>>> #endif
>>>>
>>>> Like how it's done with DSP stuff.
>>>
>>> right, but then you'd miss the omapdrm_reserve_vram() call at startup..
>>
>> Why?
>
> I guess drm.o is ending up in a module, but the code that calls that
> (in common.c) is not in a module, so you get an unresolved symbol at
> link

Right... But that code can be moved as well. Just like with the bridge.
Rob Clark Jan. 24, 2012, 3:54 p.m. UTC | #13
On Tue, Jan 24, 2012 at 9:33 AM, Felipe Contreras
<felipe.contreras@gmail.com> wrote:
> On Mon, Jan 16, 2012 at 11:24 PM, Rob Clark <rob.clark@linaro.org> wrote:
>> On Mon, Jan 16, 2012 at 2:37 PM, Felipe Contreras
>> <felipe.contreras@gmail.com> wrote:
>>> On Mon, Jan 16, 2012 at 7:01 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>> On Mon, Jan 16, 2012 at 10:59 AM, Felipe Contreras
>>>> <felipe.contreras@gmail.com> wrote:
>>>>> #if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
>>>>> extern void omapdrm_reserve_vram(void);
>>>>> #else
>>>>> static inline void omapdrm_reserve_vram(void) { }
>>>>> #endif
>>>>>
>>>>> Like how it's done with DSP stuff.
>>>>
>>>> right, but then you'd miss the omapdrm_reserve_vram() call at startup..
>>>
>>> Why?
>>
>> I guess drm.o is ending up in a module, but the code that calls that
>> (in common.c) is not in a module, so you get an unresolved symbol at
>> link
>
> Right... But that code can be moved as well. Just like with the bridge.

Hmm, looks like for dsp bridge the memory reservation is moved to
devices.c.  Although I'm not entirely sure how that is better... and
there is precedent for both approaches, ie. omapfb works like omapdrm,
and tidspbridge works as you suggest.

seems a bit bikeshed'ish to me

BR,
-R

> --
> Felipe Contreras
Felipe Contreras Jan. 25, 2012, 2:17 a.m. UTC | #14
On Tue, Jan 24, 2012 at 5:54 PM, Rob Clark <rob.clark@linaro.org> wrote:
> On Tue, Jan 24, 2012 at 9:33 AM, Felipe Contreras
> <felipe.contreras@gmail.com> wrote:
>> On Mon, Jan 16, 2012 at 11:24 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>> On Mon, Jan 16, 2012 at 2:37 PM, Felipe Contreras
>>> <felipe.contreras@gmail.com> wrote:
>>>> On Mon, Jan 16, 2012 at 7:01 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>> On Mon, Jan 16, 2012 at 10:59 AM, Felipe Contreras
>>>>> <felipe.contreras@gmail.com> wrote:
>>>>>> #if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
>>>>>> extern void omapdrm_reserve_vram(void);
>>>>>> #else
>>>>>> static inline void omapdrm_reserve_vram(void) { }
>>>>>> #endif
>>>>>>
>>>>>> Like how it's done with DSP stuff.
>>>>>
>>>>> right, but then you'd miss the omapdrm_reserve_vram() call at startup..
>>>>
>>>> Why?
>>>
>>> I guess drm.o is ending up in a module, but the code that calls that
>>> (in common.c) is not in a module, so you get an unresolved symbol at
>>> link
>>
>> Right... But that code can be moved as well. Just like with the bridge.
>
> Hmm, looks like for dsp bridge the memory reservation is moved to
> devices.c.  Although I'm not entirely sure how that is better... and
> there is precedent for both approaches, ie. omapfb works like omapdrm,
> and tidspbridge works as you suggest.
>
> seems a bit bikeshed'ish to me

I will send a patch to fix omapfb, perhaps after that it will be a bit
more clear how it should be done :) (if it gets accepted)
Rob Clark Jan. 25, 2012, 2:32 a.m. UTC | #15
On Tue, Jan 24, 2012 at 8:17 PM, Felipe Contreras
<felipe.contreras@gmail.com> wrote:
> On Tue, Jan 24, 2012 at 5:54 PM, Rob Clark <rob.clark@linaro.org> wrote:
>> On Tue, Jan 24, 2012 at 9:33 AM, Felipe Contreras
>> <felipe.contreras@gmail.com> wrote:
>>> On Mon, Jan 16, 2012 at 11:24 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>> On Mon, Jan 16, 2012 at 2:37 PM, Felipe Contreras
>>>> <felipe.contreras@gmail.com> wrote:
>>>>> On Mon, Jan 16, 2012 at 7:01 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>>> On Mon, Jan 16, 2012 at 10:59 AM, Felipe Contreras
>>>>>> <felipe.contreras@gmail.com> wrote:
>>>>>>> #if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
>>>>>>> extern void omapdrm_reserve_vram(void);
>>>>>>> #else
>>>>>>> static inline void omapdrm_reserve_vram(void) { }
>>>>>>> #endif
>>>>>>>
>>>>>>> Like how it's done with DSP stuff.
>>>>>>
>>>>>> right, but then you'd miss the omapdrm_reserve_vram() call at startup..
>>>>>
>>>>> Why?
>>>>
>>>> I guess drm.o is ending up in a module, but the code that calls that
>>>> (in common.c) is not in a module, so you get an unresolved symbol at
>>>> link
>>>
>>> Right... But that code can be moved as well. Just like with the bridge.
>>
>> Hmm, looks like for dsp bridge the memory reservation is moved to
>> devices.c.  Although I'm not entirely sure how that is better... and
>> there is precedent for both approaches, ie. omapfb works like omapdrm,
>> and tidspbridge works as you suggest.
>>
>> seems a bit bikeshed'ish to me
>
> I will send a patch to fix omapfb, perhaps after that it will be a bit
> more clear how it should be done :) (if it gets accepted)

ok, but the thing I don't like is it splits up the drm device setup
from the omapdrm_reserve_vram() part (and similarly for omapfb),

but if this is the consensus of how it should be done, well.. when in
Rome, and all that

BR,
-R

> --
> Felipe Contreras
> _______________________________________________
> dri-devel mailing list
> dri-devel@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/dri-devel
Rob Clark Jan. 25, 2012, 1:51 p.m. UTC | #16
On Tue, Jan 24, 2012 at 8:32 PM, Rob Clark <rob.clark@linaro.org> wrote:
> On Tue, Jan 24, 2012 at 8:17 PM, Felipe Contreras
> <felipe.contreras@gmail.com> wrote:
>> On Tue, Jan 24, 2012 at 5:54 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>> On Tue, Jan 24, 2012 at 9:33 AM, Felipe Contreras
>>> <felipe.contreras@gmail.com> wrote:
>>>> On Mon, Jan 16, 2012 at 11:24 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>> On Mon, Jan 16, 2012 at 2:37 PM, Felipe Contreras
>>>>> <felipe.contreras@gmail.com> wrote:
>>>>>> On Mon, Jan 16, 2012 at 7:01 PM, Rob Clark <rob.clark@linaro.org> wrote:
>>>>>>> On Mon, Jan 16, 2012 at 10:59 AM, Felipe Contreras
>>>>>>> <felipe.contreras@gmail.com> wrote:
>>>>>>>> #if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
>>>>>>>> extern void omapdrm_reserve_vram(void);
>>>>>>>> #else
>>>>>>>> static inline void omapdrm_reserve_vram(void) { }
>>>>>>>> #endif
>>>>>>>>
>>>>>>>> Like how it's done with DSP stuff.
>>>>>>>
>>>>>>> right, but then you'd miss the omapdrm_reserve_vram() call at startup..
>>>>>>
>>>>>> Why?
>>>>>
>>>>> I guess drm.o is ending up in a module, but the code that calls that
>>>>> (in common.c) is not in a module, so you get an unresolved symbol at
>>>>> link
>>>>
>>>> Right... But that code can be moved as well. Just like with the bridge.
>>>
>>> Hmm, looks like for dsp bridge the memory reservation is moved to
>>> devices.c.  Although I'm not entirely sure how that is better... and
>>> there is precedent for both approaches, ie. omapfb works like omapdrm,
>>> and tidspbridge works as you suggest.
>>>
>>> seems a bit bikeshed'ish to me
>>
>> I will send a patch to fix omapfb, perhaps after that it will be a bit
>> more clear how it should be done :) (if it gets accepted)
>
> ok, but the thing I don't like is it splits up the drm device setup
> from the omapdrm_reserve_vram() part (and similarly for omapfb),
>
> but if this is the consensus of how it should be done, well.. when in
> Rome, and all that

oh, sorry, I am mistaken, the oampdrm_reserve_vram() cannot be split
into devices.c, since you need the 'struct device *' to register the
CMA region.  I'd ask that you don't patch omapfb to "fix" it because
this would have to be undone once CMA is merged (since we should then
remove omap_vram and other carveout mechanism and use CMA instead)

BR,
-R

> BR,
> -R
>
>> --
>> Felipe Contreras
>> _______________________________________________
>> dri-devel mailing list
>> dri-devel@lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/dri-devel
Greg KH Feb. 9, 2012, 5:28 p.m. UTC | #17
On Fri, Jan 13, 2012 at 02:41:59PM -0600, Rob Clark wrote:
> From: Rob Clark <rob@ti.com>
> 
> Register OMAP DRM/KMS platform device, and reserve a CMA region for
> the device to use for buffer allocation.
> 
> v1: initial patch
> v2: move platform data structs into plat-omap to avoid having to
>     #include headers from drivers/staging and cleanups
> 
> Signed-off-by: Rob Clark <rob@ti.com>
> ---
> Note: after applying this patch there will be duplicate copies of the
> platform data structs (until the 2nd patch is applied).  But I tested
> to ensure this does not cause build breaks.  So the 2nd patch which
> should go thru staging tree is safe to be held until this patch hits
> Linus's master branch.
> 
>  arch/arm/plat-omap/Makefile           |    2 +-
>  arch/arm/plat-omap/common.c           |    3 +-
>  arch/arm/plat-omap/drm.c              |   83 +++++++++++++++++++++++++++++++++
>  arch/arm/plat-omap/include/plat/drm.h |   70 +++++++++++++++++++++++++++
>  4 files changed, 156 insertions(+), 2 deletions(-)
>  create mode 100644 arch/arm/plat-omap/drm.c
>  create mode 100644 arch/arm/plat-omap/include/plat/drm.h

Did this ever get applied?  As I can't apply 2/2 without it, please feel
free to add:
	Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

to 2/2 when some dri/omap developer commits this one.

thanks,

greg k-h
Rob Clark Feb. 9, 2012, 5:41 p.m. UTC | #18
On Thu, Feb 9, 2012 at 11:28 AM, Greg KH <gregkh@linuxfoundation.org> wrote:
> On Fri, Jan 13, 2012 at 02:41:59PM -0600, Rob Clark wrote:
>> From: Rob Clark <rob@ti.com>
>>
>> Register OMAP DRM/KMS platform device, and reserve a CMA region for
>> the device to use for buffer allocation.
>>
>> v1: initial patch
>> v2: move platform data structs into plat-omap to avoid having to
>>     #include headers from drivers/staging and cleanups
>>
>> Signed-off-by: Rob Clark <rob@ti.com>
>> ---
>> Note: after applying this patch there will be duplicate copies of the
>> platform data structs (until the 2nd patch is applied).  But I tested
>> to ensure this does not cause build breaks.  So the 2nd patch which
>> should go thru staging tree is safe to be held until this patch hits
>> Linus's master branch.
>>
>>  arch/arm/plat-omap/Makefile           |    2 +-
>>  arch/arm/plat-omap/common.c           |    3 +-
>>  arch/arm/plat-omap/drm.c              |   83 +++++++++++++++++++++++++++++++++
>>  arch/arm/plat-omap/include/plat/drm.h |   70 +++++++++++++++++++++++++++
>>  4 files changed, 156 insertions(+), 2 deletions(-)
>>  create mode 100644 arch/arm/plat-omap/drm.c
>>  create mode 100644 arch/arm/plat-omap/include/plat/drm.h
>
> Did this ever get applied?

Not yet, there was requested some omap hwmod related changes for how
the driver gets SoC version specific information (irq, base addr), so
I'll resubmit again the device file registration (and the 2nd patch)
but that might have to be for 3.4

You can drop the existing 2/2 patch

BR,
-R


> As I can't apply 2/2 without it, please feel
> free to add:
>        Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>
> to 2/2 when some dri/omap developer commits this one.
>
> thanks,
>
> greg k-h
Andy Gross Feb. 17, 2012, 9:14 p.m. UTC | #19
On Mon, Jan 23, 2012 at 11:24 AM, Cousson, Benoit <b-cousson@ti.com> wrote:

>
>> +       if (oh) {
>> +               dmm_platdata.base = omap_hwmod_get_mpu_rt_va(oh);
>> +               dmm_platdata.irq = oh->mpu_irqs[0].irq;
>>
>
> These are internal hwmod attributes that should not be retrieved here.
> They are included in the device resources and this is up to the driver to
> get them using the platform_get_resource.


Yeah this can be removed and I can switch to platform_get_resource.



>  +
>> +               if (dmm_platdata.base)
>> +                       omapdrm_platdata.dmm_pdata =&dmm_platdata;
>>
>
> pdata is supposed to be used for storing board level information, and we
> are in the process of removing all of them for device tree adaptation. So
> you should not use that at all in this case if this is not strictly needed.


Noted.  I'll just remove it.  I was planning ahead when I added this, but I
can cross that bridge when I get there.


>
>  +       }
>> +
>> +       return platform_device_register(&**omap_drm_device);
>>
>
> This is not the proper way to init a device nowadays.
>
> If you want to take advantage of the pm functionality, you should create
> an omap_device.
> Please have a look at the other OMAP devices creation code (GPIO, UART...).


It was my understanding that you needed a hwmod entry that corresponded to
the device if you wanted to use omap_device_build().  In the case of
omap_drm, we don't have a hwmod entry.  We do have an entry for DMM.


Regards,

Andy Gross
diff mbox

Patch

diff --git a/arch/arm/plat-omap/Makefile b/arch/arm/plat-omap/Makefile
index 9a58461..b86e6cb 100644
--- a/arch/arm/plat-omap/Makefile
+++ b/arch/arm/plat-omap/Makefile
@@ -4,7 +4,7 @@ 
 
 # Common support
 obj-y := common.o sram.o clock.o devices.o dma.o mux.o \
-	 usb.o fb.o counter_32k.o
+	 usb.o fb.o counter_32k.o drm.o
 obj-m :=
 obj-n :=
 obj-  :=
diff --git a/arch/arm/plat-omap/common.c b/arch/arm/plat-omap/common.c
index 06383b5..0d87dab 100644
--- a/arch/arm/plat-omap/common.c
+++ b/arch/arm/plat-omap/common.c
@@ -21,10 +21,10 @@ 
 #include <plat/board.h>
 #include <plat/vram.h>
 #include <plat/dsp.h>
+#include <plat/drm.h>
 
 #include <plat/omap-secure.h>
 
-
 #define NO_LENGTH_CHECK 0xffffffff
 
 struct omap_board_config_kernel *omap_board_config __initdata;
@@ -65,6 +65,7 @@  const void *__init omap_get_var_config(u16 tag, size_t *len)
 
 void __init omap_reserve(void)
 {
+	omapdrm_reserve_vram();
 	omapfb_reserve_sdram_memblock();
 	omap_vram_reserve_sdram_memblock();
 	omap_dsp_reserve_sdram_memblock();
diff --git a/arch/arm/plat-omap/drm.c b/arch/arm/plat-omap/drm.c
new file mode 100644
index 0000000..aa0ba69
--- /dev/null
+++ b/arch/arm/plat-omap/drm.c
@@ -0,0 +1,83 @@ 
+/*
+ * DRM/KMS device registration for TI OMAP platforms
+ *
+ * Copyright (C) 2012 Texas Instruments
+ * Author: Rob Clark <rob.clark@linaro.org>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published by
+ * the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/init.h>
+#include <linux/platform_device.h>
+#include <linux/dma-mapping.h>
+#ifdef CONFIG_CMA
+#  include <linux/dma-contiguous.h>
+#endif
+
+#include <plat/omap_device.h>
+#include <plat/omap_hwmod.h>
+
+#include <plat/drm.h>
+
+#if defined(CONFIG_DRM_OMAP) || (CONFIG_DRM_OMAP_MODULE)
+
+static struct omap_drm_platform_data omapdrm_platdata;
+static struct omap_dmm_platform_data dmm_platdata;
+
+static struct platform_device omap_drm_device = {
+		.dev = {
+			.coherent_dma_mask = DMA_BIT_MASK(32),
+			.platform_data = &omapdrm_platdata,
+		},
+		.name = "omapdrm",
+		.id = 0,
+};
+
+static int __init omap_init_gpu(void)
+{
+	struct omap_hwmod *oh = NULL;
+
+	/* lookup and populate the DMM information, if present - OMAP4+ */
+	oh = omap_hwmod_lookup("dmm");
+
+	if (oh) {
+		dmm_platdata.base = omap_hwmod_get_mpu_rt_va(oh);
+		dmm_platdata.irq = oh->mpu_irqs[0].irq;
+
+		if (dmm_platdata.base)
+			omapdrm_platdata.dmm_pdata = &dmm_platdata;
+	}
+
+	return platform_device_register(&omap_drm_device);
+}
+
+arch_initcall(omap_init_gpu);
+
+void omapdrm_reserve_vram(void)
+{
+#ifdef CONFIG_CMA
+	/*
+	 * Create private 32MiB contiguous memory area for omapdrm.0 device
+	 * TODO revisit size.. if uc/wc buffers are allocated from CMA pages
+	 * then the amount of memory we need goes up..
+	 */
+	dma_declare_contiguous(&omap_drm_device.dev, 32 * SZ_1M, 0, 0);
+#else
+#  warning "CMA is not enabled, there may be limitations about scanout buffer allocations on OMAP3 and earlier"
+#endif
+}
+
+#endif
diff --git a/arch/arm/plat-omap/include/plat/drm.h b/arch/arm/plat-omap/include/plat/drm.h
new file mode 100644
index 0000000..e29be29
--- /dev/null
+++ b/arch/arm/plat-omap/include/plat/drm.h
@@ -0,0 +1,70 @@ 
+/*
+ * DRM/KMS device registration for TI OMAP platforms
+ *
+ * Copyright (C) 2012 Texas Instruments
+ * Author: Rob Clark <rob.clark@linaro.org>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License version 2 as published by
+ * the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __PLAT_OMAP_DRM_H__
+#define __PLAT_OMAP_DRM_H__
+
+/*
+ * Optional platform data to configure the default configuration of which
+ * pipes/overlays/CRTCs are used.. if this is not provided, then instead the
+ * first CONFIG_DRM_OMAP_NUM_CRTCS are used, and they are each connected to
+ * one manager, with priority given to managers that are connected to
+ * detected devices.  Remaining overlays are used as video planes.  This
+ * should be a good default behavior for most cases, but yet there still
+ * might be times when you wish to do something different.
+ */
+struct omap_kms_platform_data {
+	/* overlays to use as CRTCs: */
+	int ovl_cnt;
+	const int *ovl_ids;
+
+	/* overlays to use as video planes: */
+	int pln_cnt;
+	const int *pln_ids;
+
+	int mgr_cnt;
+	const int *mgr_ids;
+
+	int dev_cnt;
+	const char **dev_names;
+};
+
+struct omap_drm_platform_data {
+	struct omap_kms_platform_data *kms_pdata;
+	struct omap_dmm_platform_data *dmm_pdata;
+};
+
+struct omap_dmm_platform_data {
+	void __iomem *base;
+	int irq;
+};
+
+#if defined(CONFIG_DRM_OMAP) || defined(CONFIG_DRM_OMAP_MODULE)
+
+void omapdrm_reserve_vram(void);
+
+#else
+
+static inline void omapdrm_reserve_vram(void)
+{
+}
+
+#endif
+
+#endif /* __PLAT_OMAP_DRM_H__ */