[RFC v2 1/5] drm/exynos: add ipp subsystem

Eunchul Kim chulspro.kim at samsung.com
Mon Dec 10 23:57:14 PST 2012


Thank's for comments.

I answer your comment, please check that.

Thank's

BR
Eunchul Kim.

On 12/11/2012 04:17 PM, Inki Dae wrote:
>
>
>> -----Original Message-----
>> From: Eunchul Kim [mailto:chulspro.kim at samsung.com]
>> Sent: Tuesday, December 11, 2012 2:55 PM
>> To: Inki Dae
>> Cc: dri-devel at lists.freedesktop.org; jy0.jeon at samsung.com;
>> yj44.cho at samsung.com; jmock.shin at samsung.com; jaejoon.seo at samsung.com;
>> kyungmin.park at samsung.com
>> Subject: Re: [RFC v2 1/5] drm/exynos: add ipp subsystem
>>
>> Thank's your comments.
>>
>> I answer your requestion. please check that.
>>
>> Thank's
>>
>> BR
>> Eunchul Kim
>>
>> On 12/11/2012 01:37 PM, Inki Dae wrote:
>>>
>>>
>>>> -----Original Message-----
>>>> From: Eunchul Kim [mailto:chulspro.kim at samsung.com]
>>>> Sent: Monday, December 10, 2012 10:46 PM
>>>> To: dri-devel at lists.freedesktop.org; inki.dae at samsung.com
>>>> Cc: jy0.jeon at samsung.com; yj44.cho at samsung.com; jmock.shin at samsung.com;
>>>> jaejoon.seo at samsung.com; kyungmin.park at samsung.com;
>>>> chulspro.kim at samsung.com
>>>> Subject: [RFC v2 1/5] drm/exynos: add ipp subsystem
>>>>
>>>> IPP stand for Image Post Processing and supports image scaler/rotator
>>>> /crop/flip/csc(color space conversion) and input/output DMA operations
>>>> using ipp drivers.
>>>> also supports writeback and display output operations.
>>>> ipp driver include FIMC, Rotator, GSC, SC, so on.
>>>> and ipp is integration device driver for each hardware.
>>>>
>>>> Signed-off-by: Eunchul Kim <chulspro.kim at samsung.com>
>>>> Signed-off-by: Jinyoung Jeon <jy0.jeon at samsung.com>
>>>> ---
>>>>    drivers/gpu/drm/exynos/Kconfig          |    6 +
>>>>    drivers/gpu/drm/exynos/Makefile         |    1 +
>>>>    drivers/gpu/drm/exynos/exynos_drm_drv.c |   24 +
>>>>    drivers/gpu/drm/exynos/exynos_drm_drv.h |    7 +
>>>>    drivers/gpu/drm/exynos/exynos_drm_ipp.c | 1944
>>>> +++++++++++++++++++++++++++++++
>>>>    drivers/gpu/drm/exynos/exynos_drm_ipp.h |  266 +++++
>>>>    include/uapi/drm/exynos_drm.h           |  189 +++
>>>>    7 files changed, 2437 insertions(+), 0 deletions(-)
>>>>    create mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.c
>>>>    create mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.h
>>>>
>>>> diff --git a/drivers/gpu/drm/exynos/Kconfig
>>>> b/drivers/gpu/drm/exynos/Kconfig
>>>> index 4ea8cdc..bcf1c9d 100644
>>>> --- a/drivers/gpu/drm/exynos/Kconfig
>>>> +++ b/drivers/gpu/drm/exynos/Kconfig
>>>> @@ -45,3 +45,9 @@ config DRM_EXYNOS_G2D
>>>>    	depends on DRM_EXYNOS && !VIDEO_SAMSUNG_S5P_G2D
>>>>    	help
>>>>    	  Choose this option if you want to use Exynos G2D for DRM.
>>>> +
>>>> +config DRM_EXYNOS_IPP
>>>> +	bool "Exynos DRM IPP"
>>>> +	depends on DRM_EXYNOS
>>>> +	help
>>>> +	  Choose this option if you want to use IPP feature for DRM.
>>>> diff --git a/drivers/gpu/drm/exynos/Makefile
>>>> b/drivers/gpu/drm/exynos/Makefile
>>>> index 26813b8..6c536ce 100644
>>>> --- a/drivers/gpu/drm/exynos/Makefile
>>>> +++ b/drivers/gpu/drm/exynos/Makefile
>>>> @@ -16,5 +16,6 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI)	+=
>>>> exynos_hdmi.o exynos_mixer.o \
>>>>    					   exynos_drm_hdmi.o
>>>>    exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI)	+= exynos_drm_vidi.o
>>>>    exynosdrm-$(CONFIG_DRM_EXYNOS_G2D)	+= exynos_drm_g2d.o
>>>> +exynosdrm-$(CONFIG_DRM_EXYNOS_IPP)	+= exynos_drm_ipp.o
>>>>
>>>>    obj-$(CONFIG_DRM_EXYNOS)		+= exynosdrm.o
>>>> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c
>>>> b/drivers/gpu/drm/exynos/exynos_drm_drv.c
>>>> index 4a1168d..0eb8a97 100644
>>>> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
>>>> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
>>>> @@ -40,6 +40,7 @@
>>>>    #include "exynos_drm_vidi.h"
>>>>    #include "exynos_drm_dmabuf.h"
>>>>    #include "exynos_drm_g2d.h"
>>>> +#include "exynos_drm_ipp.h"
>>>>    #include "exynos_drm_iommu.h"
>>>>
>>>>    #define DRIVER_NAME	"exynos"
>>>> @@ -249,6 +250,14 @@ static struct drm_ioctl_desc exynos_ioctls[] = {
>>>>    			exynos_g2d_set_cmdlist_ioctl, DRM_UNLOCKED |
>>> DRM_AUTH),
>>>>    	DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC,
>>>>    			exynos_g2d_exec_ioctl, DRM_UNLOCKED | DRM_AUTH),
>>>> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_PROPERTY,
>>>> +			exynos_drm_ipp_get_property, DRM_UNLOCKED |
>>> DRM_AUTH),
>>>> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_SET_PROPERTY,
>>>> +			exynos_drm_ipp_set_property, DRM_UNLOCKED |
>>> DRM_AUTH),
>>>> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_QUEUE_BUF,
>>>> +			exynos_drm_ipp_queue_buf, DRM_UNLOCKED | DRM_AUTH),
>>>> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_CMD_CTRL,
>>>> +			exynos_drm_ipp_cmd_ctrl, DRM_UNLOCKED | DRM_AUTH),
>>>>    };
>>>>
>>>>    static const struct file_operations exynos_drm_driver_fops = {
>>>> @@ -363,6 +372,12 @@ static int __init exynos_drm_init(void)
>>>>    		goto out_g2d;
>>>>    #endif
>>>>
>>>> +#ifdef CONFIG_DRM_EXYNOS_IPP
>>>> +	ret = platform_driver_register(&ipp_driver);
>>>> +	if (ret < 0)
>>>> +		goto out_ipp;
>>>> +#endif
>>>> +
>>>>    	ret = platform_driver_register(&exynos_drm_platform_driver);
>>>>    	if (ret < 0)
>>>>    		goto out_drm;
>>>> @@ -380,6 +395,11 @@ out:
>>>>    	platform_driver_unregister(&exynos_drm_platform_driver);
>>>>
>>>>    out_drm:
>>>> +#ifdef CONFIG_DRM_EXYNOS_IPP
>>>> +	platform_driver_unregister(&ipp_driver);
>>>> +out_ipp:
>>>> +#endif
>>>> +
>>>>    #ifdef CONFIG_DRM_EXYNOS_G2D
>>>>    	platform_driver_unregister(&g2d_driver);
>>>>    out_g2d:
>>>> @@ -416,6 +436,10 @@ static void __exit exynos_drm_exit(void)
>>>>
>>>>    	platform_driver_unregister(&exynos_drm_platform_driver);
>>>>
>>>> +#ifdef CONFIG_DRM_EXYNOS_IPP
>>>> +	platform_driver_unregister(&ipp_driver);
>>>> +#endif
>>>> +
>>>>    #ifdef CONFIG_DRM_EXYNOS_G2D
>>>>    	platform_driver_unregister(&g2d_driver);
>>>>    #endif
>>>> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h
>>>> b/drivers/gpu/drm/exynos/exynos_drm_drv.h
>>>> index a9db025..a365788 100644
>>>> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.h
>>>> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h
>>>> @@ -235,8 +235,14 @@ struct exynos_drm_g2d_private {
>>>>    	unsigned int		gem_nr;
>>>>    };
>>>>
>>>> +struct exynos_drm_ipp_private {
>>>> +	struct device	*dev;
>>>> +	struct list_head	event_list;
>>>> +};
>>>> +
>>>>    struct drm_exynos_file_private {
>>>>    	struct exynos_drm_g2d_private	*g2d_priv;
>>>> +	struct exynos_drm_ipp_private	*ipp_priv;
>>>>    };
>>>>
>>>>    /*
>>>> @@ -346,4 +352,5 @@ extern struct platform_driver mixer_driver;
>>>>    extern struct platform_driver exynos_drm_common_hdmi_driver;
>>>>    extern struct platform_driver vidi_driver;
>>>>    extern struct platform_driver g2d_driver;
>>>> +extern struct platform_driver ipp_driver;
>>>>    #endif
>>>> diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.c
>>>> b/drivers/gpu/drm/exynos/exynos_drm_ipp.c
>>>> new file mode 100644
>>>> index 0000000..2d84477
>>>> --- /dev/null
>>>> +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.c
>>>> @@ -0,0 +1,1944 @@
>>>> +/*
>>>> + * Copyright (C) 2012 Samsung Electronics Co.Ltd
>>>> + * Authors:
>>>> + *	Eunchul Kim <chulspro.kim at samsung.com>
>>>> + *	Jinyoung Jeon <jy0.jeon at samsung.com>
>>>> + *	Sangmin Lee <lsmin.lee at samsung.com>
>>>> + *
>>>> + * This program is free software; you can redistribute  it and/or
>> modify
>>>> it
>>>> + * under  the terms of  the GNU General  Public License as published
> by
>>>> the
>>>> + * Free Software Foundation;  either version 2 of the  License, or (at
>>>> your
>>>> + * option) any later version.
>>>> + *
>>>> + */
>>>> +#include <linux/kernel.h>
>>>> +#include <linux/module.h>
>>>> +#include <linux/platform_device.h>
>>>> +#include <linux/clk.h>
>>>> +#include <linux/pm_runtime.h>
>>>> +#include <plat/map-base.h>
>>>> +
>>>> +#include <drm/drmP.h>
>>>> +#include <drm/exynos_drm.h>
>>>> +#include "exynos_drm_drv.h"
>>>> +#include "exynos_drm_gem.h"
>>>> +#include "exynos_drm_ipp.h"
>>>> +
>>>> +/*
>>>> + * IPP is stand for Image Post Processing and
>>>> + * supports image scaler/rotator and input/output DMA operations.
>>>> + * using FIMC, GSC, Rotator, so on.
>>>> + * IPP is integration device driver of same attribute h/w
>>>> + */
>>>> +
>>>> +#define get_ipp_context(dev)
>>>> 	platform_get_drvdata(to_platform_device(dev))
>>>> +
>>>> +/*
>>>> + * A structure of event.
>>>> + *
>>>> + * @base: base of event.
>>>> + * @event: ipp event.
>>>> + */
>>>> +struct drm_exynos_ipp_send_event {
>>>> +	struct drm_pending_event	base;
>>>> +	struct drm_exynos_ipp_event	event;
>>>> +};
>>>> +
>>>> +/*
>>>> + * A structure of memory node.
>>>> + *
>>>> + * @list: list head to memory queue information.
>>>> + * @ops_id: id of operations.
>>>> + * @prop_id: id of property.
>>>> + * @buf_id: id of buffer.
>>>> + * @buf_info: gem objects and dma address, size.
>>>> + * @filp: a pointer to drm_file.
>>>> + */
>>>> +struct drm_exynos_ipp_mem_node {
>>>> +	struct list_head	list;
>>>> +	enum drm_exynos_ops_id	ops_id;
>>>> +	u32	prop_id;
>>>> +	u32	buf_id;
>>>> +	struct drm_exynos_ipp_buf_info	buf_info;
>>>> +	struct drm_file		*filp;
>>>> +};
>>>> +
>>>> +/*
>>>> + * A structure of ipp context.
>>>> + *
>>>> + * @subdrv: prepare initialization using subdrv.
>>>> + * @ipp_lock: lock for synchronization of access to ipp_idr.
>>>> + * @prop_lock: lock for synchronization of access to prop_idr.
>>>> + * @ipp_idr: ipp driver idr.
>>>> + * @prop_idr: property idr.
>>>> + * @event_workq: event work queue.
>>>> + * @cmd_workq: command work queue.
>>>> + */
>>>> +struct ipp_context {
>>>> +	struct exynos_drm_subdrv	subdrv;
>>>> +	struct mutex	ipp_lock;
>>>> +	struct mutex	prop_lock;
>>>> +	struct idr	ipp_idr;
>>>> +	struct idr	prop_idr;
>>>> +	struct workqueue_struct	*event_workq;
>>>> +	struct workqueue_struct	*cmd_workq;
>>>> +};
>>>> +
>>>> +static LIST_HEAD(exynos_drm_ippdrv_list);
>>>> +static BLOCKING_NOTIFIER_HEAD(exynos_drm_ippnb_list);
>>>> +
>>>> +int exynos_drm_ippdrv_register(struct exynos_drm_ippdrv *ippdrv)
>>>> +{
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>
>>> Add mutex_lock.
>>
>> - I already answer about why not use mutex in this scenario.
>>    We already serialized probe routine. so, I think we don't need mutex
>> lock.
>>    and please check exynos_drm_core.c file subdrv_register().
>>    If need mutext in subdrv_register()?, than I will changed it.
>>    please one more comments about it.
>>
>>>
>>>> +	list_add_tail(&ippdrv->drv_list, &exynos_drm_ippdrv_list);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(exynos_drm_ippdrv_register);
>>>> +
>>>> +int exynos_drm_ippdrv_unregister(struct exynos_drm_ippdrv *ippdrv)
>>>> +{
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	list_del(&ippdrv->drv_list);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(exynos_drm_ippdrv_unregister);
>>>> +
>>>> +static int ipp_create_id(struct idr *id_idr, struct mutex *lock, void
>>>> *obj,
>>>> +		u32 *idp)
>>>> +{
>>>> +	int ret = -EINVAL;
>>>
>>> Just int ret;
>>
>> - done.
>>
>>>
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +again:
>>>> +	/* ensure there is space available to allocate a handle */
>>>> +	if (idr_pre_get(id_idr, GFP_KERNEL) == 0)
>>>> +		return -ENOMEM;
>>>> +
>>>> +	/* do the allocation under our mutexlock */
>>>> +	mutex_lock(lock);
>>>> +	ret = idr_get_new_above(id_idr, obj, 1, (int *)idp);
>>>> +	mutex_unlock(lock);
>>>> +	if (ret == -EAGAIN)
>>>> +		goto again;
>>>> +
>>>> +	return ret;
>>>> +}
>>>> +
>>>> +static void *ipp_find_id(struct idr *id_idr, struct mutex *lock, u32
>> id)
>>>> +{
>>>> +	void *obj;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:id[%d]\n", __func__, id);
>>>> +
>>>> +	mutex_lock(lock);
>>>> +
>>>> +	/* find object using handle */
>>>> +	obj = idr_find(id_idr, id);
>>>> +	if (obj == NULL) {
>>>> +		mutex_unlock(lock);
>>>> +		return NULL;
>>>
>>> Return ERR_PTR(error);
>>
>> - done ERR_PTR(-ENODEV); is it OK ?
>>
>>>
>>>> +	}
>>>> +
>>>> +	mutex_unlock(lock);
>>>> +
>>>> +	return obj;
>>>> +}
>>>> +
>>>> +static struct exynos_drm_ippdrv *ipp_find_driver(struct ipp_context
>> *ctx,
>>>> +		struct drm_exynos_ipp_property *property)
>>>> +{
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +	u32 ipp_id = property->ipp_id;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:ipp_id[%d]\n", __func__, ipp_id);
>>>> +
>>>> +	if (ipp_id) {
>>>> +		/* find ipp driver */
>>>> +		ippdrv = ipp_find_id(&ctx->ipp_idr, &ctx->ipp_lock,
>>>> +			ipp_id);
>>>> +		if (!ippdrv) {
>>>> +			DRM_ERROR("not found ipp%d driver.\n", ipp_id);
>>>> +			goto err_null;
>>>> +		}
>>>> +
>>>> +		/* check dedicated state */
>>>
>>> Add more comments to what is dedicated?
>>
>> - Writeback operation and Output interface is dedicated. this operation
>> not support multiple operation. I added comment.
>>
>>>
>>>> +		if (ippdrv->dedicated) {
>>>> +			DRM_ERROR("used choose device.\n");
>>>> +			goto err_null;
>>>> +		}
>>>> +
>>>> +		if (property->cmd != IPP_CMD_M2M
>>>> +			&& !pm_runtime_suspended(ippdrv->dev)) {
>>>> +			DRM_ERROR("can't run dedicatedly.\n");
>>>> +			goto err_null;
>>>> +		}
>>>> +
>>>> +		/* check property */
>>>
>>> Unnecessary comment.
>>
>> - done.
>>
>>>
>>>> +		if (ippdrv->check_property &&
>>>> +		    ippdrv->check_property(ippdrv->dev, property)) {
>>>> +			DRM_ERROR("not support property.\n");
>>>> +			goto err_null;
>>>> +		}
>>>> +
>>>> +		return ippdrv;
>>>> +	} else {
>>>> +		/* get ipp driver entry */
>>>> +		list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list,
>>>> drv_list) {
>>>> +			/* check dedicated state */
>>>
>>> Add more comments to what is dedicated?
>>
>> - done.
>>
>>>
>>>> +			if (ippdrv->dedicated)
>>>> +				continue;
>>>> +
>>>> +			if (property->cmd != IPP_CMD_M2M
>>>> +				&& !pm_runtime_suspended(ippdrv->dev)) {
>>>> +				DRM_DEBUG_KMS("%s:can't run dedicatedly.\n",
>>>> +					__func__);
>>>> +				continue;
>>>> +			}
>>>> +
>>>> +			/* check property */
>>>
>>> Unnecessary comment.
>>
>> - done.
>>
>>>
>>>> +			if (ippdrv->check_property &&
>>>> +			    ippdrv->check_property(ippdrv->dev, property)) {
>>>> +				DRM_DEBUG_KMS("%s:not support property.\n",
>>>> +					__func__);
>>>> +				continue;
>>>> +			}
>>>> +
>>>> +			return ippdrv;
>>>> +		}
>>>> +
>>>> +		DRM_ERROR("not support ipp driver operations.\n");
>>>> +	}
>>>> +
>>>> +	return ERR_PTR(-ENODEV);
>>>> +
>>>> +err_null:
>>>> +	return NULL;
>>>> +}
>>>> +
>>>> +static struct exynos_drm_ippdrv *ipp_find_drv_node(u32 prop_id)
>>>> +{
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +	struct drm_exynos_ipp_cmd_node *c_node;
>>>> +	int count = 0;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:prop_id[%d]\n", __func__, prop_id);
>>>> +
>>>> +	if (list_empty(&exynos_drm_ippdrv_list)) {
>>>> +		DRM_DEBUG_KMS("%s:ippdrv_list is empty.\n",
>>>> +			__func__);
>>>> +		return NULL;
>>>
>>> Return ERR_PTR(error);
>>
>> - done.
>>
>>>
>>>> +	}
>>>> +
>>>> +	list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
>>>> +		DRM_DEBUG_KMS("%s:count[%d]ippdrv[0x%x]\n",
>>>> +			__func__, count++, (int)ippdrv);
>>>> +
>>>> +		if (!list_empty(&ippdrv->cmd_list)) {
>>>> +			list_for_each_entry(c_node, &ippdrv->cmd_list, list)
>>> {
>>>> +				if (c_node->property.prop_id == prop_id)
>>>> +					return ippdrv;
>>>> +			}
>>>> +		}
>>>> +	}
>>>> +
>>>> +	return NULL;
>>>
>>> Return ERR_PTR(error);
>>
>> - done.
>>
>>>
>>>> +}
>>>> +
>>>> +int exynos_drm_ipp_get_property(struct drm_device *drm_dev, void
> *data,
>>>> +		struct drm_file *file)
>>>> +{
>>>> +	struct drm_exynos_file_private *file_priv = file->driver_priv;
>>>> +	struct exynos_drm_ipp_private *priv = file_priv->ipp_priv;
>>>> +	struct device *dev = priv->dev;
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +	struct drm_exynos_ipp_prop_list *prop_list = data;
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +	int count = 0;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	if (!ctx) {
>>>> +		DRM_ERROR("invalid context.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	if (!prop_list) {
>>>> +		DRM_ERROR("invalid property parameter.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:ipp_id[%d]\n", __func__, prop_list->ipp_id);
>>>> +
>>>> +	if (prop_list->ipp_id == 0) {
>>>> +		list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list,
>>>> drv_list)
>>>
>>> Where is ippdrv used at? Remove the above line if ippdrv isn't used.
>>
>> - When userland appl check ippdrv count. we give current ippdrv count to
>> userland.
>>     same with connector, encoder list.
>>     we need this code.
>>
>>>
>>>> +			count++;
>>>> +		prop_list->count = count;
>>>
>>> What does prop_list->count mean? Add comment.
>>
>> - done.
>>
>>>
>>>> +	} else {
>>>
>>> Declare struct exynos_drm_ippdrv *ippdrv at here and remove the above
>> one.
>>>
>>>> +		ippdrv = ipp_find_id(&ctx->ipp_idr, &ctx->ipp_lock,
>>>> +						   prop_list->ipp_id);
>>>> +
>>>> +		if (!ippdrv) {
>>>> +			DRM_ERROR("not found ipp%d driver.\n",
>>>> +					prop_list->ipp_id);
>>>> +			return -EINVAL;
>>>> +		}
>>>> +
>>>> +		prop_list = ippdrv->prop_list;
>>>> +	}
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(exynos_drm_ipp_get_property);
>>>> +
>>>> +int exynos_drm_ipp_set_property(struct drm_device *drm_dev, void
> *data,
>>>> +		struct drm_file *file)
>>>> +{
>>>> +	struct drm_exynos_file_private *file_priv = file->driver_priv;
>>>> +	struct exynos_drm_ipp_private *priv = file_priv->ipp_priv;
>>>> +	struct device *dev = priv->dev;
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +	struct drm_exynos_ipp_property *property = data;
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +	struct drm_exynos_ipp_cmd_node *c_node;
>>>> +	struct drm_exynos_ipp_config *config;
>>>> +	struct drm_exynos_pos *pos;
>>>> +	struct drm_exynos_sz *sz;
>>>> +	int ret, i;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	if (!ctx) {
>>>> +		DRM_ERROR("invalid context.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	if (!property) {
>>>> +		DRM_ERROR("invalid property parameter.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	for_each_ipp_ops(i) {
>>>> +		config = &property->config[i];
>>>> +		pos = &config->pos;
>>>> +		sz = &config->sz;
>>>> +
>>>> +		DRM_DEBUG_KMS("%s:prop_id[%d]ops[%s]fmt[0x%x]\n",
>>>> +			__func__, property->prop_id,
>>>> +			i ? "dst" : "src", config->fmt);
>>>> +
>>>> +		DRM_DEBUG_KMS("%s:pos[%d %d %d %d]sz[%d %d]f[%d]r[%d]\n",
>>>> +			__func__, pos->x, pos->y, pos->w, pos->h,
>>>> +			sz->hsize, sz->vsize, config->flip, config->degree);
>>>> +	}
>>>> +
>>
>> - I added comment in this.
>>
>>>> +	if (property->prop_id) {
>>>> +		ippdrv = ipp_find_drv_node(property->prop_id);
>>>> +		if (!ippdrv) {
>>>> +			DRM_ERROR("failed to get ipp driver.\n");
>>>> +			return -EINVAL;
>>>> +		}
>>>> +
>>>
>>> Add comments. what does the below codes mean?
>>
>> - done. and When we use generated prop_id from userland.
>>     this scenario support.(e.g pause state)
>>
>>>
>>>> +		list_for_each_entry(c_node, &ippdrv->cmd_list, list) {
>>>> +			if ((c_node->property.prop_id ==
>>>> +				property->prop_id) &&
>>>> +				(c_node->state == IPP_STATE_STOP)) {
>>>> +				DRM_DEBUG_KMS("%s:found
>>> cmd[%d]ippdrv[0x%x]\n",
>>>> +					__func__, property->cmd,
>>> (int)ippdrv);
>>>> +
>>>> +				c_node->property = *property;
>>>> +				return 0;
>>>> +			}
>>>> +		}
>>>> +
>>>> +		DRM_ERROR("failed to search property.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	/* find ipp driver using ipp id */
>>>> +	ippdrv = ipp_find_driver(ctx, property);
>>>> +	if (IS_ERR_OR_NULL(ippdrv)) {
>>>> +		DRM_ERROR("failed to get ipp driver.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	/* allocate command node */
>>>> +	c_node = kzalloc(sizeof(*c_node), GFP_KERNEL);
>>>> +	if (!c_node) {
>>>> +		DRM_ERROR("failed to allocate map node.\n");
>>>> +		return -ENOMEM;
>>>> +	}
>>>> +
>>>> +	/* create property id */
>>>> +	ret = ipp_create_id(&ctx->prop_idr, &ctx->prop_lock, c_node,
>>>> +		&property->prop_id);
>>>> +	if (ret) {
>>>> +		DRM_ERROR("failed to create id.\n");
>>>> +		goto err_clear;
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:created prop_id[%d]cmd[%d]ippdrv[0x%x]\n",
>>>> +		__func__, property->prop_id, property->cmd, (int)ippdrv);
>>>> +
>>>> +	/* stored property information and ippdrv in private data */
>>>> +	c_node->priv = priv;
>>>> +	c_node->property = *property;
>>>> +	c_node->state = IPP_STATE_IDLE;
>>>> +
>>>> +	c_node->start_work = kzalloc(sizeof(*c_node->start_work),
>>>> +		GFP_KERNEL);
>>>> +	if (!c_node->start_work) {
>>>> +		DRM_ERROR("failed to alloc start_work.\n");
>>>> +		ret = -ENOMEM;
>>>> +		goto err_clear;
>>>> +	}
>>>> +
>>>> +	INIT_WORK((struct work_struct *)c_node->start_work,
>>>> +		ipp_sched_cmd);
>>>> +
>>>> +	c_node->stop_work = kzalloc(sizeof(*c_node->stop_work),
>>>> +		GFP_KERNEL);
>>>> +	if (!c_node->stop_work) {
>>>> +		DRM_ERROR("failed to alloc stop_work.\n");
>>>> +		ret = -ENOMEM;
>>>> +		goto err_free_start;
>>>> +	}
>>>> +
>>>> +	INIT_WORK((struct work_struct *)c_node->stop_work,
>>>> +		ipp_sched_cmd);
>>>> +
>>>> +	c_node->event_work = kzalloc(sizeof(*c_node->event_work),
>>>> +		GFP_KERNEL);
>>>> +	if (!c_node->event_work) {
>>>> +		DRM_ERROR("failed to alloc event_work.\n");
>>>> +		ret = -ENOMEM;
>>>> +		goto err_free_stop;
>>>> +	}
>>>> +
>>>> +	INIT_WORK((struct work_struct *)c_node->event_work,
>>>> +		ipp_sched_event);
>>>> +
>>>> +	/* init ioctl lock */
>>>> +	mutex_init(&c_node->cmd_lock);
>>>> +	mutex_init(&c_node->mem_lock);
>>>> +	mutex_init(&c_node->event_lock);
>>>> +	init_completion(&c_node->start_complete);
>>>> +	init_completion(&c_node->stop_complete);
>>>> +
>>>> +	for_each_ipp_ops(i)
>>>> +		INIT_LIST_HEAD(&c_node->mem_list[i]);
>>>> +
>>>> +	INIT_LIST_HEAD(&c_node->event_list);
>>>> +	list_splice_init(&priv->event_list, &c_node->event_list);
>>>> +	list_add_tail(&c_node->list, &ippdrv->cmd_list);
>>>> +
>>>> +	/* make dedicated state without m2m */
>>>> +	if (property->cmd != IPP_CMD_M2M)
>>>> +		ippdrv->dedicated = true;
>>>> +
>>>> +	return 0;
>>>> +
>>>> +err_free_stop:
>>>> +	kfree(c_node->stop_work);
>>>> +err_free_start:
>>>> +	kfree(c_node->start_work);
>>>> +err_clear:
>>>> +	kfree(c_node);
>>>> +	return ret;
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(exynos_drm_ipp_set_property);
>>>> +
>>>> +static struct drm_exynos_ipp_mem_node
>>>> +		*ipp_find_mem_node(struct drm_exynos_ipp_cmd_node *c_node,
>>>> +		struct drm_exynos_ipp_queue_buf *qbuf)
>>>> +{
>>>> +	struct drm_exynos_ipp_mem_node *m_node;
>>>> +	struct list_head *head;
>>>> +	int count = 0;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:buf_id[%d]\n", __func__, qbuf->buf_id);
>>>> +
>>>> +	/* source/destination memory list */
>>>> +	head = &c_node->mem_list[qbuf->ops_id];
>>>> +
>>>> +	/* find memory node entry */
>>>> +	list_for_each_entry(m_node, head, list) {
>>>> +		DRM_DEBUG_KMS("%s:count[%d]m_node[0x%x]\n",
>>>> +			__func__, count++, (int)m_node);
>>>> +
>>>> +		/* compare buffer id */
>>>> +		if (m_node->buf_id == qbuf->buf_id)
>>>> +			return m_node;
>>>> +	}
>>>> +
>>>> +	return NULL;
>>>> +}
>>>> +
>>>> +static int ipp_check_mem_list(struct drm_exynos_ipp_cmd_node *c_node)
>>>> +{
>>>> +	struct drm_exynos_ipp_property *property = &c_node->property;
>>>> +	struct drm_exynos_ipp_mem_node *m_node;
>>>> +	struct list_head *head;
>>>> +	int ret, i, count[EXYNOS_DRM_OPS_MAX] = { 0, };
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	mutex_lock(&c_node->mem_lock);
>>>> +
>>>> +	for_each_ipp_ops(i) {
>>>> +		/* source/destination memory list */
>>>> +		head = &c_node->mem_list[i];
>>>> +
>>>> +		if (list_empty(head)) {
>>>> +			DRM_DEBUG_KMS("%s:%s memory empty.\n", __func__,
>>>> +				i ? "dst" : "src");
>>>> +			continue;
>>>> +		}
>>>> +
>>>> +		/* find memory node entry */
>>>> +		list_for_each_entry(m_node, head, list) {
>>>> +			DRM_DEBUG_KMS("%s:%s,count[%d]m_node[0x%x]\n",
>>>> __func__,
>>>> +				i ? "dst" : "src", count[i], (int)m_node);
>>>> +			count[i]++;
>>>> +		}
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:min[%d]max[%d]\n", __func__,
>>>> +		min(count[EXYNOS_DRM_OPS_SRC], count[EXYNOS_DRM_OPS_DST]),
>>>> +		max(count[EXYNOS_DRM_OPS_SRC], count[EXYNOS_DRM_OPS_DST]));
>>>> +
>>>> +
>>>
>>> Add comments to why do the below codes need?
>>
>> - done. and M2M operations should be need paired memory.
>>     so, we use minimum value.
>>     other case use max value.(e.g writeback, output)
>>
>>>
>>>> +	if (property->cmd == IPP_CMD_M2M)
>>>> +		ret = min(count[EXYNOS_DRM_OPS_SRC],
>>>> +			count[EXYNOS_DRM_OPS_DST]);
>>>> +	else
>>>> +		ret = max(count[EXYNOS_DRM_OPS_SRC],
>>>> +			count[EXYNOS_DRM_OPS_DST]);
>>>> +
>>>> +	mutex_unlock(&c_node->mem_lock);
>>>> +
>>>> +	return ret;
>>>> +}
>>>> +
>>>> +static void ipp_clean_cmd_node(struct drm_exynos_ipp_cmd_node *c_node)
>>>> +{
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	/* delete list */
>>>> +	list_del(&c_node->list);
>>>> +
>>>> +	/* destroy mutex */
>>>> +	mutex_destroy(&c_node->cmd_lock);
>>>> +	mutex_destroy(&c_node->mem_lock);
>>>> +	mutex_destroy(&c_node->event_lock);
>>>> +
>>>> +	/* free command node */
>>>> +	kfree(c_node->start_work);
>>>> +	kfree(c_node->stop_work);
>>>> +	kfree(c_node->event_work);
>>>> +	kfree(c_node);
>>>> +}
>>>> +
>>>> +static int ipp_set_mem_node(struct exynos_drm_ippdrv *ippdrv,
>>>> +		struct drm_exynos_ipp_cmd_node *c_node,
>>>> +		struct drm_exynos_ipp_mem_node *m_node)
>>>> +{
>>>> +	struct exynos_drm_ipp_ops *ops = NULL;
>>>> +	int ret = 0;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:node[0x%x]\n", __func__, (int)m_node);
>>>> +
>>>> +	if (!m_node) {
>>>> +		DRM_ERROR("invalid queue node.\n");
>>>> +		return -EFAULT;
>>>> +	}
>>>> +
>>>> +	mutex_lock(&c_node->mem_lock);
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:ops_id[%d]\n", __func__, m_node->ops_id);
>>>> +
>>>> +	/* get operations callback */
>>>> +	ops = ippdrv->ops[m_node->ops_id];
>>>> +	if (!ops) {
>>>> +		DRM_ERROR("not support ops.\n");
>>>> +		ret = -EFAULT;
>>>> +		goto err_unlock;
>>>> +	}
>>>> +
>>>> +	/* set address and enable irq */
>>>> +	if (ops->set_addr) {
>>>> +		ret = ops->set_addr(ippdrv->dev, &m_node->buf_info,
>>>> +			m_node->buf_id, IPP_BUF_ENQUEUE);
>>>> +		if (ret) {
>>>> +			DRM_ERROR("failed to set addr.\n");
>>>> +			goto err_unlock;
>>>> +		}
>>>> +	}
>>>> +
>>>> +err_unlock:
>>>> +	mutex_unlock(&c_node->mem_lock);
>>>> +	return ret;
>>>> +}
>>>> +
>>>> +static struct drm_exynos_ipp_mem_node
>>>> +		*ipp_get_mem_node(struct drm_device *drm_dev,
>>>> +		struct drm_file *file,
>>>> +		struct drm_exynos_ipp_cmd_node *c_node,
>>>> +		struct drm_exynos_ipp_queue_buf *qbuf)
>>>> +{
>>>> +	struct drm_exynos_ipp_mem_node *m_node;
>>>> +	struct drm_exynos_ipp_buf_info buf_info;
>>>> +	void *addr;
>>>> +	int i;
>>>> +
>>>> +	mutex_lock(&c_node->mem_lock);
>>>> +
>>>> +	m_node = kzalloc(sizeof(*m_node), GFP_KERNEL);
>>>> +	if (!m_node) {
>>>> +		DRM_ERROR("failed to allocate queue node.\n");
>>>> +		goto err_unlock;
>>>> +	}
>>>> +
>>>> +	/* clear base address for error handling */
>>>> +	memset(&buf_info, 0x0, sizeof(buf_info));
>>>> +
>>>> +	/* operations, buffer id */
>>>> +	m_node->ops_id = qbuf->ops_id;
>>>> +	m_node->prop_id = qbuf->prop_id;
>>>> +	m_node->buf_id = qbuf->buf_id;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:m_node[0x%x]ops_id[%d]\n", __func__,
>>>> +		(int)m_node, qbuf->ops_id);
>>>> +	DRM_DEBUG_KMS("%s:prop_id[%d]buf_id[%d]\n", __func__,
>>>> +		qbuf->prop_id, m_node->buf_id);
>>>> +
>>>> +	for_each_ipp_planar(i) {
>>>> +		DRM_DEBUG_KMS("%s:i[%d]handle[0x%x]\n", __func__,
>>>> +			i, qbuf->handle[i]);
>>>> +
>>>> +		/* get dma address by handle */
>>>> +		if (qbuf->handle[i] != 0) {
>>>> +			addr = exynos_drm_gem_get_dma_addr(drm_dev,
>>>> +					qbuf->handle[i], file);
>>>> +			if (!addr) {
>>>> +				DRM_ERROR("failed to get addr.\n");
>>>> +				goto err_clear;
>>>> +			}
>>>> +
>>>> +			buf_info.handles[i] = qbuf->handle[i];
>>>> +			buf_info.base[i] = *(dma_addr_t *) addr;
>>>> +			DRM_DEBUG_KMS("%s:i[%d]base[0x%x]hd[0x%x]\n",
>>>> +				__func__, i, buf_info.base[i],
>>>> +				(int)buf_info.handles[i]);
>>>> +		}
>>>> +	}
>>>> +
>>>> +	m_node->filp = file;
>>>> +	m_node->buf_info = buf_info;
>>>> +	list_add_tail(&m_node->list, &c_node->mem_list[qbuf->ops_id]);
>>>> +
>>>> +	mutex_unlock(&c_node->mem_lock);
>>>> +	return m_node;
>>>> +
>>>> +err_clear:
>>>> +	kfree(m_node);
>>>> +
>>>> +err_unlock:
>>>> +	mutex_unlock(&c_node->mem_lock);
>>>> +
>>>> +	return NULL;
>>>
>>> Return ERR_PTR(error);
>>
>> - done.
>>
>>>
>>>> +}
>>>> +
>>>> +static int ipp_put_mem_node(struct drm_device *drm_dev,
>>>> +		struct drm_exynos_ipp_cmd_node *c_node,
>>>> +		struct drm_exynos_ipp_mem_node *m_node)
>>>> +{
>>>> +	int i, ret = 0;
>>>> +
>>>
>>> Remove ret;
>>
>> - done.
>>
>>>
>>>> +	DRM_DEBUG_KMS("%s:node[0x%x]\n", __func__, (int)m_node);
>>>> +
>>>> +	if (!m_node) {
>>>> +		DRM_ERROR("invalid dequeue node.\n");
>>>> +		return -EFAULT;
>>>> +	}
>>>> +
>>>> +	if (list_empty(&m_node->list)) {
>>>> +		DRM_ERROR("empty memory node.\n");
>>>> +		return -ENOMEM;
>>>> +	}
>>>> +
>>>> +	mutex_lock(&c_node->mem_lock);
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:ops_id[%d]\n", __func__, m_node->ops_id);
>>>> +
>>>> +	/* put gem buffer */
>>>> +	for_each_ipp_planar(i) {
>>>> +		unsigned long handle = m_node->buf_info.handles[i];
>>>> +		if (handle)
>>>> +			exynos_drm_gem_put_dma_addr(drm_dev, handle,
>>>> +							m_node->filp);
>>>> +	}
>>>> +
>>>> +	/* delete list in queue */
>>>> +	list_del(&m_node->list);
>>>> +	kfree(m_node);
>>>> +
>>>> +	mutex_unlock(&c_node->mem_lock);
>>>> +	return ret;
>>>
>>> Just return 0;
>>
>> - done.
>>
>>>
>>>> +}
>>>> +
>>>> +static void ipp_free_event(struct drm_pending_event *event)
>>>> +{
>>>> +	kfree(event);
>>>> +}
>>>> +
>>>> +static int ipp_get_event(struct drm_device *drm_dev,
>>>> +		struct drm_file *file,
>>>> +		struct drm_exynos_ipp_cmd_node *c_node,
>>>> +		struct drm_exynos_ipp_queue_buf *qbuf)
>>>> +{
>>>> +	struct drm_exynos_ipp_send_event *e;
>>>> +	unsigned long flags;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:ops_id[%d]buf_id[%d]\n", __func__,
>>>> +		qbuf->ops_id, qbuf->buf_id);
>>>> +
>>>> +	e = kzalloc(sizeof(*e), GFP_KERNEL);
>>>> +
>>>> +	if (!e) {
>>>> +		DRM_ERROR("failed to allocate event.\n");
>>>> +		spin_lock_irqsave(&drm_dev->event_lock, flags);
>>>
>>> Why do you use spin_lock_irqsave?
>>
>> - removed it.
>>
>>>
>>>> +		file->event_space += sizeof(e->event);
>>>> +		spin_unlock_irqrestore(&drm_dev->event_lock, flags);
>>>> +		return -ENOMEM;
>>>> +	}
>>>> +
>>>> +	/* make event */
>>>> +	e->event.base.type = DRM_EXYNOS_IPP_EVENT;
>>>> +	e->event.base.length = sizeof(e->event);
>>>> +	e->event.user_data = qbuf->user_data;
>>>> +	e->event.prop_id = qbuf->prop_id;
>>>> +	e->event.buf_id[EXYNOS_DRM_OPS_DST] = qbuf->buf_id;
>>>> +	e->base.event = &e->event.base;
>>>> +	e->base.file_priv = file;
>>>> +	e->base.destroy = ipp_free_event;
>>>> +	list_add_tail(&e->base.link, &c_node->event_list);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +static void ipp_put_event(struct drm_exynos_ipp_cmd_node *c_node,
>>>> +		struct drm_exynos_ipp_queue_buf *qbuf)
>>>> +{
>>>> +	struct drm_exynos_ipp_send_event *e, *te;
>>>> +	int count = 0;
>>>> +
>>>> +	if (list_empty(&c_node->event_list)) {
>>>> +		DRM_DEBUG_KMS("%s:event_list is empty.\n", __func__);
>>>> +		return;
>>>> +	}
>>>> +
>>>> +	list_for_each_entry_safe(e, te, &c_node->event_list, base.link) {
>>>> +		DRM_DEBUG_KMS("%s:count[%d]e[0x%x]\n",
>>>> +			__func__, count++, (int)e);
>>>> +
>>>
>>> Add comments to why a event should be released if qbuf is NULL and what
>> does
>>> it mean if qbuf isn't NULL?
>>
>> - done. qbug == NULL means all event deletion in Stop state.
>>
>>>
>>>> +		if (!qbuf) {
>>>> +			/* delete list */
>>>> +			list_del(&e->base.link);
>>>> +			kfree(e);
>>>> +		} else if (e->event.buf_id[EXYNOS_DRM_OPS_DST]
>>>> +			== qbuf->buf_id) {
>>>> +			/* delete list */
>>>> +			list_del(&e->base.link);
>>>> +			kfree(e);
>>>> +			return;
>>>> +		}
>>>> +	}
>>>> +
>>>> +	return;
>>>> +}
>>>> +
>>>> +void ipp_handle_cmd_work(struct device *dev,
>>>> +		struct exynos_drm_ippdrv *ippdrv,
>>>> +		struct drm_exynos_ipp_cmd_work *cmd_work,
>>>> +		struct drm_exynos_ipp_cmd_node *c_node)
>>>> +{
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +
>>>> +	cmd_work->ippdrv = ippdrv;
>>>> +	cmd_work->c_node = c_node;
>>>> +	queue_work(ctx->cmd_workq, (struct work_struct *)cmd_work);
>>>> +}
>>>> +
>>>> +int exynos_drm_ipp_queue_buf(struct drm_device *drm_dev, void *data,
>>>> +		struct drm_file *file)
>>>> +{
>>>> +	struct drm_exynos_file_private *file_priv = file->driver_priv;
>>>> +	struct exynos_drm_ipp_private *priv = file_priv->ipp_priv;
>>>> +	struct device *dev = priv->dev;
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +	struct drm_exynos_ipp_queue_buf *qbuf = data;
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +	struct drm_exynos_ipp_property *property;
>>>> +	struct exynos_drm_ipp_ops *ops;
>>>> +	struct drm_exynos_ipp_cmd_node *c_node;
>>>> +	struct drm_exynos_ipp_mem_node *m_node, *tm_node;
>>>> +	int ret;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	if (!qbuf) {
>>>> +		DRM_ERROR("invalid buf parameter.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	ippdrv = ipp_find_drv_node(qbuf->prop_id);
>>>> +
>>>> +	if (!ippdrv) {
>>>> +		DRM_ERROR("failed to get ipp driver.\n");
>>>> +		return -EFAULT;
>>>> +	}
>>>> +
>>>> +	if (qbuf->ops_id >= EXYNOS_DRM_OPS_MAX) {
>>>> +		DRM_ERROR("invalid ops parameter.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	ops = ippdrv->ops[qbuf->ops_id];
>>>> +	if (!ops) {
>>>> +		DRM_ERROR("failed to get ops.\n");
>>>> +		return -EFAULT;
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:prop_id[%d]ops_id[%s]buf_id[%d]buf_type[%d]\n",
>>>> +		__func__, qbuf->prop_id, qbuf->ops_id ? "dst" : "src",
>>>> +		qbuf->buf_id, qbuf->buf_type);
>>>> +
>>>> +	/* find command node */
>>>> +	c_node = ipp_find_id(&ctx->prop_idr, &ctx->prop_lock,
>>>> +		qbuf->prop_id);
>>>> +	if (!c_node) {
>>>> +		DRM_ERROR("failed to get command node.\n");
>>>> +		return -EFAULT;
>>>> +	}
>>>> +
>>>> +	property = &c_node->property;
>>>> +	if (!property) {
>>>> +		DRM_ERROR("invalid property parameter.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	/* buffer control */
>>>> +	switch (qbuf->buf_type) {
>>>> +	case IPP_BUF_ENQUEUE:
>>>> +		/* get memory node */
>>>> +		m_node = ipp_get_mem_node(drm_dev, file, c_node, qbuf);
>>>> +		if (!m_node) {
>>>> +			DRM_ERROR("failed to get m_node.\n");
>>>> +			return -EINVAL;
>>>> +		}
>>>> +
>>>> +		if (qbuf->ops_id == EXYNOS_DRM_OPS_DST) {
>>>
>>> What if qbuf->ops_id is EXYNOS_DRM_OPS_SRC?
>>
>> - SRC operation no need make event and start operation.
>
> Add the above comment then. This is thing only you know.

- I already added comments. done.

>
>>
>>>
>>>> +			/* get event */
>>>> +			ret = ipp_get_event(drm_dev, file, c_node, qbuf);
>>>> +			if (ret) {
>>>> +				DRM_ERROR("failed to get event.\n");
>>>> +				goto err_clean_node;
>>>> +			}
>>>> +
>>>> +			if (c_node->state != IPP_STATE_START) {
>>>> +				DRM_DEBUG_KMS("%s:bypass for invalid
>>> state.\n"
>>>> +					, __func__);
>>>> +				break;
>>>> +			}
>>>> +
>>>> +			if (!ipp_check_mem_list(c_node)) {
>>>> +				DRM_DEBUG_KMS("%s:empty memory.\n"
>>>> +					, __func__);
>>>> +				break;
>>>> +			}
>>>> +
>>>> +			/*
>>>> +			 * If set source, destination buffer and enable pm
>>>> +			 * m2m operations need start operations in queue
>>>> +			 */
>>>
>>> This comment says that it sets source buffer but this case is that
>>> qbuf->ops_id has EXYNOS_DRM_OPS_SRC. It seems to be inconsistent with
>> your
>>> comment.
>>
>> - What mean that? In this case source, destination buffer already set.
>>     so, we start in queue buf operation in M2M case.
>
> The above comment mentions also source buffer, "If set source,...". Source
> buffer is set in this routine? If not so, make sure that comment.

- OK, done.

>
>>
>>>
>>>> +			if (property->cmd == IPP_CMD_M2M) {
>>>> +				struct drm_exynos_ipp_cmd_work *cmd_work =
>>>> +					c_node->start_work;
>>>> +
>>>> +				cmd_work->ctrl = IPP_CTRL_PLAY;
>>>> +				ipp_handle_cmd_work(dev, ippdrv, cmd_work,
>>>> +					c_node);
>>>> +			} else {
>>>> +				if (ops->set_addr) {
>>>> +					ret = ops->set_addr(ippdrv->dev,
>>>> +						&m_node->buf_info,
>>>> +						m_node->buf_id,
>>> qbuf->buf_type);
>>>> +					if (ret) {
>>>> +						DRM_ERROR(
>>>> +							"failed to set
>>> addr.\n");
>>>> +						goto err_clean_node;
>>>> +					}
>>>> +				}
>>>> +			}
>>>> +		}
>>>> +		break;
>>>> +	case IPP_BUF_DEQUEUE:
>>>> +		mutex_lock(&c_node->cmd_lock);
>>>> +
>>>> +		if (qbuf->ops_id == EXYNOS_DRM_OPS_DST)
>>>> +			ipp_put_event(c_node, qbuf);
>>>> +
>>>> +		if (!list_empty(&c_node->mem_list[qbuf->ops_id])) {
>>>> +			list_for_each_entry_safe(m_node, tm_node,
>>>> +				&c_node->mem_list[qbuf->ops_id], list) {
>>>> +				if (m_node->buf_id == qbuf->buf_id &&
>>>> +					m_node->ops_id == qbuf->ops_id) {
>>>> +					ipp_put_mem_node(drm_dev,
>>>> +						c_node, m_node);
>>>> +				}
>>>> +			}
>>>> +		}
>>>> +		mutex_unlock(&c_node->cmd_lock);
>>>> +		break;
>>>> +	default:
>>>> +		DRM_ERROR("invalid buffer control.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	return 0;
>>>> +
>>>> +err_clean_node:
>>>> +	DRM_ERROR("clean memory nodes.\n");
>>>> +
>>>> +	if (!list_empty(&c_node->mem_list[qbuf->ops_id])) {
>>>> +		/* delete list */
>>>> +		list_for_each_entry_safe(m_node, tm_node,
>>>> +			&c_node->mem_list[qbuf->ops_id], list) {
>>>> +			if (m_node->buf_id == qbuf->buf_id &&
>>>> +				m_node->ops_id == qbuf->ops_id)
>>>> +				ipp_put_mem_node(drm_dev, c_node, m_node);
>>>> +		}
>>>> +	}
>>>> +
>>>> +	return ret;
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(exynos_drm_ipp_queue_buf);
>>>
>>> Separate this function into some sub functions. This function uses ugly
>>> codes like below,
>>
>> - This comment is very difficult in this time.
>>     We need to support various operation in queue buf ioctl.
>>     I will change your requestion at the future.
>>
>>> 	switch
>>> 	case
>>> 		if
>>> 			if
>>> 				if
>>> 					if
>>>
>>>
>>>> +
>>>> +static bool exynos_drm_ipp_check_valid(struct device *dev,
>>>> +		enum drm_exynos_ipp_ctrl ctrl, enum drm_exynos_ipp_state
>>>> state)
>>>> +{
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	if (ctrl != IPP_CTRL_PLAY) {
>>>> +		if (pm_runtime_suspended(dev)) {
>>>> +			DRM_ERROR("pm:runtime_suspended.\n");
>>>> +			goto err_status;
>>>> +		}
>>>> +	}
>>>> +
>>>> +	switch (ctrl) {
>>>> +	case IPP_CTRL_PLAY:
>>>> +		if (state != IPP_STATE_IDLE)
>>>> +			goto err_status;
>>>> +		break;
>>>> +	case IPP_CTRL_STOP:
>>>> +		if (state == IPP_STATE_STOP)
>>>> +			goto err_status;
>>>> +		break;
>>>> +	case IPP_CTRL_PAUSE:
>>>> +		if (state != IPP_STATE_START)
>>>> +			goto err_status;
>>>> +		break;
>>>> +	case IPP_CTRL_RESUME:
>>>> +		if (state != IPP_STATE_STOP)
>>>> +			goto err_status;
>>>> +		break;
>>>> +	default:
>>>> +		DRM_ERROR("invalid state.\n");
>>>> +		goto err_status;
>>>> +		break;
>>>> +	}
>>>> +
>>>> +	return true;
>>>> +
>>>> +err_status:
>>>> +	DRM_ERROR("invalid status:ctrl[%d]state[%d]\n", ctrl, state);
>>>> +	return false;
>>>> +}
>>>> +
>>>> +int exynos_drm_ipp_cmd_ctrl(struct drm_device *drm_dev, void *data,
>>>> +		struct drm_file *file)
>>>> +{
>>>> +	struct drm_exynos_file_private *file_priv = file->driver_priv;
>>>> +	struct exynos_drm_ipp_private *priv = file_priv->ipp_priv;
>>>> +	struct exynos_drm_ippdrv *ippdrv = NULL;
>>>> +	struct device *dev = priv->dev;
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +	struct drm_exynos_ipp_cmd_ctrl *cmd_ctrl = data;
>>>> +	struct drm_exynos_ipp_cmd_work *cmd_work;
>>>> +	struct drm_exynos_ipp_cmd_node *c_node;
>>>> +	int ret = 0;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	if (!ctx) {
>>>> +		DRM_ERROR("invalid context.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	if (!cmd_ctrl) {
>>>> +		DRM_ERROR("invalid control parameter.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:ctrl[%d]prop_id[%d]\n", __func__,
>>>> +		cmd_ctrl->ctrl, cmd_ctrl->prop_id);
>>>> +
>>>> +	ippdrv = ipp_find_drv_node(cmd_ctrl->prop_id);
>>>> +
>>>> +	if (!ippdrv) {
>>>> +		DRM_ERROR("failed to get ipp driver.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	c_node = ipp_find_id(&ctx->prop_idr, &ctx->prop_lock,
>>>> +		cmd_ctrl->prop_id);
>>>> +
>>>> +	if (!c_node) {
>>>> +		DRM_ERROR("invalid command node list.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	if (!exynos_drm_ipp_check_valid(ippdrv->dev, cmd_ctrl->ctrl,
>>>> +	    c_node->state)) {
>>>> +		DRM_ERROR("invalid state.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	switch (cmd_ctrl->ctrl) {
>>>> +	case IPP_CTRL_PLAY:
>>>> +		if (pm_runtime_suspended(ippdrv->dev))
>>>> +			pm_runtime_get_sync(ippdrv->dev);
>>>> +		c_node->state = IPP_STATE_START;
>>>> +
>>>> +		cmd_work = c_node->start_work;
>>>> +		cmd_work->ctrl = cmd_ctrl->ctrl;
>>>> +		ipp_handle_cmd_work(dev, ippdrv, cmd_work, c_node);
>>>> +		break;
>>>> +	case IPP_CTRL_STOP:
>>>> +		cancel_work_sync((struct work_struct *)c_node->start_work);
>>>> +		cancel_work_sync((struct work_struct *)c_node->event_work);
>>>> +		c_node->state = IPP_STATE_STOP;
>>>> +
>>>> +		cmd_work = c_node->stop_work;
>>>> +		cmd_work->ctrl = cmd_ctrl->ctrl;
>>>> +		ipp_handle_cmd_work(dev, ippdrv, cmd_work, c_node);
>>>> +
>>>> +		if (!wait_for_completion_timeout(&c_node->stop_complete,
>>>> +		    msecs_to_jiffies(200))) {
>>>> +			DRM_ERROR("timeout stop:prop_id[%d]\n",
>>>> +				c_node->property.prop_id);
>>>> +		}
>>>> +
>>>> +		ippdrv->dedicated = false;
>>>> +		ipp_clean_cmd_node(c_node);
>>>> +
>>>> +		if (list_empty(&ippdrv->cmd_list))
>>>> +			pm_runtime_put_sync(ippdrv->dev);
>>>> +		break;
>>>> +	case IPP_CTRL_PAUSE:
>>>> +		cancel_work_sync((struct work_struct *)c_node->start_work);
>>>> +		cancel_work_sync((struct work_struct *)c_node->event_work);
>>>> +		c_node->state = IPP_STATE_STOP;
>>>> +
>>>> +		cmd_work = c_node->stop_work;
>>>> +		cmd_work->ctrl = cmd_ctrl->ctrl;
>>>> +		ipp_handle_cmd_work(dev, ippdrv, cmd_work, c_node);
>>>> +
>>>> +		if (!wait_for_completion_timeout(&c_node->stop_complete,
>>>> +		    msecs_to_jiffies(200))) {
>>>> +			DRM_ERROR("timeout stop:prop_id[%d]\n",
>>>> +				c_node->property.prop_id);
>>>> +		}
>>>> +		break;
>>>> +	case IPP_CTRL_RESUME:
>>>> +		c_node->state = IPP_STATE_START;
>>>> +
>>>> +		cmd_work = c_node->start_work;
>>>> +		cmd_work->ctrl = cmd_ctrl->ctrl;
>>>> +		ipp_handle_cmd_work(dev, ippdrv, cmd_work, c_node);
>>>> +		break;
>>>> +	default:
>>>> +		/* ToDo: expand ctrl operation */
>>>> +		DRM_ERROR("could not support this state currently.\n");
>>>> +		goto err_clear;
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:done ctrl[%d]prop_id[%d]\n", __func__,
>>>> +		cmd_ctrl->ctrl, cmd_ctrl->prop_id);
>>>> +
>>>> +	return 0;
>>>> +
>>>> +err_clear:
>>>> +	return ret;
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(exynos_drm_ipp_cmd_ctrl);
>>>> +
>>>> +int exynos_drm_ippnb_register(struct notifier_block *nb)
>>>> +{
>>>> +	return blocking_notifier_chain_register(
>>>> +			&exynos_drm_ippnb_list, nb);
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(exynos_drm_ippnb_register);
>>>> +
>>>> +int exynos_drm_ippnb_unregister(struct notifier_block *nb)
>>>> +{
>>>> +	return blocking_notifier_chain_unregister(
>>>> +			&exynos_drm_ippnb_list, nb);
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(exynos_drm_ippnb_unregister);
>>>> +
>>>> +int exynos_drm_ippnb_send_event(unsigned long val, void *v)
>>>> +{
>>>> +	return blocking_notifier_call_chain(
>>>> +			&exynos_drm_ippnb_list, val, v);
>>>> +}
>>>> +EXPORT_SYMBOL_GPL(exynos_drm_ippnb_send_event);
>>>> +
>>>> +static int ipp_set_property(struct exynos_drm_ippdrv *ippdrv,
>>>> +		struct drm_exynos_ipp_property *property)
>>>> +{
>>>> +	struct exynos_drm_ipp_ops *ops = NULL;
>>>> +	int ret, i, swap = 0;
>>>> +
>>>> +	if (!property) {
>>>> +		DRM_ERROR("invalid property parameter.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:prop_id[%d]\n", __func__, property->prop_id);
>>>> +
>>>> +	/* reset h/w block */
>>>> +	if (ippdrv->reset &&
>>>> +		ippdrv->reset(ippdrv->dev)) {
>>>> +		DRM_ERROR("failed to reset.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	/* set source,destination operations */
>>>> +	for_each_ipp_ops(i) {
>>>> +		/* ToDo: integrate property and config */
>>>> +		struct drm_exynos_ipp_config *config =
>>>> +			&property->config[i];
>>>> +
>>>> +		ops = ippdrv->ops[i];
>>>> +		if (!ops || !config) {
>>>> +			DRM_ERROR("not support ops and config.\n");
>>>> +			return -EINVAL;
>>>> +		}
>>>> +
>>>> +		/* set format */
>>>> +		if (ops->set_fmt) {
>>>> +			ret = ops->set_fmt(ippdrv->dev, config->fmt);
>>>> +			if (ret) {
>>>> +				DRM_ERROR("not support format.\n");
>>>> +				return ret;
>>>> +			}
>>>> +		}
>>>> +
>>>> +		/* set transform for rotation, flip */
>>>> +		if (ops->set_transf) {
>>>> +			swap = ops->set_transf(ippdrv->dev, config->degree,
>>>> +				config->flip);
>>>> +			if (swap < 0) {
>>>> +				DRM_ERROR("not support tranf.\n");
>>>> +				return -EINVAL;
>>>> +			}
>>>> +		}
>>>> +
>>>> +		/* set size */
>>>> +		if (ops->set_size) {
>>>> +			ret = ops->set_size(ippdrv->dev, swap, &config->pos,
>>>> +				&config->sz);
>>>> +			if (ret) {
>>>> +				DRM_ERROR("not support size.\n");
>>>> +				return ret;
>>>> +			}
>>>> +		}
>>>> +	}
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +static int ipp_start_property(struct exynos_drm_ippdrv *ippdrv,
>>>> +		struct drm_exynos_ipp_cmd_node *c_node)
>>>> +{
>>>> +	struct drm_exynos_ipp_mem_node *m_node;
>>>> +	struct drm_exynos_ipp_property *property = &c_node->property;
>>>> +	struct list_head *head;
>>>> +	int ret, i;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:prop_id[%d]\n", __func__, property->prop_id);
>>>> +
>>>> +	/* store command info in ippdrv */
>>>> +	ippdrv->cmd = c_node;
>>>> +
>>>> +	if (!ipp_check_mem_list(c_node)) {
>>>> +		DRM_DEBUG_KMS("%s:empty memory.\n", __func__);
>>>> +		return -ENOMEM;
>>>> +	}
>>>> +
>>>> +	/* set current property in ippdrv */
>>>> +	ret = ipp_set_property(ippdrv, property);
>>>> +	if (ret) {
>>>> +		DRM_ERROR("failed to set property.\n");
>>>> +		ippdrv->cmd = NULL;
>>>> +		return ret;
>>>> +	}
>>>> +
>>>> +	/* check command */
>>>> +	switch (property->cmd) {
>>>> +	case IPP_CMD_M2M:
>>>> +		for_each_ipp_ops(i) {
>>>> +			/* source/destination memory list */
>>>> +			head = &c_node->mem_list[i];
>>>> +
>>>> +			m_node = list_first_entry(head,
>>>> +				struct drm_exynos_ipp_mem_node, list);
>>>> +			if (!m_node) {
>>>> +				DRM_ERROR("failed to get node.\n");
>>>> +				ret = -EFAULT;
>>>> +				return ret;
>>>> +			}
>>>> +
>>>> +			DRM_DEBUG_KMS("%s:m_node[0x%x]\n",
>>>> +				__func__, (int)m_node);
>>>> +
>>>> +			ret = ipp_set_mem_node(ippdrv, c_node, m_node);
>>>> +			if (ret) {
>>>> +				DRM_ERROR("failed to set m node.\n");
>>>> +				return ret;
>>>> +			}
>>>> +		}
>>>> +		break;
>>>> +	case IPP_CMD_WB:
>>>> +		/* destination memory list */
>>>> +		head = &c_node->mem_list[EXYNOS_DRM_OPS_DST];
>>>> +
>>>> +		list_for_each_entry(m_node, head, list) {
>>>> +			ret = ipp_set_mem_node(ippdrv, c_node, m_node);
>>>> +			if (ret) {
>>>> +				DRM_ERROR("failed to set m node.\n");
>>>> +				return ret;
>>>> +			}
>>>> +		}
>>>> +		break;
>>>> +	case IPP_CMD_OUTPUT:
>>>> +		/* source memory list */
>>>> +		head = &c_node->mem_list[EXYNOS_DRM_OPS_SRC];
>>>> +
>>>> +		list_for_each_entry(m_node, head, list) {
>>>> +			ret = ipp_set_mem_node(ippdrv, c_node, m_node);
>>>> +			if (ret) {
>>>> +				DRM_ERROR("failed to set m node.\n");
>>>> +				return ret;
>>>> +			}
>>>> +		}
>>>> +		break;
>>>> +	default:
>>>> +		DRM_ERROR("invalid operations.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:cmd[%d]\n", __func__, property->cmd);
>>>> +
>>>> +	/* start operations */
>>>> +	if (ippdrv->start) {
>>>> +		ret = ippdrv->start(ippdrv->dev, property->cmd);
>>>> +		if (ret) {
>>>> +			DRM_ERROR("failed to start ops.\n");
>>>> +			return ret;
>>>> +		}
>>>> +	}
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +static int ipp_stop_property(struct drm_device *drm_dev,
>>>> +		struct exynos_drm_ippdrv *ippdrv,
>>>> +		struct drm_exynos_ipp_cmd_node *c_node)
>>>> +{
>>>> +	struct drm_exynos_ipp_mem_node *m_node, *tm_node;
>>>> +	struct drm_exynos_ipp_property *property = &c_node->property;
>>>> +	struct list_head *head;
>>>> +	int ret, i;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:prop_id[%d]\n", __func__, property->prop_id);
>>>> +
>>>> +	/* put event */
>>>> +	ipp_put_event(c_node, NULL);
>>>> +
>>>> +	/* check command */
>>>> +	switch (property->cmd) {
>>>> +	case IPP_CMD_M2M:
>>>> +		for_each_ipp_ops(i) {
>>>> +			/* source/destination memory list */
>>>> +			head = &c_node->mem_list[i];
>>>> +
>>>> +			if (list_empty(head)) {
>>>> +				DRM_DEBUG_KMS("%s:mem_list is empty.\n",
>>>> +					__func__);
>>>> +				break;
>>>> +			}
>>>> +
>>>> +			list_for_each_entry_safe(m_node, tm_node,
>>>> +				head, list) {
>>>> +				ret = ipp_put_mem_node(drm_dev, c_node,
>>>> +					m_node);
>>>> +				if (ret) {
>>>> +					DRM_ERROR("failed to put
>>> m_node.\n");
>>>> +					goto err_clear;
>>>> +				}
>>>> +			}
>>>> +		}
>>>> +		break;
>>>> +	case IPP_CMD_WB:
>>>> +		/* destination memory list */
>>>> +		head = &c_node->mem_list[EXYNOS_DRM_OPS_DST];
>>>> +
>>>> +		if (list_empty(head)) {
>>>> +			DRM_DEBUG_KMS("%s:mem_list is empty.\n", __func__);
>>>> +			break;
>>>> +		}
>>>> +
>>>> +		list_for_each_entry_safe(m_node, tm_node, head, list) {
>>>> +			ret = ipp_put_mem_node(drm_dev, c_node, m_node);
>>>> +			if (ret) {
>>>> +				DRM_ERROR("failed to put m_node.\n");
>>>> +				goto err_clear;
>>>> +			}
>>>> +		}
>>>> +		break;
>>>> +	case IPP_CMD_OUTPUT:
>>>> +		/* source memory list */
>>>> +		head = &c_node->mem_list[EXYNOS_DRM_OPS_SRC];
>>>> +
>>>> +		if (list_empty(head)) {
>>>> +			DRM_DEBUG_KMS("%s:mem_list is empty.\n", __func__);
>>>> +			break;
>>>> +		}
>>>> +
>>>> +		list_for_each_entry_safe(m_node, tm_node, head, list) {
>>>> +			ret = ipp_put_mem_node(drm_dev, c_node, m_node);
>>>> +			if (ret) {
>>>> +				DRM_ERROR("failed to put m_node.\n");
>>>> +				goto err_clear;
>>>> +			}
>>>> +		}
>>>> +		break;
>>>> +	default:
>>>> +		DRM_ERROR("invalid operations.\n");
>>>> +		ret = -EINVAL;
>>>> +		goto err_clear;
>>>> +	}
>>>> +
>>>> +err_clear:
>>>> +	/* stop operations */
>>>> +	if (ippdrv->stop)
>>>> +		ippdrv->stop(ippdrv->dev, property->cmd);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +void ipp_sched_cmd(struct work_struct *work)
>>>
>>> Use int type instead of void and return proper error type.
>>
>> - This is just work thread handler api. Who handle this error ?
>>     I think that is no needed.
>>
>
> Just ignore it. This is work queue handler.
>
>>>
>>>> +{
>>>> +	struct drm_exynos_ipp_cmd_work *cmd_work =
>>>> +		(struct drm_exynos_ipp_cmd_work *)work;
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +	struct drm_exynos_ipp_cmd_node *c_node;
>>>> +	struct drm_exynos_ipp_property *property;
>>>> +	int ret;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	ippdrv = cmd_work->ippdrv;
>>>> +	if (!ippdrv) {
>>>> +		DRM_ERROR("invalid ippdrv list.\n");
>>>> +		return;
>>>> +	}
>>>> +
>>>> +	c_node = cmd_work->c_node;
>>>> +	if (!c_node) {
>>>> +		DRM_ERROR("invalid command node list.\n");
>>>> +		return;
>>>> +	}
>>>> +
>>>> +	mutex_lock(&c_node->cmd_lock);
>>>> +
>>>> +	property = &c_node->property;
>>>> +	if (!property) {
>>>> +		DRM_ERROR("failed to get property:prop_id[%d]\n",
>>>> +			c_node->property.prop_id);
>>>> +		goto err_unlock;
>>>> +	}
>>>> +
>>>> +	switch (cmd_work->ctrl) {
>>>> +	case IPP_CTRL_PLAY:
>>>> +	case IPP_CTRL_RESUME:
>>>> +		ret = ipp_start_property(ippdrv, c_node);
>>>> +		if (ret) {
>>>> +			DRM_ERROR("failed to start property:prop_id[%d]\n",
>>>> +				c_node->property.prop_id);
>>>> +			goto err_unlock;
>>>> +		}
>>>> +
>>>> +		/*
>>>> +		 * M2M case supports wait_completion of transfer.
>>>> +		 * because M2M case supports single unit operation
>>>> +		 * with multiple queue.
>>>> +		 * M2M need to wait completion of data transfer.
>>>> +		 */
>>>> +		if (property->cmd == IPP_CMD_M2M) {
>>>> +			if (!wait_for_completion_timeout
>>>> +			    (&c_node->start_complete,
>>> msecs_to_jiffies(200))) {
>>>> +				DRM_ERROR("timeout event:prop_id[%d]\n",
>>>> +					c_node->property.prop_id);
>>>> +				goto err_unlock;
>>>> +			}
>>>> +		}
>>>> +		break;
>>>> +	case IPP_CTRL_STOP:
>>>> +	case IPP_CTRL_PAUSE:
>>>> +		ret = ipp_stop_property(ippdrv->drm_dev, ippdrv,
>>>> +			c_node);
>>>> +		if (ret) {
>>>> +			DRM_ERROR("failed to stop property.\n");
>>>> +			goto err_unlock;
>>>> +		}
>>>> +
>>>> +		complete(&c_node->stop_complete);
>>>> +		break;
>>>> +	default:
>>>> +		DRM_ERROR("unknown control type\n");
>>>> +		break;
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:ctrl[%d] done.\n", __func__, cmd_work->ctrl);
>>>> +
>>>> +err_unlock:
>>>> +	mutex_unlock(&c_node->cmd_lock);
>>>> +}
>>>> +
>>>> +static int ipp_send_event(struct exynos_drm_ippdrv *ippdrv,
>>>> +		struct drm_exynos_ipp_cmd_node *c_node, int *buf_id)
>>>> +{
>>>> +	struct drm_device *drm_dev = ippdrv->drm_dev;
>>>> +	struct drm_exynos_ipp_property *property = &c_node->property;
>>>> +	struct drm_exynos_ipp_mem_node *m_node;
>>>> +	struct drm_exynos_ipp_queue_buf qbuf;
>>>> +	struct drm_exynos_ipp_send_event *e;
>>>> +	struct list_head *head;
>>>> +	struct timeval now;
>>>> +	unsigned long flags;
>>>> +	u32 tbuf_id[EXYNOS_DRM_OPS_MAX] = {0, };
>>>> +	int ret, i;
>>>> +
>>>> +	for_each_ipp_ops(i)
>>>> +		DRM_DEBUG_KMS("%s:%s buf_id[%d]\n", __func__,
>>>> +			i ? "dst" : "src", buf_id[i]);
>>>> +
>>>> +	if (!drm_dev) {
>>>> +		DRM_ERROR("failed to get drm_dev.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	if (!property) {
>>>> +		DRM_ERROR("failed to get property.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	if (list_empty(&c_node->event_list)) {
>>>> +		DRM_DEBUG_KMS("%s:event list is empty.\n", __func__);
>>>> +		return 0;
>>>> +	}
>>>> +
>>>> +	if (!ipp_check_mem_list(c_node)) {
>>>> +		DRM_DEBUG_KMS("%s:empty memory.\n", __func__);
>>>> +		return 0;
>>>> +	}
>>>> +
>>>> +	/* check command */
>>>> +	switch (property->cmd) {
>>>> +	case IPP_CMD_M2M:
>>>> +		for_each_ipp_ops(i) {
>>>> +			/* source/destination memory list */
>>>> +			head = &c_node->mem_list[i];
>>>> +
>>>> +			m_node = list_first_entry(head,
>>>> +				struct drm_exynos_ipp_mem_node, list);
>>>> +			if (!m_node) {
>>>> +				DRM_ERROR("empty memory node.\n");
>>>> +				return -ENOMEM;
>>>> +			}
>>>> +
>>>> +			tbuf_id[i] = m_node->buf_id;
>>>> +			DRM_DEBUG_KMS("%s:%s buf_id[%d]\n", __func__,
>>>> +				i ? "dst" : "src", tbuf_id[i]);
>>>> +
>>>> +			ret = ipp_put_mem_node(drm_dev, c_node, m_node);
>>>> +			if (ret)
>>>> +				DRM_ERROR("failed to put m_node.\n");
>>>> +		}
>>>> +		break;
>>>> +	case IPP_CMD_WB:
>>>> +		/* clear buf for finding */
>>>> +		memset(&qbuf, 0x0, sizeof(qbuf));
>>>> +		qbuf.ops_id = EXYNOS_DRM_OPS_DST;
>>>> +		qbuf.buf_id = buf_id[EXYNOS_DRM_OPS_DST];
>>>> +
>>>> +		/* get memory node entry */
>>>> +		m_node = ipp_find_mem_node(c_node, &qbuf);
>>>> +		if (!m_node) {
>>>> +			DRM_ERROR("empty memory node.\n");
>>>> +			return -ENOMEM;
>>>> +		}
>>>> +
>>>> +		tbuf_id[EXYNOS_DRM_OPS_DST] = m_node->buf_id;
>>>> +
>>>> +		ret = ipp_put_mem_node(drm_dev, c_node, m_node);
>>>> +		if (ret)
>>>> +			DRM_ERROR("failed to put m_node.\n");
>>>> +		break;
>>>> +	case IPP_CMD_OUTPUT:
>>>> +		/* source memory list */
>>>> +		head = &c_node->mem_list[EXYNOS_DRM_OPS_SRC];
>>>> +
>>>> +		m_node = list_first_entry(head,
>>>> +			struct drm_exynos_ipp_mem_node, list);
>>>> +		if (!m_node) {
>>>> +			DRM_ERROR("empty memory node.\n");
>>>> +			return -ENOMEM;
>>>> +		}
>>>> +
>>>> +		tbuf_id[EXYNOS_DRM_OPS_SRC] = m_node->buf_id;
>>>> +
>>>> +		ret = ipp_put_mem_node(drm_dev, c_node, m_node);
>>>> +		if (ret)
>>>> +			DRM_ERROR("failed to put m_node.\n");
>>>> +		break;
>>>> +	default:
>>>> +		DRM_ERROR("invalid operations.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	/* ToDo: Fix buffer id */
>>>> +	if (tbuf_id[EXYNOS_DRM_OPS_DST] != buf_id[EXYNOS_DRM_OPS_DST])
>>>> +		DRM_ERROR("failed to match buf_id[%d %d]prop_id[%d]\n",
>>>> +			tbuf_id[1], buf_id[1], property->prop_id);
>>>> +
>>>> +	/*
>>>> +	 * command node have event list of destination buffer
>>>> +	 * If destination buffer enqueue to mem list,
>>>> +	 * than we make event and link to event list tail.
>>>
>>> Typo. s/than/then
>>
>> - done.
>>
>>>
>>>> +	 * so, we get first event for first enqueued buffer.
>>>> +	 */
>>>> +	e = list_first_entry(&c_node->event_list,
>>>> +		struct drm_exynos_ipp_send_event, base.link);
>>>> +
>>>> +	if (!e) {
>>>> +		DRM_ERROR("empty event.\n");
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>> +	do_gettimeofday(&now);
>>>> +	DRM_DEBUG_KMS("%s:tv_sec[%ld]tv_usec[%ld]\n"
>>>> +		, __func__, now.tv_sec, now.tv_usec);
>>>> +	e->event.tv_sec = now.tv_sec;
>>>> +	e->event.tv_usec = now.tv_usec;
>>>> +	e->event.prop_id = property->prop_id;
>>>> +
>>>> +	/* set buffer id about source destination */
>>>> +	for_each_ipp_ops(i)
>>>> +		e->event.buf_id[i] = tbuf_id[i];
>>>> +	/* ToDo: compare index. If needed */
>>>> +
>>>> +	spin_lock_irqsave(&drm_dev->event_lock, flags);
>>>> +	list_move_tail(&e->base.link, &e->base.file_priv->event_list);
>>>> +	wake_up_interruptible(&e->base.file_priv->event_wait);
>>>> +	spin_unlock_irqrestore(&drm_dev->event_lock, flags);
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:done cmd[%d]prop_id[%d]buf_id[%d]\n", __func__,
>>>> +		property->cmd, property->prop_id,
>>>> tbuf_id[EXYNOS_DRM_OPS_DST]);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +void ipp_sched_event(struct work_struct *work)
>>>> +{
>>>> +	struct drm_exynos_ipp_event_work *event_work =
>>>> +		(struct drm_exynos_ipp_event_work *)work;
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +	struct drm_exynos_ipp_cmd_node *c_node;
>>>> +	int ret;
>>>> +
>>>> +	if (!event_work) {
>>>> +		DRM_ERROR("failed to get event_work.\n");
>>>> +		return;
>>>> +	}
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:buf_id[%d]\n", __func__,
>>>> +		event_work->buf_id[EXYNOS_DRM_OPS_DST]);
>>>> +
>>>> +	ippdrv = event_work->ippdrv;
>>>> +	if (!ippdrv) {
>>>> +		DRM_ERROR("failed to get ipp driver.\n");
>>>> +		return;
>>>> +	}
>>>> +
>>>> +	c_node = ippdrv->cmd;
>>>> +	if (!c_node) {
>>>> +		DRM_ERROR("failed to get command node.\n");
>>>> +		return;
>>>> +	}
>>>> +
>>>> +	/*
>>>> +	 * IPP supports command thread, event thread synchronization.
>>>> +	 * If IPP close immediately from user land, than IPP make
>>>
>>> Typo. s/than/then
>>
>> - done.
>>
>>>
>>>> +	 * synchronization with command thread, so make complete event.
>>>> +	 * or going out operations.
>>>> +	 */
>>>> +	if (c_node->state != IPP_STATE_START) {
>>>> +		DRM_DEBUG_KMS("%s:bypass state[%d]prop_id[%d]\n",
>>>> +			__func__, c_node->state, c_node->property.prop_id);
>>>> +		goto err_completion;
>>>> +	}
>>>> +
>>>> +	mutex_lock(&c_node->event_lock);
>>>> +
>>>> +	ret = ipp_send_event(ippdrv, c_node, event_work->buf_id);
>>>> +	if (ret) {
>>>> +		DRM_ERROR("failed to send event.\n");
>>>> +		goto err_completion;
>>>> +	}
>>>> +
>>>> +err_completion:
>>>> +	if (c_node->property.cmd == IPP_CMD_M2M)
>>>> +		complete(&c_node->start_complete);
>>>> +
>>>> +	mutex_unlock(&c_node->event_lock);
>>>> +}
>>>> +
>>>> +static int ipp_subdrv_probe(struct drm_device *drm_dev, struct device
>>>> *dev)
>>>> +{
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +	int ret, count = 0;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	/* get ipp driver entry */
>>>> +	list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
>>>> +		ippdrv->drm_dev = drm_dev;
>>>> +
>>>> +		ret = ipp_create_id(&ctx->ipp_idr, &ctx->ipp_lock, ippdrv,
>>>> +			&ippdrv->ipp_id);
>>>> +		if (ret) {
>>>> +			DRM_ERROR("failed to create id.\n");
>>>> +			goto err_idr;
>>>> +		}
>>>> +
>>>> +		DRM_DEBUG_KMS("%s:count[%d]ippdrv[0x%x]ipp_id[%d]\n",
>>>> __func__,
>>>> +			count++, (int)ippdrv, ippdrv->ipp_id);
>>>> +
>>>> +		if (ippdrv->ipp_id == 0) {
>>>> +			DRM_ERROR("failed to get ipp_id[%d]\n",
>>>> +				ippdrv->ipp_id);
>>>> +			goto err_idr;
>>>> +		}
>>>> +
>>>> +		/* store parent device for node */
>>>> +		ippdrv->parent_dev = dev;
>>>> +
>>>> +		/* store event work queue and handler */
>>>> +		ippdrv->event_workq = ctx->event_workq;
>>>> +		ippdrv->sched_event = ipp_sched_event;
>>>> +		INIT_LIST_HEAD(&ippdrv->cmd_list);
>>>> +	}
>>>> +
>>>> +	return 0;
>>>> +
>>>> +err_idr:
>>>> +	idr_remove_all(&ctx->ipp_idr);
>>>> +	idr_remove_all(&ctx->prop_idr);
>>>> +	idr_destroy(&ctx->ipp_idr);
>>>> +	idr_destroy(&ctx->prop_idr);
>>>> +
>>>> +	return ret;
>>>> +}
>>>> +
>>>> +static void ipp_subdrv_remove(struct drm_device *drm_dev, struct
>> device
>>>> *dev)
>>>> +{
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	/* get ipp driver entry */
>>>> +	list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
>>>> +		ippdrv->drm_dev = NULL;
>>>> +		exynos_drm_ippdrv_unregister(ippdrv);
>>>> +	}
>>>> +
>>>> +	/* ToDo: free notifier callback list if needed */
>>>> +}
>>>> +
>>>> +static int ipp_subdrv_open(struct drm_device *drm_dev, struct device
>>> *dev,
>>>> +		struct drm_file *file)
>>>> +{
>>>> +	struct drm_exynos_file_private *file_priv = file->driver_priv;
>>>> +	struct exynos_drm_ipp_private *priv;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	/* ToDo: multi device open */
>>>> +
>>>> +	priv = kzalloc(sizeof(*priv), GFP_KERNEL);
>>>> +	if (!priv) {
>>>> +		DRM_ERROR("failed to allocate priv.\n");
>>>> +		return -ENOMEM;
>>>> +	}
>>>> +	priv->dev = dev;
>>>> +	file_priv->ipp_priv = priv;
>>>> +
>>>> +	INIT_LIST_HEAD(&priv->event_list);
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:done priv[0x%x]\n", __func__, (int)priv);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +static void ipp_subdrv_close(struct drm_device *drm_dev, struct device
>>>> *dev,
>>>> +		struct drm_file *file)
>>>> +{
>>>> +	struct drm_exynos_file_private *file_priv = file->driver_priv;
>>>> +	struct exynos_drm_ipp_private *priv = file_priv->ipp_priv;
>>>> +	struct exynos_drm_ippdrv *ippdrv = NULL;
>>>> +	struct drm_exynos_ipp_cmd_node *c_node, *tc_node;
>>>> +	int count = 0;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s:for priv[0x%x]\n", __func__, (int)priv);
>>>> +
>>>> +	if (list_empty(&exynos_drm_ippdrv_list)) {
>>>> +		DRM_DEBUG_KMS("%s:ippdrv_list is empty.\n", __func__);
>>>> +		goto err_clear;
>>>> +	}
>>>> +
>>>> +	list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
>>>> +		if (list_empty(&ippdrv->cmd_list))
>>>> +			continue;
>>>> +
>>>> +		list_for_each_entry_safe(c_node, tc_node,
>>>> +			&ippdrv->cmd_list, list) {
>>>> +			DRM_DEBUG_KMS("%s:count[%d]ippdrv[0x%x]\n",
>>>> +				__func__, count++, (int)ippdrv);
>>>> +
>>>> +			if (c_node->priv == priv) {
>>>> +				/*
>>>> +				 * userland goto unnormal state. process
>>> killed.
>>>> +				 * and close the file.
>>>> +				 * so, IPP didn't called stop cmd ctrl.
>>>> +				 * so, we are make stop operation in this
>>> state.
>>>> +				 */
>>>> +				if (c_node->state == IPP_STATE_START) {
>>>> +					ipp_stop_property(drm_dev, ippdrv,
>>>> +						c_node);
>>>> +					c_node->state = IPP_STATE_STOP;
>>>> +				}
>>>> +
>>>> +				ippdrv->dedicated = false;
>>>> +				ipp_clean_cmd_node(c_node);
>>>> +				if (list_empty(&ippdrv->cmd_list))
>>>> +					pm_runtime_put_sync(ippdrv->dev);
>>>> +			}
>>>> +		}
>>>> +	}
>>>> +
>>>> +err_clear:
>>>> +	kfree(priv);
>>>> +
>>>> +	return;
>>>> +}
>>>> +
>>>> +static int __devinit ipp_probe(struct platform_device *pdev)
>>>> +{
>>>> +	struct device *dev = &pdev->dev;
>>>> +	struct ipp_context *ctx;
>>>> +	struct exynos_drm_subdrv *subdrv;
>>>> +	int ret = -EINVAL;
>>>
>>> Just use int ret;
>>
>> - done.
>>
>>>
>>>> +
>>>> +	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
>>>> +	if (!ctx)
>>>> +		return -ENOMEM;
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	mutex_init(&ctx->ipp_lock);
>>>> +	mutex_init(&ctx->prop_lock);
>>>> +
>>>> +	idr_init(&ctx->ipp_idr);
>>>> +	idr_init(&ctx->prop_idr);
>>>> +
>>>> +	/*
>>>> +	 * create single thread for ipp event
>>>> +	 * IPP supports event thread for IPP drivers.
>>>> +	 * IPP driver send event_work to this thread.
>>>> +	 * and IPP event thread send event to user process.
>>>> +	 */
>>>> +	ctx->event_workq = create_singlethread_workqueue("ipp_event");
>>>> +	if (!ctx->event_workq) {
>>>> +		dev_err(dev, "failed to create event workqueue\n");
>>>> +		ret = -EINVAL;
>>>> +		goto err_clear;
>>>> +	}
>>>> +
>>>> +	/*
>>>> +	 * create single thread for ipp command
>>>> +	 * IPP supports command thread for user process.
>>>> +	 * user process make command node using set property ioctl.
>>>> +	 * and make start_work and send this work to command thread.
>>>> +	 * and than this command thread start property.
>>>> +	 */
>>>> +	ctx->cmd_workq = create_singlethread_workqueue("ipp_cmd");
>>>> +	if (!ctx->cmd_workq) {
>>>> +		dev_err(dev, "failed to create cmd workqueue\n");
>>>> +		ret = -EINVAL;
>>>> +		goto err_event_workq;
>>>> +	}
>>>> +
>>>> +	/* set sub driver informations */
>>>> +	subdrv = &ctx->subdrv;
>>>> +	subdrv->dev = dev;
>>>> +	subdrv->probe = ipp_subdrv_probe;
>>>> +	subdrv->remove = ipp_subdrv_remove;
>>>> +	subdrv->open = ipp_subdrv_open;
>>>> +	subdrv->close = ipp_subdrv_close;
>>>> +
>>>> +	platform_set_drvdata(pdev, ctx);
>>>> +
>>>> +	ret = exynos_drm_subdrv_register(subdrv);
>>>> +	if (ret < 0) {
>>>> +		DRM_ERROR("failed to register drm ipp device.\n");
>>>> +		goto err_cmd_workq;
>>>> +	}
>>>> +
>>>> +	dev_info(&pdev->dev, "drm ipp registered successfully.\n");
>>>> +
>>>> +	return 0;
>>>> +
>>>> +err_cmd_workq:
>>>> +	destroy_workqueue(ctx->cmd_workq);
>>>> +err_event_workq:
>>>> +	destroy_workqueue(ctx->event_workq);
>>>> +err_clear:
>>>> +	kfree(ctx);
>>>> +
>>>> +	return ret;
>>>> +}
>>>> +
>>>> +static int __devexit ipp_remove(struct platform_device *pdev)
>>>> +{
>>>> +	struct ipp_context *ctx = platform_get_drvdata(pdev);
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	/* unregister sub driver */
>>>> +	exynos_drm_subdrv_unregister(&ctx->subdrv);
>>>> +
>>>> +	/* remove,destroy ipp idr */
>>>> +	idr_remove_all(&ctx->ipp_idr);
>>>> +	idr_remove_all(&ctx->prop_idr);
>>>> +	idr_destroy(&ctx->ipp_idr);
>>>> +	idr_destroy(&ctx->prop_idr);
>>>> +
>>>> +	/* destroy command, event work queue */
>>>> +	destroy_workqueue(ctx->cmd_workq);
>>>> +	destroy_workqueue(ctx->event_workq);
>>>> +
>>>> +	kfree(ctx);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +static int ipp_power_ctrl(struct ipp_context *ctx, bool enable)
>>>> +{
>>>> +	/* ToDo: Need to implement power and sysmmu ctrl. */
>>>> +	DRM_DEBUG_KMS("%s:enable[%d]\n", __func__, enable);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +
>>>> +#ifdef CONFIG_PM_SLEEP
>>>> +static int ipp_suspend(struct device *dev)
>>>> +{
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	if (pm_runtime_suspended(dev))
>>>> +		return 0;
>>>> +
>>>> +	return ipp_power_ctrl(ctx, false);
>>>> +}
>>>> +
>>>> +static int ipp_resume(struct device *dev)
>>>> +{
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	if (!pm_runtime_suspended(dev))
>>>> +		return ipp_power_ctrl(ctx, true);
>>>> +
>>>> +	return 0;
>>>> +}
>>>> +#endif
>>>> +
>>>> +#ifdef CONFIG_PM_RUNTIME
>>>> +static int ipp_runtime_suspend(struct device *dev)
>>>> +{
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	return ipp_power_ctrl(ctx, false);
>>>> +}
>>>> +
>>>> +static int ipp_runtime_resume(struct device *dev)
>>>> +{
>>>> +	struct ipp_context *ctx = get_ipp_context(dev);
>>>> +
>>>> +	DRM_DEBUG_KMS("%s\n", __func__);
>>>> +
>>>> +	return ipp_power_ctrl(ctx, true);
>>>> +}
>>>> +#endif
>>>> +
>>>> +static const struct dev_pm_ops ipp_pm_ops = {
>>>> +	SET_SYSTEM_SLEEP_PM_OPS(ipp_suspend, ipp_resume)
>>>> +	SET_RUNTIME_PM_OPS(ipp_runtime_suspend, ipp_runtime_resume, NULL)
>>>> +};
>>>> +
>>>> +struct platform_driver ipp_driver = {
>>>> +	.probe		= ipp_probe,
>>>> +	.remove		= __devexit_p(ipp_remove),
>>>> +	.driver		= {
>>>> +		.name	= "exynos-drm-ipp",
>>>> +		.owner	= THIS_MODULE,
>>>> +		.pm	= &ipp_pm_ops,
>>>> +	},
>>>> +};
>>>> +
>>>> diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.h
>>>> b/drivers/gpu/drm/exynos/exynos_drm_ipp.h
>>>> new file mode 100644
>>>> index 0000000..baab1f0
>>>> --- /dev/null
>>>> +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.h
>>>> @@ -0,0 +1,266 @@
>>>> +/*
>>>> + * Copyright (c) 2012 Samsung Electronics Co., Ltd.
>>>> + *
>>>> + * Authors:
>>>> + *	Eunchul Kim <chulspro.kim at samsung.com>
>>>> + *	Jinyoung Jeon <jy0.jeon at samsung.com>
>>>> + *	Sangmin Lee <lsmin.lee at samsung.com>
>>>> + *
>>>> + * Permission is hereby granted, free of charge, to any person
>> obtaining
>>>> a
>>>> + * copy of this software and associated documentation files (the
>>>> "Software"),
>>>> + * to deal in the Software without restriction, including without
>>>> limitation
>>>> + * the rights to use, copy, modify, merge, publish, distribute,
>>>> sublicense,
>>>> + * and/or sell copies of the Software, and to permit persons to whom
>> the
>>>> + * Software is furnished to do so, subject to the following
> conditions:
>>>> + *
>>>> + * The above copyright notice and this permission notice (including
>> the
>>>> next
>>>> + * paragraph) shall be included in all copies or substantial portions
>> of
>>>> the
>>>> + * Software.
>>>> + *
>>>> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
>>>> EXPRESS OR
>>>> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
>>>> MERCHANTABILITY,
>>>> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT
>>>> SHALL
>>>> + * VA LINUX SYSTEMS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
>> DAMAGES
>>>> OR
>>>> + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
>> OTHERWISE,
>>>> + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE
>> OR
>>>> + * OTHER DEALINGS IN THE SOFTWARE.
>>>> + */
>>>> +
>>>> +#ifndef _EXYNOS_DRM_IPP_H_
>>>> +#define _EXYNOS_DRM_IPP_H_
>>>> +
>>>> +#define for_each_ipp_ops(pos)	\
>>>> +	for (pos = 0; pos < EXYNOS_DRM_OPS_MAX; pos++)
>>>> +#define for_each_ipp_planar(pos)	\
>>>> +	for (pos = 0; pos < EXYNOS_DRM_PLANAR_MAX; pos++)
>>>> +
>>>> +#define IPP_GET_LCD_WIDTH	_IOR('F', 302, int)
>>>> +#define IPP_GET_LCD_HEIGHT	_IOR('F', 303, int)
>>>> +#define IPP_SET_WRITEBACK	_IOW('F', 304, u32)
>>>> +
>>>> +/* definition of state */
>>>> +enum drm_exynos_ipp_state {
>>>> +	IPP_STATE_IDLE,
>>>> +	IPP_STATE_START,
>>>> +	IPP_STATE_STOP,
>>>> +};
>>>> +
>>>> +/*
>>>> + * A structure of command work information.
>>>> + * @work: work structure.
>>>> + * @ippdrv: current work ippdrv.
>>>> + * @c_node: command node information.
>>>> + * @ctrl: command control.
>>>> + */
>>>> +struct drm_exynos_ipp_cmd_work {
>>>> +	struct work_struct	work;
>>>> +	struct exynos_drm_ippdrv	*ippdrv;
>>>> +	struct drm_exynos_ipp_cmd_node *c_node;
>>>> +	enum drm_exynos_ipp_ctrl	ctrl;
>>>> +};
>>>> +
>>>> +/*
>>>> + * A structure of command node.
>>>> + *
>>>> + * @priv: IPP private infomation.
>>>> + * @list: list head to command queue information.
>>>> + * @event_list: list head of event.
>>>> + * @mem_list: list head to source,destination memory queue
> information.
>>>> + * @cmd_lock: lock for synchronization of access to ioctl.
>>>> + * @mem_lock: lock for synchronization of access to memory nodes.
>>>> + * @event_lock: lock for synchronization of access to scheduled event.
>>>> + * @start_complete: completion of start of command.
>>>> + * @stop_complete: completion of stop of command.
>>>> + * @property: property information.
>>>> + * @start_work: start command work structure.
>>>> + * @stop_work: stop command work structure.
>>>> + * @event_work: event work structure.
>>>> + * @state: state of command node.
>>>> + */
>>>> +struct drm_exynos_ipp_cmd_node {
>>>> +	struct exynos_drm_ipp_private *priv;
>>>> +	struct list_head	list;
>>>> +	struct list_head	event_list;
>>>> +	struct list_head	mem_list[EXYNOS_DRM_OPS_MAX];
>>>> +	struct mutex	cmd_lock;
>>>> +	struct mutex	mem_lock;
>>>> +	struct mutex	event_lock;
>>>> +	struct completion	start_complete;
>>>> +	struct completion	stop_complete;
>>>> +	struct drm_exynos_ipp_property	property;
>>>> +	struct drm_exynos_ipp_cmd_work *start_work;
>>>> +	struct drm_exynos_ipp_cmd_work *stop_work;
>>>> +	struct drm_exynos_ipp_event_work *event_work;
>>>> +	enum drm_exynos_ipp_state	state;
>>>> +};
>>>> +
>>>> +/*
>>>> + * A structure of buffer information.
>>>> + *
>>>> + * @gem_objs: Y, Cb, Cr each gem object.
>>>> + * @base: Y, Cb, Cr each planar address.
>>>> + */
>>>> +struct drm_exynos_ipp_buf_info {
>>>> +	unsigned long	handles[EXYNOS_DRM_PLANAR_MAX];
>>>> +	dma_addr_t	base[EXYNOS_DRM_PLANAR_MAX];
>>>> +};
>>>> +
>>>> +/*
>>>> + * A structure of wb setting infomation.
>>>> + *
>>>> + * @enable: enable flag for wb.
>>>> + * @refresh: HZ of the refresh rate.
>>>> + */
>>>> +struct drm_exynos_ipp_set_wb {
>>>> +	__u32	enable;
>>>> +	__u32	refresh;
>>>> +};
>>>> +
>>>> +/*
>>>> + * A structure of event work information.
>>>> + *
>>>> + * @work: work structure.
>>>> + * @ippdrv: current work ippdrv.
>>>> + * @buf_id: id of src, dst buffer.
>>>> + */
>>>> +struct drm_exynos_ipp_event_work {
>>>> +	struct work_struct	work;
>>>> +	struct exynos_drm_ippdrv *ippdrv;
>>>> +	u32	buf_id[EXYNOS_DRM_OPS_MAX];
>>>> +};
>>>> +
>>>> +/*
>>>> + * A structure of source,destination operations.
>>>> + *
>>>> + * @set_fmt: set format of image.
>>>> + * @set_transf: set transform(rotations, flip).
>>>> + * @set_size: set size of region.
>>>> + * @set_addr: set address for dma.
>>>> + */
>>>> +struct exynos_drm_ipp_ops {
>>>> +	int (*set_fmt)(struct device *dev, u32 fmt);
>>>> +	int (*set_transf)(struct device *dev,
>>>> +		enum drm_exynos_degree degree,
>>>> +		enum drm_exynos_flip flip);
>>>> +	int (*set_size)(struct device *dev, int swap,
>>>> +		struct drm_exynos_pos *pos, struct drm_exynos_sz *sz);
>>>> +	int (*set_addr)(struct device *dev,
>>>> +			 struct drm_exynos_ipp_buf_info *buf_info, u32
>>> buf_id,
>>>> +		enum drm_exynos_ipp_buf_type buf_type);
>>>> +};
>>>> +
>>>> +/*
>>>> + * A structure of ipp driver.
>>>> + *
>>>> + * @drv_list: list head for registed sub driver information.
>>>> + * @parent_dev: parent device information.
>>>> + * @dev: platform device.
>>>> + * @drm_dev: drm device.
>>>> + * @ipp_id: id of ipp driver.
>>>> + * @dedicated: dedicated ipp device.
>>>> + * @ops: source, destination operations.
>>>> + * @event_workq: event work queue.
>>>> + * @cmd: current command information.
>>>> + * @cmd_list: list head for command information.
>>>> + * @prop_list: property informations of current ipp driver.
>>>> + * @check_property: check property about format, size, buffer.
>>>> + * @reset: reset ipp block.
>>>> + * @start: ipp each device start.
>>>> + * @stop: ipp each device stop.
>>>> + * @sched_event: work schedule handler.
>>>> + */
>>>> +struct exynos_drm_ippdrv {
>>>> +	struct list_head	drv_list;
>>>> +	struct device	*parent_dev;
>>>> +	struct device	*dev;
>>>> +	struct drm_device	*drm_dev;
>>>> +	u32	ipp_id;
>>>> +	bool	dedicated;
>>>> +	struct exynos_drm_ipp_ops	*ops[EXYNOS_DRM_OPS_MAX];
>>>> +	struct workqueue_struct	*event_workq;
>>>> +	struct drm_exynos_ipp_cmd_node *cmd;
>>>> +	struct list_head	cmd_list;
>>>> +	struct drm_exynos_ipp_prop_list *prop_list;
>>>> +
>>>> +	int (*check_property)(struct device *dev,
>>>> +		struct drm_exynos_ipp_property *property);
>>>> +	int (*reset)(struct device *dev);
>>>> +	int (*start)(struct device *dev, enum drm_exynos_ipp_cmd cmd);
>>>> +	void (*stop)(struct device *dev, enum drm_exynos_ipp_cmd cmd);
>>>> +	void (*sched_event)(struct work_struct *work);
>>>> +};
>>>> +
>>>> +#ifdef CONFIG_DRM_EXYNOS_IPP
>>>> +extern int exynos_drm_ippdrv_register(struct exynos_drm_ippdrv
>> *ippdrv);
>>>> +extern int exynos_drm_ippdrv_unregister(struct exynos_drm_ippdrv
>>> *ippdrv);
>>>> +extern int exynos_drm_ipp_get_property(struct drm_device *drm_dev,
>> void
>>>> *data,
>>>> +					 struct drm_file *file);
>>>> +extern int exynos_drm_ipp_set_property(struct drm_device *drm_dev,
>> void
>>>> *data,
>>>> +					 struct drm_file *file);
>>>> +extern int exynos_drm_ipp_queue_buf(struct drm_device *drm_dev, void
>>>> *data,
>>>> +					 struct drm_file *file);
>>>> +extern int exynos_drm_ipp_cmd_ctrl(struct drm_device *drm_dev, void
>>> *data,
>>>> +					 struct drm_file *file);
>>>> +extern int exynos_drm_ippnb_register(struct notifier_block *nb);
>>>> +extern int exynos_drm_ippnb_unregister(struct notifier_block *nb);
>>>> +extern int exynos_drm_ippnb_send_event(unsigned long val, void *v);
>>>> +extern void ipp_sched_cmd(struct work_struct *work);
>>>> +extern void ipp_sched_event(struct work_struct *work);
>>>> +
>>>> +#else
>>>> +static inline int exynos_drm_ippdrv_register(struct exynos_drm_ippdrv
>>>> *ippdrv)
>>>> +{
>>>> +	return -ENODEV;
>>>> +}
>>>> +
>>>> +static inline int exynos_drm_ippdrv_unregister(struct
>> exynos_drm_ippdrv
>>>> *ippdrv)
>>>> +{
>>>> +	return -ENODEV;
>>>> +}
>>>> +
>>>> +static inline int exynos_drm_ipp_get_property(struct drm_device
>> *drm_dev,
>>>> +						void *data,
>>>> +						struct drm_file *file_priv)
>>>> +{
>>>> +	return -ENOTTY;
>>>> +}
>>>> +
>>>> +static inline int exynos_drm_ipp_set_property(struct drm_device
>> *drm_dev,
>>>> +						void *data,
>>>> +						struct drm_file *file_priv)
>>>> +{
>>>> +	return -ENOTTY;
>>>> +}
>>>> +
>>>> +static inline int exynos_drm_ipp_queue_buf(struct drm_device *drm_dev,
>>>> +						void *data,
>>>> +						struct drm_file *file)
>>>> +{
>>>> +	return -ENOTTY;
>>>> +}
>>>> +
>>>> +static inline int exynos_drm_ipp_cmd_ctrl(struct drm_device *drm_dev,
>>>> +						void *data,
>>>> +						struct drm_file *file)
>>>> +{
>>>> +	return -ENOTTY;
>>>> +}
>>>> +
>>>> +static inline int exynos_drm_ippnb_register(struct notifier_block *nb)
>>>> +{
>>>> +	return -ENODEV;
>>>> +}
>>>> +
>>>> +static inline int exynos_drm_ippnb_unregister(struct notifier_block
>> *nb)
>>>> +{
>>>> +	return -ENODEV;
>>>> +}
>>>> +
>>>> +static inline int exynos_drm_ippnb_send_event(unsigned long val, void
>> *v)
>>>> +{
>>>> +	return -ENOTTY;
>>>> +}
>>>> +#endif
>>>> +
>>>> +#endif /* _EXYNOS_DRM_IPP_H_ */
>>>> +
>>>> diff --git a/include/uapi/drm/exynos_drm.h
>> b/include/uapi/drm/exynos_drm.h
>>>> index c0494d5..0e33aec 100644
>>>> --- a/include/uapi/drm/exynos_drm.h
>>>> +++ b/include/uapi/drm/exynos_drm.h
>>>> @@ -154,6 +154,169 @@ struct drm_exynos_g2d_exec {
>>>>    	__u64					async;
>>>>    };
>>>>
>>>> +enum drm_exynos_ops_id {
>>>> +	EXYNOS_DRM_OPS_SRC,
>>>> +	EXYNOS_DRM_OPS_DST,
>>>> +	EXYNOS_DRM_OPS_MAX,
>>>> +};
>>>> +
>>>> +struct drm_exynos_sz {
>>>> +	__u32	hsize;
>>>> +	__u32	vsize;
>>>> +};
>>>> +
>>>> +struct drm_exynos_pos {
>>>> +	__u32	x;
>>>> +	__u32	y;
>>>> +	__u32	w;
>>>> +	__u32	h;
>>>> +};
>>>> +
>>>> +enum drm_exynos_flip {
>>>> +	EXYNOS_DRM_FLIP_NONE = (0 << 0),
>>>
>>> Just use EXYNOS_DRM_FLIP_NONE;
>>>
>>>> +	EXYNOS_DRM_FLIP_VERTICAL = (1 << 0),
>>>> +	EXYNOS_DRM_FLIP_HORIZONTAL = (1 << 1),
>>>> +};
>>>> +
>>>> +enum drm_exynos_degree {
>>>> +	EXYNOS_DRM_DEGREE_0,
>>>> +	EXYNOS_DRM_DEGREE_90,
>>>> +	EXYNOS_DRM_DEGREE_180,
>>>> +	EXYNOS_DRM_DEGREE_270,
>>>> +};
>>>> +
>>>> +enum drm_exynos_planer {
>>>> +	EXYNOS_DRM_PLANAR_Y,
>>>> +	EXYNOS_DRM_PLANAR_CB,
>>>> +	EXYNOS_DRM_PLANAR_CR,
>>>> +	EXYNOS_DRM_PLANAR_MAX,
>>>> +};
>>>> +
>>>> +/**
>>>> + * A structure for ipp supported property list.
>>>> + *
>>>> + * @version: version of this structure.
>>>> + * @ipp_id: id of ipp driver.
>>>> + * @count: count of ipp driver.
>>>> + * @writeback: flag of writeback supporting.
>>>> + * @flip: flag of flip supporting.
>>>> + * @degree: flag of degree information.
>>>> + * @csc: flag of csc supporting.
>>>> + * @crop: flag of crop supporting.
>>>> + * @scale: flag of scale supporting.
>>>> + * @refresh_min: min hz of refresh.
>>>> + * @refresh_max: max hz of refresh.
>>>> + * @crop_min: crop min resolution.
>>>> + * @crop_max: crop max resolution.
>>>> + * @scale_min: scale min resolution.
>>>> + * @scale_max: scale max resolution.
>>>> + */
>>>> +struct drm_exynos_ipp_prop_list {
>>>> +	__u32	version;
>>>> +	__u32	ipp_id;
>>>> +	__u32	count;
>>>> +	__u32	writeback;
>>>> +	__u32	flip;
>>>> +	__u32	degree;
>>>> +	__u32	csc;
>>>> +	__u32	crop;
>>>> +	__u32	scale;
>>>> +	__u32	refresh_min;
>>>> +	__u32	refresh_max;
>>>> +	__u32	reserved;
>>>> +	struct drm_exynos_sz	crop_min;
>>>> +	struct drm_exynos_sz	crop_max;
>>>> +	struct drm_exynos_sz	scale_min;
>>>> +	struct drm_exynos_sz	scale_max;
>>>> +};
>>>> +
>>>> +/**
>>>> + * A structure for ipp config.
>>>> + *
>>>> + * @ops_id: property of operation directions.
>>>> + * @flip: property of mirror, flip.
>>>> + * @degree: property of rotation degree.
>>>> + * @fmt: property of image format.
>>>> + * @sz: property of image size.
>>>> + * @pos: property of image position(src-cropped,dst-scaler).
>>>> + */
>>>> +struct drm_exynos_ipp_config {
>>>> +	enum drm_exynos_ops_id ops_id;
>>>> +	enum drm_exynos_flip	flip;
>>>> +	enum drm_exynos_degree	degree;
>>>> +	__u32	fmt;
>>>> +	struct drm_exynos_sz	sz;
>>>> +	struct drm_exynos_pos	pos;
>>>> +};
>>>> +
>>>> +enum drm_exynos_ipp_cmd {
>>>> +	IPP_CMD_NONE,
>>>> +	IPP_CMD_M2M,
>>>> +	IPP_CMD_WB,
>>>> +	IPP_CMD_OUTPUT,
>>>> +	IPP_CMD_MAX,
>>>> +};
>>>> +
>>>> +/**
>>>> + * A structure for ipp property.
>>>> + *
>>>> + * @config: source, destination config.
>>>> + * @cmd: definition of command.
>>>> + * @ipp_id: id of ipp driver.
>>>> + * @prop_id: id of property.
>>>> + */
>>>> +struct drm_exynos_ipp_property {
>>>> +	struct drm_exynos_ipp_config config[EXYNOS_DRM_OPS_MAX];
>>>> +	enum drm_exynos_ipp_cmd	cmd;
>>>> +	__u32	ipp_id;
>>>> +	__u32	prop_id;
>>>> +	__u32	reserved;
>>>> +};
>>>> +
>>>> +enum drm_exynos_ipp_buf_type {
>>>> +	IPP_BUF_ENQUEUE,
>>>> +	IPP_BUF_DEQUEUE,
>>>> +};
>>>> +
>>>> +/**
>>>> + * A structure for ipp buffer operations.
>>>> + *
>>>> + * @ops_id: operation directions.
>>>> + * @buf_type: definition of buffer.
>>>> + * @prop_id: id of property.
>>>> + * @buf_id: id of buffer.
>>>> + * @handle: Y, Cb, Cr each planar handle.
>>>> + * @user_data: user data.
>>>> + */
>>>> +struct drm_exynos_ipp_queue_buf {
>>>> +	enum drm_exynos_ops_id	ops_id;
>>>> +	enum drm_exynos_ipp_buf_type	buf_type;
>>>> +	__u32	prop_id;
>>>> +	__u32	buf_id;
>>>> +	__u32	handle[EXYNOS_DRM_PLANAR_MAX];
>>>> +	__u32	reserved;
>>>> +	__u64	user_data;
>>>> +};
>>>> +
>>>> +enum drm_exynos_ipp_ctrl {
>>>> +	IPP_CTRL_PLAY,
>>>> +	IPP_CTRL_STOP,
>>>> +	IPP_CTRL_PAUSE,
>>>> +	IPP_CTRL_RESUME,
>>>> +	IPP_CTRL_MAX,
>>>> +};
>>>> +
>>>> +/**
>>>> + * A structure for ipp start/stop operations.
>>>> + *
>>>> + * @prop_id: id of property.
>>>> + * @ctrl: definition of control.
>>>> + */
>>>> +struct drm_exynos_ipp_cmd_ctrl {
>>>> +	__u32	prop_id;
>>>> +	enum drm_exynos_ipp_ctrl	ctrl;
>>>> +};
>>>> +
>>>>    #define DRM_EXYNOS_GEM_CREATE		0x00
>>>>    #define DRM_EXYNOS_GEM_MAP_OFFSET	0x01
>>>>    #define DRM_EXYNOS_GEM_MMAP		0x02
>>>> @@ -166,6 +329,12 @@ struct drm_exynos_g2d_exec {
>>>>    #define DRM_EXYNOS_G2D_SET_CMDLIST	0x21
>>>>    #define DRM_EXYNOS_G2D_EXEC		0x22
>>>>
>>>> +/* IPP - Image Post Processing */
>>>> +#define DRM_EXYNOS_IPP_GET_PROPERTY	0x30
>>>> +#define DRM_EXYNOS_IPP_SET_PROPERTY	0x31
>>>> +#define DRM_EXYNOS_IPP_QUEUE_BUF	0x32
>>>> +#define DRM_EXYNOS_IPP_CMD_CTRL	0x33
>>>> +
>>>>    #define DRM_IOCTL_EXYNOS_GEM_CREATE
> DRM_IOWR(DRM_COMMAND_BASE +
>>>> \
>>>>    		DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create)
>>>>
>>>> @@ -188,8 +357,18 @@ struct drm_exynos_g2d_exec {
>>>>    #define DRM_IOCTL_EXYNOS_G2D_EXEC
> DRM_IOWR(DRM_COMMAND_BASE +
>>>> \
>>>>    		DRM_EXYNOS_G2D_EXEC, struct drm_exynos_g2d_exec)
>>>>
>>>> +#define DRM_IOCTL_EXYNOS_IPP_GET_PROPERTY	DRM_IOWR(DRM_COMMAND_BASE +
>>>> \
>>>> +		DRM_EXYNOS_IPP_GET_PROPERTY, struct
>>> drm_exynos_ipp_prop_list)
>>>> +#define DRM_IOCTL_EXYNOS_IPP_SET_PROPERTY	DRM_IOWR(DRM_COMMAND_BASE +
>>>> \
>>>> +		DRM_EXYNOS_IPP_SET_PROPERTY, struct drm_exynos_ipp_property)
>>>> +#define DRM_IOCTL_EXYNOS_IPP_QUEUE_BUF	DRM_IOWR(DRM_COMMAND_BASE +
>>> \
>>>> +		DRM_EXYNOS_IPP_QUEUE_BUF, struct drm_exynos_ipp_queue_buf)
>>>> +#define DRM_IOCTL_EXYNOS_IPP_CMD_CTRL
>>> DRM_IOWR(DRM_COMMAND_BASE +
>>>> \
>>>> +		DRM_EXYNOS_IPP_CMD_CTRL, struct drm_exynos_ipp_cmd_ctrl)
>>>> +
>>>>    /* EXYNOS specific events */
>>>>    #define DRM_EXYNOS_G2D_EVENT		0x80000000
>>>> +#define DRM_EXYNOS_IPP_EVENT		0x80000001
>>>>
>>>>    struct drm_exynos_g2d_event {
>>>>    	struct drm_event	base;
>>>> @@ -200,4 +379,14 @@ struct drm_exynos_g2d_event {
>>>>    	__u32			reserved;
>>>>    };
>>>>
>>>> +struct drm_exynos_ipp_event {
>>>> +	struct drm_event	base;
>>>> +	__u64			user_data;
>>>> +	__u32			tv_sec;
>>>> +	__u32			tv_usec;
>>>> +	__u32			prop_id;
>>>> +	__u32			reserved;
>>>> +	__u32			buf_id[EXYNOS_DRM_OPS_MAX];
>>>> +};
>>>> +
>>>>    #endif /* _UAPI_EXYNOS_DRM_H_ */
>>>> --
>>>> 1.7.0.4
>>>
>>>
>
>



More information about the dri-devel mailing list