[PATCH 0/6] Exynos DRM: rewrite IPP subsystem and userspace API

[PATCH 0/6] Exynos DRM: rewrite IPP subsystem and userspace API

[Marek Szyprowski]

Dear all,

This patchset performs complete rewrite of Exynos DRM IPP subsystem and
its userspace API.

Why such rewrite is needed? Exynos DRM IPP API is over-engineered in
general, but not really extensible on the other side. It is also buggy,
with significant design flaws:
- Userspace API covers memory-2-memory picture operations together with
  CRTC writeback and duplicating features, which belongs to video plane.
- Lack of support of the all required image formats (for example NV12
  Samsung-tiled cannot be used due to lack of pixel format modifier
  support).
- Userspace API designed only to mimic hardware behaviour, not easy to
  understand.
- Lack of proper input validation in the core, drivers also didn't do that
  correctly, so it was possible to set incorrect parameters and easil
  trigger IOMMU fault or memory trash.
- Drivers were partially disfunctional or supported only a subset of modes.

Due to the above limitations and issues the Exynos DRM IPP API was not
used by any of the open-source projects. I assume that it is safe to remove
this broken API without any damage to open-source community. All remaining
users (mainly Tizen project related) will be updated to the new version.

This patchset changes Exynos DRM IPP subsystem to something useful. The
userspace API is much simpler, state-less and easy to understand. Also
the code of the core and driver is significantly smaller and easier to
understand.

Patches were tested on Exynos4412 based Odroid U3 and Exynos5422
Odroid XU3 boards, on top of Linux next-20170911 kernel.

Best regards
Marek Szyprowski
Samsung R&D Institute Poland


My previous works in this area:

"[RFC v2 0/2] Exynos DRM: add Picture Processor extension"
https://www.spinics.net/lists/dri-devel/msg140669.html
- removed usage of DRM objects and properties - replaced them with simple
  list of parameters with predefined IDs

"[RFC 0/4] Exynos DRM: add Picture Processor extension"
https://www.spinics.net/lists/linux-samsung-soc/msg59323.html
- moved this feature from DRM core to Exynos DRM driver
- changed name from framebuffer processor to picture processor
- simplified code to cover only things needed by Exynos drivers
- implemented simple fifo task scheduler
- cleaned up rotator driver conversion (removed IPP remainings)

"[RFC 0/2] New feature: Framebuffer processors"
https://www.spinics.net/lists/linux-samsung-soc/msg54810.html
- generic approach implemented in DRM core, rejected


Patch summary:

Marek Szyprowski (6):
  drm/exynos: ipp: Remove Exynos DRM IPP subsystem
  drm/exynos: ipp: Add IPP v2 framework
  drm/exynos: rotator: Convert driver to IPP v2 core API
  drm/exynos: gsc: Convert driver to IPP v2 core API
  drm/exynos: Add generic support for devices shared with V4L2 subsystem
  drm/exynos: fimc: Convert driver to IPP v2 core API

 drivers/gpu/drm/exynos/Kconfig              |   12 +-
 drivers/gpu/drm/exynos/exynos_drm_drv.c     |   33 +-
 drivers/gpu/drm/exynos/exynos_drm_drv.h     |    4 +-
 drivers/gpu/drm/exynos/exynos_drm_fimc.c    |  893 +++--------
 drivers/gpu/drm/exynos/exynos_drm_fimc.h    |   23 -
 drivers/gpu/drm/exynos/exynos_drm_gsc.c     |  853 +++-------
 drivers/gpu/drm/exynos/exynos_drm_gsc.h     |   24 -
 drivers/gpu/drm/exynos/exynos_drm_ipp.c     | 2239 ++++++++-------------------
 drivers/gpu/drm/exynos/exynos_drm_ipp.h     |  352 ++---
 drivers/gpu/drm/exynos/exynos_drm_rotator.c |  735 ++-------
 drivers/gpu/drm/exynos/exynos_drm_rotator.h |   19 -
 include/uapi/drm/exynos_drm.h               |  315 ++--
 12 files changed, 1627 insertions(+), 3875 deletions(-)
 delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_fimc.h
 delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_gsc.h
 delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_rotator.h

-- 
1.9.1

[PATCH 1/6] drm/exynos: ipp: Remove Exynos DRM IPP subsystem

[Marek Szyprowski]

Exynos IPP will be rewritten, so remove current IPP core code and mark
existing drivers as BROKEN.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/gpu/drm/exynos/Kconfig          |   11 +-
 drivers/gpu/drm/exynos/Makefile         |    1 -
 drivers/gpu/drm/exynos/exynos_drm_drv.c |   12 -
 drivers/gpu/drm/exynos/exynos_drm_drv.h |    2 -
 drivers/gpu/drm/exynos/exynos_drm_ipp.c | 1806 -------------------------------
 drivers/gpu/drm/exynos/exynos_drm_ipp.h |  252 -----
 include/uapi/drm/exynos_drm.h           |  192 +---
 7 files changed, 4 insertions(+), 2272 deletions(-)
 delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.c
 delete 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 305dc3d4ff77..88cff0e039b6 100644
--- a/drivers/gpu/drm/exynos/Kconfig
+++ b/drivers/gpu/drm/exynos/Kconfig
@@ -94,26 +94,21 @@ config DRM_EXYNOS_G2D
 	help
 	  Choose this option if you want to use Exynos G2D for DRM.
 
-config DRM_EXYNOS_IPP
-	bool "Image Post Processor"
-	help
-	  Choose this option if you want to use IPP feature for DRM.
-
 config DRM_EXYNOS_FIMC
 	bool "FIMC"
-	depends on DRM_EXYNOS_IPP && MFD_SYSCON
+	depends on BROKEN && MFD_SYSCON
 	help
 	  Choose this option if you want to use Exynos FIMC for DRM.
 
 config DRM_EXYNOS_ROTATOR
 	bool "Rotator"
-	depends on DRM_EXYNOS_IPP
+	depends on BROKEN
 	help
 	  Choose this option if you want to use Exynos Rotator for DRM.
 
 config DRM_EXYNOS_GSC
 	bool "GScaler"
-	depends on DRM_EXYNOS_IPP && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n
+	depends on BROKEN && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n
 	help
 	  Choose this option if you want to use Exynos GSC for DRM.
 
diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile
index f663490e949d..09bb002c9555 100644
--- a/drivers/gpu/drm/exynos/Makefile
+++ b/drivers/gpu/drm/exynos/Makefile
@@ -17,7 +17,6 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER)	+= exynos_mixer.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI)	+= exynos_hdmi.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI)	+= exynos_drm_vidi.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_G2D)	+= exynos_drm_g2d.o
-exynosdrm-$(CONFIG_DRM_EXYNOS_IPP)	+= exynos_drm_ipp.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC)	+= exynos_drm_fimc.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR)	+= exynos_drm_rotator.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_GSC)	+= exynos_drm_gsc.o
diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
index b1f7299600f0..7f19054ebce3 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
@@ -28,7 +28,6 @@
 #include "exynos_drm_plane.h"
 #include "exynos_drm_vidi.h"
 #include "exynos_drm_g2d.h"
-#include "exynos_drm_ipp.h"
 #include "exynos_drm_iommu.h"
 
 #define DRIVER_NAME	"exynos"
@@ -115,14 +114,6 @@ static void exynos_drm_lastclose(struct drm_device *dev)
 			DRM_AUTH | DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl,
 			DRM_AUTH | DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_PROPERTY, exynos_drm_ipp_get_property,
-			DRM_AUTH | DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_SET_PROPERTY, exynos_drm_ipp_set_property,
-			DRM_AUTH | DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_QUEUE_BUF, exynos_drm_ipp_queue_buf,
-			DRM_AUTH | DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_CMD_CTRL, exynos_drm_ipp_cmd_ctrl,
-			DRM_AUTH | DRM_RENDER_ALLOW),
 };
 
 static const struct file_operations exynos_drm_driver_fops = {
@@ -272,9 +263,6 @@ struct exynos_drm_driver_info {
 	}, {
 		DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC),
 	}, {
-		DRV_PTR(ipp_driver, CONFIG_DRM_EXYNOS_IPP),
-		DRM_VIRTUAL_DEVICE
-	}, {
 		&exynos_drm_platform_driver,
 		DRM_VIRTUAL_DEVICE
 	}
diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h b/drivers/gpu/drm/exynos/exynos_drm_drv.h
index cf131c2aa23e..21f4271c012a 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_drv.h
+++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h
@@ -185,7 +185,6 @@ struct exynos_drm_g2d_private {
 
 struct drm_exynos_file_private {
 	struct exynos_drm_g2d_private	*g2d_priv;
-	struct device			*ipp_dev;
 };
 
 /*
@@ -292,6 +291,5 @@ int exynos_atomic_commit(struct drm_device *dev, struct drm_atomic_state *state,
 extern struct platform_driver fimc_driver;
 extern struct platform_driver rotator_driver;
 extern struct platform_driver gsc_driver;
-extern struct platform_driver ipp_driver;
 extern struct platform_driver mic_driver;
 #endif
diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.c b/drivers/gpu/drm/exynos/exynos_drm_ipp.c
deleted file mode 100644
index 3edda18cc2d2..000000000000
--- a/drivers/gpu/drm/exynos/exynos_drm_ipp.c
+++ /dev/null
@@ -1,1806 +0,0 @@
-/*
- * Copyright (C) 2012 Samsung Electronics Co.Ltd
- * Authors:
- *	Eunchul Kim <chulspro.kim@samsung.com>
- *	Jinyoung Jeon <jy0.jeon@samsung.com>
- *	Sangmin Lee <lsmin.lee@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/platform_device.h>
-#include <linux/types.h>
-#include <linux/clk.h>
-#include <linux/pm_runtime.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"
-#include "exynos_drm_iommu.h"
-
-/*
- * IPP stands 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
- */
-
-/*
- * TODO
- * 1. expand command control id.
- * 2. integrate	property and config.
- * 3. removed send_event id check routine.
- * 4. compare send_event id if needed.
- * 5. free subdrv_remove notifier callback list if needed.
- * 6. need to check subdrv_open about multi-open.
- * 7. need to power_on implement power and sysmmu ctrl.
- */
-
-#define get_ipp_context(dev)	platform_get_drvdata(to_platform_device(dev))
-#define ipp_is_m2m_cmd(c)	(c == IPP_CMD_M2M)
-
-/*
- * 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;
-};
-
-/*
- * 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 DEFINE_MUTEX(exynos_drm_ippdrv_lock);
-static BLOCKING_NOTIFIER_HEAD(exynos_drm_ippnb_list);
-
-int exynos_drm_ippdrv_register(struct exynos_drm_ippdrv *ippdrv)
-{
-	mutex_lock(&exynos_drm_ippdrv_lock);
-	list_add_tail(&ippdrv->drv_list, &exynos_drm_ippdrv_list);
-	mutex_unlock(&exynos_drm_ippdrv_lock);
-
-	return 0;
-}
-
-int exynos_drm_ippdrv_unregister(struct exynos_drm_ippdrv *ippdrv)
-{
-	mutex_lock(&exynos_drm_ippdrv_lock);
-	list_del(&ippdrv->drv_list);
-	mutex_unlock(&exynos_drm_ippdrv_lock);
-
-	return 0;
-}
-
-static int ipp_create_id(struct idr *id_idr, struct mutex *lock, void *obj)
-{
-	int ret;
-
-	mutex_lock(lock);
-	ret = idr_alloc(id_idr, obj, 1, 0, GFP_KERNEL);
-	mutex_unlock(lock);
-
-	return ret;
-}
-
-static void ipp_remove_id(struct idr *id_idr, struct mutex *lock, u32 id)
-{
-	mutex_lock(lock);
-	idr_remove(id_idr, id);
-	mutex_unlock(lock);
-}
-
-static void *ipp_find_obj(struct idr *id_idr, struct mutex *lock, u32 id)
-{
-	void *obj;
-
-	mutex_lock(lock);
-	obj = idr_find(id_idr, id);
-	mutex_unlock(lock);
-
-	return obj;
-}
-
-static int ipp_check_driver(struct exynos_drm_ippdrv *ippdrv,
-			    struct drm_exynos_ipp_property *property)
-{
-	if (ippdrv->dedicated || (!ipp_is_m2m_cmd(property->cmd) &&
-				  !pm_runtime_suspended(ippdrv->dev)))
-		return -EBUSY;
-
-	if (ippdrv->check_property &&
-	    ippdrv->check_property(ippdrv->dev, property))
-		return -EINVAL;
-
-	return 0;
-}
-
-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;
-	int ret;
-
-	if (ipp_id) {
-		ippdrv = ipp_find_obj(&ctx->ipp_idr, &ctx->ipp_lock, ipp_id);
-		if (!ippdrv) {
-			DRM_DEBUG("ipp%d driver not found\n", ipp_id);
-			return ERR_PTR(-ENODEV);
-		}
-
-		ret = ipp_check_driver(ippdrv, property);
-		if (ret < 0) {
-			DRM_DEBUG("ipp%d driver check error %d\n", ipp_id, ret);
-			return ERR_PTR(ret);
-		}
-
-		return ippdrv;
-	} else {
-		list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
-			ret = ipp_check_driver(ippdrv, property);
-			if (ret == 0)
-				return ippdrv;
-		}
-
-		DRM_DEBUG("cannot find driver suitable for given property.\n");
-	}
-
-	return ERR_PTR(-ENODEV);
-}
-
-static struct exynos_drm_ippdrv *ipp_find_drv_by_handle(u32 prop_id)
-{
-	struct exynos_drm_ippdrv *ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node;
-	int count = 0;
-
-	DRM_DEBUG_KMS("prop_id[%d]\n", prop_id);
-
-	/*
-	 * This case is search ipp driver by prop_id handle.
-	 * sometimes, ipp subsystem find driver by prop_id.
-	 * e.g PAUSE state, queue buf, command control.
-	 */
-	list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
-		DRM_DEBUG_KMS("count[%d]ippdrv[%pK]\n", count++, ippdrv);
-
-		mutex_lock(&ippdrv->cmd_lock);
-		list_for_each_entry(c_node, &ippdrv->cmd_list, list) {
-			if (c_node->property.prop_id == prop_id) {
-				mutex_unlock(&ippdrv->cmd_lock);
-				return ippdrv;
-			}
-		}
-		mutex_unlock(&ippdrv->cmd_lock);
-	}
-
-	return ERR_PTR(-ENODEV);
-}
-
-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 device *dev = file_priv->ipp_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;
-
-	if (!ctx) {
-		DRM_ERROR("invalid context.\n");
-		return -EINVAL;
-	}
-
-	if (!prop_list) {
-		DRM_ERROR("invalid property parameter.\n");
-		return -EINVAL;
-	}
-
-	DRM_DEBUG_KMS("ipp_id[%d]\n", prop_list->ipp_id);
-
-	if (!prop_list->ipp_id) {
-		list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list)
-			count++;
-
-		/*
-		 * Supports ippdrv list count for user application.
-		 * First step user application getting ippdrv count.
-		 * and second step getting ippdrv capability using ipp_id.
-		 */
-		prop_list->count = count;
-	} else {
-		/*
-		 * Getting ippdrv capability by ipp_id.
-		 * some device not supported wb, output interface.
-		 * so, user application detect correct ipp driver
-		 * using this ioctl.
-		 */
-		ippdrv = ipp_find_obj(&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 -ENODEV;
-		}
-
-		*prop_list = ippdrv->prop_list;
-	}
-
-	return 0;
-}
-
-static void ipp_print_property(struct drm_exynos_ipp_property *property,
-		int idx)
-{
-	struct drm_exynos_ipp_config *config = &property->config[idx];
-	struct drm_exynos_pos *pos = &config->pos;
-	struct drm_exynos_sz *sz = &config->sz;
-
-	DRM_DEBUG_KMS("prop_id[%d]ops[%s]fmt[0x%x]\n",
-		property->prop_id, idx ? "dst" : "src", config->fmt);
-
-	DRM_DEBUG_KMS("pos[%d %d %d %d]sz[%d %d]f[%d]r[%d]\n",
-		pos->x, pos->y, pos->w, pos->h,
-		sz->hsize, sz->vsize, config->flip, config->degree);
-}
-
-static struct drm_exynos_ipp_cmd_work *ipp_create_cmd_work(void)
-{
-	struct drm_exynos_ipp_cmd_work *cmd_work;
-
-	cmd_work = kzalloc(sizeof(*cmd_work), GFP_KERNEL);
-	if (!cmd_work)
-		return ERR_PTR(-ENOMEM);
-
-	INIT_WORK((struct work_struct *)cmd_work, ipp_sched_cmd);
-
-	return cmd_work;
-}
-
-static struct drm_exynos_ipp_event_work *ipp_create_event_work(void)
-{
-	struct drm_exynos_ipp_event_work *event_work;
-
-	event_work = kzalloc(sizeof(*event_work), GFP_KERNEL);
-	if (!event_work)
-		return ERR_PTR(-ENOMEM);
-
-	INIT_WORK(&event_work->work, ipp_sched_event);
-
-	return event_work;
-}
-
-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 device *dev = file_priv->ipp_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;
-	u32 prop_id;
-	int ret, i;
-
-	if (!ctx) {
-		DRM_ERROR("invalid context.\n");
-		return -EINVAL;
-	}
-
-	if (!property) {
-		DRM_ERROR("invalid property parameter.\n");
-		return -EINVAL;
-	}
-
-	prop_id = property->prop_id;
-
-	/*
-	 * This is log print for user application property.
-	 * user application set various property.
-	 */
-	for_each_ipp_ops(i)
-		ipp_print_property(property, i);
-
-	/*
-	 * In case prop_id is not zero try to set existing property.
-	 */
-	if (prop_id) {
-		c_node = ipp_find_obj(&ctx->prop_idr, &ctx->prop_lock, prop_id);
-
-		if (!c_node || c_node->filp != file) {
-			DRM_DEBUG_KMS("prop_id[%d] not found\n", prop_id);
-			return -EINVAL;
-		}
-
-		if (c_node->state != IPP_STATE_STOP) {
-			DRM_DEBUG_KMS("prop_id[%d] not stopped\n", prop_id);
-			return -EINVAL;
-		}
-
-		c_node->property = *property;
-
-		return 0;
-	}
-
-	/* find ipp driver using ipp id */
-	ippdrv = ipp_find_driver(ctx, property);
-	if (IS_ERR(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)
-		return -ENOMEM;
-
-	ret = ipp_create_id(&ctx->prop_idr, &ctx->prop_lock, c_node);
-	if (ret < 0) {
-		DRM_ERROR("failed to create id.\n");
-		goto err_clear;
-	}
-	property->prop_id = ret;
-
-	DRM_DEBUG_KMS("created prop_id[%d]cmd[%d]ippdrv[%pK]\n",
-		property->prop_id, property->cmd, ippdrv);
-
-	/* stored property information and ippdrv in private data */
-	c_node->property = *property;
-	c_node->state = IPP_STATE_IDLE;
-	c_node->filp = file;
-
-	c_node->start_work = ipp_create_cmd_work();
-	if (IS_ERR(c_node->start_work)) {
-		DRM_ERROR("failed to create start work.\n");
-		ret = PTR_ERR(c_node->start_work);
-		goto err_remove_id;
-	}
-
-	c_node->stop_work = ipp_create_cmd_work();
-	if (IS_ERR(c_node->stop_work)) {
-		DRM_ERROR("failed to create stop work.\n");
-		ret = PTR_ERR(c_node->stop_work);
-		goto err_free_start;
-	}
-
-	c_node->event_work = ipp_create_event_work();
-	if (IS_ERR(c_node->event_work)) {
-		DRM_ERROR("failed to create event work.\n");
-		ret = PTR_ERR(c_node->event_work);
-		goto err_free_stop;
-	}
-
-	mutex_init(&c_node->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);
-	mutex_lock(&ippdrv->cmd_lock);
-	list_add_tail(&c_node->list, &ippdrv->cmd_list);
-	mutex_unlock(&ippdrv->cmd_lock);
-
-	/* make dedicated state without m2m */
-	if (!ipp_is_m2m_cmd(property->cmd))
-		ippdrv->dedicated = true;
-
-	return 0;
-
-err_free_stop:
-	kfree(c_node->stop_work);
-err_free_start:
-	kfree(c_node->start_work);
-err_remove_id:
-	ipp_remove_id(&ctx->prop_idr, &ctx->prop_lock, property->prop_id);
-err_clear:
-	kfree(c_node);
-	return ret;
-}
-
-static int ipp_validate_mem_node(struct drm_device *drm_dev,
-				 struct drm_exynos_ipp_mem_node *m_node,
-				 struct drm_exynos_ipp_cmd_node *c_node)
-{
-	struct drm_exynos_ipp_config *ipp_cfg;
-	unsigned int num_plane;
-	unsigned long size, buf_size = 0, plane_size, img_size = 0;
-	unsigned int bpp, width, height;
-	int i;
-
-	ipp_cfg = &c_node->property.config[m_node->ops_id];
-	num_plane = drm_format_num_planes(ipp_cfg->fmt);
-
-	/**
-	 * This is a rather simplified validation of a memory node.
-	 * It basically verifies provided gem object handles
-	 * and the buffer sizes with respect to current configuration.
-	 * This is not the best that can be done
-	 * but it seems more than enough
-	 */
-	for (i = 0; i < num_plane; ++i) {
-		width = ipp_cfg->sz.hsize;
-		height = ipp_cfg->sz.vsize;
-		bpp = drm_format_plane_cpp(ipp_cfg->fmt, i);
-
-		/*
-		 * The result of drm_format_plane_cpp() for chroma planes must
-		 * be used with drm_format_xxxx_chroma_subsampling() for
-		 * correct result.
-		 */
-		if (i > 0) {
-			width /= drm_format_horz_chroma_subsampling(
-								ipp_cfg->fmt);
-			height /= drm_format_vert_chroma_subsampling(
-								ipp_cfg->fmt);
-		}
-		plane_size = width * height * bpp;
-		img_size += plane_size;
-
-		if (m_node->buf_info.handles[i]) {
-			size = exynos_drm_gem_get_size(drm_dev,
-					m_node->buf_info.handles[i],
-					c_node->filp);
-			if (plane_size > size) {
-				DRM_ERROR(
-					"buffer %d is smaller than required\n",
-					i);
-				return -EINVAL;
-			}
-
-			buf_size += size;
-		}
-	}
-
-	if (buf_size < img_size) {
-		DRM_ERROR("size of buffers(%lu) is smaller than image(%lu)\n",
-			buf_size, img_size);
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
-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;
-
-	DRM_DEBUG_KMS("node[%pK]\n", m_node);
-
-	if (!m_node) {
-		DRM_ERROR("invalid dequeue node.\n");
-		return -EFAULT;
-	}
-
-	DRM_DEBUG_KMS("ops_id[%d]\n", 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,
-							c_node->filp);
-	}
-
-	list_del(&m_node->list);
-	kfree(m_node);
-
-	return 0;
-}
-
-static struct drm_exynos_ipp_mem_node
-		*ipp_get_mem_node(struct drm_device *drm_dev,
-		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;
-	int i;
-
-	m_node = kzalloc(sizeof(*m_node), GFP_KERNEL);
-	if (!m_node)
-		return ERR_PTR(-ENOMEM);
-
-	buf_info = &m_node->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;
-	INIT_LIST_HEAD(&m_node->list);
-
-	DRM_DEBUG_KMS("m_node[%pK]ops_id[%d]\n", m_node, qbuf->ops_id);
-	DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]\n", qbuf->prop_id, m_node->buf_id);
-
-	for_each_ipp_planar(i) {
-		DRM_DEBUG_KMS("i[%d]handle[0x%x]\n", i, qbuf->handle[i]);
-
-		/* get dma address by handle */
-		if (qbuf->handle[i]) {
-			dma_addr_t *addr;
-
-			addr = exynos_drm_gem_get_dma_addr(drm_dev,
-					qbuf->handle[i], c_node->filp);
-			if (IS_ERR(addr)) {
-				DRM_ERROR("failed to get addr.\n");
-				ipp_put_mem_node(drm_dev, c_node, m_node);
-				return ERR_PTR(-EFAULT);
-			}
-
-			buf_info->handles[i] = qbuf->handle[i];
-			buf_info->base[i] = *addr;
-			DRM_DEBUG_KMS("i[%d]base[%pad]hd[0x%lx]\n", i,
-				      &buf_info->base[i], buf_info->handles[i]);
-		}
-	}
-
-	mutex_lock(&c_node->mem_lock);
-	if (ipp_validate_mem_node(drm_dev, m_node, c_node)) {
-		ipp_put_mem_node(drm_dev, c_node, m_node);
-		mutex_unlock(&c_node->mem_lock);
-		return ERR_PTR(-EFAULT);
-	}
-	list_add_tail(&m_node->list, &c_node->mem_list[qbuf->ops_id]);
-	mutex_unlock(&c_node->mem_lock);
-
-	return m_node;
-}
-
-static void ipp_clean_mem_nodes(struct drm_device *drm_dev,
-			       struct drm_exynos_ipp_cmd_node *c_node, int ops)
-{
-	struct drm_exynos_ipp_mem_node *m_node, *tm_node;
-	struct list_head *head = &c_node->mem_list[ops];
-
-	mutex_lock(&c_node->mem_lock);
-
-	list_for_each_entry_safe(m_node, tm_node, head, list) {
-		int ret;
-
-		ret = ipp_put_mem_node(drm_dev, c_node, m_node);
-		if (ret)
-			DRM_ERROR("failed to put m_node.\n");
-	}
-
-	mutex_unlock(&c_node->mem_lock);
-}
-
-static int ipp_get_event(struct drm_device *drm_dev,
-		struct drm_exynos_ipp_cmd_node *c_node,
-		struct drm_exynos_ipp_queue_buf *qbuf)
-{
-	struct drm_exynos_ipp_send_event *e;
-	int ret;
-
-	DRM_DEBUG_KMS("ops_id[%d]buf_id[%d]\n", qbuf->ops_id, qbuf->buf_id);
-
-	e = kzalloc(sizeof(*e), GFP_KERNEL);
-	if (!e)
-		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;
-
-	ret = drm_event_reserve_init(drm_dev, c_node->filp, &e->base, &e->event.base);
-	if (ret) {
-		kfree(e);
-		return ret;
-	}
-
-	mutex_lock(&c_node->event_lock);
-	list_add_tail(&e->base.link, &c_node->event_list);
-	mutex_unlock(&c_node->event_lock);
-
-	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;
-
-	mutex_lock(&c_node->event_lock);
-	list_for_each_entry_safe(e, te, &c_node->event_list, base.link) {
-		DRM_DEBUG_KMS("count[%d]e[%pK]\n", count++, e);
-
-		/*
-		 * qbuf == NULL condition means all event deletion.
-		 * stop operations want to delete all event list.
-		 * another case delete only same buf id.
-		 */
-		if (!qbuf) {
-			/* delete list */
-			list_del(&e->base.link);
-			kfree(e);
-		}
-
-		/* compare buffer id */
-		if (qbuf && (qbuf->buf_id ==
-		    e->event.buf_id[EXYNOS_DRM_OPS_DST])) {
-			/* delete list */
-			list_del(&e->base.link);
-			kfree(e);
-			goto out_unlock;
-		}
-	}
-
-out_unlock:
-	mutex_unlock(&c_node->event_lock);
-	return;
-}
-
-static void ipp_clean_cmd_node(struct ipp_context *ctx,
-				struct drm_exynos_ipp_cmd_node *c_node)
-{
-	int i;
-
-	/* cancel works */
-	cancel_work_sync(&c_node->start_work->work);
-	cancel_work_sync(&c_node->stop_work->work);
-	cancel_work_sync(&c_node->event_work->work);
-
-	/* put event */
-	ipp_put_event(c_node, NULL);
-
-	for_each_ipp_ops(i)
-		ipp_clean_mem_nodes(ctx->subdrv.drm_dev, c_node, i);
-
-	/* delete list */
-	list_del(&c_node->list);
-
-	ipp_remove_id(&ctx->prop_idr, &ctx->prop_lock,
-			c_node->property.prop_id);
-
-	/* destroy mutex */
-	mutex_destroy(&c_node->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 bool ipp_check_mem_list(struct drm_exynos_ipp_cmd_node *c_node)
-{
-	switch (c_node->property.cmd) {
-	case IPP_CMD_WB:
-		return !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_DST]);
-	case IPP_CMD_OUTPUT:
-		return !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_SRC]);
-	case IPP_CMD_M2M:
-	default:
-		return !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_SRC]) &&
-		       !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_DST]);
-	}
-}
-
-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("buf_id[%d]\n", qbuf->buf_id);
-
-	/* source/destination memory list */
-	head = &c_node->mem_list[qbuf->ops_id];
-
-	/* find memory node from memory list */
-	list_for_each_entry(m_node, head, list) {
-		DRM_DEBUG_KMS("count[%d]m_node[%pK]\n", count++, m_node);
-
-		/* compare buffer id */
-		if (m_node->buf_id == qbuf->buf_id)
-			return m_node;
-	}
-
-	return NULL;
-}
-
-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("node[%pK]\n", m_node);
-
-	if (!m_node) {
-		DRM_ERROR("invalid queue node.\n");
-		return -EFAULT;
-	}
-
-	DRM_DEBUG_KMS("ops_id[%d]\n", m_node->ops_id);
-
-	/* get operations callback */
-	ops = ippdrv->ops[m_node->ops_id];
-	if (!ops) {
-		DRM_ERROR("not support ops.\n");
-		return -EFAULT;
-	}
-
-	/* 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");
-			return ret;
-		}
-	}
-
-	return ret;
-}
-
-static 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, &cmd_work->work);
-}
-
-static int ipp_queue_buf_with_run(struct device *dev,
-		struct drm_exynos_ipp_cmd_node *c_node,
-		struct drm_exynos_ipp_mem_node *m_node,
-		struct drm_exynos_ipp_queue_buf *qbuf)
-{
-	struct exynos_drm_ippdrv *ippdrv;
-	struct drm_exynos_ipp_property *property;
-	struct exynos_drm_ipp_ops *ops;
-	int ret;
-
-	ippdrv = ipp_find_drv_by_handle(qbuf->prop_id);
-	if (IS_ERR(ippdrv)) {
-		DRM_ERROR("failed to get ipp driver.\n");
-		return -EFAULT;
-	}
-
-	ops = ippdrv->ops[qbuf->ops_id];
-	if (!ops) {
-		DRM_ERROR("failed to get ops.\n");
-		return -EFAULT;
-	}
-
-	property = &c_node->property;
-
-	if (c_node->state != IPP_STATE_START) {
-		DRM_DEBUG_KMS("bypass for invalid state.\n");
-		return 0;
-	}
-
-	mutex_lock(&c_node->mem_lock);
-	if (!ipp_check_mem_list(c_node)) {
-		mutex_unlock(&c_node->mem_lock);
-		DRM_DEBUG_KMS("empty memory.\n");
-		return 0;
-	}
-
-	/*
-	 * If set destination buffer and enabled clock,
-	 * then m2m operations need start operations at queue_buf
-	 */
-	if (ipp_is_m2m_cmd(property->cmd)) {
-		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 {
-		ret = ipp_set_mem_node(ippdrv, c_node, m_node);
-		if (ret) {
-			mutex_unlock(&c_node->mem_lock);
-			DRM_ERROR("failed to set m node.\n");
-			return ret;
-		}
-	}
-	mutex_unlock(&c_node->mem_lock);
-
-	return 0;
-}
-
-static void ipp_clean_queue_buf(struct drm_device *drm_dev,
-		struct drm_exynos_ipp_cmd_node *c_node,
-		struct drm_exynos_ipp_queue_buf *qbuf)
-{
-	struct drm_exynos_ipp_mem_node *m_node, *tm_node;
-
-	/* delete list */
-	mutex_lock(&c_node->mem_lock);
-	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->mem_lock);
-}
-
-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 device *dev = file_priv->ipp_dev;
-	struct ipp_context *ctx = get_ipp_context(dev);
-	struct drm_exynos_ipp_queue_buf *qbuf = data;
-	struct drm_exynos_ipp_cmd_node *c_node;
-	struct drm_exynos_ipp_mem_node *m_node;
-	int ret;
-
-	if (!qbuf) {
-		DRM_ERROR("invalid buf parameter.\n");
-		return -EINVAL;
-	}
-
-	if (qbuf->ops_id >= EXYNOS_DRM_OPS_MAX) {
-		DRM_ERROR("invalid ops parameter.\n");
-		return -EINVAL;
-	}
-
-	DRM_DEBUG_KMS("prop_id[%d]ops_id[%s]buf_id[%d]buf_type[%d]\n",
-		qbuf->prop_id, qbuf->ops_id ? "dst" : "src",
-		qbuf->buf_id, qbuf->buf_type);
-
-	/* find command node */
-	c_node = ipp_find_obj(&ctx->prop_idr, &ctx->prop_lock,
-		qbuf->prop_id);
-	if (!c_node || c_node->filp != file) {
-		DRM_ERROR("failed to get command node.\n");
-		return -ENODEV;
-	}
-
-	/* buffer control */
-	switch (qbuf->buf_type) {
-	case IPP_BUF_ENQUEUE:
-		/* get memory node */
-		m_node = ipp_get_mem_node(drm_dev, c_node, qbuf);
-		if (IS_ERR(m_node)) {
-			DRM_ERROR("failed to get m_node.\n");
-			return PTR_ERR(m_node);
-		}
-
-		/*
-		 * first step get event for destination buffer.
-		 * and second step when M2M case run with destination buffer
-		 * if needed.
-		 */
-		if (qbuf->ops_id == EXYNOS_DRM_OPS_DST) {
-			/* get event for destination buffer */
-			ret = ipp_get_event(drm_dev, c_node, qbuf);
-			if (ret) {
-				DRM_ERROR("failed to get event.\n");
-				goto err_clean_node;
-			}
-
-			/*
-			 * M2M case run play control for streaming feature.
-			 * other case set address and waiting.
-			 */
-			ret = ipp_queue_buf_with_run(dev, c_node, m_node, qbuf);
-			if (ret) {
-				DRM_ERROR("failed to run command.\n");
-				goto err_clean_node;
-			}
-		}
-		break;
-	case IPP_BUF_DEQUEUE:
-		mutex_lock(&c_node->lock);
-
-		/* put event for destination buffer */
-		if (qbuf->ops_id == EXYNOS_DRM_OPS_DST)
-			ipp_put_event(c_node, qbuf);
-
-		ipp_clean_queue_buf(drm_dev, c_node, qbuf);
-
-		mutex_unlock(&c_node->lock);
-		break;
-	default:
-		DRM_ERROR("invalid buffer control.\n");
-		return -EINVAL;
-	}
-
-	return 0;
-
-err_clean_node:
-	DRM_ERROR("clean memory nodes.\n");
-
-	ipp_clean_queue_buf(drm_dev, c_node, qbuf);
-	return ret;
-}
-
-static bool exynos_drm_ipp_check_valid(struct device *dev,
-		enum drm_exynos_ipp_ctrl ctrl, enum drm_exynos_ipp_state state)
-{
-	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;
-	}
-
-	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_ippdrv *ippdrv = NULL;
-	struct device *dev = file_priv->ipp_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;
-
-	if (!ctx) {
-		DRM_ERROR("invalid context.\n");
-		return -EINVAL;
-	}
-
-	if (!cmd_ctrl) {
-		DRM_ERROR("invalid control parameter.\n");
-		return -EINVAL;
-	}
-
-	DRM_DEBUG_KMS("ctrl[%d]prop_id[%d]\n",
-		cmd_ctrl->ctrl, cmd_ctrl->prop_id);
-
-	ippdrv = ipp_find_drv_by_handle(cmd_ctrl->prop_id);
-	if (IS_ERR(ippdrv)) {
-		DRM_ERROR("failed to get ipp driver.\n");
-		return PTR_ERR(ippdrv);
-	}
-
-	c_node = ipp_find_obj(&ctx->prop_idr, &ctx->prop_lock,
-		cmd_ctrl->prop_id);
-	if (!c_node || c_node->filp != file) {
-		DRM_ERROR("invalid command node list.\n");
-		return -ENODEV;
-	}
-
-	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:
-		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(300))) {
-			DRM_ERROR("timeout stop:prop_id[%d]\n",
-				c_node->property.prop_id);
-		}
-
-		c_node->state = IPP_STATE_STOP;
-		ippdrv->dedicated = false;
-		mutex_lock(&ippdrv->cmd_lock);
-		ipp_clean_cmd_node(ctx, c_node);
-
-		if (list_empty(&ippdrv->cmd_list))
-			pm_runtime_put_sync(ippdrv->dev);
-		mutex_unlock(&ippdrv->cmd_lock);
-		break;
-	case IPP_CTRL_PAUSE:
-		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);
-		}
-
-		c_node->state = IPP_STATE_STOP;
-		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:
-		DRM_ERROR("could not support this state currently.\n");
-		return -EINVAL;
-	}
-
-	DRM_DEBUG_KMS("done ctrl[%d]prop_id[%d]\n",
-		cmd_ctrl->ctrl, cmd_ctrl->prop_id);
-
-	return 0;
-}
-
-int exynos_drm_ippnb_register(struct notifier_block *nb)
-{
-	return blocking_notifier_chain_register(
-		&exynos_drm_ippnb_list, nb);
-}
-
-int exynos_drm_ippnb_unregister(struct notifier_block *nb)
-{
-	return blocking_notifier_chain_unregister(
-		&exynos_drm_ippnb_list, nb);
-}
-
-int exynos_drm_ippnb_send_event(unsigned long val, void *v)
-{
-	return blocking_notifier_call_chain(
-		&exynos_drm_ippnb_list, val, v);
-}
-
-static int ipp_set_property(struct exynos_drm_ippdrv *ippdrv,
-		struct drm_exynos_ipp_property *property)
-{
-	struct exynos_drm_ipp_ops *ops = NULL;
-	bool swap = false;
-	int ret, i;
-
-	if (!property) {
-		DRM_ERROR("invalid property parameter.\n");
-		return -EINVAL;
-	}
-
-	DRM_DEBUG_KMS("prop_id[%d]\n", property->prop_id);
-
-	/* reset h/w block */
-	if (ippdrv->reset &&
-	    ippdrv->reset(ippdrv->dev)) {
-		return -EINVAL;
-	}
-
-	/* set source,destination operations */
-	for_each_ipp_ops(i) {
-		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)
-				return ret;
-		}
-
-		/* set transform for rotation, flip */
-		if (ops->set_transf) {
-			ret = ops->set_transf(ippdrv->dev, config->degree,
-				config->flip, &swap);
-			if (ret)
-				return ret;
-		}
-
-		/* set size */
-		if (ops->set_size) {
-			ret = ops->set_size(ippdrv->dev, swap, &config->pos,
-				&config->sz);
-			if (ret)
-				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("prop_id[%d]\n", property->prop_id);
-
-	/* store command info in ippdrv */
-	ippdrv->c_node = c_node;
-
-	mutex_lock(&c_node->mem_lock);
-	if (!ipp_check_mem_list(c_node)) {
-		DRM_DEBUG_KMS("empty memory.\n");
-		ret = -ENOMEM;
-		goto err_unlock;
-	}
-
-	/* set current property in ippdrv */
-	ret = ipp_set_property(ippdrv, property);
-	if (ret) {
-		DRM_ERROR("failed to set property.\n");
-		ippdrv->c_node = NULL;
-		goto err_unlock;
-	}
-
-	/* 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);
-
-			DRM_DEBUG_KMS("m_node[%pK]\n", m_node);
-
-			ret = ipp_set_mem_node(ippdrv, c_node, m_node);
-			if (ret) {
-				DRM_ERROR("failed to set m node.\n");
-				goto err_unlock;
-			}
-		}
-		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");
-				goto err_unlock;
-			}
-		}
-		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");
-				goto err_unlock;
-			}
-		}
-		break;
-	default:
-		DRM_ERROR("invalid operations.\n");
-		ret = -EINVAL;
-		goto err_unlock;
-	}
-	mutex_unlock(&c_node->mem_lock);
-
-	DRM_DEBUG_KMS("cmd[%d]\n", property->cmd);
-
-	/* start operations */
-	if (ippdrv->start) {
-		ret = ippdrv->start(ippdrv->dev, property->cmd);
-		if (ret) {
-			DRM_ERROR("failed to start ops.\n");
-			ippdrv->c_node = NULL;
-			return ret;
-		}
-	}
-
-	return 0;
-
-err_unlock:
-	mutex_unlock(&c_node->mem_lock);
-	ippdrv->c_node = NULL;
-	return ret;
-}
-
-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_property *property = &c_node->property;
-	int i;
-
-	DRM_DEBUG_KMS("prop_id[%d]\n", property->prop_id);
-
-	/* stop operations */
-	if (ippdrv->stop)
-		ippdrv->stop(ippdrv->dev, property->cmd);
-
-	/* check command */
-	switch (property->cmd) {
-	case IPP_CMD_M2M:
-		for_each_ipp_ops(i)
-			ipp_clean_mem_nodes(drm_dev, c_node, i);
-		break;
-	case IPP_CMD_WB:
-		ipp_clean_mem_nodes(drm_dev, c_node, EXYNOS_DRM_OPS_DST);
-		break;
-	case IPP_CMD_OUTPUT:
-		ipp_clean_mem_nodes(drm_dev, c_node, EXYNOS_DRM_OPS_SRC);
-		break;
-	default:
-		DRM_ERROR("invalid operations.\n");
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
-void ipp_sched_cmd(struct work_struct *work)
-{
-	struct drm_exynos_ipp_cmd_work *cmd_work =
-		container_of(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;
-
-	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->lock);
-
-	property = &c_node->property;
-
-	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 (ipp_is_m2m_cmd(property->cmd)) {
-			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("ctrl[%d] done.\n", cmd_work->ctrl);
-
-err_unlock:
-	mutex_unlock(&c_node->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;
-	u32 tbuf_id[EXYNOS_DRM_OPS_MAX] = {0, };
-	int ret, i;
-
-	for_each_ipp_ops(i)
-		DRM_DEBUG_KMS("%s buf_id[%d]\n", 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;
-	}
-
-	mutex_lock(&c_node->event_lock);
-	if (list_empty(&c_node->event_list)) {
-		DRM_DEBUG_KMS("event list is empty.\n");
-		ret = 0;
-		goto err_event_unlock;
-	}
-
-	mutex_lock(&c_node->mem_lock);
-	if (!ipp_check_mem_list(c_node)) {
-		DRM_DEBUG_KMS("empty memory.\n");
-		ret = 0;
-		goto err_mem_unlock;
-	}
-
-	/* 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);
-
-			tbuf_id[i] = m_node->buf_id;
-			DRM_DEBUG_KMS("%s buf_id[%d]\n",
-				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");
-			ret = -ENOMEM;
-			goto err_mem_unlock;
-		}
-
-		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);
-
-		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");
-		ret = -EINVAL;
-		goto err_mem_unlock;
-	}
-	mutex_unlock(&c_node->mem_lock);
-
-	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,
-	 * then we make event and link to event list tail.
-	 * 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);
-
-	do_gettimeofday(&now);
-	DRM_DEBUG_KMS("tv_sec[%ld]tv_usec[%ld]\n", 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];
-
-	drm_send_event(drm_dev, &e->base);
-	mutex_unlock(&c_node->event_lock);
-
-	DRM_DEBUG_KMS("done cmd[%d]prop_id[%d]buf_id[%d]\n",
-		property->cmd, property->prop_id, tbuf_id[EXYNOS_DRM_OPS_DST]);
-
-	return 0;
-
-err_mem_unlock:
-	mutex_unlock(&c_node->mem_lock);
-err_event_unlock:
-	mutex_unlock(&c_node->event_lock);
-	return ret;
-}
-
-void ipp_sched_event(struct work_struct *work)
-{
-	struct drm_exynos_ipp_event_work *event_work =
-		container_of(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("buf_id[%d]\n", 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->c_node;
-	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, then IPP make
-	 * synchronization with command thread, so make complete event.
-	 * or going out operations.
-	 */
-	if (c_node->state != IPP_STATE_START) {
-		DRM_DEBUG_KMS("bypass state[%d]prop_id[%d]\n",
-			c_node->state, c_node->property.prop_id);
-		goto err_completion;
-	}
-
-	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 (ipp_is_m2m_cmd(c_node->property.cmd))
-		complete(&c_node->start_complete);
-}
-
-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;
-
-	/* 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);
-		if (ret < 0) {
-			DRM_ERROR("failed to create id.\n");
-			goto err;
-		}
-		ippdrv->prop_list.ipp_id = ret;
-
-		DRM_DEBUG_KMS("count[%d]ippdrv[%pK]ipp_id[%d]\n",
-			count++, ippdrv, ret);
-
-		/* 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);
-		mutex_init(&ippdrv->cmd_lock);
-
-		ret = drm_iommu_attach_device(drm_dev, ippdrv->dev);
-		if (ret) {
-			DRM_ERROR("failed to activate iommu\n");
-			goto err;
-		}
-	}
-
-	return 0;
-
-err:
-	/* get ipp driver entry */
-	list_for_each_entry_continue_reverse(ippdrv, &exynos_drm_ippdrv_list,
-						drv_list) {
-		drm_iommu_detach_device(drm_dev, ippdrv->dev);
-
-		ipp_remove_id(&ctx->ipp_idr, &ctx->ipp_lock,
-				ippdrv->prop_list.ipp_id);
-	}
-
-	return ret;
-}
-
-static void ipp_subdrv_remove(struct drm_device *drm_dev, struct device *dev)
-{
-	struct exynos_drm_ippdrv *ippdrv, *t;
-	struct ipp_context *ctx = get_ipp_context(dev);
-
-	/* get ipp driver entry */
-	list_for_each_entry_safe(ippdrv, t, &exynos_drm_ippdrv_list, drv_list) {
-		drm_iommu_detach_device(drm_dev, ippdrv->dev);
-
-		ipp_remove_id(&ctx->ipp_idr, &ctx->ipp_lock,
-				ippdrv->prop_list.ipp_id);
-
-		ippdrv->drm_dev = NULL;
-		exynos_drm_ippdrv_unregister(ippdrv);
-	}
-}
-
-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;
-
-	file_priv->ipp_dev = dev;
-
-	DRM_DEBUG_KMS("done priv[%pK]\n", dev);
-
-	return 0;
-}
-
-static void ipp_subdrv_close(struct drm_device *drm_dev, struct device *dev,
-		struct drm_file *file)
-{
-	struct exynos_drm_ippdrv *ippdrv = NULL;
-	struct ipp_context *ctx = get_ipp_context(dev);
-	struct drm_exynos_ipp_cmd_node *c_node, *tc_node;
-	int count = 0;
-
-	list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
-		mutex_lock(&ippdrv->cmd_lock);
-		list_for_each_entry_safe(c_node, tc_node,
-			&ippdrv->cmd_list, list) {
-			DRM_DEBUG_KMS("count[%d]ippdrv[%pK]\n",
-				count++, ippdrv);
-
-			if (c_node->filp == file) {
-				/*
-				 * 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(ctx, c_node);
-				if (list_empty(&ippdrv->cmd_list))
-					pm_runtime_put_sync(ippdrv->dev);
-			}
-		}
-		mutex_unlock(&ippdrv->cmd_lock);
-	}
-
-	return;
-}
-
-static int ipp_probe(struct platform_device *pdev)
-{
-	struct device *dev = &pdev->dev;
-	struct ipp_context *ctx;
-	struct exynos_drm_subdrv *subdrv;
-	int ret;
-
-	ctx = devm_kzalloc(dev, sizeof(*ctx), GFP_KERNEL);
-	if (!ctx)
-		return -ENOMEM;
-
-	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");
-		return -EINVAL;
-	}
-
-	/*
-	 * 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 then 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(dev, "drm ipp registered successfully.\n");
-
-	return 0;
-
-err_cmd_workq:
-	destroy_workqueue(ctx->cmd_workq);
-err_event_workq:
-	destroy_workqueue(ctx->event_workq);
-	return ret;
-}
-
-static int ipp_remove(struct platform_device *pdev)
-{
-	struct ipp_context *ctx = platform_get_drvdata(pdev);
-
-	/* unregister sub driver */
-	exynos_drm_subdrv_unregister(&ctx->subdrv);
-
-	/* remove,destroy ipp idr */
-	idr_destroy(&ctx->ipp_idr);
-	idr_destroy(&ctx->prop_idr);
-
-	mutex_destroy(&ctx->ipp_lock);
-	mutex_destroy(&ctx->prop_lock);
-
-	/* destroy command, event work queue */
-	destroy_workqueue(ctx->cmd_workq);
-	destroy_workqueue(ctx->event_workq);
-
-	return 0;
-}
-
-struct platform_driver ipp_driver = {
-	.probe		= ipp_probe,
-	.remove		= ipp_remove,
-	.driver		= {
-		.name	= "exynos-drm-ipp",
-		.owner	= THIS_MODULE,
-	},
-};
-
diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.h b/drivers/gpu/drm/exynos/exynos_drm_ipp.h
deleted file mode 100644
index 2a61547a39d0..000000000000
--- a/drivers/gpu/drm/exynos/exynos_drm_ipp.h
+++ /dev/null
@@ -1,252 +0,0 @@
-/*
- * Copyright (c) 2012 Samsung Electronics Co., Ltd.
- *
- * Authors:
- *	Eunchul Kim <chulspro.kim@samsung.com>
- *	Jinyoung Jeon <jy0.jeon@samsung.com>
- *	Sangmin Lee <lsmin.lee@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.
- */
-
-#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.
- *
- * @list: list head to command queue information.
- * @event_list: list head of event.
- * @mem_list: list head to source,destination memory queue information.
- * @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.
- * @filp: associated file pointer.
- */
-struct drm_exynos_ipp_cmd_node {
-	struct list_head	list;
-	struct list_head	event_list;
-	struct list_head	mem_list[EXYNOS_DRM_OPS_MAX];
-	struct mutex	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;
-	struct drm_file	*filp;
-};
-
-/*
- * A structure of buffer information.
- *
- * @handles: Y, Cb, Cr each gem object handle.
- * @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 information.
- *
- * @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, bool *swap);
-	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.
- * @dedicated: dedicated ipp device.
- * @ops: source, destination operations.
- * @event_workq: event work queue.
- * @c_node: current command information.
- * @cmd_list: list head for command information.
- * @cmd_lock: lock for synchronization of access to cmd_list.
- * @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;
-	bool	dedicated;
-	struct exynos_drm_ipp_ops	*ops[EXYNOS_DRM_OPS_MAX];
-	struct workqueue_struct	*event_workq;
-	struct drm_exynos_ipp_cmd_node *c_node;
-	struct list_head	cmd_list;
-	struct mutex	cmd_lock;
-	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 cb3e9f9d029f..544db2cecb36 100644
--- a/include/uapi/drm/exynos_drm.h
+++ b/include/uapi/drm/exynos_drm.h
@@ -134,172 +134,6 @@ 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),
-	EXYNOS_DRM_FLIP_VERTICAL = (1 << 0),
-	EXYNOS_DRM_FLIP_HORIZONTAL = (1 << 1),
-	EXYNOS_DRM_FLIP_BOTH = EXYNOS_DRM_FLIP_VERTICAL |
-			EXYNOS_DRM_FLIP_HORIZONTAL,
-};
-
-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 {
-	__u32 ops_id;
-	__u32 flip;
-	__u32 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.
- * @refresh_rate: refresh rate.
- */
-struct drm_exynos_ipp_property {
-	struct drm_exynos_ipp_config config[EXYNOS_DRM_OPS_MAX];
-	__u32	cmd;
-	__u32	ipp_id;
-	__u32	prop_id;
-	__u32	refresh_rate;
-};
-
-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 {
-	__u32	ops_id;
-	__u32	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;
-	__u32	ctrl;
-};
-
 #define DRM_EXYNOS_GEM_CREATE		0x00
 #define DRM_EXYNOS_GEM_MAP		0x01
 /* Reserved 0x03 ~ 0x05 for exynos specific gem ioctl */
@@ -311,11 +145,7 @@ struct drm_exynos_ipp_cmd_ctrl {
 #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
+/* Reserved 0x30 ~ 0x33 for obsolete Exynos IPP ioctls */
 
 #define DRM_IOCTL_EXYNOS_GEM_CREATE		DRM_IOWR(DRM_COMMAND_BASE + \
 		DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create)
@@ -334,18 +164,8 @@ struct drm_exynos_ipp_cmd_ctrl {
 #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;
@@ -356,16 +176,6 @@ 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];
-};
-
 #if defined(__cplusplus)
 }
 #endif
-- 
1.9.1

[PATCH 2/6] drm/exynos: ipp: Add IPP v2 framework

[Marek Szyprowski]

This patch adds Exynos IPP v2 subsystem and userspace API.

New userspace API is focused ONLY on memory-to-memory image processing.
The two remainging IPP operation modes (framebuffer writeback and
local-path output with image processing) can be implemented using
standard DRM features: writeback connectors and additional DRM planes with
scaling features.

V2 IPP userspace API is not compatible with old IPP ioctls. This is a
significant change, but the old IPP subsystem in mainline Linux kernel was
partially disfunctional anyway and not used in any open-source project.

V2 IPP userspace API is based on stateless approach, which much better fits
to memory-to-memory image processing model. It also provides support for
all image formats, which are both already defined in DRM API and supported
by the existing IPP hardware modules.

The API consists of the following ioctls:
- DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image
  processing modules,
- DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image
  formats of given IPP module,
- DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for
  selected image format of given IPP module,
- DRM_IOCTL_EXYNOS_IPP_COMMIT: to perform operation described by the
  provided structures (source and destination buffers, operation rectangle,
  transformation, etc).

The proposed userspace API is extensible. In the future more advanced image
processing operations can be defined to support for example blending.

Userspace API is fully functional also on DRM render nodes, so it is not
limited to the root/privileged client.

Internal driver API also has been completely rewritten. New IPP core
performs all possible input validation, checks and object life-time
control. The drivers can focus only on writing configuration to hardware
registers. Stateless nature of DRM_IOCTL_EXYNOS_IPP_COMMIT ioctl simplifies
the driver API. Minimal driver needs to provide a single callback for
starting processing and an array with supported image formats.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/gpu/drm/exynos/Kconfig          |   3 +
 drivers/gpu/drm/exynos/Makefile         |   1 +
 drivers/gpu/drm/exynos/exynos_drm_drv.c |   9 +
 drivers/gpu/drm/exynos/exynos_drm_ipp.c | 893 ++++++++++++++++++++++++++++++++
 drivers/gpu/drm/exynos/exynos_drm_ipp.h | 188 +++++++
 include/uapi/drm/exynos_drm.h           | 227 ++++++++
 6 files changed, 1321 insertions(+)
 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 88cff0e039b6..2e097a81df12 100644
--- a/drivers/gpu/drm/exynos/Kconfig
+++ b/drivers/gpu/drm/exynos/Kconfig
@@ -94,6 +94,9 @@ config DRM_EXYNOS_G2D
 	help
 	  Choose this option if you want to use Exynos G2D for DRM.
 
+config DRM_EXYNOS_IPP
+	bool
+
 config DRM_EXYNOS_FIMC
 	bool "FIMC"
 	depends on BROKEN && MFD_SYSCON
diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile
index 09bb002c9555..f663490e949d 100644
--- a/drivers/gpu/drm/exynos/Makefile
+++ b/drivers/gpu/drm/exynos/Makefile
@@ -17,6 +17,7 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER)	+= exynos_mixer.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI)	+= exynos_hdmi.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI)	+= exynos_drm_vidi.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_G2D)	+= exynos_drm_g2d.o
+exynosdrm-$(CONFIG_DRM_EXYNOS_IPP)	+= exynos_drm_ipp.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC)	+= exynos_drm_fimc.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR)	+= exynos_drm_rotator.o
 exynosdrm-$(CONFIG_DRM_EXYNOS_GSC)	+= exynos_drm_gsc.o
diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
index 7f19054ebce3..645046c5943a 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
@@ -26,6 +26,7 @@
 #include "exynos_drm_fb.h"
 #include "exynos_drm_gem.h"
 #include "exynos_drm_plane.h"
+#include "exynos_drm_ipp.h"
 #include "exynos_drm_vidi.h"
 #include "exynos_drm_g2d.h"
 #include "exynos_drm_iommu.h"
@@ -114,6 +115,14 @@ static void exynos_drm_lastclose(struct drm_device *dev)
 			DRM_AUTH | DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl,
 			DRM_AUTH | DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_RESOURCES, exynos_drm_ipp_get_res_ioctl,
+			DRM_AUTH | DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_CAPS, exynos_drm_ipp_get_caps_ioctl,
+			DRM_AUTH | DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_LIMITS, exynos_drm_ipp_get_limits_ioctl,
+			DRM_AUTH | DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_COMMIT, exynos_drm_ipp_commit_ioctl,
+			DRM_AUTH | DRM_RENDER_ALLOW),
 };
 
 static const struct file_operations exynos_drm_driver_fops = {
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 000000000000..11be6a1bf839
--- /dev/null
+++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.c
@@ -0,0 +1,892 @@
+/*
+ * Copyright (C) 2017 Samsung Electronics Co.Ltd
+ * Authors:
+ *	Marek Szyprowski <m.szyprowski@samsung.com>
+ *
+ * Exynos DRM Image Post Processing (IPP) related functions
+ *
+ * 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 shall be included in
+ * all copies or substantial portions of the Software.
+ */
+
+
+#include <drm/drmP.h>
+#include <drm/drm_mode.h>
+#include <uapi/drm/exynos_drm.h>
+
+#include "exynos_drm_drv.h"
+#include "exynos_drm_gem.h"
+#include "exynos_drm_ipp.h"
+
+static int num_ipp;
+static LIST_HEAD(ipp_list);
+
+struct drm_pending_exynos_ipp_event {
+	struct drm_pending_event base;
+	struct drm_exynos_ipp_event event;
+};
+
+/**
+ * exynos_drm_ipp_register - Register a new picture processor hardware module
+ * @dev: DRM device
+ * @ipp: ipp module to init
+ * @funcs: callbacks for the new ipp object
+ * @caps: bitmask of ipp capabilities (%DRM_EXYNOS_IPP_CAP_*)
+ * @formats: array of supported formats (%DRM_FORMAT_*)
+ * @name: name (for debugging purposes)
+ *
+ * Initializes a ipp module.
+ *
+ * Returns:
+ * Zero on success, error code on failure.
+ */
+int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp,
+		const struct exynos_drm_ipp_funcs *funcs, unsigned int caps,
+		const struct exynos_drm_ipp_formats *formats, const char *name)
+{
+	WARN_ON(!ipp);
+	WARN_ON(!funcs);
+	WARN_ON(!formats);
+
+	spin_lock_init(&ipp->lock);
+	INIT_LIST_HEAD(&ipp->todo_list);
+	init_waitqueue_head(&ipp->done_wq);
+	ipp->dev = dev;
+	ipp->funcs = funcs;
+	ipp->capabilities = caps;
+	ipp->name = name;
+	ipp->formats = formats;
+	while (formats++->fourcc)
+		ipp->num_formats++;
+
+	list_add_tail(&ipp->head, &ipp_list);
+	ipp->id = num_ipp++;
+
+	DRM_DEBUG_DRIVER("Registered ipp %d\n", ipp->id);
+
+	return 0;
+}
+
+/**
+ * exynos_drm_ipp_unregister - Unregister the picture processor module
+ * @dev: DRM device
+ * @ipp: ipp module
+ */
+void exynos_drm_ipp_unregister(struct drm_device *dev,
+			       struct exynos_drm_ipp *ipp)
+{
+	BUG_ON(ipp->task);
+	BUG_ON(!list_empty(&ipp->todo_list));
+	list_del(&ipp->head);
+}
+
+/**
+ * exynos_drm_ipp_ioctl_get_res_ioctl - enumerate all ipp modules
+ * @dev: DRM device
+ * @data: ioctl data
+ * @file_priv: DRM file info
+ *
+ * Construct a list of ipp ids.
+ *
+ * Called by the user via ioctl.
+ *
+ * Returns:
+ * Zero on success, negative errno on failure.
+ */
+int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data,
+				 struct drm_file *file_priv)
+{
+	struct drm_exynos_ioctl_ipp_get_res *resp = data;
+	struct exynos_drm_ipp *ipp;
+	uint32_t __user *ipp_ptr = (uint32_t __user *)
+						(unsigned long)resp->ipp_id_ptr;
+	unsigned int count = num_ipp, copied = 0;
+
+	/*
+	 * This ioctl is called twice, once to determine how much space is
+	 * needed, and the 2nd time to fill it.
+	 */
+	if (count && resp->count_ipps >= count) {
+		list_for_each_entry(ipp, &ipp_list, head) {
+			if (put_user(ipp->id, ipp_ptr + copied))
+				return -EFAULT;
+			copied++;
+		}
+	}
+	resp->count_ipps = count;
+
+	return 0;
+}
+
+static inline struct exynos_drm_ipp *__ipp_get(uint32_t id)
+{
+	struct exynos_drm_ipp *ipp;
+
+	list_for_each_entry(ipp, &ipp_list, head)
+		if (ipp->id == id)
+			return ipp;
+	return NULL;
+}
+
+/**
+ * exynos_drm_ipp_ioctl_get_caps - get ipp module capabilities and formats
+ * @dev: DRM device
+ * @data: ioctl data
+ * @file_priv: DRM file info
+ *
+ * Construct a structure describing ipp module capabilities.
+ *
+ * Called by the user via ioctl.
+ *
+ * Returns:
+ * Zero on success, negative errno on failure.
+ */
+int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data,
+				  struct drm_file *file_priv)
+{
+	struct drm_exynos_ioctl_ipp_get_caps *resp = data;
+	void __user *ptr = (void __user *)(unsigned long)resp->formats_ptr;
+	struct exynos_drm_ipp *ipp;
+	int i;
+
+	ipp = __ipp_get(resp->ipp_id);
+	if (!ipp)
+		return -ENOENT;
+
+	resp->ipp_id = ipp->id;
+	resp->capabilities = ipp->capabilities;
+
+	/*
+	 * This ioctl is called twice, once to determine how much space is
+	 * needed, and the 2nd time to fill it.
+	 */
+	if (resp->formats_count >= ipp->num_formats) {
+		for (i = 0; i < ipp->num_formats; i++) {
+			struct exynos_drm_ipp_format tmp = {
+				.fourcc = ipp->formats[i].fourcc,
+				.type = ipp->formats[i].type,
+				.modifier = ipp->formats[i].modifier,
+			};
+
+			if (copy_to_user(ptr, &tmp, sizeof(tmp)))
+				return -EFAULT;
+			ptr += sizeof(tmp);
+		}
+	}
+	resp->formats_count = ipp->num_formats;
+
+	return 0;
+}
+
+static inline const struct exynos_drm_ipp_formats *__ipp_format_get(
+			uint32_t fourcc, uint64_t mod, unsigned long type,
+			const struct exynos_drm_ipp_formats *f)
+{
+	for (; f->fourcc; f++)
+		if ((f->type & type) && f->fourcc == fourcc &&
+		    f->modifier == mod)
+			return f;
+	return NULL;
+}
+
+/**
+ * exynos_drm_ipp_get_limits_ioctl - get ipp module limits
+ * @dev: DRM device
+ * @data: ioctl data
+ * @file_priv: DRM file info
+ *
+ * Construct a structure describing ipp module limitations for provided
+ * picture format.
+ *
+ * Called by the user via ioctl.
+ *
+ * Returns:
+ * Zero on success, negative errno on failure.
+ */
+int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data,
+				    struct drm_file *file_priv)
+{
+	struct drm_exynos_ioctl_ipp_get_limits *resp = data;
+	void __user *ptr = (void __user *)(unsigned long)resp->limits_ptr;
+	const struct exynos_drm_ipp_formats *format;
+	const struct drm_exynos_ipp_limit *limits, *l;
+	struct exynos_drm_ipp *ipp;
+	int count = 0;
+
+	if (resp->type != DRM_EXYNOS_IPP_FORMAT_SOURCE &&
+	    resp->type != DRM_EXYNOS_IPP_FORMAT_DESTINATION)
+		return -EINVAL;
+
+	ipp = __ipp_get(resp->ipp_id);
+	if (!ipp)
+		return -ENOENT;
+
+	format = __ipp_format_get(resp->fourcc, resp->modifier, resp->type,
+			      ipp->formats);
+	if (!format)
+		return -EINVAL;
+
+	limits = format->limits;
+	for (l = limits; l->type; l++)
+		count++;
+
+	/*
+	 * This ioctl is called twice, once to determine how much space is
+	 * needed, and the 2nd time to fill it.
+	 */
+	if (count && resp->limits_size >= count * sizeof(*l)) {
+		for (l = limits; l->type; l++, ptr += sizeof(*l))
+			if (copy_to_user((void __user *)ptr, l, sizeof(*l)))
+				return -EFAULT;
+	}
+	resp->limits_size = count * sizeof(*l);
+
+	return 0;
+}
+
+static inline struct exynos_drm_ipp_task *
+	exynos_drm_ipp_task_alloc(struct exynos_drm_ipp *ipp)
+{
+	struct exynos_drm_ipp_task *task;
+
+	task = kzalloc(sizeof(*task), GFP_KERNEL);
+	if (!task)
+		return NULL;
+
+	task->dev = ipp->dev;
+	task->ipp = ipp;
+
+	/* some defaults */
+	task->src.rect.w = task->dst.rect.w = UINT_MAX;
+	task->src.rect.h = task->dst.rect.h = UINT_MAX;
+	task->transform.rotation = DRM_MODE_ROTATE_0;
+
+	DRM_DEBUG_DRIVER("Allocated task %pK\n", task);
+
+	return task;
+}
+
+struct exynos_drm_param_map {
+	unsigned int id;
+	unsigned int size;
+	unsigned int offset;
+} exynos_drm_ipp_params_maps[] = {
+	{
+		DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
+		sizeof(struct drm_exynos_ipp_task_buffer),
+		offsetof(struct exynos_drm_ipp_task, src.buf),
+	}, {
+		DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
+		sizeof(struct drm_exynos_ipp_task_buffer),
+		offsetof(struct exynos_drm_ipp_task, dst.buf),
+	}, {
+		DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
+		sizeof(struct drm_exynos_ipp_task_rect),
+		offsetof(struct exynos_drm_ipp_task, src.rect),
+	}, {
+		DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
+		sizeof(struct drm_exynos_ipp_task_rect),
+		offsetof(struct exynos_drm_ipp_task, dst.rect),
+	}, {
+		DRM_EXYNOS_IPP_TASK_TRANSFORM,
+		sizeof(struct drm_exynos_ipp_task_transform),
+		offsetof(struct exynos_drm_ipp_task, transform),
+	}, {
+		DRM_EXYNOS_IPP_TASK_ALPHA,
+		sizeof(struct drm_exynos_ipp_task_alpha),
+		offsetof(struct exynos_drm_ipp_task, alpha),
+	},
+};
+
+static int exynos_drm_ipp_task_set(struct exynos_drm_ipp_task *task,
+				  struct drm_exynos_ioctl_ipp_commit *arg)
+{
+	unsigned int size = arg->params_size;
+	void *p, *params;
+	int ret = 0;
+
+	if (size > PAGE_SIZE)
+		return -ERANGE;
+
+	p = kmalloc(size, GFP_KERNEL);
+	if (!p)
+		return -ENOMEM;
+
+	if (copy_from_user(p, (void __user *)(unsigned long)arg->params_ptr,
+			   size)) {
+		ret = -EFAULT;
+		DRM_DEBUG_DRIVER("Failed to copy configuration from userspace\n");
+		goto free;
+	}
+
+	params = p;
+	while (size) {
+		struct exynos_drm_param_map *map = exynos_drm_ipp_params_maps;
+		u32 *id = params;
+		int i;
+
+		for (i = 0; i < ARRAY_SIZE(exynos_drm_ipp_params_maps); i++)
+			if (map[i].id == *id)
+				break;
+
+		if (i < ARRAY_SIZE(exynos_drm_ipp_params_maps) &&
+		    map[i].size <= size) {
+			memcpy((void *)task + map[i].offset, params,
+			       map[i].size);
+			params += map[i].size;
+			size -= map[i].size;
+		} else {
+			ret = -EINVAL;
+			goto free;
+		}
+	}
+
+	DRM_DEBUG_DRIVER("Got task %pK configuration from userspace\n", task);
+free:
+	kfree(p);
+	return ret;
+}
+
+static int exynos_drm_ipp_task_setup_buffer(struct exynos_drm_ipp_buffer *buf,
+					    struct drm_file *filp)
+{
+	int ret = 0;
+	int i;
+
+	/* basic checks */
+	if (buf->buf.width == 0 || buf->buf.height == 0)
+		return -EINVAL;
+	buf->format = drm_format_info(buf->buf.fourcc);
+	for (i = 0; i < buf->format->num_planes; i++) {
+		unsigned int width = (i == 0) ? buf->buf.width :
+			     DIV_ROUND_UP(buf->buf.width, buf->format->hsub);
+
+		if (buf->buf.pitch[i] == 0)
+			buf->buf.pitch[i] = width * buf->format->cpp[i];
+		if (buf->buf.pitch[i] < width * buf->format->cpp[i])
+			return -EINVAL;
+		if (!buf->buf.gem_id[i])
+			return -ENOENT;
+	}
+
+	/* get GEM buffers and check their size */
+	for (i = 0; i < buf->format->num_planes; i++) {
+		unsigned int height = (i == 0) ? buf->buf.height :
+			     DIV_ROUND_UP(buf->buf.height, buf->format->vsub);
+		unsigned long size = height * buf->buf.pitch[i] +
+				     buf->buf.offset[i];
+		struct drm_gem_object *obj = drm_gem_object_lookup(filp,
+							    buf->buf.gem_id[i]);
+		if (!obj) {
+			ret = -ENOENT;
+			goto gem_free;
+		}
+		buf->exynos_gem[i] = to_exynos_gem(obj);
+
+		if (size > buf->exynos_gem[i]->size) {
+			i++;
+			ret = -EINVAL;
+			goto gem_free;
+		}
+		buf->dma_addr[i] = buf->exynos_gem[i]->dma_addr +
+				   buf->buf.offset[i];
+	}
+
+	return 0;
+gem_free:
+	while (i--) {
+		drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base);
+		buf->exynos_gem[i] = NULL;
+	}
+	return ret;
+}
+
+static void exynos_drm_ipp_task_release_buffer(struct exynos_drm_ipp_buffer *buf)
+{
+	int i;
+
+	if (!buf->exynos_gem[0])
+		return;
+	for (i = 0; i < buf->format->num_planes; i++)
+		drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base);
+}
+
+static void exynos_drm_ipp_task_free(struct exynos_drm_ipp *ipp,
+				 struct exynos_drm_ipp_task *task)
+{
+	DRM_DEBUG_DRIVER("Freeing task %pK\n", task);
+
+	exynos_drm_ipp_task_release_buffer(&task->src);
+	exynos_drm_ipp_task_release_buffer(&task->dst);
+	if (task->event)
+		drm_event_cancel_free(ipp->dev, &task->event->base);
+	kfree(task);
+}
+
+struct drm_ipp_limit {
+	struct drm_exynos_ipp_limit_val h;
+	struct drm_exynos_ipp_limit_val v;
+};
+
+enum drm_ipp_size_id {
+	IPP_LIMIT_BUFFER, IPP_LIMIT_AREA, IPP_LIMIT_ROTATED, IPP_LIMIT_MAX
+};
+
+static const enum drm_ipp_size_id limit_id_fallback[IPP_LIMIT_MAX][4] = {
+	[IPP_LIMIT_BUFFER]  = { DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER },
+	[IPP_LIMIT_AREA]    = { DRM_EXYNOS_IPP_LIMIT_SIZE_AREA,
+				DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER },
+	[IPP_LIMIT_ROTATED] = { DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED,
+				DRM_EXYNOS_IPP_LIMIT_SIZE_AREA,
+				DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER },
+};
+
+static inline void __limit_set_val(unsigned int *ptr, unsigned int val)
+{
+	if (!*ptr)
+		*ptr = val;
+}
+
+static void __get_size_limit(const struct drm_exynos_ipp_limit *limits,
+			     enum drm_ipp_size_id id, struct drm_ipp_limit *res)
+{
+	const struct drm_exynos_ipp_limit *l = limits;
+	int i = 0;
+
+	memset(res, 0, sizeof(*res));
+	for (i = 0; limit_id_fallback[id][i]; i++)
+		for (l = limits; l->type; l++) {
+			if (((l->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) !=
+			      DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE) ||
+			    ((l->type & DRM_EXYNOS_IPP_LIMIT_SIZE_MASK) !=
+						     limit_id_fallback[id][i]))
+				continue;
+			__limit_set_val(&res->h.min, l->h.min);
+			__limit_set_val(&res->h.max, l->h.max);
+			__limit_set_val(&res->h.align, l->h.align);
+			__limit_set_val(&res->v.min, l->v.min);
+			__limit_set_val(&res->v.max, l->v.max);
+			__limit_set_val(&res->v.align, l->v.align);
+		}
+}
+
+static inline bool __align_check(unsigned int val, unsigned int align)
+{
+	if (align && (val & (align - 1)))
+		return false;
+	return true;
+}
+
+static inline bool __size_limit_check(unsigned int val,
+				 struct drm_exynos_ipp_limit_val *l)
+{
+	if ((l->min && val < l->min) || (l->max && val > l->max) ||
+	    !__align_check(val, l->align))
+		return false;
+	return true;
+}
+
+static int exynos_drm_ipp_check_size_limits(struct exynos_drm_ipp_buffer *buf,
+	const struct drm_exynos_ipp_limit *limits, bool rotate, bool swap)
+{
+	enum drm_ipp_size_id id = rotate ? IPP_LIMIT_ROTATED : IPP_LIMIT_AREA;
+	struct drm_ipp_limit l;
+	struct drm_exynos_ipp_limit_val *lh = &l.h, *lv = &l.v;
+
+	if (!limits)
+		return 0;
+
+	__get_size_limit(limits, IPP_LIMIT_BUFFER, &l);
+	if (!__size_limit_check(buf->buf.width, &l.h) ||
+	    !__size_limit_check(buf->buf.height, &l.v))
+		return -EINVAL;
+
+	if (swap) {
+		lv = &l.h;
+		lh = &l.v;
+	}
+	__get_size_limit(limits, id, &l);
+	if (!__size_limit_check(buf->rect.w, lh) ||
+	    !__align_check(buf->rect.x, lh->align) ||
+	    !__size_limit_check(buf->rect.h, lv) ||
+	    !__align_check(buf->rect.y, lv->align))
+		return -EINVAL;
+
+	return 0;
+}
+
+static int exynos_drm_ipp_check_scale_limits(struct drm_exynos_ipp_task_rect *src,
+	struct drm_exynos_ipp_task_rect *dst,
+	const struct drm_exynos_ipp_limit *limits, bool swap)
+{
+	const struct drm_exynos_ipp_limit_val *lh, *lv;
+
+	if (!limits)
+		return 0;
+
+	for (; limits->type; limits++)
+		if ((limits->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) ==
+		    DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE)
+			break;
+	if (!limits->type)
+		return 0;
+
+	lh = (!swap) ? &limits->h : &limits->v;
+	lv = (!swap) ? &limits->v : &limits->h;
+
+	if ((lh->max && dst->w > src->w * lh->max) ||
+	    (lh->min && (dst->w << 16) < src->w * lh->min) ||
+	    (lv->max && dst->h > src->h * lv->max) ||
+	    (lv->min && (dst->h << 16) < src->h * lv->min))
+		return -EINVAL;
+
+	return 0;
+}
+
+static int exynos_drm_ipp_task_check(struct exynos_drm_ipp_task *task,
+				    struct drm_file *filp)
+{
+	struct exynos_drm_ipp *ipp = task->ipp;
+	const struct exynos_drm_ipp_formats *src_format, *dst_format;
+	struct exynos_drm_ipp_buffer *src = &task->src, *dst = &task->dst;
+	unsigned int rotation = task->transform.rotation;
+	int ret = 0;
+	bool scale = false, swap = false;
+
+	DRM_DEBUG_DRIVER("Checking %pK\n", task);
+
+	if (src->rect.w == UINT_MAX)
+		src->rect.w = src->buf.width;
+	if (src->rect.h == UINT_MAX)
+		src->rect.h = src->buf.height;
+	if (dst->rect.w == UINT_MAX)
+		dst->rect.w = dst->buf.width;
+	if (dst->rect.h == UINT_MAX)
+		dst->rect.h = dst->buf.height;
+
+	if (src->rect.x + src->rect.w > (src->buf.width) ||
+	    src->rect.y + src->rect.h > (src->buf.height) ||
+	    dst->rect.x + dst->rect.w > (dst->buf.width) ||
+	    dst->rect.y + dst->rect.h > (dst->buf.height))
+		return -EINVAL;
+
+	DRM_DEBUG_DRIVER("Task %pK: basic checks done\n", task);
+
+	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CROP) &&
+	    (src->rect.x || src->rect.y || dst->rect.x || dst->rect.y))
+		return -EINVAL;
+
+	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_ROTATE) &&
+	    rotation != DRM_MODE_ROTATE_0)
+		return -EINVAL;
+
+	if (drm_rotation_90_or_270(rotation))
+		swap = true;
+
+	if ((!swap && (src->rect.w != dst->rect.w || src->rect.h != dst->rect.h)) ||
+	    (swap && (src->rect.w != dst->rect.h || src->rect.h != dst->rect.w)))
+		scale = true;
+
+	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_SCALE) && scale)
+		return -EINVAL;
+
+	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CONVERT) &&
+	    src->buf.fourcc != dst->buf.fourcc)
+		return -EINVAL;
+
+	DRM_DEBUG_DRIVER("Task %pK: capability checks done\n", task);
+
+	src_format = __ipp_format_get(src->buf.fourcc, src->buf.modifier,
+				  DRM_EXYNOS_IPP_FORMAT_SOURCE, ipp->formats);
+	if (!src_format)
+		return -EINVAL;
+	ret = exynos_drm_ipp_check_size_limits(src, src_format->limits,
+					  rotation != DRM_MODE_ROTATE_0, false);
+	if (ret)
+		return ret;
+	ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect,
+						src_format->limits, swap);
+	if (ret)
+		return ret;
+
+	DRM_DEBUG_DRIVER("Task %pK: source image checks done\n", task);
+
+	dst_format = __ipp_format_get(dst->buf.fourcc, dst->buf.modifier,
+			       DRM_EXYNOS_IPP_FORMAT_DESTINATION, ipp->formats);
+	if (!dst_format)
+		return -EINVAL;
+	ret = exynos_drm_ipp_check_size_limits(dst, dst_format->limits,
+					   rotation != DRM_MODE_ROTATE_0, swap);
+	if (ret)
+		return ret;
+	ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect,
+						dst_format->limits, swap);
+	if (ret)
+		return ret;
+
+	DRM_DEBUG_DRIVER("Task %pK: destination image checks done\n", task);
+
+	ret = exynos_drm_ipp_task_setup_buffer(src, filp);
+	if (ret)
+		return ret;
+
+	DRM_DEBUG_DRIVER("Task %pK: source buffer checks done\n", task);
+
+	ret = exynos_drm_ipp_task_setup_buffer(dst, filp);
+	if (ret)
+		return ret;
+
+	DRM_DEBUG_DRIVER("Task %pK: destination buffer checks done\n", task);
+
+	if (ipp->funcs->check)
+		ret = ipp->funcs->check(ipp, task);
+
+	DRM_DEBUG_DRIVER("Task %pK: all checks done.\n", task);
+
+	return ret;
+}
+
+static int exynos_drm_ipp_event_create(struct exynos_drm_ipp_task *task,
+			struct drm_file *file_priv, uint64_t user_data)
+{
+	struct drm_pending_exynos_ipp_event *e = NULL;
+	int ret;
+
+	e = kzalloc(sizeof(*e), GFP_KERNEL);
+	if (!e)
+		return -ENOMEM;
+
+	e->event.base.type = DRM_EXYNOS_IPP_EVENT;
+	e->event.base.length = sizeof(e->event);
+	e->event.user_data = user_data;
+
+	ret = drm_event_reserve_init(task->dev, file_priv, &e->base,
+				     &e->event.base);
+	if (ret)
+		goto free;
+
+	task->event = e;
+	return 0;
+free:
+	kfree(e);
+	return ret;
+}
+
+static void exynos_drm_ipp_event_send(struct exynos_drm_ipp_task *task)
+{
+	struct timeval now = ktime_to_timeval(ktime_get());
+
+	task->event->event.tv_sec = now.tv_sec;
+	task->event->event.tv_usec = now.tv_usec;
+	task->event->event.sequence = atomic_inc_return(&task->ipp->sequence);
+
+	drm_send_event(task->dev, &task->event->base);
+}
+
+static int exynos_drm_ipp_task_cleanup(struct exynos_drm_ipp_task *task)
+{
+	int ret = task->ret;
+
+	if (ret == 0 && task->event) {
+		exynos_drm_ipp_event_send(task);
+		/* ensure event won't be canceled on task free */
+		task->event = NULL;
+	}
+
+	exynos_drm_ipp_task_free(task->ipp, task);
+	return ret;
+}
+
+static void exynos_drm_ipp_cleanup_work(struct work_struct *work)
+{
+	struct exynos_drm_ipp_task *task = container_of(work,
+				struct exynos_drm_ipp_task, cleanup_work);
+
+	exynos_drm_ipp_task_cleanup(task);
+}
+
+static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp);
+
+/**
+ * exynos_drm_ipp_task_done - finish given task and set return code
+ * @task: ipp task to finish
+ * @ret: error code or 0 if operation has been performed successfully
+ */
+void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret)
+{
+	struct exynos_drm_ipp *ipp = task->ipp;
+	unsigned long flags;
+
+	DRM_DEBUG_DRIVER("ipp: %d, task %pK done\n", ipp->id, task);
+
+	spin_lock_irqsave(&ipp->lock, flags);
+	if (ipp->task == task)
+		ipp->task = NULL;
+	task->flags |= DRM_EXYNOS_IPP_TASK_DONE;
+	task->ret = ret;
+	spin_unlock_irqrestore(&ipp->lock, flags);
+
+	exynos_drm_ipp_next_task(ipp);
+	wake_up(&ipp->done_wq);
+
+	if (task->flags & DRM_EXYNOS_IPP_TASK_ASYNC) {
+		INIT_WORK(&task->cleanup_work, exynos_drm_ipp_cleanup_work);
+		schedule_work(&task->cleanup_work);
+	}
+}
+
+static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp)
+{
+	struct exynos_drm_ipp_task *task;
+	unsigned long flags;
+	int ret;
+
+	DRM_DEBUG_DRIVER("ipp: %d, try to run new task\n", ipp->id);
+
+	spin_lock_irqsave(&ipp->lock, flags);
+
+	if (ipp->task || list_empty(&ipp->todo_list)) {
+		spin_unlock_irqrestore(&ipp->lock, flags);
+		return;
+	}
+
+	task = list_first_entry(&ipp->todo_list, struct exynos_drm_ipp_task,
+				head);
+	list_del_init(&task->head);
+	ipp->task = task;
+
+	spin_unlock_irqrestore(&ipp->lock, flags);
+
+	DRM_DEBUG_DRIVER("ipp: %d, selected task %pK to run\n", ipp->id, task);
+
+	ret = ipp->funcs->commit(ipp, task);
+	if (ret)
+		exynos_drm_ipp_task_done(task, ret);
+}
+
+static void exynos_drm_ipp_schedule_task(struct exynos_drm_ipp *ipp,
+				     struct exynos_drm_ipp_task *task)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&ipp->lock, flags);
+	list_add(&task->head, &ipp->todo_list);
+	spin_unlock_irqrestore(&ipp->lock, flags);
+
+	exynos_drm_ipp_next_task(ipp);
+}
+
+static void exynos_drm_ipp_task_abort(struct exynos_drm_ipp *ipp,
+				  struct exynos_drm_ipp_task *task)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&ipp->lock, flags);
+	if (task->flags & DRM_EXYNOS_IPP_TASK_DONE) {
+		/* already completed task */
+		exynos_drm_ipp_task_cleanup(task);
+	} else if (ipp->task != task) {
+		/* task has not been scheduled for execution yet */
+		list_del_init(&task->head);
+		exynos_drm_ipp_task_cleanup(task);
+	} else {
+		/*
+		 * currently processed task, call abort() and perform
+		 * cleanup with async worker
+		 */
+		task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC;
+		spin_unlock_irqrestore(&ipp->lock, flags);
+		if (ipp->funcs->abort)
+			ipp->funcs->abort(ipp, task);
+		return;
+	}
+	spin_unlock_irqrestore(&ipp->lock, flags);
+}
+
+/**
+ * exynos_drm_ipp_commit_ioctl - perform image processing operation
+ * @dev: DRM device
+ * @data: ioctl data
+ * @file_priv: DRM file info
+ *
+ * Construct a ipp task from the set of properties provided from the user
+ * and try to schedule it to framebuffer processor hardware.
+ *
+ * Called by the user via ioctl.
+ *
+ * Returns:
+ * Zero on success, negative errno on failure.
+ */
+int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, void *data,
+			  struct drm_file *file_priv)
+{
+	struct drm_exynos_ioctl_ipp_commit *arg = data;
+	struct exynos_drm_ipp *ipp;
+	struct exynos_drm_ipp_task *task;
+	int ret = 0;
+
+	if ((arg->flags & ~DRM_EXYNOS_IPP_FLAGS) || arg->reserved)
+		return -EINVAL;
+
+	/* can't test and expect an event at the same time */
+	if ((arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY) &&
+			(arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT))
+		return -EINVAL;
+
+	ipp = __ipp_get(arg->ipp_id);
+	if (!ipp)
+		return -ENOENT;
+
+	task = exynos_drm_ipp_task_alloc(ipp);
+	if (!task)
+		return -ENOMEM;
+
+	ret = exynos_drm_ipp_task_set(task, arg);
+	if (ret)
+		goto free;
+
+	ret = exynos_drm_ipp_task_check(task, file_priv);
+	if (ret || arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY)
+		goto free;
+
+	if (arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT) {
+		ret = exynos_drm_ipp_event_create(task, file_priv,
+						 arg->user_data);
+		if (ret)
+			goto free;
+	}
+
+	/*
+	 * Queue task for processing on the hardware. task object will be
+	 * then freed after exynos_drm_ipp_task_done()
+	 */
+	if (arg->flags & DRM_EXYNOS_IPP_FLAG_NONBLOCK) {
+		DRM_DEBUG_DRIVER("ipp: %d, nonblocking processing task %pK\n",
+				 ipp->id, task);
+
+		task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC;
+		exynos_drm_ipp_schedule_task(task->ipp, task);
+		ret = 0;
+	} else {
+		DRM_DEBUG_DRIVER("ipp: %d, processing task %pK\n", ipp->id, task);
+		exynos_drm_ipp_schedule_task(ipp, task);
+		ret = wait_event_interruptible(ipp->done_wq,
+					task->flags & DRM_EXYNOS_IPP_TASK_DONE);
+		if (ret)
+			exynos_drm_ipp_task_abort(ipp, task);
+		else
+			ret = exynos_drm_ipp_task_cleanup(task);
+	}
+	return ret;
+free:
+	exynos_drm_ipp_task_free(ipp, task);
+
+	return ret;
+}
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 000000000000..532a60698a6a
--- /dev/null
+++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.h
@@ -0,0 +1,188 @@
+/*
+ * Copyright (c) 2017 Samsung Electronics Co., Ltd.
+ *
+ * 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.
+ */
+
+#ifndef _EXYNOS_DRM_IPP_H_
+#define _EXYNOS_DRM_IPP_H_
+
+#include <drm/drmP.h>
+
+struct exynos_drm_ipp;
+struct exynos_drm_ipp_task;
+
+/**
+ * struct exynos_drm_ipp_funcs - exynos_drm_ipp control functions
+ */
+struct exynos_drm_ipp_funcs {
+	/**
+	 * @check:
+	 *
+	 * This is the optional hook to validate an ipp task. This function
+	 * must reject any task which the hardware or driver doesn't support.
+	 * This includes but is of course not limited to:
+	 *
+	 *  - Checking that the buffers, scaling and placement requirements
+	 *    and so on are within the limits of the hardware not specified by
+	 *    limits array.
+	 *
+	 *  - The driver does not need to repeat basic input validation like
+	 *    done in the exynos_drm_ipp_task_check() function. The core does
+	 *    that before calling this hook.
+	 *
+	 * RETURNS:
+	 *
+	 * 0 on success or one of the below negative error codes:
+	 *
+	 *  - -EINVAL, if any of the above constraints are violated.
+	 */
+	int (*check)(struct exynos_drm_ipp *ipp,
+		     struct exynos_drm_ipp_task *task);
+
+	/**
+	 * @commit:
+	 *
+	 * This is the main entry point to start framebuffer processing
+	 * in the hardware. The exynos_drm_ipp_task has been already validated.
+	 * This function must not wait until the device finishes processing.
+	 * When the driver finishes processing, it has to call
+	 * exynos_exynos_drm_ipp_task_done() function.
+	 *
+	 * RETURNS:
+	 *
+	 * 0 on success or negative error codes in case of failure.
+	 */
+	int (*commit)(struct exynos_drm_ipp *ipp,
+		      struct exynos_drm_ipp_task *task);
+
+	/**
+	 * @abort:
+	 *
+	 * Informs the driver that it has to abort the currently running
+	 * task as soon as possible (i.e. as soon as it can stop the device
+	 * safely), even if the task would not have been finished by then.
+	 * After the driver performs the necessary steps, it has to call
+	 * exynos_drm_ipp_task_done() (as if the task ended normally).
+	 * This function does not have to (and will usually not) wait
+	 * until the device enters a state when it can be stopped.
+	 */
+	void (*abort)(struct exynos_drm_ipp *ipp,
+		      struct exynos_drm_ipp_task *task);
+};
+
+/**
+ * struct exynos_drm_ipp - central picture processor module structure
+ */
+struct exynos_drm_ipp {
+	struct drm_device *dev;
+	struct list_head head;
+	unsigned int id;
+
+	const char *name;
+	const struct exynos_drm_ipp_funcs *funcs;
+	unsigned int capabilities;
+	const struct exynos_drm_ipp_formats *formats;
+	unsigned int num_formats;
+	atomic_t sequence;
+
+	spinlock_t lock;
+	struct exynos_drm_ipp_task *task;
+	struct list_head todo_list;
+	wait_queue_head_t done_wq;
+};
+
+struct exynos_drm_ipp_buffer {
+	struct drm_exynos_ipp_task_buffer buf;
+	struct drm_exynos_ipp_task_rect rect;
+
+	struct exynos_drm_gem *exynos_gem[MAX_FB_BUFFER];
+	const struct drm_format_info *format;
+	dma_addr_t dma_addr[MAX_FB_BUFFER];
+};
+
+/**
+ * struct exynos_drm_ipp_task - a structure describing transformation that
+ * has to be performed by the picture processor hardware module
+ */
+struct exynos_drm_ipp_task {
+	struct drm_device *dev;
+	struct exynos_drm_ipp *ipp;
+	struct list_head head;
+
+	struct exynos_drm_ipp_buffer src;
+	struct exynos_drm_ipp_buffer dst;
+
+	struct drm_exynos_ipp_task_transform transform;
+	struct drm_exynos_ipp_task_alpha alpha;
+
+	struct work_struct cleanup_work;
+	unsigned int flags;
+	int ret;
+
+	struct drm_pending_exynos_ipp_event *event;
+};
+
+#define DRM_EXYNOS_IPP_TASK_DONE	(1 << 0)
+#define DRM_EXYNOS_IPP_TASK_ASYNC	(1 << 1)
+
+struct exynos_drm_ipp_formats {
+	u32 fourcc;
+	u32 type;
+	u64 modifier;
+	const struct drm_exynos_ipp_limit *limits;
+};
+
+/* helper macros to set exynos_drm_ipp_formats structure and limits*/
+#define IPP_SRCDST_MFORMAT(f, m, l) \
+     .fourcc = DRM_FORMAT_##f, .modifier = m, .limits = l, \
+     .type = (DRM_EXYNOS_IPP_FORMAT_SOURCE | DRM_EXYNOS_IPP_FORMAT_DESTINATION)
+
+#define IPP_SRCDST_FORMAT(f, l) IPP_SRCDST_MFORMAT(f, 0, l)
+
+#define IPP_SIZE_LIMIT(l, val...)	\
+     .type = (DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE | DRM_EXYNOS_IPP_LIMIT_SIZE_##l), \
+     val
+
+int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp,
+		const struct exynos_drm_ipp_funcs *funcs, unsigned int caps,
+		const struct exynos_drm_ipp_formats *formats, const char *name);
+void exynos_drm_ipp_unregister(struct drm_device *dev, struct exynos_drm_ipp *ipp);
+
+void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret);
+
+#ifdef CONFIG_DRM_EXYNOS_IPP
+int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data,
+				 struct drm_file *file_priv);
+int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data,
+				  struct drm_file *file_priv);
+int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data,
+				    struct drm_file *file_priv);
+int exynos_drm_ipp_commit_ioctl(struct drm_device *dev,
+				void *data, struct drm_file *file_priv);
+#else
+static inline int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev,
+	 void *data, struct drm_file *file_priv)
+{
+	return -ENODEV;
+}
+static inline int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev,
+	 void *data, struct drm_file *file_priv)
+{
+	return -ENODEV;
+}
+static inline int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev,
+	 void *data, struct drm_file *file_priv)
+{
+	return -ENODEV;
+}
+static inline int exynos_drm_ipp_commit_ioctl(struct drm_device *dev,
+	 void *data, struct drm_file *file_priv)
+{
+	return -ENODEV;
+}
+#endif
+#endif
diff --git a/include/uapi/drm/exynos_drm.h b/include/uapi/drm/exynos_drm.h
index 544db2cecb36..239e09022d75 100644
--- a/include/uapi/drm/exynos_drm.h
+++ b/include/uapi/drm/exynos_drm.h
@@ -134,6 +134,208 @@ struct drm_exynos_g2d_exec {
 	__u64					async;
 };
 
+/* Exynos DRM IPP v2 API */
+
+/**
+ * Enumerate available IPP hardware modules.
+ *
+ * @count_ipps: size of ipp_id array / number of ipp modules (set by driver)
+ * @reserved: padding
+ * @ipp_id_ptr: pointer to ipp_id array or NULL
+ */
+struct drm_exynos_ioctl_ipp_get_res {
+	__u32 count_ipps;
+	__u32 reserved;
+	__u64 ipp_id_ptr;
+};
+
+enum drm_exynos_ipp_format_type {
+	DRM_EXYNOS_IPP_FORMAT_SOURCE		= 0x01,
+	DRM_EXYNOS_IPP_FORMAT_DESTINATION	= 0x02,
+};
+
+struct exynos_drm_ipp_format {
+	__u32 fourcc;
+	__u32 type;
+	__u64 modifier;
+};
+
+enum drm_exynos_ipp_capability {
+	DRM_EXYNOS_IPP_CAP_CROP		= 0x01,
+	DRM_EXYNOS_IPP_CAP_ROTATE	= 0x02,
+	DRM_EXYNOS_IPP_CAP_SCALE	= 0x04,
+	DRM_EXYNOS_IPP_CAP_CONVERT	= 0x08,
+};
+
+/**
+ * Get IPP hardware capabilities and supported image formats.
+ *
+ * @ipp_id: id of IPP module to query
+ * @capabilities: bitmask of drm_exynos_ipp_capability (set by driver)
+ * @reserved: padding
+ * @formats_count: size of formats array (in entries) / number of filled
+ *		   formats (set by driver)
+ * @formats_ptr: pointer to formats array or NULL
+ */
+struct drm_exynos_ioctl_ipp_get_caps {
+	__u32 ipp_id;
+	__u32 capabilities;
+	__u32 reserved;
+	__u32 formats_count;
+	__u64 formats_ptr;
+};
+
+enum drm_exynos_ipp_limit_type {
+	/* size (horizontal/vertial) limits, in pixels (min, max, alignment) */
+	DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE		= 0x0001,
+	/* scale ratio (horizonta/vertial), 16.16 fixed point (min, max) */
+	DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE		= 0x0002,
+
+	/* image buffer area */
+	DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER	= 0x0001 << 16,
+	/* src/dst rectangle area */
+	DRM_EXYNOS_IPP_LIMIT_SIZE_AREA		= 0x0002 << 16,
+	/* src/dst rectangle area when rotation enabled */
+	DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED	= 0x0003 << 16,
+
+	DRM_EXYNOS_IPP_LIMIT_TYPE_MASK		= 0x000f,
+	DRM_EXYNOS_IPP_LIMIT_SIZE_MASK		= 0x000f << 16,
+};
+
+struct drm_exynos_ipp_limit_val {
+	__u32 min, max, align;
+};
+
+/**
+ * IPP module limitation.
+ *
+ * @type: limit type (see drm_exynos_ipp_limit_type enum)
+ * @h, @v: horizontal and vertical limits
+ */
+struct drm_exynos_ipp_limit {
+	__u32 type;
+	struct drm_exynos_ipp_limit_val h;
+	struct drm_exynos_ipp_limit_val v;
+};
+
+/**
+ * Get IPP limits for given image format.
+ *
+ * @ipp_id: id of IPP module to query
+ * @fourcc: image format code (see DRM_FORMAT_* in drm_fourcc.h)
+ * @modifier: image format modifier (see DRM_FORMAT_MOD_* in drm_fourcc.h)
+ * @type: source/destination identifier (drm_exynos_ipp_format_flag enum)
+ * @limits_size: size of limits array (in bytes) / size of filled entries
+ *		 (set by driver)
+ * @limits_ptr: pointer to limits array or NULL
+ */
+struct drm_exynos_ioctl_ipp_get_limits {
+	__u32 ipp_id;
+	__u32 fourcc;
+	__u64 modifier;
+	__u32 type;
+	__u32 limits_size;
+	__u64 limits_ptr;
+};
+
+enum drm_exynos_ipp_task_id {
+	/* buffer described by struct drm_exynos_ipp_task_buffer */
+	DRM_EXYNOS_IPP_TASK_BUFFER		= 0x0001,
+	/* rectangle described by struct drm_exynos_ipp_task_rect */
+	DRM_EXYNOS_IPP_TASK_RECTANGLE		= 0x0002,
+	/* transformation described by struct drm_exynos_ipp_task_transform */
+	DRM_EXYNOS_IPP_TASK_TRANSFORM		= 0x0003,
+	/* alpha configuration described by struct drm_exynos_ipp_task_alpha */
+	DRM_EXYNOS_IPP_TASK_ALPHA		= 0x0004,
+
+	/* source image data (for buffer and rectangle chunks) */
+	DRM_EXYNOS_IPP_TASK_TYPE_SOURCE		= 0x0001 << 16,
+	/* destination image data (for buffer and rectangle chunks) */
+	DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION	= 0x0002 << 16,
+};
+
+/**
+ * Memory buffer with image data.
+ *
+ * @id: must be DRM_EXYNOS_IPP_TASK_BUFFER
+ * other parameters are same as for AddFB2 generic DRM ioctl
+ */
+struct drm_exynos_ipp_task_buffer {
+	__u32	id;
+	__u32	fourcc;
+	__u32	width, height;
+	__u32	gem_id[4];
+	__u32	offset[4];
+	__u32	pitch[4];
+	__u64	modifier;
+};
+
+/**
+ * Rectangle for processing.
+ *
+ * @id: must be DRM_EXYNOS_IPP_TASK_RECTANGLE
+ * @x,@y: left corner in pixels
+ * @w,@h: width/height in pixels
+ */
+struct drm_exynos_ipp_task_rect {
+	__u32	id;
+	__u32	x, y, w, h;
+};
+
+/**
+ * Image tranformation description.
+ *
+ * @id: must be DRM_EXYNOS_IPP_TASK_TRANSFORM
+ * @rotation: DRM_MODE_ROTATE_* and DRM_MODE_REFLECT_* values
+ */
+struct drm_exynos_ipp_task_transform {
+	__u32	id;
+	__u32	rotation;
+};
+
+/**
+ * Image global alpha configuration for formats without alpha values.
+ *
+ * @id: must be DRM_EXYNOS_IPP_TASK_ALPHA
+ * @value: global alpha value (0-255)
+ */
+struct drm_exynos_ipp_task_alpha {
+	__u32	id;
+	__u32	value;
+};
+
+enum drm_exynos_ipp_flag {
+	/* generate DRM event after processing */
+	DRM_EXYNOS_IPP_FLAG_EVENT	= 0x01,
+	/* dry run, only check task parameters */
+	DRM_EXYNOS_IPP_FLAG_TEST_ONLY	= 0x02,
+	/* non-blocking processing */
+	DRM_EXYNOS_IPP_FLAG_NONBLOCK	= 0x04,
+};
+
+#define DRM_EXYNOS_IPP_FLAGS (DRM_EXYNOS_IPP_FLAG_EVENT |\
+		DRM_EXYNOS_IPP_FLAG_TEST_ONLY | DRM_EXYNOS_IPP_FLAG_NONBLOCK)
+
+/**
+ * Perform image processing described by array of drm_exynos_ipp_task_*
+ * structures (parameters array).
+ *
+ * @ipp_id: id of IPP module to run the task
+ * @flags: bitmask of drm_exynos_ipp_flag values
+ * @reserved: padding
+ * @params_size: size of parameters array (in bytes)
+ * @params_ptr: pointer to parameters array or NULL
+ * @user_data: (optional) data for drm event
+ */
+struct drm_exynos_ioctl_ipp_commit {
+	__u32 ipp_id;
+	__u32 flags;
+	__u32 reserved;
+	__u32 params_size;
+	__u64 params_ptr;
+	__u64 user_data;
+};
+
 #define DRM_EXYNOS_GEM_CREATE		0x00
 #define DRM_EXYNOS_GEM_MAP		0x01
 /* Reserved 0x03 ~ 0x05 for exynos specific gem ioctl */
@@ -146,6 +348,11 @@ struct drm_exynos_g2d_exec {
 #define DRM_EXYNOS_G2D_EXEC		0x22
 
 /* Reserved 0x30 ~ 0x33 for obsolete Exynos IPP ioctls */
+/* IPP - Image Post Processing */
+#define DRM_EXYNOS_IPP_GET_RESOURCES	0x40
+#define DRM_EXYNOS_IPP_GET_CAPS		0x41
+#define DRM_EXYNOS_IPP_GET_LIMITS	0x42
+#define DRM_EXYNOS_IPP_COMMIT		0x43
 
 #define DRM_IOCTL_EXYNOS_GEM_CREATE		DRM_IOWR(DRM_COMMAND_BASE + \
 		DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create)
@@ -164,8 +371,18 @@ 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_RESOURCES	DRM_IOWR(DRM_COMMAND_BASE + \
+		DRM_EXYNOS_IPP_GET_RESOURCES, struct drm_exynos_ioctl_ipp_get_res)
+#define DRM_IOCTL_EXYNOS_IPP_GET_CAPS		DRM_IOWR(DRM_COMMAND_BASE + \
+		DRM_EXYNOS_IPP_GET_CAPS, struct drm_exynos_ioctl_ipp_get_caps)
+#define DRM_IOCTL_EXYNOS_IPP_GET_LIMITS		DRM_IOWR(DRM_COMMAND_BASE + \
+		DRM_EXYNOS_IPP_GET_LIMITS, struct drm_exynos_ioctl_ipp_get_limits)
+#define DRM_IOCTL_EXYNOS_IPP_COMMIT		DRM_IOWR(DRM_COMMAND_BASE + \
+		DRM_EXYNOS_IPP_COMMIT, struct drm_exynos_ioctl_ipp_commit)
+
 /* EXYNOS specific events */
 #define DRM_EXYNOS_G2D_EVENT		0x80000000
+#define DRM_EXYNOS_IPP_EVENT		0x80000002
 
 struct drm_exynos_g2d_event {
 	struct drm_event	base;
@@ -176,6 +393,16 @@ 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			ipp_id;
+	__u32			sequence;
+	__u64			reserved;
+};
+
 #if defined(__cplusplus)
 }
 #endif
-- 
1.9.1

[PATCH 3/6] drm/exynos: rotator: Convert driver to IPP v2 core API

[Marek Szyprowski]

This patch adapts Exynos DRM rotator driver to new IPP v2 core API.
The side effect of this conversion is a switch to driver component API
to register properly in the Exynos DRM core.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/gpu/drm/exynos/Kconfig              |   2 +-
 drivers/gpu/drm/exynos/exynos_drm_drv.c     |   1 +
 drivers/gpu/drm/exynos/exynos_drm_rotator.c | 735 +++++++---------------------
 drivers/gpu/drm/exynos/exynos_drm_rotator.h |  19 -
 4 files changed, 172 insertions(+), 585 deletions(-)
 delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_rotator.h

diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig
index 2e097a81df12..c10c9ca0d8b4 100644
--- a/drivers/gpu/drm/exynos/Kconfig
+++ b/drivers/gpu/drm/exynos/Kconfig
@@ -105,7 +105,7 @@ config DRM_EXYNOS_FIMC
 
 config DRM_EXYNOS_ROTATOR
 	bool "Rotator"
-	depends on BROKEN
+	select DRM_EXYNOS_IPP
 	help
 	  Choose this option if you want to use Exynos Rotator for DRM.
 
diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
index 645046c5943a..277e444b0be6 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
@@ -269,6 +269,7 @@ struct exynos_drm_driver_info {
 		DRV_PTR(fimc_driver, CONFIG_DRM_EXYNOS_FIMC),
 	}, {
 		DRV_PTR(rotator_driver, CONFIG_DRM_EXYNOS_ROTATOR),
+		DRM_COMPONENT_DRIVER
 	}, {
 		DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC),
 	}, {
diff --git a/drivers/gpu/drm/exynos/exynos_drm_rotator.c b/drivers/gpu/drm/exynos/exynos_drm_rotator.c
index 79282a820ecc..8d4182871244 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_rotator.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_rotator.c
@@ -10,6 +10,7 @@
  */
 
 #include <linux/kernel.h>
+#include <linux/component.h>
 #include <linux/err.h>
 #include <linux/interrupt.h>
 #include <linux/io.h>
@@ -22,29 +23,16 @@
 #include <drm/exynos_drm.h>
 #include "regs-rotator.h"
 #include "exynos_drm_drv.h"
+#include "exynos_drm_iommu.h"
 #include "exynos_drm_ipp.h"
 
 /*
  * Rotator supports image crop/rotator and input/output DMA operations.
  * input DMA reads image data from the memory.
  * output DMA writes image data to memory.
- *
- * M2M operation : supports crop/scale/rotation/csc so on.
- * Memory ----> Rotator H/W ----> Memory.
- */
-
-/*
- * TODO
- * 1. check suspend/resume api if needed.
- * 2. need to check use case platform_device_id.
- * 3. check src/dst size with, height.
- * 4. need to add supported list in prop_list.
  */
 
-#define get_rot_context(dev)	platform_get_drvdata(to_platform_device(dev))
-#define get_ctx_from_ippdrv(ippdrv)	container_of(ippdrv,\
-					struct rot_context, ippdrv);
-#define rot_read(offset)		readl(rot->regs + (offset))
+#define rot_read(offset)	readl(rot->regs + (offset))
 #define rot_write(cfg, offset)	writel(cfg, rot->regs + (offset))
 
 enum rot_irq_status {
@@ -53,53 +41,21 @@ enum rot_irq_status {
 };
 
 /*
- * A structure of limitation.
- *
- * @min_w: minimum width.
- * @min_h: minimum height.
- * @max_w: maximum width.
- * @max_h: maximum height.
- * @align: align size.
- */
-struct rot_limit {
-	u32	min_w;
-	u32	min_h;
-	u32	max_w;
-	u32	max_h;
-	u32	align;
-};
-
-/*
- * A structure of limitation table.
- *
- * @ycbcr420_2p: case of YUV.
- * @rgb888: case of RGB.
- */
-struct rot_limit_table {
-	struct rot_limit	ycbcr420_2p;
-	struct rot_limit	rgb888;
-};
-
-/*
  * A structure of rotator context.
  * @ippdrv: prepare initialization using ippdrv.
- * @regs_res: register resources.
  * @regs: memory mapped io registers.
  * @clock: rotator gate clock.
  * @limit_tbl: limitation of rotator.
  * @irq: irq number.
- * @cur_buf_id: current operation buffer id.
- * @suspended: suspended state.
  */
 struct rot_context {
-	struct exynos_drm_ippdrv	ippdrv;
-	struct resource	*regs_res;
+	struct exynos_drm_ipp ipp;
+	struct drm_device *drm_dev;
+	struct device	*dev;
 	void __iomem	*regs;
 	struct clk	*clock;
-	struct rot_limit_table	*limit_tbl;
-	int	irq;
-	int	cur_buf_id[EXYNOS_DRM_OPS_MAX];
-	bool	suspended;
+	const struct exynos_drm_ipp_formats *formats;
+	struct exynos_drm_ipp_task	*task;
 };
 
 static void rotator_reg_set_irq(struct rot_context *rot, bool enable)
@@ -114,15 +70,6 @@ static void rotator_reg_set_irq(struct rot_context *rot, bool enable)
 	rot_write(val, ROT_CONFIG);
 }
 
-static u32 rotator_reg_get_fmt(struct rot_context *rot)
-{
-	u32 val = rot_read(ROT_CONTROL);
-
-	val &= ROT_CONTROL_FMT_MASK;
-
-	return val;
-}
-
 static enum rot_irq_status rotator_reg_get_irq_status(struct rot_context *rot)
 {
 	u32 val = rot_read(ROT_STATUS);
@@ -138,9 +85,6 @@ static enum rot_irq_status rotator_reg_get_irq_status(struct rot_context *rot)
 static irqreturn_t rotator_irq_handler(int irq, void *arg)
 {
 	struct rot_context *rot = arg;
-	struct exynos_drm_ippdrv *ippdrv = &rot->ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
-	struct drm_exynos_ipp_event_work *event_work = c_node->event_work;
 	enum rot_irq_status irq_status;
 	u32 val;
 
@@ -152,56 +96,20 @@ static irqreturn_t rotator_irq_handler(int irq, void *arg)
 	val |= ROT_STATUS_IRQ_PENDING((u32)irq_status);
 	rot_write(val, ROT_STATUS);
 
-	if (irq_status == ROT_IRQ_STATUS_COMPLETE) {
-		event_work->ippdrv = ippdrv;
-		event_work->buf_id[EXYNOS_DRM_OPS_DST] =
-			rot->cur_buf_id[EXYNOS_DRM_OPS_DST];
-		queue_work(ippdrv->event_workq, &event_work->work);
-	} else {
-		DRM_ERROR("the SFR is set illegally\n");
+	if (rot->task) {
+		struct exynos_drm_ipp_task *task = rot->task;
+
+		rot->task = NULL;
+		pm_runtime_put(rot->dev);
+		exynos_drm_ipp_task_done(task,
+			irq_status == ROT_IRQ_STATUS_COMPLETE ? 0 : -EINVAL);
 	}
 
 	return IRQ_HANDLED;
 }
 
-static void rotator_align_size(struct rot_context *rot, u32 fmt, u32 *hsize,
-		u32 *vsize)
-{
-	struct rot_limit_table *limit_tbl = rot->limit_tbl;
-	struct rot_limit *limit;
-	u32 mask, val;
-
-	/* Get size limit */
-	if (fmt == ROT_CONTROL_FMT_RGB888)
-		limit = &limit_tbl->rgb888;
-	else
-		limit = &limit_tbl->ycbcr420_2p;
-
-	/* Get mask for rounding to nearest aligned val */
-	mask = ~((1 << limit->align) - 1);
-
-	/* Set aligned width */
-	val = ROT_ALIGN(*hsize, limit->align, mask);
-	if (val < limit->min_w)
-		*hsize = ROT_MIN(limit->min_w, mask);
-	else if (val > limit->max_w)
-		*hsize = ROT_MAX(limit->max_w, mask);
-	else
-		*hsize = val;
-
-	/* Set aligned height */
-	val = ROT_ALIGN(*vsize, limit->align, mask);
-	if (val < limit->min_h)
-		*vsize = ROT_MIN(limit->min_h, mask);
-	else if (val > limit->max_h)
-		*vsize = ROT_MAX(limit->max_h, mask);
-	else
-		*vsize = val;
-}
-
-static int rotator_src_set_fmt(struct device *dev, u32 fmt)
+static int rotator_src_set_fmt(struct rot_context *rot, u32 fmt)
 {
-	struct rot_context *rot = dev_get_drvdata(dev);
 	u32 val;
 
 	val = rot_read(ROT_CONTROL);
@@ -214,9 +122,6 @@ static int rotator_src_set_fmt(struct device *dev, u32 fmt)
 	case DRM_FORMAT_XRGB8888:
 		val |= ROT_CONTROL_FMT_RGB888;
 		break;
-	default:
-		DRM_ERROR("invalid image format\n");
-		return -EINVAL;
 	}
 
 	rot_write(val, ROT_CONTROL);
@@ -224,505 +129,176 @@ static int rotator_src_set_fmt(struct device *dev, u32 fmt)
 	return 0;
 }
 
-static inline bool rotator_check_reg_fmt(u32 fmt)
-{
-	if ((fmt == ROT_CONTROL_FMT_YCBCR420_2P) ||
-	    (fmt == ROT_CONTROL_FMT_RGB888))
-		return true;
-
-	return false;
-}
-
-static int rotator_src_set_size(struct device *dev, int swap,
-		struct drm_exynos_pos *pos,
-		struct drm_exynos_sz *sz)
+static int rotator_src_set_buf(struct rot_context *rot,
+			       struct exynos_drm_ipp_buffer *buf)
 {
-	struct rot_context *rot = dev_get_drvdata(dev);
-	u32 fmt, hsize, vsize;
 	u32 val;
 
-	/* Get format */
-	fmt = rotator_reg_get_fmt(rot);
-	if (!rotator_check_reg_fmt(fmt)) {
-		DRM_ERROR("invalid format.\n");
-		return -EINVAL;
-	}
-
-	/* Align buffer size */
-	hsize = sz->hsize;
-	vsize = sz->vsize;
-	rotator_align_size(rot, fmt, &hsize, &vsize);
-
 	/* Set buffer size configuration */
-	val = ROT_SET_BUF_SIZE_H(vsize) | ROT_SET_BUF_SIZE_W(hsize);
+	val = ROT_SET_BUF_SIZE_H(buf->buf.height) |
+	      ROT_SET_BUF_SIZE_W(buf->buf.pitch[0] / buf->format->cpp[0]);
 	rot_write(val, ROT_SRC_BUF_SIZE);
 
 	/* Set crop image position configuration */
-	val = ROT_CROP_POS_Y(pos->y) | ROT_CROP_POS_X(pos->x);
+	val = ROT_CROP_POS_Y(buf->rect.y) | ROT_CROP_POS_X(buf->rect.x);
 	rot_write(val, ROT_SRC_CROP_POS);
-	val = ROT_SRC_CROP_SIZE_H(pos->h) | ROT_SRC_CROP_SIZE_W(pos->w);
+	val = ROT_SRC_CROP_SIZE_H(buf->rect.h) |
+	      ROT_SRC_CROP_SIZE_W(buf->rect.w);
 	rot_write(val, ROT_SRC_CROP_SIZE);
 
-	return 0;
-}
-
-static int rotator_src_set_addr(struct device *dev,
-		struct drm_exynos_ipp_buf_info *buf_info,
-		u32 buf_id, enum drm_exynos_ipp_buf_type buf_type)
-{
-	struct rot_context *rot = dev_get_drvdata(dev);
-	dma_addr_t addr[EXYNOS_DRM_PLANAR_MAX];
-	u32 val, fmt, hsize, vsize;
-	int i;
-
-	/* Set current buf_id */
-	rot->cur_buf_id[EXYNOS_DRM_OPS_SRC] = buf_id;
-
-	switch (buf_type) {
-	case IPP_BUF_ENQUEUE:
-		/* Set address configuration */
-		for_each_ipp_planar(i)
-			addr[i] = buf_info->base[i];
-
-		/* Get format */
-		fmt = rotator_reg_get_fmt(rot);
-		if (!rotator_check_reg_fmt(fmt)) {
-			DRM_ERROR("invalid format.\n");
-			return -EINVAL;
-		}
-
-		/* Re-set cb planar for NV12 format */
-		if ((fmt == ROT_CONTROL_FMT_YCBCR420_2P) &&
-		    !addr[EXYNOS_DRM_PLANAR_CB]) {
-
-			val = rot_read(ROT_SRC_BUF_SIZE);
-			hsize = ROT_GET_BUF_SIZE_W(val);
-			vsize = ROT_GET_BUF_SIZE_H(val);
-
-			/* Set cb planar */
-			addr[EXYNOS_DRM_PLANAR_CB] =
-				addr[EXYNOS_DRM_PLANAR_Y] + hsize * vsize;
-		}
-
-		for_each_ipp_planar(i)
-			rot_write(addr[i], ROT_SRC_BUF_ADDR(i));
-		break;
-	case IPP_BUF_DEQUEUE:
-		for_each_ipp_planar(i)
-			rot_write(0x0, ROT_SRC_BUF_ADDR(i));
-		break;
-	default:
-		/* Nothing to do */
-		break;
-	}
+	/* Set buffer DMA address */
+	rot_write(buf->dma_addr[0], ROT_SRC_BUF_ADDR(0));
+	rot_write(buf->dma_addr[1], ROT_SRC_BUF_ADDR(1));
 
 	return 0;
 }
 
-static int rotator_dst_set_transf(struct device *dev,
-		enum drm_exynos_degree degree,
-		enum drm_exynos_flip flip, bool *swap)
+static int rotator_dst_set_transf(struct rot_context *rot,
+				  unsigned int rotation)
 {
-	struct rot_context *rot = dev_get_drvdata(dev);
 	u32 val;
 
 	/* Set transform configuration */
 	val = rot_read(ROT_CONTROL);
 	val &= ~ROT_CONTROL_FLIP_MASK;
 
-	switch (flip) {
-	case EXYNOS_DRM_FLIP_VERTICAL:
-		val |= ROT_CONTROL_FLIP_VERTICAL;
-		break;
-	case EXYNOS_DRM_FLIP_HORIZONTAL:
+	if (rotation & DRM_MODE_REFLECT_X)
 		val |= ROT_CONTROL_FLIP_HORIZONTAL;
-		break;
-	default:
-		/* Flip None */
-		break;
-	}
+	if (rotation & DRM_MODE_REFLECT_Y)
+		val |= ROT_CONTROL_FLIP_VERTICAL;
 
 	val &= ~ROT_CONTROL_ROT_MASK;
 
-	switch (degree) {
-	case EXYNOS_DRM_DEGREE_90:
+	if (rotation & DRM_MODE_ROTATE_90)
 		val |= ROT_CONTROL_ROT_90;
-		break;
-	case EXYNOS_DRM_DEGREE_180:
+	else if (rotation & DRM_MODE_ROTATE_180)
 		val |= ROT_CONTROL_ROT_180;
-		break;
-	case EXYNOS_DRM_DEGREE_270:
+	else if (rotation & DRM_MODE_ROTATE_270)
 		val |= ROT_CONTROL_ROT_270;
-		break;
-	default:
-		/* Rotation 0 Degree */
-		break;
-	}
 
 	rot_write(val, ROT_CONTROL);
 
-	/* Check degree for setting buffer size swap */
-	if ((degree == EXYNOS_DRM_DEGREE_90) ||
-	    (degree == EXYNOS_DRM_DEGREE_270))
-		*swap = true;
-	else
-		*swap = false;
-
 	return 0;
 }
 
-static int rotator_dst_set_size(struct device *dev, int swap,
-		struct drm_exynos_pos *pos,
-		struct drm_exynos_sz *sz)
+static int rotator_dst_set_buf(struct rot_context *rot,
+			       struct exynos_drm_ipp_buffer *buf)
 {
-	struct rot_context *rot = dev_get_drvdata(dev);
-	u32 val, fmt, hsize, vsize;
-
-	/* Get format */
-	fmt = rotator_reg_get_fmt(rot);
-	if (!rotator_check_reg_fmt(fmt)) {
-		DRM_ERROR("invalid format.\n");
-		return -EINVAL;
-	}
-
-	/* Align buffer size */
-	hsize = sz->hsize;
-	vsize = sz->vsize;
-	rotator_align_size(rot, fmt, &hsize, &vsize);
+	u32 val;
 
 	/* Set buffer size configuration */
-	val = ROT_SET_BUF_SIZE_H(vsize) | ROT_SET_BUF_SIZE_W(hsize);
+	val = ROT_SET_BUF_SIZE_H(buf->buf.height) |
+	      ROT_SET_BUF_SIZE_W(buf->buf.pitch[0] / buf->format->cpp[0]);
 	rot_write(val, ROT_DST_BUF_SIZE);
 
 	/* Set crop image position configuration */
-	val = ROT_CROP_POS_Y(pos->y) | ROT_CROP_POS_X(pos->x);
+	val = ROT_CROP_POS_Y(buf->rect.y) | ROT_CROP_POS_X(buf->rect.x);
 	rot_write(val, ROT_DST_CROP_POS);
 
-	return 0;
-}
-
-static int rotator_dst_set_addr(struct device *dev,
-		struct drm_exynos_ipp_buf_info *buf_info,
-		u32 buf_id, enum drm_exynos_ipp_buf_type buf_type)
-{
-	struct rot_context *rot = dev_get_drvdata(dev);
-	dma_addr_t addr[EXYNOS_DRM_PLANAR_MAX];
-	u32 val, fmt, hsize, vsize;
-	int i;
-
-	/* Set current buf_id */
-	rot->cur_buf_id[EXYNOS_DRM_OPS_DST] = buf_id;
-
-	switch (buf_type) {
-	case IPP_BUF_ENQUEUE:
-		/* Set address configuration */
-		for_each_ipp_planar(i)
-			addr[i] = buf_info->base[i];
-
-		/* Get format */
-		fmt = rotator_reg_get_fmt(rot);
-		if (!rotator_check_reg_fmt(fmt)) {
-			DRM_ERROR("invalid format.\n");
-			return -EINVAL;
-		}
-
-		/* Re-set cb planar for NV12 format */
-		if ((fmt == ROT_CONTROL_FMT_YCBCR420_2P) &&
-		    !addr[EXYNOS_DRM_PLANAR_CB]) {
-			/* Get buf size */
-			val = rot_read(ROT_DST_BUF_SIZE);
-
-			hsize = ROT_GET_BUF_SIZE_W(val);
-			vsize = ROT_GET_BUF_SIZE_H(val);
-
-			/* Set cb planar */
-			addr[EXYNOS_DRM_PLANAR_CB] =
-				addr[EXYNOS_DRM_PLANAR_Y] + hsize * vsize;
-		}
-
-		for_each_ipp_planar(i)
-			rot_write(addr[i], ROT_DST_BUF_ADDR(i));
-		break;
-	case IPP_BUF_DEQUEUE:
-		for_each_ipp_planar(i)
-			rot_write(0x0, ROT_DST_BUF_ADDR(i));
-		break;
-	default:
-		/* Nothing to do */
-		break;
-	}
+	/* Set buffer DMA address */
+	rot_write(buf->dma_addr[0], ROT_DST_BUF_ADDR(0));
+	rot_write(buf->dma_addr[1], ROT_DST_BUF_ADDR(1));
 
 	return 0;
 }
 
-static struct exynos_drm_ipp_ops rot_src_ops = {
-	.set_fmt	=	rotator_src_set_fmt,
-	.set_size	=	rotator_src_set_size,
-	.set_addr	=	rotator_src_set_addr,
-};
-
-static struct exynos_drm_ipp_ops rot_dst_ops = {
-	.set_transf	=	rotator_dst_set_transf,
-	.set_size	=	rotator_dst_set_size,
-	.set_addr	=	rotator_dst_set_addr,
-};
-
-static int rotator_init_prop_list(struct exynos_drm_ippdrv *ippdrv)
+static int rotator_start(struct rot_context *rot)
 {
-	struct drm_exynos_ipp_prop_list *prop_list = &ippdrv->prop_list;
-
-	prop_list->version = 1;
-	prop_list->flip = (1 << EXYNOS_DRM_FLIP_VERTICAL) |
-				(1 << EXYNOS_DRM_FLIP_HORIZONTAL);
-	prop_list->degree = (1 << EXYNOS_DRM_DEGREE_0) |
-				(1 << EXYNOS_DRM_DEGREE_90) |
-				(1 << EXYNOS_DRM_DEGREE_180) |
-				(1 << EXYNOS_DRM_DEGREE_270);
-	prop_list->csc = 0;
-	prop_list->crop = 0;
-	prop_list->scale = 0;
+	u32 val;
 
-	return 0;
-}
+	/* Set interrupt enable */
+	rotator_reg_set_irq(rot, true);
 
-static inline bool rotator_check_drm_fmt(u32 fmt)
-{
-	switch (fmt) {
-	case DRM_FORMAT_XRGB8888:
-	case DRM_FORMAT_NV12:
-		return true;
-	default:
-		DRM_DEBUG_KMS("not support format\n");
-		return false;
-	}
-}
+	val = rot_read(ROT_CONTROL);
+	val |= ROT_CONTROL_START;
+	rot_write(val, ROT_CONTROL);
 
-static inline bool rotator_check_drm_flip(enum drm_exynos_flip flip)
-{
-	switch (flip) {
-	case EXYNOS_DRM_FLIP_NONE:
-	case EXYNOS_DRM_FLIP_VERTICAL:
-	case EXYNOS_DRM_FLIP_HORIZONTAL:
-	case EXYNOS_DRM_FLIP_BOTH:
-		return true;
-	default:
-		DRM_DEBUG_KMS("invalid flip\n");
-		return false;
-	}
+	return 0;
 }
 
-static int rotator_ippdrv_check_property(struct device *dev,
-		struct drm_exynos_ipp_property *property)
+static int rotator_commit(struct exynos_drm_ipp *ipp,
+			  struct exynos_drm_ipp_task *task)
 {
-	struct drm_exynos_ipp_config *src_config =
-					&property->config[EXYNOS_DRM_OPS_SRC];
-	struct drm_exynos_ipp_config *dst_config =
-					&property->config[EXYNOS_DRM_OPS_DST];
-	struct drm_exynos_pos *src_pos = &src_config->pos;
-	struct drm_exynos_pos *dst_pos = &dst_config->pos;
-	struct drm_exynos_sz *src_sz = &src_config->sz;
-	struct drm_exynos_sz *dst_sz = &dst_config->sz;
-	bool swap = false;
-
-	/* Check format configuration */
-	if (src_config->fmt != dst_config->fmt) {
-		DRM_DEBUG_KMS("not support csc feature\n");
-		return -EINVAL;
-	}
-
-	if (!rotator_check_drm_fmt(dst_config->fmt)) {
-		DRM_DEBUG_KMS("invalid format\n");
-		return -EINVAL;
-	}
-
-	/* Check transform configuration */
-	if (src_config->degree != EXYNOS_DRM_DEGREE_0) {
-		DRM_DEBUG_KMS("not support source-side rotation\n");
-		return -EINVAL;
-	}
-
-	switch (dst_config->degree) {
-	case EXYNOS_DRM_DEGREE_90:
-	case EXYNOS_DRM_DEGREE_270:
-		swap = true;
-	case EXYNOS_DRM_DEGREE_0:
-	case EXYNOS_DRM_DEGREE_180:
-		/* No problem */
-		break;
-	default:
-		DRM_DEBUG_KMS("invalid degree\n");
-		return -EINVAL;
-	}
-
-	if (src_config->flip != EXYNOS_DRM_FLIP_NONE) {
-		DRM_DEBUG_KMS("not support source-side flip\n");
-		return -EINVAL;
-	}
-
-	if (!rotator_check_drm_flip(dst_config->flip)) {
-		DRM_DEBUG_KMS("invalid flip\n");
-		return -EINVAL;
-	}
+	struct rot_context *rot =
+			container_of(ipp, struct rot_context, ipp);
 
-	/* Check size configuration */
-	if ((src_pos->x + src_pos->w > src_sz->hsize) ||
-		(src_pos->y + src_pos->h > src_sz->vsize)) {
-		DRM_DEBUG_KMS("out of source buffer bound\n");
-		return -EINVAL;
-	}
+	pm_runtime_get_sync(rot->dev);
+	rot->task = task;
 
-	if (swap) {
-		if ((dst_pos->x + dst_pos->h > dst_sz->vsize) ||
-			(dst_pos->y + dst_pos->w > dst_sz->hsize)) {
-			DRM_DEBUG_KMS("out of destination buffer bound\n");
-			return -EINVAL;
-		}
-
-		if ((src_pos->w != dst_pos->h) || (src_pos->h != dst_pos->w)) {
-			DRM_DEBUG_KMS("not support scale feature\n");
-			return -EINVAL;
-		}
-	} else {
-		if ((dst_pos->x + dst_pos->w > dst_sz->hsize) ||
-			(dst_pos->y + dst_pos->h > dst_sz->vsize)) {
-			DRM_DEBUG_KMS("out of destination buffer bound\n");
-			return -EINVAL;
-		}
-
-		if ((src_pos->w != dst_pos->w) || (src_pos->h != dst_pos->h)) {
-			DRM_DEBUG_KMS("not support scale feature\n");
-			return -EINVAL;
-		}
-	}
+	rotator_src_set_fmt(rot, task->src.buf.fourcc);
+	rotator_src_set_buf(rot, &task->src);
+	rotator_dst_set_transf(rot, task->transform.rotation);
+	rotator_dst_set_buf(rot, &task->dst);
+	rotator_start(rot);
 
 	return 0;
 }
 
-static int rotator_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd)
+static const struct exynos_drm_ipp_funcs ipp_funcs = {
+	.commit = rotator_commit,
+};
+
+static int rotator_bind(struct device *dev, struct device *master, void *data)
 {
 	struct rot_context *rot = dev_get_drvdata(dev);
-	u32 val;
+	struct drm_device *drm_dev = data;
+	struct exynos_drm_ipp *ipp = &rot->ipp;
 
-	if (rot->suspended) {
-		DRM_ERROR("suspended state\n");
-		return -EPERM;
-	}
-
-	if (cmd != IPP_CMD_M2M) {
-		DRM_ERROR("not support cmd: %d\n", cmd);
-		return -EINVAL;
-	}
-
-	/* Set interrupt enable */
-	rotator_reg_set_irq(rot, true);
+	rot->drm_dev = drm_dev;
+	drm_iommu_attach_device(drm_dev, dev);
 
-	val = rot_read(ROT_CONTROL);
-	val |= ROT_CONTROL_START;
+	exynos_drm_ipp_register(drm_dev, ipp, &ipp_funcs,
+			   DRM_EXYNOS_IPP_CAP_CROP | DRM_EXYNOS_IPP_CAP_ROTATE,
+			   rot->formats, "rotator");
 
-	rot_write(val, ROT_CONTROL);
+	dev_info(dev, "The exynos rotator has been probed successfully\n");
 
 	return 0;
 }
 
-static struct rot_limit_table rot_limit_tbl_4210 = {
-	.ycbcr420_2p = {
-		.min_w = 32,
-		.min_h = 32,
-		.max_w = SZ_64K,
-		.max_h = SZ_64K,
-		.align = 3,
-	},
-	.rgb888 = {
-		.min_w = 8,
-		.min_h = 8,
-		.max_w = SZ_16K,
-		.max_h = SZ_16K,
-		.align = 2,
-	},
-};
-
-static struct rot_limit_table rot_limit_tbl_4x12 = {
-	.ycbcr420_2p = {
-		.min_w = 32,
-		.min_h = 32,
-		.max_w = SZ_32K,
-		.max_h = SZ_32K,
-		.align = 3,
-	},
-	.rgb888 = {
-		.min_w = 8,
-		.min_h = 8,
-		.max_w = SZ_8K,
-		.max_h = SZ_8K,
-		.align = 2,
-	},
-};
+static void rotator_unbind(struct device *dev, struct device *master,
+			void *data)
+{
+	struct rot_context *rot = dev_get_drvdata(dev);
+	struct drm_device *drm_dev = data;
+	struct exynos_drm_ipp *ipp = &rot->ipp;
 
-static struct rot_limit_table rot_limit_tbl_5250 = {
-	.ycbcr420_2p = {
-		.min_w = 32,
-		.min_h = 32,
-		.max_w = SZ_32K,
-		.max_h = SZ_32K,
-		.align = 3,
-	},
-	.rgb888 = {
-		.min_w = 8,
-		.min_h = 8,
-		.max_w = SZ_8K,
-		.max_h = SZ_8K,
-		.align = 1,
-	},
-};
+	exynos_drm_ipp_unregister(drm_dev, ipp);
+	drm_iommu_detach_device(rot->drm_dev, rot->dev);
+}
 
-static const struct of_device_id exynos_rotator_match[] = {
-	{
-		.compatible = "samsung,exynos4210-rotator",
-		.data = &rot_limit_tbl_4210,
-	},
-	{
-		.compatible = "samsung,exynos4212-rotator",
-		.data = &rot_limit_tbl_4x12,
-	},
-	{
-		.compatible = "samsung,exynos5250-rotator",
-		.data = &rot_limit_tbl_5250,
-	},
-	{},
+static const struct component_ops rotator_component_ops = {
+	.bind	= rotator_bind,
+	.unbind = rotator_unbind,
 };
-MODULE_DEVICE_TABLE(of, exynos_rotator_match);
 
 static int rotator_probe(struct platform_device *pdev)
 {
 	struct device *dev = &pdev->dev;
+	struct resource	*regs_res;
 	struct rot_context *rot;
-	struct exynos_drm_ippdrv *ippdrv;
+	int irq;
 	int ret;
 
-	if (!dev->of_node) {
-		dev_err(dev, "cannot find of_node.\n");
-		return -ENODEV;
-	}
-
 	rot = devm_kzalloc(dev, sizeof(*rot), GFP_KERNEL);
 	if (!rot)
 		return -ENOMEM;
 
-	rot->limit_tbl = (struct rot_limit_table *)
-				of_device_get_match_data(dev);
-	rot->regs_res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
-	rot->regs = devm_ioremap_resource(dev, rot->regs_res);
+	rot->formats = of_device_get_match_data(dev);
+	rot->dev = dev;
+	regs_res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	rot->regs = devm_ioremap_resource(dev, regs_res);
 	if (IS_ERR(rot->regs))
 		return PTR_ERR(rot->regs);
 
-	rot->irq = platform_get_irq(pdev, 0);
-	if (rot->irq < 0) {
+	irq = platform_get_irq(pdev, 0);
+	if (irq < 0) {
 		dev_err(dev, "failed to get irq\n");
-		return rot->irq;
+		return irq;
 	}
 
-	ret = devm_request_threaded_irq(dev, rot->irq, NULL,
-			rotator_irq_handler, IRQF_ONESHOT, "drm_rotator", rot);
+	ret = devm_request_irq(dev, irq, rotator_irq_handler, 0, dev_name(dev),
+			       rot);
 	if (ret < 0) {
 		dev_err(dev, "failed to request irq\n");
 		return ret;
@@ -735,34 +311,15 @@ static int rotator_probe(struct platform_device *pdev)
 	}
 
 	pm_runtime_enable(dev);
-
-	ippdrv = &rot->ippdrv;
-	ippdrv->dev = dev;
-	ippdrv->ops[EXYNOS_DRM_OPS_SRC] = &rot_src_ops;
-	ippdrv->ops[EXYNOS_DRM_OPS_DST] = &rot_dst_ops;
-	ippdrv->check_property = rotator_ippdrv_check_property;
-	ippdrv->start = rotator_ippdrv_start;
-	ret = rotator_init_prop_list(ippdrv);
-	if (ret < 0) {
-		dev_err(dev, "failed to init property list.\n");
-		goto err_ippdrv_register;
-	}
-
-	DRM_DEBUG_KMS("ippdrv[%pK]\n", ippdrv);
-
 	platform_set_drvdata(pdev, rot);
 
-	ret = exynos_drm_ippdrv_register(ippdrv);
-	if (ret < 0) {
-		dev_err(dev, "failed to register drm rotator device\n");
-		goto err_ippdrv_register;
-	}
-
-	dev_info(dev, "The exynos rotator is probed successfully\n");
+	ret = component_add(dev, &rotator_component_ops);
+	if (ret)
+		goto err_component;
 
 	return 0;
 
-err_ippdrv_register:
+err_component:
 	pm_runtime_disable(dev);
 	return ret;
 }
@@ -770,45 +327,93 @@ static int rotator_probe(struct platform_device *pdev)
 static int rotator_remove(struct platform_device *pdev)
 {
 	struct device *dev = &pdev->dev;
-	struct rot_context *rot = dev_get_drvdata(dev);
-	struct exynos_drm_ippdrv *ippdrv = &rot->ippdrv;
-
-	exynos_drm_ippdrv_unregister(ippdrv);
 
+	component_del(dev, &rotator_component_ops);
 	pm_runtime_disable(dev);
 
 	return 0;
 }
 
 #ifdef CONFIG_PM
-static int rotator_clk_crtl(struct rot_context *rot, bool enable)
-{
-	if (enable) {
-		clk_prepare_enable(rot->clock);
-		rot->suspended = false;
-	} else {
-		clk_disable_unprepare(rot->clock);
-		rot->suspended = true;
-	}
-
-	return 0;
-}
-
 static int rotator_runtime_suspend(struct device *dev)
 {
 	struct rot_context *rot = dev_get_drvdata(dev);
 
-	return  rotator_clk_crtl(rot, false);
+	clk_disable_unprepare(rot->clock);
+	return 0;
 }
 
 static int rotator_runtime_resume(struct device *dev)
 {
 	struct rot_context *rot = dev_get_drvdata(dev);
 
-	return  rotator_clk_crtl(rot, true);
+	return clk_prepare_enable(rot->clock);
 }
 #endif
 
+static const struct drm_exynos_ipp_limit rotator_4210_rbg888_limits[] = {
+	{ IPP_SIZE_LIMIT(BUFFER, .h = {8, SZ_16K}, .v = {8, SZ_16K}) },
+	{ IPP_SIZE_LIMIT(AREA, .h.align = 4, .v.align = 4) },
+	{ }
+};
+
+static const struct drm_exynos_ipp_limit rotator_4412_rbg888_limits[] = {
+	{ IPP_SIZE_LIMIT(BUFFER, .h = {8, SZ_8K}, .v = {8, SZ_8K}) },
+	{ IPP_SIZE_LIMIT(AREA, .h.align = 4, .v.align = 4) },
+	{ }
+};
+
+static const struct drm_exynos_ipp_limit rotator_5250_rbg888_limits[] = {
+	{ IPP_SIZE_LIMIT(BUFFER, .h = {8, SZ_8K}, .v = {8, SZ_8K}) },
+	{ IPP_SIZE_LIMIT(AREA, .h.align = 2, .v.align = 2) },
+	{ }
+};
+
+static const struct drm_exynos_ipp_limit rotator_4210_yuv_limits[] = {
+	{ IPP_SIZE_LIMIT(BUFFER, .h = {32, SZ_64K}, .v = {32, SZ_64K}) },
+	{ IPP_SIZE_LIMIT(AREA, .h.align = 8, .v.align = 8) },
+	{ }
+};
+
+static const struct drm_exynos_ipp_limit rotator_4412_yuv_limits[] = {
+	{ IPP_SIZE_LIMIT(BUFFER, .h = {32, SZ_32K}, .v = {32, SZ_32K}) },
+	{ IPP_SIZE_LIMIT(AREA, .h.align = 8, .v.align = 8) },
+	{ }
+};
+
+static const struct exynos_drm_ipp_formats rotator_4210_formats[] = {
+	{ IPP_SRCDST_FORMAT(XRGB8888, rotator_4210_rbg888_limits) },
+	{ IPP_SRCDST_FORMAT(NV12, rotator_4210_yuv_limits) },
+	{ }
+};
+
+static const struct exynos_drm_ipp_formats rotator_4412_formats[] = {
+	{ IPP_SRCDST_FORMAT(XRGB8888, rotator_4412_rbg888_limits) },
+	{ IPP_SRCDST_FORMAT(NV12, rotator_4412_yuv_limits) },
+	{ }
+};
+
+static const struct exynos_drm_ipp_formats rotator_5250_formats[] = {
+	{ IPP_SRCDST_FORMAT(XRGB8888, rotator_5250_rbg888_limits) },
+	{ IPP_SRCDST_FORMAT(NV12, rotator_4412_yuv_limits) },
+	{ }
+};
+
+static const struct of_device_id exynos_rotator_match[] = {
+	{
+		.compatible = "samsung,exynos4210-rotator",
+		.data = rotator_4210_formats,
+	}, {
+		.compatible = "samsung,exynos4212-rotator",
+		.data = rotator_4412_formats,
+	}, {
+		.compatible = "samsung,exynos5250-rotator",
+		.data = rotator_5250_formats,
+	}, {
+	},
+};
+MODULE_DEVICE_TABLE(of, exynos_rotator_match);
+
 static const struct dev_pm_ops rotator_pm_ops = {
 	SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
 				pm_runtime_force_resume)
@@ -820,7 +425,7 @@ struct platform_driver rotator_driver = {
 	.probe		= rotator_probe,
 	.remove		= rotator_remove,
 	.driver		= {
-		.name	= "exynos-rot",
+		.name	= "exynos-rotator",
 		.owner	= THIS_MODULE,
 		.pm	= &rotator_pm_ops,
 		.of_match_table = exynos_rotator_match,
diff --git a/drivers/gpu/drm/exynos/exynos_drm_rotator.h b/drivers/gpu/drm/exynos/exynos_drm_rotator.h
deleted file mode 100644
index 71a0b4c0c1e8..000000000000
--- a/drivers/gpu/drm/exynos/exynos_drm_rotator.h
+++ /dev/null
@@ -1,19 +0,0 @@
-/*
- * Copyright (c) 2012 Samsung Electronics Co., Ltd.
- *
- * Authors:
- *	YoungJun Cho <yj44.cho@samsung.com>
- *	Eunchul Kim <chulspro.kim@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.
- */
-
-#ifndef	_EXYNOS_DRM_ROTATOR_H_
-#define	_EXYNOS_DRM_ROTATOR_H_
-
-/* TODO */
-
-#endif
-- 
1.9.1

[PATCH 4/6] drm/exynos: gsc: Convert driver to IPP v2 core API

[Marek Szyprowski]

This patch adapts Exynos DRM rotator driver to new IPP v2 core API.
The side effect of this conversion is a switch to driver component API
to register properly in the Exynos DRM core.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/gpu/drm/exynos/Kconfig          |   3 +-
 drivers/gpu/drm/exynos/exynos_drm_drv.c |   1 +
 drivers/gpu/drm/exynos/exynos_drm_gsc.c | 853 ++++++++------------------------
 drivers/gpu/drm/exynos/exynos_drm_gsc.h |  24 -
 4 files changed, 198 insertions(+), 683 deletions(-)
 delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_gsc.h

diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig
index c10c9ca0d8b4..4bb9edb00601 100644
--- a/drivers/gpu/drm/exynos/Kconfig
+++ b/drivers/gpu/drm/exynos/Kconfig
@@ -111,7 +111,8 @@ config DRM_EXYNOS_ROTATOR
 
 config DRM_EXYNOS_GSC
 	bool "GScaler"
-	depends on BROKEN && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n
+	depends on ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n
+	select DRM_EXYNOS_IPP
 	help
 	  Choose this option if you want to use Exynos GSC for DRM.
 
diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
index 277e444b0be6..27242af103ea 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
@@ -272,6 +272,7 @@ struct exynos_drm_driver_info {
 		DRM_COMPONENT_DRIVER
 	}, {
 		DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC),
+		DRM_COMPONENT_DRIVER
 	}, {
 		&exynos_drm_platform_driver,
 		DRM_VIRTUAL_DEVICE
diff --git a/drivers/gpu/drm/exynos/exynos_drm_gsc.c b/drivers/gpu/drm/exynos/exynos_drm_gsc.c
index 0506b2b17ac1..1293441a4212 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_gsc.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_gsc.c
@@ -12,6 +12,7 @@
  *
  */
 #include <linux/kernel.h>
+#include <linux/component.h>
 #include <linux/platform_device.h>
 #include <linux/clk.h>
 #include <linux/pm_runtime.h>
@@ -22,8 +23,8 @@
 #include <drm/exynos_drm.h>
 #include "regs-gsc.h"
 #include "exynos_drm_drv.h"
+#include "exynos_drm_iommu.h"
 #include "exynos_drm_ipp.h"
-#include "exynos_drm_gsc.h"
 
 /*
  * GSC stands for General SCaler and
@@ -31,26 +32,9 @@
  * input DMA reads image data from the memory.
  * output DMA writes image data to memory.
  * GSC supports image rotation and image effect functions.
- *
- * M2M operation : supports crop/scale/rotation/csc so on.
- * Memory ----> GSC H/W ----> Memory.
- * Writeback operation : supports cloned screen with FIMD.
- * FIMD ----> GSC H/W ----> Memory.
- * Output operation : supports direct display using local path.
- * Memory ----> GSC H/W ----> FIMD, Mixer.
  */
 
-/*
- * TODO
- * 1. check suspend/resume api if needed.
- * 2. need to check use case platform_device_id.
- * 3. check src/dst size with, height.
- * 4. added check_prepare api for right register.
- * 5. need to add supported list in prop_list.
- * 6. check prescaler/scaler optimization.
- */
 
-#define GSC_MAX_DEVS	4
 #define GSC_MAX_SRC		4
 #define GSC_MAX_DST		16
 #define GSC_RESET_TIMEOUT	50
@@ -65,8 +49,6 @@
 #define GSC_SC_DOWN_RATIO_4_8		131072
 #define GSC_SC_DOWN_RATIO_3_8		174762
 #define GSC_SC_DOWN_RATIO_2_8		262144
-#define GSC_REFRESH_MIN	12
-#define GSC_REFRESH_MAX	60
 #define GSC_CROP_MAX	8192
 #define GSC_CROP_MIN	32
 #define GSC_SCALE_MAX	4224
@@ -79,8 +61,6 @@
 #define GSC_COEF_DEPTH	3
 
 #define get_gsc_context(dev)	platform_get_drvdata(to_platform_device(dev))
-#define get_ctx_from_ippdrv(ippdrv)	container_of(ippdrv,\
-					struct gsc_context, ippdrv);
 #define gsc_read(offset)		readl(ctx->regs + (offset))
 #define gsc_write(cfg, offset)	writel(cfg, ctx->regs + (offset))
 
@@ -124,7 +104,6 @@ struct gsc_capability {
 /*
  * A structure of gsc context.
  *
- * @ippdrv: prepare initialization using ippdrv.
  * @regs_res: register resources.
  * @regs: memory mapped io registers.
  * @sysreg: handle to SYSREG block regmap.
@@ -137,11 +116,14 @@ struct gsc_capability {
  * @suspended: qos operations.
  */
 struct gsc_context {
-	struct exynos_drm_ippdrv	ippdrv;
+	struct exynos_drm_ipp ipp;
+	struct drm_device *drm_dev;
+	struct device	*dev;
+	struct exynos_drm_ipp_task	*task;
+
 	struct resource	*regs_res;
 	void __iomem	*regs;
 	struct regmap	*sysreg;
-	struct mutex	lock;
 	struct clk	*gsc_clk;
 	struct gsc_scaler	sc;
 	int	id;
@@ -438,25 +420,6 @@ static int gsc_sw_reset(struct gsc_context *ctx)
 	return 0;
 }
 
-static void gsc_set_gscblk_fimd_wb(struct gsc_context *ctx, bool enable)
-{
-	unsigned int gscblk_cfg;
-
-	if (!ctx->sysreg)
-		return;
-
-	regmap_read(ctx->sysreg, SYSREG_GSCBLK_CFG1, &gscblk_cfg);
-
-	if (enable)
-		gscblk_cfg |= GSC_BLK_DISP1WB_DEST(ctx->id) |
-				GSC_BLK_GSCL_WB_IN_SRC_SEL(ctx->id) |
-				GSC_BLK_SW_RESET_WB_DEST(ctx->id);
-	else
-		gscblk_cfg |= GSC_BLK_PXLASYNC_LO_MASK_WB(ctx->id);
-
-	regmap_write(ctx->sysreg, SYSREG_GSCBLK_CFG1, gscblk_cfg);
-}
-
 static void gsc_handle_irq(struct gsc_context *ctx, bool enable,
 		bool overflow, bool done)
 {
@@ -487,10 +450,8 @@ static void gsc_handle_irq(struct gsc_context *ctx, bool enable,
 }
 
 
-static int gsc_src_set_fmt(struct device *dev, u32 fmt)
+static int gsc_src_set_fmt(struct gsc_context *ctx, u32 fmt)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 	u32 cfg;
 
 	DRM_DEBUG_KMS("fmt[0x%x]\n", fmt);
@@ -549,7 +510,7 @@ static int gsc_src_set_fmt(struct device *dev, u32 fmt)
 			GSC_IN_YUV420_2P);
 		break;
 	default:
-		dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt);
+		dev_err(ctx->dev, "invalid target yuv order 0x%x.\n", fmt);
 		return -EINVAL;
 	}
 
@@ -558,105 +519,81 @@ static int gsc_src_set_fmt(struct device *dev, u32 fmt)
 	return 0;
 }
 
-static int gsc_src_set_transf(struct device *dev,
-		enum drm_exynos_degree degree,
-		enum drm_exynos_flip flip, bool *swap)
+static int gsc_src_set_transf(struct gsc_context *ctx, unsigned int rotation)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
+	unsigned int degree = rotation & DRM_MODE_ROTATE_MASK;
 	u32 cfg;
 
-	DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip);
-
 	cfg = gsc_read(GSC_IN_CON);
 	cfg &= ~GSC_IN_ROT_MASK;
 
 	switch (degree) {
-	case EXYNOS_DRM_DEGREE_0:
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+	case DRM_MODE_ROTATE_0:
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg |= GSC_IN_ROT_XFLIP;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg |= GSC_IN_ROT_YFLIP;
 		break;
-	case EXYNOS_DRM_DEGREE_90:
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+	case DRM_MODE_ROTATE_90:
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg |= GSC_IN_ROT_90_XFLIP;
-		else if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		else if (rotation & DRM_MODE_REFLECT_X)
 			cfg |= GSC_IN_ROT_90_YFLIP;
 		else
 			cfg |= GSC_IN_ROT_90;
 		break;
-	case EXYNOS_DRM_DEGREE_180:
+	case DRM_MODE_ROTATE_180:
 		cfg |= GSC_IN_ROT_180;
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg &= ~GSC_IN_ROT_XFLIP;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg &= ~GSC_IN_ROT_YFLIP;
 		break;
-	case EXYNOS_DRM_DEGREE_270:
+	case DRM_MODE_ROTATE_270:
 		cfg |= GSC_IN_ROT_270;
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg &= ~GSC_IN_ROT_XFLIP;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg &= ~GSC_IN_ROT_YFLIP;
 		break;
-	default:
-		dev_err(ippdrv->dev, "invalid degree value %d.\n", degree);
-		return -EINVAL;
 	}
 
 	gsc_write(cfg, GSC_IN_CON);
 
 	ctx->rotation = (cfg & GSC_IN_ROT_90) ? 1 : 0;
-	*swap = ctx->rotation;
-
 	return 0;
 }
 
-static int gsc_src_set_size(struct device *dev, int swap,
-		struct drm_exynos_pos *pos, struct drm_exynos_sz *sz)
+static int gsc_src_set_size(struct gsc_context *ctx,
+		struct exynos_drm_ipp_buffer *buf)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct drm_exynos_pos img_pos = *pos;
 	struct gsc_scaler *sc = &ctx->sc;
 	u32 cfg;
 
-	DRM_DEBUG_KMS("swap[%d]x[%d]y[%d]w[%d]h[%d]\n",
-		swap, pos->x, pos->y, pos->w, pos->h);
-
-	if (swap) {
-		img_pos.w = pos->h;
-		img_pos.h = pos->w;
-	}
-
 	/* pixel offset */
-	cfg = (GSC_SRCIMG_OFFSET_X(img_pos.x) |
-		GSC_SRCIMG_OFFSET_Y(img_pos.y));
+	cfg = (GSC_SRCIMG_OFFSET_X(buf->rect.x) |
+		GSC_SRCIMG_OFFSET_Y(buf->rect.y));
 	gsc_write(cfg, GSC_SRCIMG_OFFSET);
 
 	/* cropped size */
-	cfg = (GSC_CROPPED_WIDTH(img_pos.w) |
-		GSC_CROPPED_HEIGHT(img_pos.h));
+	cfg = (GSC_CROPPED_WIDTH(buf->rect.w) |
+		GSC_CROPPED_HEIGHT(buf->rect.h));
 	gsc_write(cfg, GSC_CROPPED_SIZE);
 
-	DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", sz->hsize, sz->vsize);
-
 	/* original size */
 	cfg = gsc_read(GSC_SRCIMG_SIZE);
 	cfg &= ~(GSC_SRCIMG_HEIGHT_MASK |
 		GSC_SRCIMG_WIDTH_MASK);
 
-	cfg |= (GSC_SRCIMG_WIDTH(sz->hsize) |
-		GSC_SRCIMG_HEIGHT(sz->vsize));
+	cfg |= (GSC_SRCIMG_WIDTH(buf->buf.width) |
+		GSC_SRCIMG_HEIGHT(buf->buf.height));
 
 	gsc_write(cfg, GSC_SRCIMG_SIZE);
 
 	cfg = gsc_read(GSC_IN_CON);
 	cfg &= ~GSC_IN_RGB_TYPE_MASK;
 
-	DRM_DEBUG_KMS("width[%d]range[%d]\n", pos->w, sc->range);
-
-	if (pos->w >= GSC_WIDTH_ITU_709)
+	if (buf->rect.w >= GSC_WIDTH_ITU_709)
 		if (sc->range)
 			cfg |= GSC_IN_RGB_HD_WIDE;
 		else
@@ -673,30 +610,15 @@ static int gsc_src_set_size(struct device *dev, int swap,
 }
 
 static int gsc_src_set_buf_seq(struct gsc_context *ctx, u32 buf_id,
-		enum drm_exynos_ipp_buf_type buf_type)
+			       bool enqueue)
 {
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	bool masked;
+	bool masked = !enqueue;
 	u32 cfg;
 	u32 mask = 0x00000001 << buf_id;
 
-	DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type);
-
 	/* mask register set */
 	cfg = gsc_read(GSC_IN_BASE_ADDR_Y_MASK);
 
-	switch (buf_type) {
-	case IPP_BUF_ENQUEUE:
-		masked = false;
-		break;
-	case IPP_BUF_DEQUEUE:
-		masked = true;
-		break;
-	default:
-		dev_err(ippdrv->dev, "invalid buf ctrl parameter.\n");
-		return -EINVAL;
-	}
-
 	/* sequence id */
 	cfg &= ~mask;
 	cfg |= masked << buf_id;
@@ -707,64 +629,19 @@ static int gsc_src_set_buf_seq(struct gsc_context *ctx, u32 buf_id,
 	return 0;
 }
 
-static int gsc_src_set_addr(struct device *dev,
-		struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id,
-		enum drm_exynos_ipp_buf_type buf_type)
+static int gsc_src_set_addr(struct gsc_context *ctx, u32 buf_id,
+			    struct exynos_drm_ipp_buffer *buf)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
-	struct drm_exynos_ipp_property *property;
-
-	if (!c_node) {
-		DRM_ERROR("failed to get c_node.\n");
-		return -EFAULT;
-	}
-
-	property = &c_node->property;
-
-	DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n",
-		property->prop_id, buf_id, buf_type);
-
-	if (buf_id > GSC_MAX_SRC) {
-		dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id);
-		return -EINVAL;
-	}
-
 	/* address register set */
-	switch (buf_type) {
-	case IPP_BUF_ENQUEUE:
-		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_Y],
-			GSC_IN_BASE_ADDR_Y(buf_id));
-		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CB],
-			GSC_IN_BASE_ADDR_CB(buf_id));
-		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CR],
-			GSC_IN_BASE_ADDR_CR(buf_id));
-		break;
-	case IPP_BUF_DEQUEUE:
-		gsc_write(0x0, GSC_IN_BASE_ADDR_Y(buf_id));
-		gsc_write(0x0, GSC_IN_BASE_ADDR_CB(buf_id));
-		gsc_write(0x0, GSC_IN_BASE_ADDR_CR(buf_id));
-		break;
-	default:
-		/* bypass */
-		break;
-	}
+	gsc_write(buf->dma_addr[0], GSC_IN_BASE_ADDR_Y(buf_id));
+	gsc_write(buf->dma_addr[1], GSC_IN_BASE_ADDR_CB(buf_id));
+	gsc_write(buf->dma_addr[2], GSC_IN_BASE_ADDR_CR(buf_id));
 
-	return gsc_src_set_buf_seq(ctx, buf_id, buf_type);
+	return gsc_src_set_buf_seq(ctx, buf_id, true);
 }
 
-static struct exynos_drm_ipp_ops gsc_src_ops = {
-	.set_fmt = gsc_src_set_fmt,
-	.set_transf = gsc_src_set_transf,
-	.set_size = gsc_src_set_size,
-	.set_addr = gsc_src_set_addr,
-};
-
-static int gsc_dst_set_fmt(struct device *dev, u32 fmt)
+static int gsc_dst_set_fmt(struct gsc_context *ctx, u32 fmt)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 	u32 cfg;
 
 	DRM_DEBUG_KMS("fmt[0x%x]\n", fmt);
@@ -819,9 +696,6 @@ static int gsc_dst_set_fmt(struct device *dev, u32 fmt)
 		cfg |= (GSC_OUT_CHROMA_ORDER_CBCR |
 			GSC_OUT_YUV420_2P);
 		break;
-	default:
-		dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt);
-		return -EINVAL;
 	}
 
 	gsc_write(cfg, GSC_OUT_CON);
@@ -829,61 +703,6 @@ static int gsc_dst_set_fmt(struct device *dev, u32 fmt)
 	return 0;
 }
 
-static int gsc_dst_set_transf(struct device *dev,
-		enum drm_exynos_degree degree,
-		enum drm_exynos_flip flip, bool *swap)
-{
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	u32 cfg;
-
-	DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip);
-
-	cfg = gsc_read(GSC_IN_CON);
-	cfg &= ~GSC_IN_ROT_MASK;
-
-	switch (degree) {
-	case EXYNOS_DRM_DEGREE_0:
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
-			cfg |= GSC_IN_ROT_XFLIP;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
-			cfg |= GSC_IN_ROT_YFLIP;
-		break;
-	case EXYNOS_DRM_DEGREE_90:
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
-			cfg |= GSC_IN_ROT_90_XFLIP;
-		else if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
-			cfg |= GSC_IN_ROT_90_YFLIP;
-		else
-			cfg |= GSC_IN_ROT_90;
-		break;
-	case EXYNOS_DRM_DEGREE_180:
-		cfg |= GSC_IN_ROT_180;
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
-			cfg &= ~GSC_IN_ROT_XFLIP;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
-			cfg &= ~GSC_IN_ROT_YFLIP;
-		break;
-	case EXYNOS_DRM_DEGREE_270:
-		cfg |= GSC_IN_ROT_270;
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
-			cfg &= ~GSC_IN_ROT_XFLIP;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
-			cfg &= ~GSC_IN_ROT_YFLIP;
-		break;
-	default:
-		dev_err(ippdrv->dev, "invalid degree value %d.\n", degree);
-		return -EINVAL;
-	}
-
-	gsc_write(cfg, GSC_IN_CON);
-
-	ctx->rotation = (cfg & GSC_IN_ROT_90) ? 1 : 0;
-	*swap = ctx->rotation;
-
-	return 0;
-}
-
 static int gsc_get_ratio_shift(u32 src, u32 dst, u32 *ratio)
 {
 	DRM_DEBUG_KMS("src[%d]dst[%d]\n", src, dst);
@@ -919,9 +738,8 @@ static void gsc_get_prescaler_shfactor(u32 hratio, u32 vratio, u32 *shfactor)
 }
 
 static int gsc_set_prescaler(struct gsc_context *ctx, struct gsc_scaler *sc,
-		struct drm_exynos_pos *src, struct drm_exynos_pos *dst)
+     struct drm_exynos_ipp_task_rect *src, struct drm_exynos_ipp_task_rect *dst)
 {
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 	u32 cfg;
 	u32 src_w, src_h, dst_w, dst_h;
 	int ret = 0;
@@ -939,13 +757,13 @@ static int gsc_set_prescaler(struct gsc_context *ctx, struct gsc_scaler *sc,
 
 	ret = gsc_get_ratio_shift(src_w, dst_w, &sc->pre_hratio);
 	if (ret) {
-		dev_err(ippdrv->dev, "failed to get ratio horizontal.\n");
+		dev_err(ctx->dev, "failed to get ratio horizontal.\n");
 		return ret;
 	}
 
 	ret = gsc_get_ratio_shift(src_h, dst_h, &sc->pre_vratio);
 	if (ret) {
-		dev_err(ippdrv->dev, "failed to get ratio vertical.\n");
+		dev_err(ctx->dev, "failed to get ratio vertical.\n");
 		return ret;
 	}
 
@@ -1039,47 +857,31 @@ static void gsc_set_scaler(struct gsc_context *ctx, struct gsc_scaler *sc)
 	gsc_write(cfg, GSC_MAIN_V_RATIO);
 }
 
-static int gsc_dst_set_size(struct device *dev, int swap,
-		struct drm_exynos_pos *pos, struct drm_exynos_sz *sz)
+static int gsc_dst_set_size(struct gsc_context *ctx,
+			    struct exynos_drm_ipp_buffer *buf)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct drm_exynos_pos img_pos = *pos;
 	struct gsc_scaler *sc = &ctx->sc;
 	u32 cfg;
 
-	DRM_DEBUG_KMS("swap[%d]x[%d]y[%d]w[%d]h[%d]\n",
-		swap, pos->x, pos->y, pos->w, pos->h);
-
-	if (swap) {
-		img_pos.w = pos->h;
-		img_pos.h = pos->w;
-	}
-
 	/* pixel offset */
-	cfg = (GSC_DSTIMG_OFFSET_X(pos->x) |
-		GSC_DSTIMG_OFFSET_Y(pos->y));
+	cfg = (GSC_DSTIMG_OFFSET_X(buf->rect.x) |
+		GSC_DSTIMG_OFFSET_Y(buf->rect.y));
 	gsc_write(cfg, GSC_DSTIMG_OFFSET);
 
 	/* scaled size */
-	cfg = (GSC_SCALED_WIDTH(img_pos.w) | GSC_SCALED_HEIGHT(img_pos.h));
+	cfg = (GSC_SCALED_WIDTH(buf->rect.w) | GSC_SCALED_HEIGHT(buf->rect.h));
 	gsc_write(cfg, GSC_SCALED_SIZE);
 
-	DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", sz->hsize, sz->vsize);
-
 	/* original size */
 	cfg = gsc_read(GSC_DSTIMG_SIZE);
-	cfg &= ~(GSC_DSTIMG_HEIGHT_MASK |
-		GSC_DSTIMG_WIDTH_MASK);
-	cfg |= (GSC_DSTIMG_WIDTH(sz->hsize) |
-		GSC_DSTIMG_HEIGHT(sz->vsize));
+	cfg &= ~(GSC_DSTIMG_HEIGHT_MASK | GSC_DSTIMG_WIDTH_MASK);
+	cfg |= (GSC_DSTIMG_WIDTH(buf->buf.width) | GSC_DSTIMG_HEIGHT(buf->buf.height));
 	gsc_write(cfg, GSC_DSTIMG_SIZE);
 
 	cfg = gsc_read(GSC_OUT_CON);
 	cfg &= ~GSC_OUT_RGB_TYPE_MASK;
 
-	DRM_DEBUG_KMS("width[%d]range[%d]\n", pos->w, sc->range);
-
-	if (pos->w >= GSC_WIDTH_ITU_709)
+	if (buf->rect.w >= GSC_WIDTH_ITU_709)
 		if (sc->range)
 			cfg |= GSC_OUT_RGB_HD_WIDE;
 		else
@@ -1112,34 +914,16 @@ static int gsc_dst_get_buf_seq(struct gsc_context *ctx)
 }
 
 static int gsc_dst_set_buf_seq(struct gsc_context *ctx, u32 buf_id,
-		enum drm_exynos_ipp_buf_type buf_type)
+			       bool enqueue)
 {
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	bool masked;
+	bool masked = !enqueue;
 	u32 cfg;
 	u32 mask = 0x00000001 << buf_id;
 	int ret = 0;
 
-	DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type);
-
-	mutex_lock(&ctx->lock);
-
 	/* mask register set */
 	cfg = gsc_read(GSC_OUT_BASE_ADDR_Y_MASK);
 
-	switch (buf_type) {
-	case IPP_BUF_ENQUEUE:
-		masked = false;
-		break;
-	case IPP_BUF_DEQUEUE:
-		masked = true;
-		break;
-	default:
-		dev_err(ippdrv->dev, "invalid buf ctrl parameter.\n");
-		ret =  -EINVAL;
-		goto err_unlock;
-	}
-
 	/* sequence id */
 	cfg &= ~mask;
 	cfg |= masked << buf_id;
@@ -1148,74 +932,27 @@ static int gsc_dst_set_buf_seq(struct gsc_context *ctx, u32 buf_id,
 	gsc_write(cfg, GSC_OUT_BASE_ADDR_CR_MASK);
 
 	/* interrupt enable */
-	if (buf_type == IPP_BUF_ENQUEUE &&
-	    gsc_dst_get_buf_seq(ctx) >= GSC_BUF_START)
+	if (enqueue && gsc_dst_get_buf_seq(ctx) >= GSC_BUF_START)
 		gsc_handle_irq(ctx, true, false, true);
 
 	/* interrupt disable */
-	if (buf_type == IPP_BUF_DEQUEUE &&
-	    gsc_dst_get_buf_seq(ctx) <= GSC_BUF_STOP)
+	if (!enqueue && gsc_dst_get_buf_seq(ctx) <= GSC_BUF_STOP)
 		gsc_handle_irq(ctx, false, false, true);
 
-err_unlock:
-	mutex_unlock(&ctx->lock);
 	return ret;
 }
 
-static int gsc_dst_set_addr(struct device *dev,
-		struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id,
-		enum drm_exynos_ipp_buf_type buf_type)
+static int gsc_dst_set_addr(struct gsc_context *ctx,
+		u32 buf_id, struct exynos_drm_ipp_buffer *buf)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
-	struct drm_exynos_ipp_property *property;
-
-	if (!c_node) {
-		DRM_ERROR("failed to get c_node.\n");
-		return -EFAULT;
-	}
-
-	property = &c_node->property;
-
-	DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n",
-		property->prop_id, buf_id, buf_type);
-
-	if (buf_id > GSC_MAX_DST) {
-		dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id);
-		return -EINVAL;
-	}
-
 	/* address register set */
-	switch (buf_type) {
-	case IPP_BUF_ENQUEUE:
-		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_Y],
-			GSC_OUT_BASE_ADDR_Y(buf_id));
-		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CB],
-			GSC_OUT_BASE_ADDR_CB(buf_id));
-		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CR],
-			GSC_OUT_BASE_ADDR_CR(buf_id));
-		break;
-	case IPP_BUF_DEQUEUE:
-		gsc_write(0x0, GSC_OUT_BASE_ADDR_Y(buf_id));
-		gsc_write(0x0, GSC_OUT_BASE_ADDR_CB(buf_id));
-		gsc_write(0x0, GSC_OUT_BASE_ADDR_CR(buf_id));
-		break;
-	default:
-		/* bypass */
-		break;
-	}
+	gsc_write(buf->dma_addr[0], GSC_OUT_BASE_ADDR_Y(buf_id));
+	gsc_write(buf->dma_addr[1], GSC_OUT_BASE_ADDR_CB(buf_id));
+	gsc_write(buf->dma_addr[2], GSC_OUT_BASE_ADDR_CR(buf_id));
 
-	return gsc_dst_set_buf_seq(ctx, buf_id, buf_type);
+	return gsc_dst_set_buf_seq(ctx, buf_id, true);
 }
 
-static struct exynos_drm_ipp_ops gsc_dst_ops = {
-	.set_fmt = gsc_dst_set_fmt,
-	.set_transf = gsc_dst_set_transf,
-	.set_size = gsc_dst_set_size,
-	.set_addr = gsc_dst_set_addr,
-};
-
 static int gsc_clk_ctrl(struct gsc_context *ctx, bool enable)
 {
 	DRM_DEBUG_KMS("enable[%d]\n", enable);
@@ -1249,20 +986,20 @@ static int gsc_get_src_buf_index(struct gsc_context *ctx)
 		}
 	}
 
+	DRM_DEBUG_KMS("cfg[0x%x]curr_index[%d]buf_id[%d]\n", cfg,
+		curr_index, buf_id);
+
 	if (buf_id == GSC_MAX_SRC) {
 		DRM_ERROR("failed to get in buffer index.\n");
 		return -EINVAL;
 	}
 
-	ret = gsc_src_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE);
+	ret = gsc_src_set_buf_seq(ctx, buf_id, false);
 	if (ret < 0) {
 		DRM_ERROR("failed to dequeue.\n");
 		return ret;
 	}
 
-	DRM_DEBUG_KMS("cfg[0x%x]curr_index[%d]buf_id[%d]\n", cfg,
-		curr_index, buf_id);
-
 	return buf_id;
 }
 
@@ -1289,7 +1026,7 @@ static int gsc_get_dst_buf_index(struct gsc_context *ctx)
 		return -EINVAL;
 	}
 
-	ret = gsc_dst_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE);
+	ret = gsc_dst_set_buf_seq(ctx, buf_id, false);
 	if (ret < 0) {
 		DRM_ERROR("failed to dequeue.\n");
 		return ret;
@@ -1304,215 +1041,54 @@ static int gsc_get_dst_buf_index(struct gsc_context *ctx)
 static irqreturn_t gsc_irq_handler(int irq, void *dev_id)
 {
 	struct gsc_context *ctx = dev_id;
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
-	struct drm_exynos_ipp_event_work *event_work =
-		c_node->event_work;
 	u32 status;
-	int buf_id[EXYNOS_DRM_OPS_MAX];
+	int err = 0;
 
 	DRM_DEBUG_KMS("gsc id[%d]\n", ctx->id);
 
 	status = gsc_read(GSC_IRQ);
 	if (status & GSC_IRQ_STATUS_OR_IRQ) {
-		dev_err(ippdrv->dev, "occurred overflow at %d, status 0x%x.\n",
+		dev_err(ctx->dev, "occurred overflow at %d, status 0x%x.\n",
 			ctx->id, status);
-		return IRQ_NONE;
+		err = -EINVAL;
 	}
 
 	if (status & GSC_IRQ_STATUS_OR_FRM_DONE) {
-		dev_dbg(ippdrv->dev, "occurred frame done at %d, status 0x%x.\n",
-			ctx->id, status);
-
-		buf_id[EXYNOS_DRM_OPS_SRC] = gsc_get_src_buf_index(ctx);
-		if (buf_id[EXYNOS_DRM_OPS_SRC] < 0)
-			return IRQ_HANDLED;
-
-		buf_id[EXYNOS_DRM_OPS_DST] = gsc_get_dst_buf_index(ctx);
-		if (buf_id[EXYNOS_DRM_OPS_DST] < 0)
-			return IRQ_HANDLED;
+		int src_buf_id, dst_buf_id;
 
-		DRM_DEBUG_KMS("buf_id_src[%d]buf_id_dst[%d]\n",
-			buf_id[EXYNOS_DRM_OPS_SRC], buf_id[EXYNOS_DRM_OPS_DST]);
-
-		event_work->ippdrv = ippdrv;
-		event_work->buf_id[EXYNOS_DRM_OPS_SRC] =
-			buf_id[EXYNOS_DRM_OPS_SRC];
-		event_work->buf_id[EXYNOS_DRM_OPS_DST] =
-			buf_id[EXYNOS_DRM_OPS_DST];
-		queue_work(ippdrv->event_workq, &event_work->work);
-	}
-
-	return IRQ_HANDLED;
-}
-
-static int gsc_init_prop_list(struct exynos_drm_ippdrv *ippdrv)
-{
-	struct drm_exynos_ipp_prop_list *prop_list = &ippdrv->prop_list;
-
-	prop_list->version = 1;
-	prop_list->writeback = 1;
-	prop_list->refresh_min = GSC_REFRESH_MIN;
-	prop_list->refresh_max = GSC_REFRESH_MAX;
-	prop_list->flip = (1 << EXYNOS_DRM_FLIP_VERTICAL) |
-				(1 << EXYNOS_DRM_FLIP_HORIZONTAL);
-	prop_list->degree = (1 << EXYNOS_DRM_DEGREE_0) |
-				(1 << EXYNOS_DRM_DEGREE_90) |
-				(1 << EXYNOS_DRM_DEGREE_180) |
-				(1 << EXYNOS_DRM_DEGREE_270);
-	prop_list->csc = 1;
-	prop_list->crop = 1;
-	prop_list->crop_max.hsize = GSC_CROP_MAX;
-	prop_list->crop_max.vsize = GSC_CROP_MAX;
-	prop_list->crop_min.hsize = GSC_CROP_MIN;
-	prop_list->crop_min.vsize = GSC_CROP_MIN;
-	prop_list->scale = 1;
-	prop_list->scale_max.hsize = GSC_SCALE_MAX;
-	prop_list->scale_max.vsize = GSC_SCALE_MAX;
-	prop_list->scale_min.hsize = GSC_SCALE_MIN;
-	prop_list->scale_min.vsize = GSC_SCALE_MIN;
-
-	return 0;
-}
-
-static inline bool gsc_check_drm_flip(enum drm_exynos_flip flip)
-{
-	switch (flip) {
-	case EXYNOS_DRM_FLIP_NONE:
-	case EXYNOS_DRM_FLIP_VERTICAL:
-	case EXYNOS_DRM_FLIP_HORIZONTAL:
-	case EXYNOS_DRM_FLIP_BOTH:
-		return true;
-	default:
-		DRM_DEBUG_KMS("invalid flip\n");
-		return false;
-	}
-}
-
-static int gsc_ippdrv_check_property(struct device *dev,
-		struct drm_exynos_ipp_property *property)
-{
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_prop_list *pp = &ippdrv->prop_list;
-	struct drm_exynos_ipp_config *config;
-	struct drm_exynos_pos *pos;
-	struct drm_exynos_sz *sz;
-	bool swap;
-	int i;
-
-	for_each_ipp_ops(i) {
-		if ((i == EXYNOS_DRM_OPS_SRC) &&
-			(property->cmd == IPP_CMD_WB))
-			continue;
-
-		config = &property->config[i];
-		pos = &config->pos;
-		sz = &config->sz;
-
-		/* check for flip */
-		if (!gsc_check_drm_flip(config->flip)) {
-			DRM_ERROR("invalid flip.\n");
-			goto err_property;
-		}
-
-		/* check for degree */
-		switch (config->degree) {
-		case EXYNOS_DRM_DEGREE_90:
-		case EXYNOS_DRM_DEGREE_270:
-			swap = true;
-			break;
-		case EXYNOS_DRM_DEGREE_0:
-		case EXYNOS_DRM_DEGREE_180:
-			swap = false;
-			break;
-		default:
-			DRM_ERROR("invalid degree.\n");
-			goto err_property;
-		}
+		dev_dbg(ctx->dev, "occurred frame done at %d, status 0x%x.\n",
+			ctx->id, status);
 
-		/* check for buffer bound */
-		if ((pos->x + pos->w > sz->hsize) ||
-			(pos->y + pos->h > sz->vsize)) {
-			DRM_ERROR("out of buf bound.\n");
-			goto err_property;
-		}
+		src_buf_id = gsc_get_src_buf_index(ctx);
+		dst_buf_id = gsc_get_dst_buf_index(ctx);
 
-		/* check for crop */
-		if ((i == EXYNOS_DRM_OPS_SRC) && (pp->crop)) {
-			if (swap) {
-				if ((pos->h < pp->crop_min.hsize) ||
-					(sz->vsize > pp->crop_max.hsize) ||
-					(pos->w < pp->crop_min.vsize) ||
-					(sz->hsize > pp->crop_max.vsize)) {
-					DRM_ERROR("out of crop size.\n");
-					goto err_property;
-				}
-			} else {
-				if ((pos->w < pp->crop_min.hsize) ||
-					(sz->hsize > pp->crop_max.hsize) ||
-					(pos->h < pp->crop_min.vsize) ||
-					(sz->vsize > pp->crop_max.vsize)) {
-					DRM_ERROR("out of crop size.\n");
-					goto err_property;
-				}
-			}
-		}
+		DRM_DEBUG_KMS("buf_id_src[%d]buf_id_dst[%d]\n",	src_buf_id,
+			      dst_buf_id);
 
-		/* check for scale */
-		if ((i == EXYNOS_DRM_OPS_DST) && (pp->scale)) {
-			if (swap) {
-				if ((pos->h < pp->scale_min.hsize) ||
-					(sz->vsize > pp->scale_max.hsize) ||
-					(pos->w < pp->scale_min.vsize) ||
-					(sz->hsize > pp->scale_max.vsize)) {
-					DRM_ERROR("out of scale size.\n");
-					goto err_property;
-				}
-			} else {
-				if ((pos->w < pp->scale_min.hsize) ||
-					(sz->hsize > pp->scale_max.hsize) ||
-					(pos->h < pp->scale_min.vsize) ||
-					(sz->vsize > pp->scale_max.vsize)) {
-					DRM_ERROR("out of scale size.\n");
-					goto err_property;
-				}
-			}
-		}
+		if (src_buf_id < 0 || dst_buf_id < 0)
+			err = -EINVAL;
 	}
 
-	return 0;
+	if (ctx->task) {
+		struct exynos_drm_ipp_task *task = ctx->task;
 
-err_property:
-	for_each_ipp_ops(i) {
-		if ((i == EXYNOS_DRM_OPS_SRC) &&
-			(property->cmd == IPP_CMD_WB))
-			continue;
-
-		config = &property->config[i];
-		pos = &config->pos;
-		sz = &config->sz;
-
-		DRM_ERROR("[%s]f[%d]r[%d]pos[%d %d %d %d]sz[%d %d]\n",
-			i ? "dst" : "src", config->flip, config->degree,
-			pos->x, pos->y, pos->w, pos->h,
-			sz->hsize, sz->vsize);
+		ctx->task = NULL;
+		pm_runtime_put(ctx->dev);
+		exynos_drm_ipp_task_done(task, err);
 	}
 
-	return -EINVAL;
+	return IRQ_HANDLED;
 }
 
-
-static int gsc_ippdrv_reset(struct device *dev)
+static int gsc_reset(struct gsc_context *ctx)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
 	struct gsc_scaler *sc = &ctx->sc;
 	int ret;
 
 	/* reset h/w block */
 	ret = gsc_sw_reset(ctx);
 	if (ret < 0) {
-		dev_err(dev, "failed to reset hardware.\n");
+		dev_err(ctx->dev, "failed to reset hardware.\n");
 		return ret;
 	}
 
@@ -1523,96 +1099,29 @@ static int gsc_ippdrv_reset(struct device *dev)
 	return 0;
 }
 
-static int gsc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd)
+static int gsc_start(struct gsc_context *ctx)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
-	struct drm_exynos_ipp_property *property;
-	struct drm_exynos_ipp_config *config;
-	struct drm_exynos_pos	img_pos[EXYNOS_DRM_OPS_MAX];
-	struct drm_exynos_ipp_set_wb set_wb;
 	u32 cfg;
-	int ret, i;
-
-	DRM_DEBUG_KMS("cmd[%d]\n", cmd);
-
-	if (!c_node) {
-		DRM_ERROR("failed to get c_node.\n");
-		return -EINVAL;
-	}
-
-	property = &c_node->property;
 
 	gsc_handle_irq(ctx, true, false, true);
 
-	for_each_ipp_ops(i) {
-		config = &property->config[i];
-		img_pos[i] = config->pos;
-	}
+	/* enable one shot */
+	cfg = gsc_read(GSC_ENABLE);
+	cfg &= ~(GSC_ENABLE_ON_CLEAR_MASK |
+		GSC_ENABLE_CLK_GATE_MODE_MASK);
+	cfg |= GSC_ENABLE_ON_CLEAR_ONESHOT;
+	gsc_write(cfg, GSC_ENABLE);
 
-	switch (cmd) {
-	case IPP_CMD_M2M:
-		/* enable one shot */
-		cfg = gsc_read(GSC_ENABLE);
-		cfg &= ~(GSC_ENABLE_ON_CLEAR_MASK |
-			GSC_ENABLE_CLK_GATE_MODE_MASK);
-		cfg |= GSC_ENABLE_ON_CLEAR_ONESHOT;
-		gsc_write(cfg, GSC_ENABLE);
-
-		/* src dma memory */
-		cfg = gsc_read(GSC_IN_CON);
-		cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK);
-		cfg |= GSC_IN_PATH_MEMORY;
-		gsc_write(cfg, GSC_IN_CON);
-
-		/* dst dma memory */
-		cfg = gsc_read(GSC_OUT_CON);
-		cfg |= GSC_OUT_PATH_MEMORY;
-		gsc_write(cfg, GSC_OUT_CON);
-		break;
-	case IPP_CMD_WB:
-		set_wb.enable = 1;
-		set_wb.refresh = property->refresh_rate;
-		gsc_set_gscblk_fimd_wb(ctx, set_wb.enable);
-		exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb);
-
-		/* src local path */
-		cfg = gsc_read(GSC_IN_CON);
-		cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK);
-		cfg |= (GSC_IN_PATH_LOCAL | GSC_IN_LOCAL_FIMD_WB);
-		gsc_write(cfg, GSC_IN_CON);
-
-		/* dst dma memory */
-		cfg = gsc_read(GSC_OUT_CON);
-		cfg |= GSC_OUT_PATH_MEMORY;
-		gsc_write(cfg, GSC_OUT_CON);
-		break;
-	case IPP_CMD_OUTPUT:
-		/* src dma memory */
-		cfg = gsc_read(GSC_IN_CON);
-		cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK);
-		cfg |= GSC_IN_PATH_MEMORY;
-		gsc_write(cfg, GSC_IN_CON);
-
-		/* dst local path */
-		cfg = gsc_read(GSC_OUT_CON);
-		cfg |= GSC_OUT_PATH_MEMORY;
-		gsc_write(cfg, GSC_OUT_CON);
-		break;
-	default:
-		ret = -EINVAL;
-		dev_err(dev, "invalid operations.\n");
-		return ret;
-	}
+	/* src dma memory */
+	cfg = gsc_read(GSC_IN_CON);
+	cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK);
+	cfg |= GSC_IN_PATH_MEMORY;
+	gsc_write(cfg, GSC_IN_CON);
 
-	ret = gsc_set_prescaler(ctx, &ctx->sc,
-		&img_pos[EXYNOS_DRM_OPS_SRC],
-		&img_pos[EXYNOS_DRM_OPS_DST]);
-	if (ret) {
-		dev_err(dev, "failed to set prescaler.\n");
-		return ret;
-	}
+	/* dst dma memory */
+	cfg = gsc_read(GSC_OUT_CON);
+	cfg |= GSC_OUT_PATH_MEMORY;
+	gsc_write(cfg, GSC_OUT_CON);
 
 	gsc_set_scaler(ctx, &ctx->sc);
 
@@ -1623,60 +1132,112 @@ static int gsc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd)
 	return 0;
 }
 
-static void gsc_ippdrv_stop(struct device *dev, enum drm_exynos_ipp_cmd cmd)
+static int gsc_commit(struct exynos_drm_ipp *ipp,
+			  struct exynos_drm_ipp_task *task)
 {
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct drm_exynos_ipp_set_wb set_wb = {0, 0};
-	u32 cfg;
+	struct gsc_context *ctx = container_of(ipp, struct gsc_context, ipp);
+
+	pm_runtime_get_sync(ctx->dev);
+	ctx->task = task;
+
+	gsc_reset(ctx);
+	gsc_src_set_fmt(ctx, task->src.buf.fourcc);
+	gsc_src_set_transf(ctx, task->transform.rotation);
+	gsc_src_set_size(ctx, &task->src);
+	gsc_src_set_addr(ctx, 0, &task->src);
+	gsc_dst_set_fmt(ctx, task->dst.buf.fourcc);
+	gsc_dst_set_size(ctx, &task->dst);
+	gsc_dst_set_addr(ctx, 0, &task->dst);
+	gsc_set_prescaler(ctx, &ctx->sc, &task->src.rect, &task->dst.rect);
+	gsc_start(ctx);
 
-	DRM_DEBUG_KMS("cmd[%d]\n", cmd);
+	return 0;
+}
 
-	switch (cmd) {
-	case IPP_CMD_M2M:
-		/* bypass */
-		break;
-	case IPP_CMD_WB:
-		gsc_set_gscblk_fimd_wb(ctx, set_wb.enable);
-		exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb);
-		break;
-	case IPP_CMD_OUTPUT:
-	default:
-		dev_err(dev, "invalid operations.\n");
-		break;
+static void gsc_abort(struct exynos_drm_ipp *ipp,
+			  struct exynos_drm_ipp_task *task)
+{
+	struct gsc_context *ctx =
+			container_of(ipp, struct gsc_context, ipp);
+
+	gsc_reset(ctx);
+	if (ctx->task) {
+		struct exynos_drm_ipp_task *task = ctx->task;
+
+		ctx->task = NULL;
+		pm_runtime_put(ctx->dev);
+		exynos_drm_ipp_task_done(task, -EIO);
 	}
+}
 
-	gsc_handle_irq(ctx, false, false, true);
+static struct exynos_drm_ipp_funcs ipp_funcs = {
+	.commit = gsc_commit,
+	.abort = gsc_abort,
+};
 
-	/* reset sequence */
-	gsc_write(0xff, GSC_OUT_BASE_ADDR_Y_MASK);
-	gsc_write(0xff, GSC_OUT_BASE_ADDR_CB_MASK);
-	gsc_write(0xff, GSC_OUT_BASE_ADDR_CR_MASK);
+static const struct exynos_drm_ipp_formats gsc_formats[] = {
+	{ IPP_SRCDST_FORMAT(XRGB8888, NULL) },
+	{ IPP_SRCDST_FORMAT(RGB565, NULL) },
+	{ IPP_SRCDST_FORMAT(BGRX8888, NULL) },
+	{ IPP_SRCDST_FORMAT(NV12, NULL) },
+	{ IPP_SRCDST_FORMAT(NV16, NULL) },
+	{ IPP_SRCDST_FORMAT(NV21, NULL) },
+	{ IPP_SRCDST_FORMAT(NV61, NULL) },
+	{ IPP_SRCDST_FORMAT(UYVY, NULL) },
+	{ IPP_SRCDST_FORMAT(VYUY, NULL) },
+	{ IPP_SRCDST_FORMAT(YUYV, NULL) },
+	{ IPP_SRCDST_FORMAT(YVYU, NULL) },
+	{ IPP_SRCDST_FORMAT(YUV420, NULL) },
+	{ IPP_SRCDST_FORMAT(YVU420, NULL) },
+	{ IPP_SRCDST_FORMAT(YUV422, NULL) },
+	{ }
+};
 
-	cfg = gsc_read(GSC_ENABLE);
-	cfg &= ~GSC_ENABLE_ON;
-	gsc_write(cfg, GSC_ENABLE);
+static int gsc_bind(struct device *dev, struct device *master, void *data)
+{
+	struct gsc_context *ctx = dev_get_drvdata(dev);
+	struct drm_device *drm_dev = data;
+	struct exynos_drm_ipp *ipp = &ctx->ipp;
+
+	ctx->drm_dev = drm_dev;
+	drm_iommu_attach_device(drm_dev, dev);
+
+	exynos_drm_ipp_register(drm_dev, ipp, &ipp_funcs,
+			   DRM_EXYNOS_IPP_CAP_CROP | DRM_EXYNOS_IPP_CAP_ROTATE |
+			   DRM_EXYNOS_IPP_CAP_SCALE | DRM_EXYNOS_IPP_CAP_CONVERT,
+			   gsc_formats, "gsc");
+
+	dev_info(dev, "The exynos gscaler is probed successfully\n");
+
+	return 0;
 }
 
+static void gsc_unbind(struct device *dev, struct device *master,
+			void *data)
+{
+	struct gsc_context *ctx = dev_get_drvdata(dev);
+
+	drm_iommu_detach_device(ctx->drm_dev, ctx->dev);
+}
+
+static const struct component_ops gsc_component_ops = {
+	.bind	= gsc_bind,
+	.unbind = gsc_unbind,
+};
+
+
 static int gsc_probe(struct platform_device *pdev)
 {
 	struct device *dev = &pdev->dev;
 	struct gsc_context *ctx;
 	struct resource *res;
-	struct exynos_drm_ippdrv *ippdrv;
 	int ret;
 
 	ctx = devm_kzalloc(dev, sizeof(*ctx), GFP_KERNEL);
 	if (!ctx)
 		return -ENOMEM;
 
-	if (dev->of_node) {
-		ctx->sysreg = syscon_regmap_lookup_by_phandle(dev->of_node,
-							"samsung,sysreg");
-		if (IS_ERR(ctx->sysreg)) {
-			dev_warn(dev, "failed to get system register.\n");
-			ctx->sysreg = NULL;
-		}
-	}
+	ctx->dev = dev;
 
 	/* clock control */
 	ctx->gsc_clk = devm_clk_get(dev, "gscl");
@@ -1699,8 +1260,8 @@ static int gsc_probe(struct platform_device *pdev)
 	}
 
 	ctx->irq = res->start;
-	ret = devm_request_threaded_irq(dev, ctx->irq, NULL, gsc_irq_handler,
-		IRQF_ONESHOT, "drm_gsc", ctx);
+	ret = devm_request_irq(dev, ctx->irq, gsc_irq_handler, 0,
+			       dev_name(dev), ctx);
 	if (ret < 0) {
 		dev_err(dev, "failed to request irq.\n");
 		return ret;
@@ -1709,38 +1270,19 @@ static int gsc_probe(struct platform_device *pdev)
 	/* context initailization */
 	ctx->id = pdev->id;
 
-	ippdrv = &ctx->ippdrv;
-	ippdrv->dev = dev;
-	ippdrv->ops[EXYNOS_DRM_OPS_SRC] = &gsc_src_ops;
-	ippdrv->ops[EXYNOS_DRM_OPS_DST] = &gsc_dst_ops;
-	ippdrv->check_property = gsc_ippdrv_check_property;
-	ippdrv->reset = gsc_ippdrv_reset;
-	ippdrv->start = gsc_ippdrv_start;
-	ippdrv->stop = gsc_ippdrv_stop;
-	ret = gsc_init_prop_list(ippdrv);
-	if (ret < 0) {
-		dev_err(dev, "failed to init property list.\n");
-		return ret;
-	}
-
-	DRM_DEBUG_KMS("id[%d]ippdrv[%pK]\n", ctx->id, ippdrv);
-
-	mutex_init(&ctx->lock);
 	platform_set_drvdata(pdev, ctx);
 
 	pm_runtime_enable(dev);
 
-	ret = exynos_drm_ippdrv_register(ippdrv);
-	if (ret < 0) {
-		dev_err(dev, "failed to register drm gsc device.\n");
-		goto err_ippdrv_register;
-	}
+	ret = component_add(dev, &gsc_component_ops);
+	if (ret)
+		goto err_pm_dis;
 
 	dev_info(dev, "drm gsc registered successfully.\n");
 
 	return 0;
 
-err_ippdrv_register:
+err_pm_dis:
 	pm_runtime_disable(dev);
 	return ret;
 }
@@ -1748,11 +1290,6 @@ static int gsc_probe(struct platform_device *pdev)
 static int gsc_remove(struct platform_device *pdev)
 {
 	struct device *dev = &pdev->dev;
-	struct gsc_context *ctx = get_gsc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-
-	exynos_drm_ippdrv_unregister(ippdrv);
-	mutex_destroy(&ctx->lock);
 
 	pm_runtime_set_suspended(dev);
 	pm_runtime_disable(dev);
diff --git a/drivers/gpu/drm/exynos/exynos_drm_gsc.h b/drivers/gpu/drm/exynos/exynos_drm_gsc.h
deleted file mode 100644
index 29ec1c5efcf2..000000000000
--- a/drivers/gpu/drm/exynos/exynos_drm_gsc.h
+++ /dev/null
@@ -1,24 +0,0 @@
-/*
- * Copyright (c) 2012 Samsung Electronics Co., Ltd.
- *
- * Authors:
- *	Eunchul Kim <chulspro.kim@samsung.com>
- *	Jinyoung Jeon <jy0.jeon@samsung.com>
- *	Sangmin Lee <lsmin.lee@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.
- */
-
-#ifndef _EXYNOS_DRM_GSC_H_
-#define _EXYNOS_DRM_GSC_H_
-
-/*
- * TODO
- * FIMD output interface notifier callback.
- * Mixer output interface notifier callback.
- */
-
-#endif /* _EXYNOS_DRM_GSC_H_ */
-- 
1.9.1

[PATCH 5/6] drm/exynos: Add generic support for devices shared with V4L2 subsystem

[Marek Szyprowski]

Some hardware modules, like FIMC in Exynos4 series are shared between
V4L2 (camera support) and DRM (memory-to-memory processing) subsystems.
This patch provides a simple check to let such drivers to be used in the
driver components framework.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/gpu/drm/exynos/exynos_drm_drv.c | 17 ++++++++++++++++-
 drivers/gpu/drm/exynos/exynos_drm_drv.h |  2 ++
 2 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
index 27242af103ea..6b46d7aab0c6 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
@@ -228,6 +228,7 @@ struct exynos_drm_driver_info {
 #define DRM_COMPONENT_DRIVER	BIT(0)	/* supports component framework */
 #define DRM_VIRTUAL_DEVICE	BIT(1)	/* create virtual platform device */
 #define DRM_DMA_DEVICE		BIT(2)	/* can be used for dma allocations */
+#define DRM_SHARED_DEVICE	BIT(3)	/* devices shared with V4L2 subsystem */
 
 #define DRV_PTR(drv, cond) (IS_ENABLED(cond) ? &drv : NULL)
 
@@ -279,6 +280,17 @@ struct exynos_drm_driver_info {
 	}
 };
 
+int exynos_drm_check_shared_device(struct device *dev)
+{
+	/*
+	 * Exynos DRM drivers handle only devices that support
+	 * the LCD Writeback data path, rest is handled by V4L2 driver
+	 */
+	if (!of_property_read_bool(dev->of_node, "samsung,lcd-wb"))
+		return -ENODEV;
+	return 0;
+}
+
 static int compare_dev(struct device *dev, void *data)
 {
 	return dev == (struct device *)data;
@@ -300,7 +312,10 @@ static struct component_match *exynos_drm_match_add(struct device *dev)
 					    &info->driver->driver,
 					    (void *)platform_bus_type.match))) {
 			put_device(p);
-			component_match_add(dev, &match, compare_dev, d);
+
+			if (!(info->flags & DRM_SHARED_DEVICE) ||
+			    exynos_drm_check_shared_device(d) == 0)
+				component_match_add(dev, &match, compare_dev, d);
 			p = d;
 		}
 		put_device(p);
diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h b/drivers/gpu/drm/exynos/exynos_drm_drv.h
index 21f4271c012a..ac4ac73a6edb 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_drv.h
+++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h
@@ -274,6 +274,8 @@ static inline int exynos_dpi_bind(struct drm_device *dev,
 }
 #endif
 
+int exynos_drm_check_shared_device(struct device *dev);
+
 int exynos_atomic_commit(struct drm_device *dev, struct drm_atomic_state *state,
 			 bool nonblock);
 int exynos_atomic_check(struct drm_device *dev, struct drm_atomic_state *state);
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 6/6] drm/exynos: fimc: Convert driver to IPP v2 core API

[Marek Szyprowski]

This patch adapts Exynos DRM rotator driver to new IPP v2 core API.
The side effect of this conversion is a switch to driver component API
to register properly in the Exynos DRM core.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 drivers/gpu/drm/exynos/Kconfig           |   3 +-
 drivers/gpu/drm/exynos/exynos_drm_drv.c  |   1 +
 drivers/gpu/drm/exynos/exynos_drm_fimc.c | 893 +++++++++----------------------
 drivers/gpu/drm/exynos/exynos_drm_fimc.h |  23 -
 4 files changed, 250 insertions(+), 670 deletions(-)
 delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_fimc.h

diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig
index 4bb9edb00601..4a4582202e60 100644
--- a/drivers/gpu/drm/exynos/Kconfig
+++ b/drivers/gpu/drm/exynos/Kconfig
@@ -99,7 +99,8 @@ config DRM_EXYNOS_IPP
 
 config DRM_EXYNOS_FIMC
 	bool "FIMC"
-	depends on BROKEN && MFD_SYSCON
+	depends on MFD_SYSCON
+	select DRM_EXYNOS_IPP
 	help
 	  Choose this option if you want to use Exynos FIMC for DRM.
 
diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
index 6b46d7aab0c6..775998693477 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
@@ -268,6 +268,7 @@ struct exynos_drm_driver_info {
 		DRV_PTR(g2d_driver, CONFIG_DRM_EXYNOS_G2D),
 	}, {
 		DRV_PTR(fimc_driver, CONFIG_DRM_EXYNOS_FIMC),
+		DRM_COMPONENT_DRIVER | DRM_SHARED_DEVICE,
 	}, {
 		DRV_PTR(rotator_driver, CONFIG_DRM_EXYNOS_ROTATOR),
 		DRM_COMPONENT_DRIVER
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimc.c b/drivers/gpu/drm/exynos/exynos_drm_fimc.c
index 5b18b5c5fdf2..f38268bb8148 100644
--- a/drivers/gpu/drm/exynos/exynos_drm_fimc.c
+++ b/drivers/gpu/drm/exynos/exynos_drm_fimc.c
@@ -12,6 +12,7 @@
  *
  */
 #include <linux/kernel.h>
+#include <linux/component.h>
 #include <linux/platform_device.h>
 #include <linux/mfd/syscon.h>
 #include <linux/regmap.h>
@@ -24,8 +25,9 @@
 #include <drm/exynos_drm.h>
 #include "regs-fimc.h"
 #include "exynos_drm_drv.h"
+#include "exynos_drm_iommu.h"
 #include "exynos_drm_ipp.h"
-#include "exynos_drm_fimc.h"
+
 
 /*
  * FIMC stands for Fully Interactive Mobile Camera and
@@ -33,23 +35,6 @@
  * input DMA reads image data from the memory.
  * output DMA writes image data to memory.
  * FIMC supports image rotation and image effect functions.
- *
- * M2M operation : supports crop/scale/rotation/csc so on.
- * Memory ----> FIMC H/W ----> Memory.
- * Writeback operation : supports cloned screen with FIMD.
- * FIMD ----> FIMC H/W ----> Memory.
- * Output operation : supports direct display using local path.
- * Memory ----> FIMC H/W ----> FIMD.
- */
-
-/*
- * TODO
- * 1. check suspend/resume api if needed.
- * 2. need to check use case platform_device_id.
- * 3. check src/dst size with, height.
- * 4. added check_prepare api for right register.
- * 5. need to add supported list in prop_list.
- * 6. check prescaler/scaler optimization.
  */
 
 #define FIMC_MAX_DEVS	4
@@ -59,29 +44,18 @@
 #define FIMC_BUF_STOP	1
 #define FIMC_BUF_START	2
 #define FIMC_WIDTH_ITU_709	1280
-#define FIMC_REFRESH_MAX	60
-#define FIMC_REFRESH_MIN	12
 #define FIMC_CROP_MAX	8192
 #define FIMC_CROP_MIN	32
 #define FIMC_SCALE_MAX	4224
 #define FIMC_SCALE_MIN	32
 
 #define get_fimc_context(dev)	platform_get_drvdata(to_platform_device(dev))
-#define get_ctx_from_ippdrv(ippdrv)	container_of(ippdrv,\
-					struct fimc_context, ippdrv);
-enum fimc_wb {
-	FIMC_WB_NONE,
-	FIMC_WB_A,
-	FIMC_WB_B,
-};
 
 enum {
 	FIMC_CLK_LCLK,
 	FIMC_CLK_GATE,
 	FIMC_CLK_WB_A,
 	FIMC_CLK_WB_B,
-	FIMC_CLK_MUX,
-	FIMC_CLK_PARENT,
 	FIMC_CLKS_MAX
 };
 
@@ -90,8 +64,6 @@ enum {
 	[FIMC_CLK_GATE]   = "fimc",
 	[FIMC_CLK_WB_A]   = "pxl_async0",
 	[FIMC_CLK_WB_B]   = "pxl_async1",
-	[FIMC_CLK_MUX]    = "mux",
-	[FIMC_CLK_PARENT] = "parent",
 };
 
 #define FIMC_DEFAULT_LCLK_FREQUENCY 133000000UL
@@ -141,13 +113,10 @@ struct fimc_capability {
 /*
  * A structure of fimc context.
  *
- * @ippdrv: prepare initialization using ippdrv.
  * @regs_res: register resources.
  * @regs: memory mapped io registers.
  * @lock: locking of operations.
  * @clocks: fimc clocks.
- * @clk_frequency: LCLK clock frequency.
- * @sysreg: handle to SYSREG block regmap.
  * @sc: scaler infomations.
  * @pol: porarity of writeback.
  * @id: fimc id.
@@ -155,13 +124,15 @@ struct fimc_capability {
  * @suspended: qos operations.
  */
 struct fimc_context {
-	struct exynos_drm_ippdrv	ippdrv;
+	struct exynos_drm_ipp ipp;
+	struct drm_device *drm_dev;
+	struct device	*dev;
+	struct exynos_drm_ipp_task	*task;
+
 	struct resource	*regs_res;
 	void __iomem	*regs;
 	spinlock_t	lock;
 	struct clk	*clocks[FIMC_CLKS_MAX];
-	u32		clk_frequency;
-	struct regmap	*sysreg;
 	struct fimc_scaler	sc;
 	int	id;
 	int	irq;
@@ -217,19 +188,10 @@ static void fimc_sw_reset(struct fimc_context *ctx)
 	fimc_write(ctx, 0x0, EXYNOS_CIFCNTSEQ);
 }
 
-static int fimc_set_camblk_fimd0_wb(struct fimc_context *ctx)
-{
-	return regmap_update_bits(ctx->sysreg, SYSREG_CAMERA_BLK,
-				  SYSREG_FIMD0WB_DEST_MASK,
-				  ctx->id << SYSREG_FIMD0WB_DEST_SHIFT);
-}
-
-static void fimc_set_type_ctrl(struct fimc_context *ctx, enum fimc_wb wb)
+static void fimc_set_type_ctrl(struct fimc_context *ctx)
 {
 	u32 cfg;
 
-	DRM_DEBUG_KMS("wb[%d]\n", wb);
-
 	cfg = fimc_read(ctx, EXYNOS_CIGCTRL);
 	cfg &= ~(EXYNOS_CIGCTRL_TESTPATTERN_MASK |
 		EXYNOS_CIGCTRL_SELCAM_ITU_MASK |
@@ -238,23 +200,10 @@ static void fimc_set_type_ctrl(struct fimc_context *ctx, enum fimc_wb wb)
 		EXYNOS_CIGCTRL_SELWB_CAMIF_MASK |
 		EXYNOS_CIGCTRL_SELWRITEBACK_MASK);
 
-	switch (wb) {
-	case FIMC_WB_A:
-		cfg |= (EXYNOS_CIGCTRL_SELWRITEBACK_A |
-			EXYNOS_CIGCTRL_SELWB_CAMIF_WRITEBACK);
-		break;
-	case FIMC_WB_B:
-		cfg |= (EXYNOS_CIGCTRL_SELWRITEBACK_B |
-			EXYNOS_CIGCTRL_SELWB_CAMIF_WRITEBACK);
-		break;
-	case FIMC_WB_NONE:
-	default:
-		cfg |= (EXYNOS_CIGCTRL_SELCAM_ITU_A |
-			EXYNOS_CIGCTRL_SELWRITEBACK_A |
-			EXYNOS_CIGCTRL_SELCAM_MIPI_A |
-			EXYNOS_CIGCTRL_SELCAM_FIMC_ITU);
-		break;
-	}
+	cfg |= (EXYNOS_CIGCTRL_SELCAM_ITU_A |
+		EXYNOS_CIGCTRL_SELWRITEBACK_A |
+		EXYNOS_CIGCTRL_SELCAM_MIPI_A |
+		EXYNOS_CIGCTRL_SELCAM_FIMC_ITU);
 
 	fimc_write(ctx, cfg, EXYNOS_CIGCTRL);
 }
@@ -296,7 +245,6 @@ static void fimc_clear_irq(struct fimc_context *ctx)
 
 static bool fimc_check_ovf(struct fimc_context *ctx)
 {
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 	u32 status, flag;
 
 	status = fimc_read(ctx, EXYNOS_CISTATUS);
@@ -310,7 +258,7 @@ static bool fimc_check_ovf(struct fimc_context *ctx)
 			EXYNOS_CIWDOFST_CLROVFIY | EXYNOS_CIWDOFST_CLROVFICB |
 			EXYNOS_CIWDOFST_CLROVFICR);
 
-		dev_err(ippdrv->dev, "occurred overflow at %d, status 0x%x.\n",
+		dev_err(ctx->dev, "occurred overflow at %d, status 0x%x.\n",
 			ctx->id, status);
 		return true;
 	}
@@ -379,7 +327,6 @@ static void fimc_handle_lastend(struct fimc_context *ctx, bool enable)
 
 static int fimc_src_set_fmt_order(struct fimc_context *ctx, u32 fmt)
 {
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 	u32 cfg;
 
 	DRM_DEBUG_KMS("fmt[0x%x]\n", fmt);
@@ -439,7 +386,7 @@ static int fimc_src_set_fmt_order(struct fimc_context *ctx, u32 fmt)
 			EXYNOS_MSCTRL_C_INT_IN_2PLANE);
 		break;
 	default:
-		dev_err(ippdrv->dev, "invalid source yuv order 0x%x.\n", fmt);
+		dev_err(ctx->dev, "invalid source yuv order 0x%x.\n", fmt);
 		return -EINVAL;
 	}
 
@@ -448,10 +395,8 @@ static int fimc_src_set_fmt_order(struct fimc_context *ctx, u32 fmt)
 	return 0;
 }
 
-static int fimc_src_set_fmt(struct device *dev, u32 fmt)
+static int fimc_src_set_fmt(struct fimc_context *ctx, u32 fmt)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 	u32 cfg;
 
 	DRM_DEBUG_KMS("fmt[0x%x]\n", fmt);
@@ -486,7 +431,7 @@ static int fimc_src_set_fmt(struct device *dev, u32 fmt)
 		cfg |= EXYNOS_MSCTRL_INFORMAT_YCBCR420;
 		break;
 	default:
-		dev_err(ippdrv->dev, "invalid source format 0x%x.\n", fmt);
+		dev_err(ctx->dev, "invalid source format 0x%x.\n", fmt);
 		return -EINVAL;
 	}
 
@@ -502,15 +447,12 @@ static int fimc_src_set_fmt(struct device *dev, u32 fmt)
 	return fimc_src_set_fmt_order(ctx, fmt);
 }
 
-static int fimc_src_set_transf(struct device *dev,
-		enum drm_exynos_degree degree,
-		enum drm_exynos_flip flip, bool *swap)
+static int fimc_src_set_transf(struct fimc_context *ctx, unsigned int rotation)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
+	unsigned int degree = rotation & DRM_MODE_ROTATE_MASK;
 	u32 cfg1, cfg2;
 
-	DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip);
+	DRM_DEBUG_KMS("rotation[%x]\n", rotation);
 
 	cfg1 = fimc_read(ctx, EXYNOS_MSCTRL);
 	cfg1 &= ~(EXYNOS_MSCTRL_FLIP_X_MIRROR |
@@ -520,61 +462,58 @@ static int fimc_src_set_transf(struct device *dev,
 	cfg2 &= ~EXYNOS_CITRGFMT_INROT90_CLOCKWISE;
 
 	switch (degree) {
-	case EXYNOS_DRM_DEGREE_0:
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+	case DRM_MODE_ROTATE_0:
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg1 |= EXYNOS_MSCTRL_FLIP_X_MIRROR;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg1 |= EXYNOS_MSCTRL_FLIP_Y_MIRROR;
 		break;
-	case EXYNOS_DRM_DEGREE_90:
+	case DRM_MODE_ROTATE_90:
 		cfg2 |= EXYNOS_CITRGFMT_INROT90_CLOCKWISE;
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg1 |= EXYNOS_MSCTRL_FLIP_X_MIRROR;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg1 |= EXYNOS_MSCTRL_FLIP_Y_MIRROR;
 		break;
-	case EXYNOS_DRM_DEGREE_180:
+	case DRM_MODE_ROTATE_180:
 		cfg1 |= (EXYNOS_MSCTRL_FLIP_X_MIRROR |
 			EXYNOS_MSCTRL_FLIP_Y_MIRROR);
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg1 &= ~EXYNOS_MSCTRL_FLIP_X_MIRROR;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg1 &= ~EXYNOS_MSCTRL_FLIP_Y_MIRROR;
 		break;
-	case EXYNOS_DRM_DEGREE_270:
+	case DRM_MODE_ROTATE_270:
 		cfg1 |= (EXYNOS_MSCTRL_FLIP_X_MIRROR |
 			EXYNOS_MSCTRL_FLIP_Y_MIRROR);
 		cfg2 |= EXYNOS_CITRGFMT_INROT90_CLOCKWISE;
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg1 &= ~EXYNOS_MSCTRL_FLIP_X_MIRROR;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg1 &= ~EXYNOS_MSCTRL_FLIP_Y_MIRROR;
 		break;
-	default:
-		dev_err(ippdrv->dev, "invalid degree value %d.\n", degree);
-		return -EINVAL;
 	}
 
 	fimc_write(ctx, cfg1, EXYNOS_MSCTRL);
 	fimc_write(ctx, cfg2, EXYNOS_CITRGFMT);
-	*swap = (cfg2 & EXYNOS_CITRGFMT_INROT90_CLOCKWISE) ? 1 : 0;
 
 	return 0;
 }
 
 static int fimc_set_window(struct fimc_context *ctx,
-		struct drm_exynos_pos *pos, struct drm_exynos_sz *sz)
+			   struct exynos_drm_ipp_buffer *buf)
 {
 	u32 cfg, h1, h2, v1, v2;
 
 	/* cropped image */
-	h1 = pos->x;
-	h2 = sz->hsize - pos->w - pos->x;
-	v1 = pos->y;
-	v2 = sz->vsize - pos->h - pos->y;
+	h1 = buf->rect.x;
+	h2 = buf->buf.width - buf->rect.w - buf->rect.x;
+	v1 = buf->rect.y;
+	v2 = buf->buf.height - buf->rect.h - buf->rect.y;
 
 	DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]hsize[%d]vsize[%d]\n",
-		pos->x, pos->y, pos->w, pos->h, sz->hsize, sz->vsize);
+		buf->rect.x, buf->rect.y, buf->rect.w, buf->rect.h,
+		buf->buf.width, buf->buf.height);
 	DRM_DEBUG_KMS("h1[%d]h2[%d]v1[%d]v2[%d]\n", h1, h2, v1, v2);
 
 	/*
@@ -596,38 +535,28 @@ static int fimc_set_window(struct fimc_context *ctx,
 	return 0;
 }
 
-static int fimc_src_set_size(struct device *dev, int swap,
-		struct drm_exynos_pos *pos, struct drm_exynos_sz *sz)
+static int fimc_src_set_size(struct fimc_context *ctx,
+			     struct exynos_drm_ipp_buffer *buf)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct drm_exynos_pos img_pos = *pos;
-	struct drm_exynos_sz img_sz = *sz;
 	u32 cfg;
 
-	DRM_DEBUG_KMS("swap[%d]hsize[%d]vsize[%d]\n",
-		swap, sz->hsize, sz->vsize);
+	DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", buf->buf.width, buf->buf.height);
 
 	/* original size */
-	cfg = (EXYNOS_ORGISIZE_HORIZONTAL(img_sz.hsize) |
-		EXYNOS_ORGISIZE_VERTICAL(img_sz.vsize));
+	cfg = (EXYNOS_ORGISIZE_HORIZONTAL(buf->buf.width) |
+		EXYNOS_ORGISIZE_VERTICAL(buf->buf.height));
 
 	fimc_write(ctx, cfg, EXYNOS_ORGISIZE);
 
-	DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]\n", pos->x, pos->y, pos->w, pos->h);
-
-	if (swap) {
-		img_pos.w = pos->h;
-		img_pos.h = pos->w;
-		img_sz.hsize = sz->vsize;
-		img_sz.vsize = sz->hsize;
-	}
+	DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]\n", buf->rect.x, buf->rect.y,
+		buf->rect.w, buf->rect.h);
 
 	/* set input DMA image size */
 	cfg = fimc_read(ctx, EXYNOS_CIREAL_ISIZE);
 	cfg &= ~(EXYNOS_CIREAL_ISIZE_HEIGHT_MASK |
 		EXYNOS_CIREAL_ISIZE_WIDTH_MASK);
-	cfg |= (EXYNOS_CIREAL_ISIZE_WIDTH(img_pos.w) |
-		EXYNOS_CIREAL_ISIZE_HEIGHT(img_pos.h));
+	cfg |= (EXYNOS_CIREAL_ISIZE_WIDTH(buf->rect.w) |
+		EXYNOS_CIREAL_ISIZE_HEIGHT(buf->rect.h));
 	fimc_write(ctx, cfg, EXYNOS_CIREAL_ISIZE);
 
 	/*
@@ -635,91 +564,36 @@ static int fimc_src_set_size(struct device *dev, int swap,
 	 * for now, we support only ITU601 8 bit mode
 	 */
 	cfg = (EXYNOS_CISRCFMT_ITU601_8BIT |
-		EXYNOS_CISRCFMT_SOURCEHSIZE(img_sz.hsize) |
-		EXYNOS_CISRCFMT_SOURCEVSIZE(img_sz.vsize));
+		EXYNOS_CISRCFMT_SOURCEHSIZE(buf->buf.width) |
+		EXYNOS_CISRCFMT_SOURCEVSIZE(buf->buf.height));
 	fimc_write(ctx, cfg, EXYNOS_CISRCFMT);
 
 	/* offset Y(RGB), Cb, Cr */
-	cfg = (EXYNOS_CIIYOFF_HORIZONTAL(img_pos.x) |
-		EXYNOS_CIIYOFF_VERTICAL(img_pos.y));
+	cfg = (EXYNOS_CIIYOFF_HORIZONTAL(buf->rect.x) |
+		EXYNOS_CIIYOFF_VERTICAL(buf->rect.y));
 	fimc_write(ctx, cfg, EXYNOS_CIIYOFF);
-	cfg = (EXYNOS_CIICBOFF_HORIZONTAL(img_pos.x) |
-		EXYNOS_CIICBOFF_VERTICAL(img_pos.y));
+	cfg = (EXYNOS_CIICBOFF_HORIZONTAL(buf->rect.x) |
+		EXYNOS_CIICBOFF_VERTICAL(buf->rect.y));
 	fimc_write(ctx, cfg, EXYNOS_CIICBOFF);
-	cfg = (EXYNOS_CIICROFF_HORIZONTAL(img_pos.x) |
-		EXYNOS_CIICROFF_VERTICAL(img_pos.y));
+	cfg = (EXYNOS_CIICROFF_HORIZONTAL(buf->rect.x) |
+		EXYNOS_CIICROFF_VERTICAL(buf->rect.y));
 	fimc_write(ctx, cfg, EXYNOS_CIICROFF);
 
-	return fimc_set_window(ctx, &img_pos, &img_sz);
+	return fimc_set_window(ctx, buf);
 }
 
-static int fimc_src_set_addr(struct device *dev,
-		struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id,
-		enum drm_exynos_ipp_buf_type buf_type)
+static int fimc_src_set_addr(struct fimc_context *ctx,
+			     struct exynos_drm_ipp_buffer *buf)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
-	struct drm_exynos_ipp_property *property;
-	struct drm_exynos_ipp_config *config;
-
-	if (!c_node) {
-		DRM_ERROR("failed to get c_node.\n");
-		return -EINVAL;
-	}
-
-	property = &c_node->property;
-
-	DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n",
-		property->prop_id, buf_id, buf_type);
-
-	if (buf_id > FIMC_MAX_SRC) {
-		dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id);
-		return -ENOMEM;
-	}
-
-	/* address register set */
-	switch (buf_type) {
-	case IPP_BUF_ENQUEUE:
-		config = &property->config[EXYNOS_DRM_OPS_SRC];
-		fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_Y],
-			EXYNOS_CIIYSA0);
-
-		if (config->fmt == DRM_FORMAT_YVU420) {
-			fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CR],
-				EXYNOS_CIICBSA0);
-			fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CB],
-				EXYNOS_CIICRSA0);
-		} else {
-			fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CB],
-				EXYNOS_CIICBSA0);
-			fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CR],
-				EXYNOS_CIICRSA0);
-		}
-		break;
-	case IPP_BUF_DEQUEUE:
-		fimc_write(ctx, 0x0, EXYNOS_CIIYSA0);
-		fimc_write(ctx, 0x0, EXYNOS_CIICBSA0);
-		fimc_write(ctx, 0x0, EXYNOS_CIICRSA0);
-		break;
-	default:
-		/* bypass */
-		break;
-	}
+	fimc_write(ctx, buf->dma_addr[0], EXYNOS_CIIYSA(0));
+	fimc_write(ctx, buf->dma_addr[1], EXYNOS_CIICBSA(0));
+	fimc_write(ctx, buf->dma_addr[2], EXYNOS_CIICRSA(0));
 
 	return 0;
 }
 
-static struct exynos_drm_ipp_ops fimc_src_ops = {
-	.set_fmt = fimc_src_set_fmt,
-	.set_transf = fimc_src_set_transf,
-	.set_size = fimc_src_set_size,
-	.set_addr = fimc_src_set_addr,
-};
-
 static int fimc_dst_set_fmt_order(struct fimc_context *ctx, u32 fmt)
 {
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 	u32 cfg;
 
 	DRM_DEBUG_KMS("fmt[0x%x]\n", fmt);
@@ -785,7 +659,7 @@ static int fimc_dst_set_fmt_order(struct fimc_context *ctx, u32 fmt)
 		cfg |= EXYNOS_CIOCTRL_YCBCR_2PLANE;
 		break;
 	default:
-		dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt);
+		dev_err(ctx->dev, "invalid target yuv order 0x%x.\n", fmt);
 		return -EINVAL;
 	}
 
@@ -794,10 +668,8 @@ static int fimc_dst_set_fmt_order(struct fimc_context *ctx, u32 fmt)
 	return 0;
 }
 
-static int fimc_dst_set_fmt(struct device *dev, u32 fmt)
+static int fimc_dst_set_fmt(struct fimc_context *ctx, u32 fmt)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 	u32 cfg;
 
 	DRM_DEBUG_KMS("fmt[0x%x]\n", fmt);
@@ -838,7 +710,7 @@ static int fimc_dst_set_fmt(struct device *dev, u32 fmt)
 			cfg |= EXYNOS_CITRGFMT_OUTFORMAT_YCBCR420;
 			break;
 		default:
-			dev_err(ippdrv->dev, "invalid target format 0x%x.\n",
+			dev_err(ctx->dev, "invalid target format 0x%x.\n",
 				fmt);
 			return -EINVAL;
 		}
@@ -856,66 +728,58 @@ static int fimc_dst_set_fmt(struct device *dev, u32 fmt)
 	return fimc_dst_set_fmt_order(ctx, fmt);
 }
 
-static int fimc_dst_set_transf(struct device *dev,
-		enum drm_exynos_degree degree,
-		enum drm_exynos_flip flip, bool *swap)
+static int fimc_dst_set_transf(struct fimc_context *ctx, unsigned int rotation)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
+	unsigned int degree = rotation & DRM_MODE_ROTATE_MASK;
 	u32 cfg;
 
-	DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip);
+	DRM_DEBUG_KMS("rotation[0x%x]\n", rotation);
 
 	cfg = fimc_read(ctx, EXYNOS_CITRGFMT);
 	cfg &= ~EXYNOS_CITRGFMT_FLIP_MASK;
 	cfg &= ~EXYNOS_CITRGFMT_OUTROT90_CLOCKWISE;
 
 	switch (degree) {
-	case EXYNOS_DRM_DEGREE_0:
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+	case DRM_MODE_ROTATE_0:
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg |= EXYNOS_CITRGFMT_FLIP_X_MIRROR;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg |= EXYNOS_CITRGFMT_FLIP_Y_MIRROR;
 		break;
-	case EXYNOS_DRM_DEGREE_90:
+	case DRM_MODE_ROTATE_90:
 		cfg |= EXYNOS_CITRGFMT_OUTROT90_CLOCKWISE;
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg |= EXYNOS_CITRGFMT_FLIP_X_MIRROR;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg |= EXYNOS_CITRGFMT_FLIP_Y_MIRROR;
 		break;
-	case EXYNOS_DRM_DEGREE_180:
+	case DRM_MODE_ROTATE_180:
 		cfg |= (EXYNOS_CITRGFMT_FLIP_X_MIRROR |
 			EXYNOS_CITRGFMT_FLIP_Y_MIRROR);
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg &= ~EXYNOS_CITRGFMT_FLIP_X_MIRROR;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg &= ~EXYNOS_CITRGFMT_FLIP_Y_MIRROR;
 		break;
-	case EXYNOS_DRM_DEGREE_270:
+	case DRM_MODE_ROTATE_270:
 		cfg |= (EXYNOS_CITRGFMT_OUTROT90_CLOCKWISE |
 			EXYNOS_CITRGFMT_FLIP_X_MIRROR |
 			EXYNOS_CITRGFMT_FLIP_Y_MIRROR);
-		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
+		if (rotation & DRM_MODE_REFLECT_X)
 			cfg &= ~EXYNOS_CITRGFMT_FLIP_X_MIRROR;
-		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
+		if (rotation & DRM_MODE_REFLECT_Y)
 			cfg &= ~EXYNOS_CITRGFMT_FLIP_Y_MIRROR;
 		break;
-	default:
-		dev_err(ippdrv->dev, "invalid degree value %d.\n", degree);
-		return -EINVAL;
 	}
 
 	fimc_write(ctx, cfg, EXYNOS_CITRGFMT);
-	*swap = (cfg & EXYNOS_CITRGFMT_OUTROT90_CLOCKWISE) ? 1 : 0;
 
 	return 0;
 }
 
 static int fimc_set_prescaler(struct fimc_context *ctx, struct fimc_scaler *sc,
-		struct drm_exynos_pos *src, struct drm_exynos_pos *dst)
+     struct drm_exynos_ipp_task_rect *src, struct drm_exynos_ipp_task_rect *dst)
 {
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 	u32 cfg, cfg_ext, shfactor;
 	u32 pre_dst_width, pre_dst_height;
 	u32 hfactor, vfactor;
@@ -942,13 +806,13 @@ static int fimc_set_prescaler(struct fimc_context *ctx, struct fimc_scaler *sc,
 	/* fimc_ippdrv_check_property assures that dividers are not null */
 	hfactor = fls(src_w / dst_w / 2);
 	if (hfactor > FIMC_SHFACTOR / 2) {
-		dev_err(ippdrv->dev, "failed to get ratio horizontal.\n");
+		dev_err(ctx->dev, "failed to get ratio horizontal.\n");
 		return -EINVAL;
 	}
 
 	vfactor = fls(src_h / dst_h / 2);
 	if (vfactor > FIMC_SHFACTOR / 2) {
-		dev_err(ippdrv->dev, "failed to get ratio vertical.\n");
+		dev_err(ctx->dev, "failed to get ratio vertical.\n");
 		return -EINVAL;
 	}
 
@@ -1019,83 +883,73 @@ static void fimc_set_scaler(struct fimc_context *ctx, struct fimc_scaler *sc)
 	fimc_write(ctx, cfg_ext, EXYNOS_CIEXTEN);
 }
 
-static int fimc_dst_set_size(struct device *dev, int swap,
-		struct drm_exynos_pos *pos, struct drm_exynos_sz *sz)
+static int fimc_dst_set_size(struct fimc_context *ctx,
+			     struct exynos_drm_ipp_buffer *buf)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct drm_exynos_pos img_pos = *pos;
-	struct drm_exynos_sz img_sz = *sz;
 	u32 cfg;
 
-	DRM_DEBUG_KMS("swap[%d]hsize[%d]vsize[%d]\n",
-		swap, sz->hsize, sz->vsize);
+	DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", buf->buf.width, buf->buf.height);
 
 	/* original size */
-	cfg = (EXYNOS_ORGOSIZE_HORIZONTAL(img_sz.hsize) |
-		EXYNOS_ORGOSIZE_VERTICAL(img_sz.vsize));
+	cfg = (EXYNOS_ORGOSIZE_HORIZONTAL(buf->buf.width) |
+		EXYNOS_ORGOSIZE_VERTICAL(buf->buf.height));
 
 	fimc_write(ctx, cfg, EXYNOS_ORGOSIZE);
 
-	DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]\n", pos->x, pos->y, pos->w, pos->h);
+	DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]\n", buf->rect.x, buf->rect.y,
+		buf->rect.w, buf->rect.h);
 
 	/* CSC ITU */
 	cfg = fimc_read(ctx, EXYNOS_CIGCTRL);
 	cfg &= ~EXYNOS_CIGCTRL_CSC_MASK;
 
-	if (sz->hsize >= FIMC_WIDTH_ITU_709)
+	if (buf->buf.width >= FIMC_WIDTH_ITU_709)
 		cfg |= EXYNOS_CIGCTRL_CSC_ITU709;
 	else
 		cfg |= EXYNOS_CIGCTRL_CSC_ITU601;
 
 	fimc_write(ctx, cfg, EXYNOS_CIGCTRL);
 
-	if (swap) {
-		img_pos.w = pos->h;
-		img_pos.h = pos->w;
-		img_sz.hsize = sz->vsize;
-		img_sz.vsize = sz->hsize;
-	}
-
 	/* target image size */
 	cfg = fimc_read(ctx, EXYNOS_CITRGFMT);
 	cfg &= ~(EXYNOS_CITRGFMT_TARGETH_MASK |
 		EXYNOS_CITRGFMT_TARGETV_MASK);
-	cfg |= (EXYNOS_CITRGFMT_TARGETHSIZE(img_pos.w) |
-		EXYNOS_CITRGFMT_TARGETVSIZE(img_pos.h));
+	cfg |= (EXYNOS_CITRGFMT_TARGETHSIZE(buf->rect.w) |
+		EXYNOS_CITRGFMT_TARGETVSIZE(buf->rect.h));
 	fimc_write(ctx, cfg, EXYNOS_CITRGFMT);
 
 	/* target area */
-	cfg = EXYNOS_CITAREA_TARGET_AREA(img_pos.w * img_pos.h);
+	cfg = EXYNOS_CITAREA_TARGET_AREA(buf->rect.w * buf->rect.h);
 	fimc_write(ctx, cfg, EXYNOS_CITAREA);
 
 	/* offset Y(RGB), Cb, Cr */
-	cfg = (EXYNOS_CIOYOFF_HORIZONTAL(img_pos.x) |
-		EXYNOS_CIOYOFF_VERTICAL(img_pos.y));
+	cfg = (EXYNOS_CIOYOFF_HORIZONTAL(buf->rect.x) |
+		EXYNOS_CIOYOFF_VERTICAL(buf->rect.y));
 	fimc_write(ctx, cfg, EXYNOS_CIOYOFF);
-	cfg = (EXYNOS_CIOCBOFF_HORIZONTAL(img_pos.x) |
-		EXYNOS_CIOCBOFF_VERTICAL(img_pos.y));
+	cfg = (EXYNOS_CIOCBOFF_HORIZONTAL(buf->rect.x) |
+		EXYNOS_CIOCBOFF_VERTICAL(buf->rect.y));
 	fimc_write(ctx, cfg, EXYNOS_CIOCBOFF);
-	cfg = (EXYNOS_CIOCROFF_HORIZONTAL(img_pos.x) |
-		EXYNOS_CIOCROFF_VERTICAL(img_pos.y));
+	cfg = (EXYNOS_CIOCROFF_HORIZONTAL(buf->rect.x) |
+		EXYNOS_CIOCROFF_VERTICAL(buf->rect.y));
 	fimc_write(ctx, cfg, EXYNOS_CIOCROFF);
 
 	return 0;
 }
 
 static void fimc_dst_set_buf_seq(struct fimc_context *ctx, u32 buf_id,
-		enum drm_exynos_ipp_buf_type buf_type)
+		bool enqueue)
 {
 	unsigned long flags;
 	u32 buf_num;
 	u32 cfg;
 
-	DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type);
+	DRM_DEBUG_KMS("buf_id[%d]enqueu[%d]\n", buf_id, enqueue);
 
 	spin_lock_irqsave(&ctx->lock, flags);
 
 	cfg = fimc_read(ctx, EXYNOS_CIFCNTSEQ);
 
-	if (buf_type == IPP_BUF_ENQUEUE)
+	if (enqueue)
 		cfg |= (1 << buf_id);
 	else
 		cfg &= ~(1 << buf_id);
@@ -1104,88 +958,31 @@ static void fimc_dst_set_buf_seq(struct fimc_context *ctx, u32 buf_id,
 
 	buf_num = hweight32(cfg);
 
-	if (buf_type == IPP_BUF_ENQUEUE && buf_num >= FIMC_BUF_START)
+	if (enqueue && buf_num >= FIMC_BUF_START)
 		fimc_mask_irq(ctx, true);
-	else if (buf_type == IPP_BUF_DEQUEUE && buf_num <= FIMC_BUF_STOP)
+	else if (!enqueue && buf_num <= FIMC_BUF_STOP)
 		fimc_mask_irq(ctx, false);
 
 	spin_unlock_irqrestore(&ctx->lock, flags);
 }
 
-static int fimc_dst_set_addr(struct device *dev,
-		struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id,
-		enum drm_exynos_ipp_buf_type buf_type)
+static int fimc_dst_set_addr(struct fimc_context *ctx,
+			     struct exynos_drm_ipp_buffer *buf)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
-	struct drm_exynos_ipp_property *property;
-	struct drm_exynos_ipp_config *config;
+	fimc_write(ctx, buf->dma_addr[0], EXYNOS_CIOYSA(0));
+	fimc_write(ctx, buf->dma_addr[1], EXYNOS_CIOCBSA(0));
+	fimc_write(ctx, buf->dma_addr[2], EXYNOS_CIOCRSA(0));
 
-	if (!c_node) {
-		DRM_ERROR("failed to get c_node.\n");
-		return -EINVAL;
-	}
-
-	property = &c_node->property;
-
-	DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n",
-		property->prop_id, buf_id, buf_type);
-
-	if (buf_id > FIMC_MAX_DST) {
-		dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id);
-		return -ENOMEM;
-	}
-
-	/* address register set */
-	switch (buf_type) {
-	case IPP_BUF_ENQUEUE:
-		config = &property->config[EXYNOS_DRM_OPS_DST];
-
-		fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_Y],
-			EXYNOS_CIOYSA(buf_id));
-
-		if (config->fmt == DRM_FORMAT_YVU420) {
-			fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CR],
-				EXYNOS_CIOCBSA(buf_id));
-			fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CB],
-				EXYNOS_CIOCRSA(buf_id));
-		} else {
-			fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CB],
-				EXYNOS_CIOCBSA(buf_id));
-			fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CR],
-				EXYNOS_CIOCRSA(buf_id));
-		}
-		break;
-	case IPP_BUF_DEQUEUE:
-		fimc_write(ctx, 0x0, EXYNOS_CIOYSA(buf_id));
-		fimc_write(ctx, 0x0, EXYNOS_CIOCBSA(buf_id));
-		fimc_write(ctx, 0x0, EXYNOS_CIOCRSA(buf_id));
-		break;
-	default:
-		/* bypass */
-		break;
-	}
-
-	fimc_dst_set_buf_seq(ctx, buf_id, buf_type);
+	fimc_dst_set_buf_seq(ctx, 0, true);
 
 	return 0;
 }
 
-static struct exynos_drm_ipp_ops fimc_dst_ops = {
-	.set_fmt = fimc_dst_set_fmt,
-	.set_transf = fimc_dst_set_transf,
-	.set_size = fimc_dst_set_size,
-	.set_addr = fimc_dst_set_addr,
-};
+static void fimc_stop(struct fimc_context *ctx);
 
 static irqreturn_t fimc_irq_handler(int irq, void *dev_id)
 {
 	struct fimc_context *ctx = dev_id;
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
-	struct drm_exynos_ipp_event_work *event_work =
-		c_node->event_work;
 	int buf_id;
 
 	DRM_DEBUG_KMS("fimc id[%d]\n", ctx->id);
@@ -1203,170 +1000,18 @@ static irqreturn_t fimc_irq_handler(int irq, void *dev_id)
 
 	DRM_DEBUG_KMS("buf_id[%d]\n", buf_id);
 
-	fimc_dst_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE);
-
-	event_work->ippdrv = ippdrv;
-	event_work->buf_id[EXYNOS_DRM_OPS_DST] = buf_id;
-	queue_work(ippdrv->event_workq, &event_work->work);
-
-	return IRQ_HANDLED;
-}
-
-static int fimc_init_prop_list(struct exynos_drm_ippdrv *ippdrv)
-{
-	struct drm_exynos_ipp_prop_list *prop_list = &ippdrv->prop_list;
-
-	prop_list->version = 1;
-	prop_list->writeback = 1;
-	prop_list->refresh_min = FIMC_REFRESH_MIN;
-	prop_list->refresh_max = FIMC_REFRESH_MAX;
-	prop_list->flip = (1 << EXYNOS_DRM_FLIP_NONE) |
-				(1 << EXYNOS_DRM_FLIP_VERTICAL) |
-				(1 << EXYNOS_DRM_FLIP_HORIZONTAL);
-	prop_list->degree = (1 << EXYNOS_DRM_DEGREE_0) |
-				(1 << EXYNOS_DRM_DEGREE_90) |
-				(1 << EXYNOS_DRM_DEGREE_180) |
-				(1 << EXYNOS_DRM_DEGREE_270);
-	prop_list->csc = 1;
-	prop_list->crop = 1;
-	prop_list->crop_max.hsize = FIMC_CROP_MAX;
-	prop_list->crop_max.vsize = FIMC_CROP_MAX;
-	prop_list->crop_min.hsize = FIMC_CROP_MIN;
-	prop_list->crop_min.vsize = FIMC_CROP_MIN;
-	prop_list->scale = 1;
-	prop_list->scale_max.hsize = FIMC_SCALE_MAX;
-	prop_list->scale_max.vsize = FIMC_SCALE_MAX;
-	prop_list->scale_min.hsize = FIMC_SCALE_MIN;
-	prop_list->scale_min.vsize = FIMC_SCALE_MIN;
-
-	return 0;
-}
+	if (ctx->task) {
+		struct exynos_drm_ipp_task *task = ctx->task;
 
-static inline bool fimc_check_drm_flip(enum drm_exynos_flip flip)
-{
-	switch (flip) {
-	case EXYNOS_DRM_FLIP_NONE:
-	case EXYNOS_DRM_FLIP_VERTICAL:
-	case EXYNOS_DRM_FLIP_HORIZONTAL:
-	case EXYNOS_DRM_FLIP_BOTH:
-		return true;
-	default:
-		DRM_DEBUG_KMS("invalid flip\n");
-		return false;
+		ctx->task = NULL;
+		pm_runtime_put(ctx->dev);
+		exynos_drm_ipp_task_done(task, 0);
 	}
-}
 
-static int fimc_ippdrv_check_property(struct device *dev,
-		struct drm_exynos_ipp_property *property)
-{
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_prop_list *pp = &ippdrv->prop_list;
-	struct drm_exynos_ipp_config *config;
-	struct drm_exynos_pos *pos;
-	struct drm_exynos_sz *sz;
-	bool swap;
-	int i;
-
-	for_each_ipp_ops(i) {
-		if ((i == EXYNOS_DRM_OPS_SRC) &&
-			(property->cmd == IPP_CMD_WB))
-			continue;
-
-		config = &property->config[i];
-		pos = &config->pos;
-		sz = &config->sz;
-
-		/* check for flip */
-		if (!fimc_check_drm_flip(config->flip)) {
-			DRM_ERROR("invalid flip.\n");
-			goto err_property;
-		}
-
-		/* check for degree */
-		switch (config->degree) {
-		case EXYNOS_DRM_DEGREE_90:
-		case EXYNOS_DRM_DEGREE_270:
-			swap = true;
-			break;
-		case EXYNOS_DRM_DEGREE_0:
-		case EXYNOS_DRM_DEGREE_180:
-			swap = false;
-			break;
-		default:
-			DRM_ERROR("invalid degree.\n");
-			goto err_property;
-		}
-
-		/* check for buffer bound */
-		if ((pos->x + pos->w > sz->hsize) ||
-			(pos->y + pos->h > sz->vsize)) {
-			DRM_ERROR("out of buf bound.\n");
-			goto err_property;
-		}
-
-		/* check for crop */
-		if ((i == EXYNOS_DRM_OPS_SRC) && (pp->crop)) {
-			if (swap) {
-				if ((pos->h < pp->crop_min.hsize) ||
-					(sz->vsize > pp->crop_max.hsize) ||
-					(pos->w < pp->crop_min.vsize) ||
-					(sz->hsize > pp->crop_max.vsize)) {
-					DRM_ERROR("out of crop size.\n");
-					goto err_property;
-				}
-			} else {
-				if ((pos->w < pp->crop_min.hsize) ||
-					(sz->hsize > pp->crop_max.hsize) ||
-					(pos->h < pp->crop_min.vsize) ||
-					(sz->vsize > pp->crop_max.vsize)) {
-					DRM_ERROR("out of crop size.\n");
-					goto err_property;
-				}
-			}
-		}
-
-		/* check for scale */
-		if ((i == EXYNOS_DRM_OPS_DST) && (pp->scale)) {
-			if (swap) {
-				if ((pos->h < pp->scale_min.hsize) ||
-					(sz->vsize > pp->scale_max.hsize) ||
-					(pos->w < pp->scale_min.vsize) ||
-					(sz->hsize > pp->scale_max.vsize)) {
-					DRM_ERROR("out of scale size.\n");
-					goto err_property;
-				}
-			} else {
-				if ((pos->w < pp->scale_min.hsize) ||
-					(sz->hsize > pp->scale_max.hsize) ||
-					(pos->h < pp->scale_min.vsize) ||
-					(sz->vsize > pp->scale_max.vsize)) {
-					DRM_ERROR("out of scale size.\n");
-					goto err_property;
-				}
-			}
-		}
-	}
+	fimc_dst_set_buf_seq(ctx, buf_id, false);
+	fimc_stop(ctx);
 
-	return 0;
-
-err_property:
-	for_each_ipp_ops(i) {
-		if ((i == EXYNOS_DRM_OPS_SRC) &&
-			(property->cmd == IPP_CMD_WB))
-			continue;
-
-		config = &property->config[i];
-		pos = &config->pos;
-		sz = &config->sz;
-
-		DRM_ERROR("[%s]f[%d]r[%d]pos[%d %d %d %d]sz[%d %d]\n",
-			i ? "dst" : "src", config->flip, config->degree,
-			pos->x, pos->y, pos->w, pos->h,
-			sz->hsize, sz->vsize);
-	}
-
-	return -EINVAL;
+	return IRQ_HANDLED;
 }
 
 static void fimc_clear_addr(struct fimc_context *ctx)
@@ -1386,10 +1031,8 @@ static void fimc_clear_addr(struct fimc_context *ctx)
 	}
 }
 
-static int fimc_ippdrv_reset(struct device *dev)
+static int fimc_reset(struct fimc_context *ctx)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-
 	/* reset h/w block */
 	fimc_sw_reset(ctx);
 
@@ -1401,78 +1044,24 @@ static int fimc_ippdrv_reset(struct device *dev)
 	return 0;
 }
 
-static int fimc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd)
+static int fimc_start(struct fimc_context *ctx)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
-	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
-	struct drm_exynos_ipp_property *property;
-	struct drm_exynos_ipp_config *config;
-	struct drm_exynos_pos	img_pos[EXYNOS_DRM_OPS_MAX];
-	struct drm_exynos_ipp_set_wb set_wb;
-	int ret, i;
 	u32 cfg0, cfg1;
 
-	DRM_DEBUG_KMS("cmd[%d]\n", cmd);
-
-	if (!c_node) {
-		DRM_ERROR("failed to get c_node.\n");
-		return -EINVAL;
-	}
-
-	property = &c_node->property;
-
 	fimc_mask_irq(ctx, true);
 
-	for_each_ipp_ops(i) {
-		config = &property->config[i];
-		img_pos[i] = config->pos;
-	}
-
-	ret = fimc_set_prescaler(ctx, &ctx->sc,
-		&img_pos[EXYNOS_DRM_OPS_SRC],
-		&img_pos[EXYNOS_DRM_OPS_DST]);
-	if (ret) {
-		dev_err(dev, "failed to set prescaler.\n");
-		return ret;
-	}
-
 	/* If set ture, we can save jpeg about screen */
 	fimc_handle_jpeg(ctx, false);
 	fimc_set_scaler(ctx, &ctx->sc);
 
-	switch (cmd) {
-	case IPP_CMD_M2M:
-		fimc_set_type_ctrl(ctx, FIMC_WB_NONE);
-		fimc_handle_lastend(ctx, false);
-
-		/* setup dma */
-		cfg0 = fimc_read(ctx, EXYNOS_MSCTRL);
-		cfg0 &= ~EXYNOS_MSCTRL_INPUT_MASK;
-		cfg0 |= EXYNOS_MSCTRL_INPUT_MEMORY;
-		fimc_write(ctx, cfg0, EXYNOS_MSCTRL);
-		break;
-	case IPP_CMD_WB:
-		fimc_set_type_ctrl(ctx, FIMC_WB_A);
-		fimc_handle_lastend(ctx, true);
-
-		/* setup FIMD */
-		ret = fimc_set_camblk_fimd0_wb(ctx);
-		if (ret < 0) {
-			dev_err(dev, "camblk setup failed.\n");
-			return ret;
-		}
+	fimc_set_type_ctrl(ctx);
+	fimc_handle_lastend(ctx, false);
 
-		set_wb.enable = 1;
-		set_wb.refresh = property->refresh_rate;
-		exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb);
-		break;
-	case IPP_CMD_OUTPUT:
-	default:
-		ret = -EINVAL;
-		dev_err(dev, "invalid operations.\n");
-		return ret;
-	}
+	/* setup dma */
+	cfg0 = fimc_read(ctx, EXYNOS_MSCTRL);
+	cfg0 &= ~EXYNOS_MSCTRL_INPUT_MASK;
+	cfg0 |= EXYNOS_MSCTRL_INPUT_MEMORY;
+	fimc_write(ctx, cfg0, EXYNOS_MSCTRL);
 
 	/* Reset status */
 	fimc_write(ctx, 0x0, EXYNOS_CISTATUS);
@@ -1498,36 +1087,20 @@ static int fimc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd)
 
 	fimc_clear_bits(ctx, EXYNOS_CIOCTRL, EXYNOS_CIOCTRL_WEAVE_MASK);
 
-	if (cmd == IPP_CMD_M2M)
-		fimc_set_bits(ctx, EXYNOS_MSCTRL, EXYNOS_MSCTRL_ENVID);
+	fimc_set_bits(ctx, EXYNOS_MSCTRL, EXYNOS_MSCTRL_ENVID);
 
 	return 0;
 }
 
-static void fimc_ippdrv_stop(struct device *dev, enum drm_exynos_ipp_cmd cmd)
+static void fimc_stop(struct fimc_context *ctx)
 {
-	struct fimc_context *ctx = get_fimc_context(dev);
-	struct drm_exynos_ipp_set_wb set_wb = {0, 0};
 	u32 cfg;
 
-	DRM_DEBUG_KMS("cmd[%d]\n", cmd);
-
-	switch (cmd) {
-	case IPP_CMD_M2M:
-		/* Source clear */
-		cfg = fimc_read(ctx, EXYNOS_MSCTRL);
-		cfg &= ~EXYNOS_MSCTRL_INPUT_MASK;
-		cfg &= ~EXYNOS_MSCTRL_ENVID;
-		fimc_write(ctx, cfg, EXYNOS_MSCTRL);
-		break;
-	case IPP_CMD_WB:
-		exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb);
-		break;
-	case IPP_CMD_OUTPUT:
-	default:
-		dev_err(dev, "invalid operations.\n");
-		break;
-	}
+	/* Source clear */
+	cfg = fimc_read(ctx, EXYNOS_MSCTRL);
+	cfg &= ~EXYNOS_MSCTRL_INPUT_MASK;
+	cfg &= ~EXYNOS_MSCTRL_ENVID;
+	fimc_write(ctx, cfg, EXYNOS_MSCTRL);
 
 	fimc_mask_irq(ctx, false);
 
@@ -1545,6 +1118,103 @@ static void fimc_ippdrv_stop(struct device *dev, enum drm_exynos_ipp_cmd cmd)
 	fimc_set_bits(ctx, EXYNOS_CIGCTRL, EXYNOS_CIGCTRL_IRQ_END_DISABLE);
 }
 
+static int fimc_commit(struct exynos_drm_ipp *ipp,
+			  struct exynos_drm_ipp_task *task)
+{
+	struct fimc_context *ctx =
+			container_of(ipp, struct fimc_context, ipp);
+
+	pm_runtime_get_sync(ctx->dev);
+	ctx->task = task;
+
+	fimc_src_set_fmt(ctx, task->src.buf.fourcc);
+	fimc_src_set_size(ctx, &task->src);
+	fimc_src_set_transf(ctx, DRM_MODE_ROTATE_0);
+	fimc_src_set_addr(ctx, &task->src);
+	fimc_dst_set_fmt(ctx, task->dst.buf.fourcc);
+	fimc_dst_set_transf(ctx, task->transform.rotation);
+	fimc_dst_set_size(ctx, &task->dst);
+	fimc_dst_set_addr(ctx, &task->dst);
+	fimc_set_prescaler(ctx, &ctx->sc, &task->src.rect, &task->dst.rect);
+	fimc_start(ctx);
+
+	return 0;
+}
+
+static void fimc_abort(struct exynos_drm_ipp *ipp,
+			  struct exynos_drm_ipp_task *task)
+{
+	struct fimc_context *ctx =
+			container_of(ipp, struct fimc_context, ipp);
+
+	fimc_reset(ctx);
+
+	if (ctx->task) {
+		struct exynos_drm_ipp_task *task = ctx->task;
+
+		ctx->task = NULL;
+		pm_runtime_put(ctx->dev);
+		exynos_drm_ipp_task_done(task, -EIO);
+	}
+}
+
+static struct exynos_drm_ipp_funcs ipp_funcs = {
+	.commit = fimc_commit,
+	.abort = fimc_abort,
+};
+
+static const struct exynos_drm_ipp_formats fimc_formats[] = {
+	{ IPP_SRCDST_FORMAT(XRGB8888, NULL) },
+	{ IPP_SRCDST_FORMAT(RGB888, NULL) },
+	{ IPP_SRCDST_FORMAT(RGB565, NULL) },
+	{ IPP_SRCDST_FORMAT(NV12, NULL) },
+	{ IPP_SRCDST_FORMAT(NV16, NULL) },
+	{ IPP_SRCDST_FORMAT(NV21, NULL) },
+	{ IPP_SRCDST_FORMAT(NV61, NULL) },
+	{ IPP_SRCDST_FORMAT(UYVY, NULL) },
+	{ IPP_SRCDST_FORMAT(VYUY, NULL) },
+	{ IPP_SRCDST_FORMAT(YUYV, NULL) },
+	{ IPP_SRCDST_FORMAT(YVYU, NULL) },
+	{ IPP_SRCDST_FORMAT(YUV420, NULL) },
+	{ IPP_SRCDST_FORMAT(YVU420, NULL) },
+	{ IPP_SRCDST_FORMAT(YUV422, NULL) },
+	{ IPP_SRCDST_FORMAT(YUV444, NULL) },
+	{ }
+};
+
+static int fimc_bind(struct device *dev, struct device *master, void *data)
+{
+	struct fimc_context *ctx = dev_get_drvdata(dev);
+	struct drm_device *drm_dev = data;
+	struct exynos_drm_ipp *ipp = &ctx->ipp;
+
+	ctx->drm_dev = drm_dev;
+	drm_iommu_attach_device(drm_dev, dev);
+
+	exynos_drm_ipp_register(drm_dev, ipp, &ipp_funcs,
+			   DRM_EXYNOS_IPP_CAP_CROP | DRM_EXYNOS_IPP_CAP_ROTATE |
+			   DRM_EXYNOS_IPP_CAP_SCALE | DRM_EXYNOS_IPP_CAP_CONVERT,
+			   fimc_formats, "fimc");
+
+	dev_info(dev, "The exynos fimc is probed successfully\n");
+
+	return 0;
+}
+
+static void fimc_unbind(struct device *dev, struct device *master,
+			void *data)
+{
+	struct fimc_context *ctx = dev_get_drvdata(dev);
+
+	drm_iommu_detach_device(ctx->drm_dev, ctx->dev);
+}
+
+static const struct component_ops fimc_component_ops = {
+	.bind	= fimc_bind,
+	.unbind = fimc_unbind,
+};
+
+
 static void fimc_put_clocks(struct fimc_context *ctx)
 {
 	int i;
@@ -1559,7 +1229,7 @@ static void fimc_put_clocks(struct fimc_context *ctx)
 
 static int fimc_setup_clocks(struct fimc_context *ctx)
 {
-	struct device *fimc_dev = ctx->ippdrv.dev;
+	struct device *fimc_dev = ctx->dev;
 	struct device *dev;
 	int ret, i;
 
@@ -1574,8 +1244,6 @@ static int fimc_setup_clocks(struct fimc_context *ctx)
 
 		ctx->clocks[i] = clk_get(dev, fimc_clock_names[i]);
 		if (IS_ERR(ctx->clocks[i])) {
-			if (i >= FIMC_CLK_MUX)
-				break;
 			ret = PTR_ERR(ctx->clocks[i]);
 			dev_err(fimc_dev, "failed to get clock: %s\n",
 						fimc_clock_names[i]);
@@ -1583,20 +1251,6 @@ static int fimc_setup_clocks(struct fimc_context *ctx)
 		}
 	}
 
-	/* Optional FIMC LCLK parent clock setting */
-	if (!IS_ERR(ctx->clocks[FIMC_CLK_PARENT])) {
-		ret = clk_set_parent(ctx->clocks[FIMC_CLK_MUX],
-				     ctx->clocks[FIMC_CLK_PARENT]);
-		if (ret < 0) {
-			dev_err(fimc_dev, "failed to set parent.\n");
-			goto e_clk_free;
-		}
-	}
-
-	ret = clk_set_rate(ctx->clocks[FIMC_CLK_LCLK], ctx->clk_frequency);
-	if (ret < 0)
-		goto e_clk_free;
-
 	ret = clk_prepare_enable(ctx->clocks[FIMC_CLK_LCLK]);
 	if (!ret)
 		return ret;
@@ -1607,20 +1261,14 @@ static int fimc_setup_clocks(struct fimc_context *ctx)
 
 static int fimc_parse_dt(struct fimc_context *ctx)
 {
-	struct device_node *node = ctx->ippdrv.dev->of_node;
+	struct device_node *node = ctx->dev->of_node;
 
-	/* Handle only devices that support the LCD Writeback data path */
-	if (!of_property_read_bool(node, "samsung,lcd-wb"))
+	if (exynos_drm_check_shared_device(ctx->dev) != 0)
 		return -ENODEV;
 
-	if (of_property_read_u32(node, "clock-frequency",
-					&ctx->clk_frequency))
-		ctx->clk_frequency = FIMC_DEFAULT_LCLK_FREQUENCY;
-
 	ctx->id = of_alias_get_id(node, "fimc");
-
 	if (ctx->id < 0) {
-		dev_err(ctx->ippdrv.dev, "failed to get node alias id.\n");
+		dev_err(ctx->dev, "failed to get node alias id.\n");
 		return -EINVAL;
 	}
 
@@ -1632,7 +1280,6 @@ static int fimc_probe(struct platform_device *pdev)
 	struct device *dev = &pdev->dev;
 	struct fimc_context *ctx;
 	struct resource *res;
-	struct exynos_drm_ippdrv *ippdrv;
 	int ret;
 
 	if (!dev->of_node) {
@@ -1644,19 +1291,12 @@ static int fimc_probe(struct platform_device *pdev)
 	if (!ctx)
 		return -ENOMEM;
 
-	ctx->ippdrv.dev = dev;
+	ctx->dev = dev;
 
 	ret = fimc_parse_dt(ctx);
 	if (ret < 0)
 		return ret;
 
-	ctx->sysreg = syscon_regmap_lookup_by_phandle(dev->of_node,
-						"samsung,sysreg");
-	if (IS_ERR(ctx->sysreg)) {
-		dev_err(dev, "syscon regmap lookup failed.\n");
-		return PTR_ERR(ctx->sysreg);
-	}
-
 	/* resource memory */
 	ctx->regs_res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
 	ctx->regs = devm_ioremap_resource(dev, ctx->regs_res);
@@ -1670,9 +1310,8 @@ static int fimc_probe(struct platform_device *pdev)
 		return -ENOENT;
 	}
 
-	ctx->irq = res->start;
-	ret = devm_request_threaded_irq(dev, ctx->irq, NULL, fimc_irq_handler,
-		IRQF_ONESHOT, "drm_fimc", ctx);
+	ret = devm_request_irq(dev, res->start, fimc_irq_handler,
+		0, dev_name(dev), ctx);
 	if (ret < 0) {
 		dev_err(dev, "failed to request irq.\n");
 		return ret;
@@ -1682,31 +1321,14 @@ static int fimc_probe(struct platform_device *pdev)
 	if (ret < 0)
 		return ret;
 
-	ippdrv = &ctx->ippdrv;
-	ippdrv->ops[EXYNOS_DRM_OPS_SRC] = &fimc_src_ops;
-	ippdrv->ops[EXYNOS_DRM_OPS_DST] = &fimc_dst_ops;
-	ippdrv->check_property = fimc_ippdrv_check_property;
-	ippdrv->reset = fimc_ippdrv_reset;
-	ippdrv->start = fimc_ippdrv_start;
-	ippdrv->stop = fimc_ippdrv_stop;
-	ret = fimc_init_prop_list(ippdrv);
-	if (ret < 0) {
-		dev_err(dev, "failed to init property list.\n");
-		goto err_put_clk;
-	}
-
-	DRM_DEBUG_KMS("id[%d]ippdrv[%pK]\n", ctx->id, ippdrv);
-
 	spin_lock_init(&ctx->lock);
 	platform_set_drvdata(pdev, ctx);
 
 	pm_runtime_enable(dev);
 
-	ret = exynos_drm_ippdrv_register(ippdrv);
-	if (ret < 0) {
-		dev_err(dev, "failed to register drm fimc device.\n");
+	ret = component_add(dev, &fimc_component_ops);
+	if (ret)
 		goto err_pm_dis;
-	}
 
 	dev_info(dev, "drm fimc registered successfully.\n");
 
@@ -1714,7 +1336,6 @@ static int fimc_probe(struct platform_device *pdev)
 
 err_pm_dis:
 	pm_runtime_disable(dev);
-err_put_clk:
 	fimc_put_clocks(ctx);
 
 	return ret;
@@ -1724,42 +1345,23 @@ static int fimc_remove(struct platform_device *pdev)
 {
 	struct device *dev = &pdev->dev;
 	struct fimc_context *ctx = get_fimc_context(dev);
-	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
 
-	exynos_drm_ippdrv_unregister(ippdrv);
+	component_del(dev, &fimc_component_ops);
+	pm_runtime_disable(dev);
 
 	fimc_put_clocks(ctx);
-	pm_runtime_set_suspended(dev);
-	pm_runtime_disable(dev);
 
 	return 0;
 }
 
 #ifdef CONFIG_PM
-static int fimc_clk_ctrl(struct fimc_context *ctx, bool enable)
-{
-	DRM_DEBUG_KMS("enable[%d]\n", enable);
-
-	if (enable) {
-		clk_prepare_enable(ctx->clocks[FIMC_CLK_GATE]);
-		clk_prepare_enable(ctx->clocks[FIMC_CLK_WB_A]);
-		ctx->suspended = false;
-	} else {
-		clk_disable_unprepare(ctx->clocks[FIMC_CLK_GATE]);
-		clk_disable_unprepare(ctx->clocks[FIMC_CLK_WB_A]);
-		ctx->suspended = true;
-	}
-
-	return 0;
-}
-
 static int fimc_runtime_suspend(struct device *dev)
 {
 	struct fimc_context *ctx = get_fimc_context(dev);
 
 	DRM_DEBUG_KMS("id[%d]\n", ctx->id);
-
-	return  fimc_clk_ctrl(ctx, false);
+	clk_disable_unprepare(ctx->clocks[FIMC_CLK_GATE]);
+	return 0;
 }
 
 static int fimc_runtime_resume(struct device *dev)
@@ -1767,8 +1369,7 @@ static int fimc_runtime_resume(struct device *dev)
 	struct fimc_context *ctx = get_fimc_context(dev);
 
 	DRM_DEBUG_KMS("id[%d]\n", ctx->id);
-
-	return  fimc_clk_ctrl(ctx, true);
+	return clk_prepare_enable(ctx->clocks[FIMC_CLK_GATE]);
 }
 #endif
 
diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimc.h b/drivers/gpu/drm/exynos/exynos_drm_fimc.h
deleted file mode 100644
index 127a424c5fdf..000000000000
--- a/drivers/gpu/drm/exynos/exynos_drm_fimc.h
+++ /dev/null
@@ -1,23 +0,0 @@
-/*
- * Copyright (c) 2012 Samsung Electronics Co., Ltd.
- *
- * Authors:
- *	Eunchul Kim <chulspro.kim@samsung.com>
- *	Jinyoung Jeon <jy0.jeon@samsung.com>
- *	Sangmin Lee <lsmin.lee@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.
- */
-
-#ifndef _EXYNOS_DRM_FIMC_H_
-#define _EXYNOS_DRM_FIMC_H_
-
-/*
- * TODO
- * FIMD output interface notifier callback.
- */
-
-#endif /* _EXYNOS_DRM_FIMC_H_ */
-- 
1.9.1

[PATCH / code example] userspace: Simple tool for testing IPP v2 API

[Marek Szyprowski]

This is simple example how Exynos DRM IPP v2 API can be used from userspace.
The code allocates two GEM buffers, fills first with test pattern and then
performs 180 degree rotation of the image data.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---
 rotate.c | 235 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 235 insertions(+)
 create mode 100644 rotate.c

diff --git a/rotate.c b/rotate.c
new file mode 100644
index 0000000..ab0c5d0
--- /dev/null
+++ b/rotate.c
@@ -0,0 +1,234 @@
+/*
+ * Copyright (C) 2017 Samsung Electronics Co.Ltd
+ * Author: Marek Szyprowski <m.szyprowski@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 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 THE
+ * AUTHORS OR COPYRIGHT HOLDERS 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.
+ *
+ */
+
+#define _FILE_OFFSET_BITS 64
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/types.h>
+#include <sys/mman.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <string.h>
+#include <errno.h>
+
+#include <drm/drm.h>
+#include <drm/drm_fourcc.h>
+#include <drm/exynos_drm.h>
+
+#include <xf86drm.h>
+
+struct bo {
+	void *ptr;
+	size_t size;
+	size_t offset;
+	size_t pitch;
+	int width;
+	int height;
+	unsigned int handle;
+	unsigned int fourcc;
+};
+
+struct bo *bo_create(int fd, unsigned int width, unsigned int height,
+			  unsigned int bpp, unsigned int fourcc)
+{
+	struct drm_exynos_gem_create create_arg = { };
+	struct drm_exynos_gem_map map_arg = { };
+	struct bo *bo;
+	int ret;
+
+	bo = calloc(1, sizeof(*bo));
+	if (bo == NULL) {
+		fprintf(stderr, "failed to allocate buffer object\n");
+		return NULL;
+	}
+
+	bo->size = width * height * bpp / 8;
+	bo->pitch = width * bpp / 8;
+	bo->width = width;
+	bo->height = height;
+
+	create_arg.size = bo->size;
+	create_arg.flags = EXYNOS_BO_NONCONTIG;
+
+	if (drmCommandWriteRead(fd, DRM_EXYNOS_GEM_CREATE, &create_arg,
+				sizeof(create_arg))) {
+		fprintf(stderr, "cannot create Exynos GEM object (%d): %m\n",
+			errno);
+		return NULL;
+	}
+
+	bo->handle = create_arg.handle;
+	bo->fourcc = fourcc;
+
+	map_arg.handle = bo->handle;
+
+	if (drmCommandWriteRead(fd, DRM_EXYNOS_GEM_MAP, &map_arg,
+				sizeof(map_arg))) {
+		fprintf(stderr, "cannot map Exynos GEM object (%d): %m\n",
+			errno);
+		return NULL;
+	}
+
+	bo->ptr = mmap(0, bo->size, PROT_READ | PROT_WRITE, MAP_SHARED, fd,
+		       map_arg.offset);
+	if (bo->ptr == MAP_FAILED)
+		return NULL;
+
+	return bo;
+}
+
+void bo_destroy(int fd, struct bo *bo)
+{
+	struct drm_gem_close arg = { };
+	int ret;
+
+	if (bo->ptr)
+		munmap(bo->ptr, bo->size);
+
+	arg.handle = bo->handle;
+	ret = drmIoctl(fd, DRM_IOCTL_GEM_CLOSE, &arg);
+	if (ret)
+		fprintf(stderr, "failed to destroy dumb buffer (%d): %m\n",
+			errno);
+
+	free(bo);
+}
+
+struct bo *bo_create_xrgb(int fd, unsigned int width, unsigned int height)
+{
+	return bo_create(fd, width, height, 32, DRM_FORMAT_XRGB8888);
+}
+
+struct exynos_drm_ipp_std_task {
+	struct drm_exynos_ipp_task_buffer buf[2];
+	struct drm_exynos_ipp_task_rect rect[2];
+	struct drm_exynos_ipp_task_transform transform;
+} __packed;
+
+int process_fb(int fd, struct bo *src_bo, int sx, int sy, int sw, int sh,
+	       struct bo *dst_bo, int dx, int dy, int dw, int dh, int rotation)
+{
+	struct exynos_drm_ipp_std_task task = {	};
+	struct drm_exynos_ioctl_ipp_commit arg = { };
+	uint32_t id = 0; /* hardcoded first available IPP module */
+
+	task.buf[0].id = DRM_EXYNOS_IPP_TASK_BUFFER |
+			 DRM_EXYNOS_IPP_TASK_TYPE_SOURCE;
+	task.buf[0].fourcc = src_bo->fourcc;
+	task.buf[0].width = src_bo->width;
+	task.buf[0].height = src_bo->height;
+	task.buf[0].pitch[0] = src_bo->pitch;
+	task.buf[0].gem_id[0] = src_bo->handle;
+
+	task.buf[1].id = DRM_EXYNOS_IPP_TASK_BUFFER |
+			 DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION;
+	task.buf[1].fourcc = dst_bo->fourcc;
+	task.buf[1].width = dst_bo->width;
+	task.buf[1].height = dst_bo->height;
+	task.buf[1].pitch[0] = dst_bo->pitch;
+	task.buf[1].gem_id[0] = dst_bo->handle;
+
+	task.rect[0].id = DRM_EXYNOS_IPP_TASK_RECTANGLE |
+			  DRM_EXYNOS_IPP_TASK_TYPE_SOURCE;
+	task.rect[0].x = sx;
+	task.rect[0].y = sy;
+	task.rect[0].w = sw;
+	task.rect[0].h = sh;
+
+	task.rect[1].id = DRM_EXYNOS_IPP_TASK_RECTANGLE |
+			  DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION;
+	task.rect[1].x = dx;
+	task.rect[1].y = dy;
+	task.rect[1].w = dw;
+	task.rect[1].h = dh;
+
+	task.transform.id = DRM_EXYNOS_IPP_TASK_TRANSFORM;
+	task.transform.rotation = rotation;
+
+	arg.flags = 0;
+	arg.ipp_id = id;
+	arg.params_size = sizeof(task);
+	arg.params_ptr = (unsigned long)(&task);
+	arg.user_data = 0;
+
+	if (drmCommandWriteRead(fd, DRM_EXYNOS_IPP_COMMIT, &arg, sizeof(arg))) {
+		fprintf(stderr, "failed to commit Exynos IPP task (%d): %m\n",
+			errno);
+		return errno;
+	}
+
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	struct bo *buf1, *buf2;
+	int fd;
+	int err;
+	int width = 640;
+	int height = 480;
+	int val = 1, x, y;
+
+	fd = open("/dev/dri/card0", O_RDWR);
+
+	buf1 = bo_create_xrgb(fd, width, height);
+	buf2 = bo_create_xrgb(fd, width, height);
+
+	/* draw test pattern to buffer1 */
+	for (y = 0; y < height; y++) {
+		uint32_t *p = buf1->ptr + buf1->pitch * y;
+
+		for (x = 0; x < width; x++) {
+			if (x > y)
+				*p++ = (val++ & 0xffffff) + 0xff000000;
+		}
+	}
+
+	err = process_fb(fd, buf1, 0, 0, width, height, buf2, 0, 0, width,
+			 height, DRM_MODE_ROTATE_180);
+	if (!err) {
+		printf("Buffer processed, checking processed buffer... ");
+
+		for (y = 0; y < height; y++) {
+			uint32_t *p1 = buf1->ptr + buf1->pitch * y;
+			uint32_t *p2 = buf2->ptr + buf1->pitch * (height - y - 1);
+
+			for (x = 0; x < width; x++)
+				if (*(p1 + x) != *(p2 + width - x - 1)) {
+					printf("failed at (%d,%d) %06x != %06x.\n",
+						x, y, *(p1 + x),
+						*(p2 + width - x - 1));
+					goto free;
+				}
+		}
+		printf("okay.\n");
+	}
+
+free:
+	bo_destroy(fd, buf1);
+	bo_destroy(fd, buf2);
+
+	return 0;
+}
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 2/6] drm/exynos: ipp: Add IPP v2 framework

[Emil Velikov]

Hi Marek,

Just a couple of small suggestions - this is by no means a proper review.

On 12 September 2017 at 09:08, Marek Szyprowski
<m.szyprowski@samsung.com> wrote:
> This patch adds Exynos IPP v2 subsystem and userspace API.
>
> New userspace API is focused ONLY on memory-to-memory image processing.
> The two remainging IPP operation modes (framebuffer writeback and
s/remainging/remaining/

> local-path output with image processing) can be implemented using
> standard DRM features: writeback connectors and additional DRM planes with
> scaling features.
>
> V2 IPP userspace API is not compatible with old IPP ioctls. This is a
> significant change, but the old IPP subsystem in mainline Linux kernel was
> partially disfunctional anyway and not used in any open-source project.
>
s/disfunctional/dysfunctional/

> V2 IPP userspace API is based on stateless approach, which much better fits
> to memory-to-memory image processing model. It also provides support for
> all image formats, which are both already defined in DRM API and supported
> by the existing IPP hardware modules.
>
> The API consists of the following ioctls:
> - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image
>   processing modules,
> - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image
>   formats of given IPP module,
> - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for
s/limitiations/limitations/


> +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data,
> +                                   struct drm_file *file_priv)
> +{
> +       struct drm_exynos_ioctl_ipp_get_limits *resp = data;
> +       void __user *ptr = (void __user *)(unsigned long)resp->limits_ptr;
We have the u64_to_user_ptr helper for these casts. Please use it
through the codebase.


> +struct exynos_drm_param_map {
static const?

> +       unsigned int id;
> +       unsigned int size;
> +       unsigned int offset;
> +} exynos_drm_ipp_params_maps[] = {
> +       {
> +               DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
> +               sizeof(struct drm_exynos_ipp_task_buffer),
> +               offsetof(struct exynos_drm_ipp_task, src.buf),
> +       }, {
> +               DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
> +               sizeof(struct drm_exynos_ipp_task_buffer),
> +               offsetof(struct exynos_drm_ipp_task, dst.buf),
> +       }, {
> +               DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
> +               sizeof(struct drm_exynos_ipp_task_rect),
> +               offsetof(struct exynos_drm_ipp_task, src.rect),
> +       }, {
> +               DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
> +               sizeof(struct drm_exynos_ipp_task_rect),
> +               offsetof(struct exynos_drm_ipp_task, dst.rect),
> +       }, {
> +               DRM_EXYNOS_IPP_TASK_TRANSFORM,
> +               sizeof(struct drm_exynos_ipp_task_transform),
> +               offsetof(struct exynos_drm_ipp_task, transform),
> +       }, {
> +               DRM_EXYNOS_IPP_TASK_ALPHA,
> +               sizeof(struct drm_exynos_ipp_task_alpha),
> +               offsetof(struct exynos_drm_ipp_task, alpha),
> +       },
> +};
> +
> +static int exynos_drm_ipp_task_set(struct exynos_drm_ipp_task *task,
> +                                 struct drm_exynos_ioctl_ipp_commit *arg)
> +{
> +       unsigned int size = arg->params_size;
> +       void *p, *params;
> +       int ret = 0;
> +
> +       if (size > PAGE_SIZE)
> +               return -ERANGE;
> +
> +       p = kmalloc(size, GFP_KERNEL);
> +       if (!p)
> +               return -ENOMEM;
> +
> +       if (copy_from_user(p, (void __user *)(unsigned long)arg->params_ptr,
> +                          size)) {
> +               ret = -EFAULT;
> +               DRM_DEBUG_DRIVER("Failed to copy configuration from userspace\n");
> +               goto free;
> +       }
> +
> +       params = p;
> +       while (size) {
> +               struct exynos_drm_param_map *map = exynos_drm_ipp_params_maps;
I'd just drop this since...

> +               u32 *id = params;
> +               int i;
> +
> +               for (i = 0; i < ARRAY_SIZE(exynos_drm_ipp_params_maps); i++)
... using ARRAY_SIZE(foo) only to access bar[] is quite misleading.

> +                       if (map[i].id == *id)
> +                               break;
> +
> +               if (i < ARRAY_SIZE(exynos_drm_ipp_params_maps) &&
> +                   map[i].size <= size) {
> +                       memcpy((void *)task + map[i].offset, params,
> +                              map[i].size);
> +                       params += map[i].size;
> +                       size -= map[i].size;
> +               } else {
> +                       ret = -EINVAL;
> +                       goto free;
> +               }
Maybe flip the condition order?
if (!foo)
   ret = -EINVAL;
   goto xx;

map[i]...

> +       }
> +
> +       DRM_DEBUG_DRIVER("Got task %pK configuration from userspace\n", task);
> +free:
> +       kfree(p);
> +       return ret;
> +}
> +


> +
> +struct drm_exynos_ipp_limit_val {
> +       __u32 min, max, align;
Having these one per line, will make it harder to miss alignment issues.


> +struct drm_exynos_ipp_limit {
> +       __u32 type;
> +       struct drm_exynos_ipp_limit_val h;
> +       struct drm_exynos_ipp_limit_val v;
If you extend the struct by appending 64bit variable it will break
compat ioctls.
One could add a note/warning above the struct?

> +struct drm_exynos_ipp_task_rect {
> +       __u32   id;
> +       __u32   x, y, w, h;
Ditto - separate lines, note about future compat ioctl breakage?

HTH
Emil
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 2/6] drm/exynos: ipp: Add IPP v2 framework

[Marek Szyprowski]

Hi Emil,


On 2017-09-12 17:02, Emil Velikov wrote:
> Hi Marek,
>
> Just a couple of small suggestions - this is by no means a proper review.

Thanks, I really appreciate your suggestions! I've tried to make userspace
API 32/64bit agnostic, but it looks that I still need to learn a bit in this
area.


> On 12 September 2017 at 09:08, Marek Szyprowski
> <m.szyprowski@samsung.com> wrote:
>> This patch adds Exynos IPP v2 subsystem and userspace API.
>>
>> New userspace API is focused ONLY on memory-to-memory image processing.
>> The two remainging IPP operation modes (framebuffer writeback and
> s/remainging/remaining/
>
>> local-path output with image processing) can be implemented using
>> standard DRM features: writeback connectors and additional DRM planes with
>> scaling features.
>>
>> V2 IPP userspace API is not compatible with old IPP ioctls. This is a
>> significant change, but the old IPP subsystem in mainline Linux kernel was
>> partially disfunctional anyway and not used in any open-source project.
>>
> s/disfunctional/dysfunctional/
>
>> V2 IPP userspace API is based on stateless approach, which much better fits
>> to memory-to-memory image processing model. It also provides support for
>> all image formats, which are both already defined in DRM API and supported
>> by the existing IPP hardware modules.
>>
>> The API consists of the following ioctls:
>> - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image
>>    processing modules,
>> - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image
>>    formats of given IPP module,
>> - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for
> s/limitiations/limitations/
>
>
>> +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data,
>> +                                   struct drm_file *file_priv)
>> +{
>> +       struct drm_exynos_ioctl_ipp_get_limits *resp = data;
>> +       void __user *ptr = (void __user *)(unsigned long)resp->limits_ptr;
> We have the u64_to_user_ptr helper for these casts. Please use it
> through the codebase.
>
>
>> +struct exynos_drm_param_map {
> static const?
>
>> +       unsigned int id;
>> +       unsigned int size;
>> +       unsigned int offset;
>> +} exynos_drm_ipp_params_maps[] = {
>> +       {
>> +               DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
>> +               sizeof(struct drm_exynos_ipp_task_buffer),
>> +               offsetof(struct exynos_drm_ipp_task, src.buf),
>> +       }, {
>> +               DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
>> +               sizeof(struct drm_exynos_ipp_task_buffer),
>> +               offsetof(struct exynos_drm_ipp_task, dst.buf),
>> +       }, {
>> +               DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
>> +               sizeof(struct drm_exynos_ipp_task_rect),
>> +               offsetof(struct exynos_drm_ipp_task, src.rect),
>> +       }, {
>> +               DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
>> +               sizeof(struct drm_exynos_ipp_task_rect),
>> +               offsetof(struct exynos_drm_ipp_task, dst.rect),
>> +       }, {
>> +               DRM_EXYNOS_IPP_TASK_TRANSFORM,
>> +               sizeof(struct drm_exynos_ipp_task_transform),
>> +               offsetof(struct exynos_drm_ipp_task, transform),
>> +       }, {
>> +               DRM_EXYNOS_IPP_TASK_ALPHA,
>> +               sizeof(struct drm_exynos_ipp_task_alpha),
>> +               offsetof(struct exynos_drm_ipp_task, alpha),
>> +       },
>> +};
>> +
>> +static int exynos_drm_ipp_task_set(struct exynos_drm_ipp_task *task,
>> +                                 struct drm_exynos_ioctl_ipp_commit *arg)
>> +{
>> +       unsigned int size = arg->params_size;
>> +       void *p, *params;
>> +       int ret = 0;
>> +
>> +       if (size > PAGE_SIZE)
>> +               return -ERANGE;
>> +
>> +       p = kmalloc(size, GFP_KERNEL);
>> +       if (!p)
>> +               return -ENOMEM;
>> +
>> +       if (copy_from_user(p, (void __user *)(unsigned long)arg->params_ptr,
>> +                          size)) {
>> +               ret = -EFAULT;
>> +               DRM_DEBUG_DRIVER("Failed to copy configuration from userspace\n");
>> +               goto free;
>> +       }
>> +
>> +       params = p;
>> +       while (size) {
>> +               struct exynos_drm_param_map *map = exynos_drm_ipp_params_maps;
> I'd just drop this since...
>
>> +               u32 *id = params;
>> +               int i;
>> +
>> +               for (i = 0; i < ARRAY_SIZE(exynos_drm_ipp_params_maps); i++)
> ... using ARRAY_SIZE(foo) only to access bar[] is quite misleading.
>
>> +                       if (map[i].id == *id)
>> +                               break;
>> +
>> +               if (i < ARRAY_SIZE(exynos_drm_ipp_params_maps) &&
>> +                   map[i].size <= size) {
>> +                       memcpy((void *)task + map[i].offset, params,
>> +                              map[i].size);
>> +                       params += map[i].size;
>> +                       size -= map[i].size;
>> +               } else {
>> +                       ret = -EINVAL;
>> +                       goto free;
>> +               }
> Maybe flip the condition order?
> if (!foo)
>     ret = -EINVAL;
>     goto xx;
>
> map[i]...
>
>> +       }
>> +
>> +       DRM_DEBUG_DRIVER("Got task %pK configuration from userspace\n", task);
>> +free:
>> +       kfree(p);
>> +       return ret;
>> +}
>> +
>
>> +
>> +struct drm_exynos_ipp_limit_val {
>> +       __u32 min, max, align;
> Having these one per line, will make it harder to miss alignment issues.
>
>
>> +struct drm_exynos_ipp_limit {
>> +       __u32 type;
>> +       struct drm_exynos_ipp_limit_val h;
>> +       struct drm_exynos_ipp_limit_val v;
> If you extend the struct by appending 64bit variable it will break
> compat ioctls.
> One could add a note/warning above the struct?
>
>> +struct drm_exynos_ipp_task_rect {
>> +       __u32   id;
>> +       __u32   x, y, w, h;
> Ditto - separate lines, note about future compat ioctl breakage?
>
> HTH
> Emil
>
>
>

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 0/6] Exynos DRM: rewrite IPP subsystem and userspace API

[Tobias Jakobi]

Hey Marek,

first of all thanks for keeping this series alive. Looks pretty good so far.

I have started to rewrite my IPP userspace API in libdrm and when this is done,
I'm going to adapt my mpv video backend to it.

With best wishes,
Tobias



Marek Szyprowski wrote:
> Dear all,
> 
> This patchset performs complete rewrite of Exynos DRM IPP subsystem and
> its userspace API.
> 
> Why such rewrite is needed? Exynos DRM IPP API is over-engineered in
> general, but not really extensible on the other side. It is also buggy,
> with significant design flaws:
> - Userspace API covers memory-2-memory picture operations together with
>   CRTC writeback and duplicating features, which belongs to video plane.
> - Lack of support of the all required image formats (for example NV12
>   Samsung-tiled cannot be used due to lack of pixel format modifier
>   support).
> - Userspace API designed only to mimic hardware behaviour, not easy to
>   understand.
> - Lack of proper input validation in the core, drivers also didn't do that
>   correctly, so it was possible to set incorrect parameters and easil
>   trigger IOMMU fault or memory trash.
> - Drivers were partially disfunctional or supported only a subset of modes.
> 
> Due to the above limitations and issues the Exynos DRM IPP API was not
> used by any of the open-source projects. I assume that it is safe to remove
> this broken API without any damage to open-source community. All remaining
> users (mainly Tizen project related) will be updated to the new version.
> 
> This patchset changes Exynos DRM IPP subsystem to something useful. The
> userspace API is much simpler, state-less and easy to understand. Also
> the code of the core and driver is significantly smaller and easier to
> understand.
> 
> Patches were tested on Exynos4412 based Odroid U3 and Exynos5422
> Odroid XU3 boards, on top of Linux next-20170911 kernel.
> 
> Best regards
> Marek Szyprowski
> Samsung R&D Institute Poland
> 
> 
> My previous works in this area:
> 
> "[RFC v2 0/2] Exynos DRM: add Picture Processor extension"
> https://www.spinics.net/lists/dri-devel/msg140669.html
> - removed usage of DRM objects and properties - replaced them with simple
>   list of parameters with predefined IDs
> 
> "[RFC 0/4] Exynos DRM: add Picture Processor extension"
> https://www.spinics.net/lists/linux-samsung-soc/msg59323.html
> - moved this feature from DRM core to Exynos DRM driver
> - changed name from framebuffer processor to picture processor
> - simplified code to cover only things needed by Exynos drivers
> - implemented simple fifo task scheduler
> - cleaned up rotator driver conversion (removed IPP remainings)
> 
> "[RFC 0/2] New feature: Framebuffer processors"
> https://www.spinics.net/lists/linux-samsung-soc/msg54810.html
> - generic approach implemented in DRM core, rejected
> 
> 
> Patch summary:
> 
> Marek Szyprowski (6):
>   drm/exynos: ipp: Remove Exynos DRM IPP subsystem
>   drm/exynos: ipp: Add IPP v2 framework
>   drm/exynos: rotator: Convert driver to IPP v2 core API
>   drm/exynos: gsc: Convert driver to IPP v2 core API
>   drm/exynos: Add generic support for devices shared with V4L2 subsystem
>   drm/exynos: fimc: Convert driver to IPP v2 core API
> 
>  drivers/gpu/drm/exynos/Kconfig              |   12 +-
>  drivers/gpu/drm/exynos/exynos_drm_drv.c     |   33 +-
>  drivers/gpu/drm/exynos/exynos_drm_drv.h     |    4 +-
>  drivers/gpu/drm/exynos/exynos_drm_fimc.c    |  893 +++--------
>  drivers/gpu/drm/exynos/exynos_drm_fimc.h    |   23 -
>  drivers/gpu/drm/exynos/exynos_drm_gsc.c     |  853 +++-------
>  drivers/gpu/drm/exynos/exynos_drm_gsc.h     |   24 -
>  drivers/gpu/drm/exynos/exynos_drm_ipp.c     | 2239 ++++++++-------------------
>  drivers/gpu/drm/exynos/exynos_drm_ipp.h     |  352 ++---
>  drivers/gpu/drm/exynos/exynos_drm_rotator.c |  735 ++-------
>  drivers/gpu/drm/exynos/exynos_drm_rotator.h |   19 -
>  include/uapi/drm/exynos_drm.h               |  315 ++--
>  12 files changed, 1627 insertions(+), 3875 deletions(-)
>  delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_fimc.h
>  delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_gsc.h
>  delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_rotator.h
> 

Re: [PATCH 2/6] drm/exynos: ipp: Add IPP v2 framework

[Tobias Jakobi]

Hello Marek,


Marek Szyprowski wrote:
> This patch adds Exynos IPP v2 subsystem and userspace API.
> 
> New userspace API is focused ONLY on memory-to-memory image processing.
> The two remainging IPP operation modes (framebuffer writeback and
> local-path output with image processing) can be implemented using
> standard DRM features: writeback connectors and additional DRM planes with
> scaling features.
> 
> V2 IPP userspace API is not compatible with old IPP ioctls. This is a
> significant change, but the old IPP subsystem in mainline Linux kernel was
> partially disfunctional anyway and not used in any open-source project.
> 
> V2 IPP userspace API is based on stateless approach, which much better fits
> to memory-to-memory image processing model. It also provides support for
> all image formats, which are both already defined in DRM API and supported
> by the existing IPP hardware modules.
> 
> The API consists of the following ioctls:
> - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image
>   processing modules,
> - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image
>   formats of given IPP module,
> - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for
>   selected image format of given IPP module,
> - DRM_IOCTL_EXYNOS_IPP_COMMIT: to perform operation described by the
>   provided structures (source and destination buffers, operation rectangle,
>   transformation, etc).
> 
> The proposed userspace API is extensible. In the future more advanced image
> processing operations can be defined to support for example blending.
> 
> Userspace API is fully functional also on DRM render nodes, so it is not
> limited to the root/privileged client.
> 
> Internal driver API also has been completely rewritten. New IPP core
> performs all possible input validation, checks and object life-time
> control. The drivers can focus only on writing configuration to hardware
> registers. Stateless nature of DRM_IOCTL_EXYNOS_IPP_COMMIT ioctl simplifies
> the driver API. Minimal driver needs to provide a single callback for
> starting processing and an array with supported image formats.
> 
> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
> ---
>  drivers/gpu/drm/exynos/Kconfig          |   3 +
>  drivers/gpu/drm/exynos/Makefile         |   1 +
>  drivers/gpu/drm/exynos/exynos_drm_drv.c |   9 +
>  drivers/gpu/drm/exynos/exynos_drm_ipp.c | 893 ++++++++++++++++++++++++++++++++
>  drivers/gpu/drm/exynos/exynos_drm_ipp.h | 188 +++++++
>  include/uapi/drm/exynos_drm.h           | 227 ++++++++
>  6 files changed, 1321 insertions(+)
>  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 88cff0e039b6..2e097a81df12 100644
> --- a/drivers/gpu/drm/exynos/Kconfig
> +++ b/drivers/gpu/drm/exynos/Kconfig
> @@ -94,6 +94,9 @@ config DRM_EXYNOS_G2D
>  	help
>  	  Choose this option if you want to use Exynos G2D for DRM.
>  
> +config DRM_EXYNOS_IPP
> +	bool
> +
>  config DRM_EXYNOS_FIMC
>  	bool "FIMC"
>  	depends on BROKEN && MFD_SYSCON
> diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile
> index 09bb002c9555..f663490e949d 100644
> --- a/drivers/gpu/drm/exynos/Makefile
> +++ b/drivers/gpu/drm/exynos/Makefile
> @@ -17,6 +17,7 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER)	+= exynos_mixer.o
>  exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI)	+= exynos_hdmi.o
>  exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI)	+= exynos_drm_vidi.o
>  exynosdrm-$(CONFIG_DRM_EXYNOS_G2D)	+= exynos_drm_g2d.o
> +exynosdrm-$(CONFIG_DRM_EXYNOS_IPP)	+= exynos_drm_ipp.o
>  exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC)	+= exynos_drm_fimc.o
>  exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR)	+= exynos_drm_rotator.o
>  exynosdrm-$(CONFIG_DRM_EXYNOS_GSC)	+= exynos_drm_gsc.o
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
> index 7f19054ebce3..645046c5943a 100644
> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
> @@ -26,6 +26,7 @@
>  #include "exynos_drm_fb.h"
>  #include "exynos_drm_gem.h"
>  #include "exynos_drm_plane.h"
> +#include "exynos_drm_ipp.h"
>  #include "exynos_drm_vidi.h"
>  #include "exynos_drm_g2d.h"
>  #include "exynos_drm_iommu.h"
> @@ -114,6 +115,14 @@ static void exynos_drm_lastclose(struct drm_device *dev)
>  			DRM_AUTH | DRM_RENDER_ALLOW),
>  	DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl,
>  			DRM_AUTH | DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_RESOURCES, exynos_drm_ipp_get_res_ioctl,
> +			DRM_AUTH | DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_CAPS, exynos_drm_ipp_get_caps_ioctl,
> +			DRM_AUTH | DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_LIMITS, exynos_drm_ipp_get_limits_ioctl,
> +			DRM_AUTH | DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_COMMIT, exynos_drm_ipp_commit_ioctl,
> +			DRM_AUTH | DRM_RENDER_ALLOW),
>  };
>  
>  static const struct file_operations exynos_drm_driver_fops = {
> 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 000000000000..11be6a1bf839
> --- /dev/null
> +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.c
> @@ -0,0 +1,892 @@
> +/*
> + * Copyright (C) 2017 Samsung Electronics Co.Ltd
> + * Authors:
> + *	Marek Szyprowski <m.szyprowski@samsung.com>
> + *
> + * Exynos DRM Image Post Processing (IPP) related functions
> + *
> + * 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 shall be included in
> + * all copies or substantial portions of the Software.
> + */
> +
> +
> +#include <drm/drmP.h>
> +#include <drm/drm_mode.h>
> +#include <uapi/drm/exynos_drm.h>
> +
> +#include "exynos_drm_drv.h"
> +#include "exynos_drm_gem.h"
> +#include "exynos_drm_ipp.h"
> +
> +static int num_ipp;
> +static LIST_HEAD(ipp_list);
> +
> +struct drm_pending_exynos_ipp_event {
> +	struct drm_pending_event base;
> +	struct drm_exynos_ipp_event event;
> +};
> +
> +/**
> + * exynos_drm_ipp_register - Register a new picture processor hardware module
> + * @dev: DRM device
> + * @ipp: ipp module to init
> + * @funcs: callbacks for the new ipp object
> + * @caps: bitmask of ipp capabilities (%DRM_EXYNOS_IPP_CAP_*)
> + * @formats: array of supported formats (%DRM_FORMAT_*)
> + * @name: name (for debugging purposes)
> + *
> + * Initializes a ipp module.
> + *
> + * Returns:
> + * Zero on success, error code on failure.
> + */
> +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp,
> +		const struct exynos_drm_ipp_funcs *funcs, unsigned int caps,
> +		const struct exynos_drm_ipp_formats *formats, const char *name)
> +{
> +	WARN_ON(!ipp);
> +	WARN_ON(!funcs);
> +	WARN_ON(!formats);
> +
> +	spin_lock_init(&ipp->lock);
> +	INIT_LIST_HEAD(&ipp->todo_list);
> +	init_waitqueue_head(&ipp->done_wq);
> +	ipp->dev = dev;
> +	ipp->funcs = funcs;
> +	ipp->capabilities = caps;
> +	ipp->name = name;
> +	ipp->formats = formats;
> +	while (formats++->fourcc)
> +		ipp->num_formats++;
> +
> +	list_add_tail(&ipp->head, &ipp_list);
> +	ipp->id = num_ipp++;
> +
> +	DRM_DEBUG_DRIVER("Registered ipp %d\n", ipp->id);
> +
> +	return 0;
> +}
> +
> +/**
> + * exynos_drm_ipp_unregister - Unregister the picture processor module
> + * @dev: DRM device
> + * @ipp: ipp module
> + */
> +void exynos_drm_ipp_unregister(struct drm_device *dev,
> +			       struct exynos_drm_ipp *ipp)
> +{
> +	BUG_ON(ipp->task);
> +	BUG_ON(!list_empty(&ipp->todo_list));
> +	list_del(&ipp->head);
> +}
> +
> +/**
> + * exynos_drm_ipp_ioctl_get_res_ioctl - enumerate all ipp modules
> + * @dev: DRM device
> + * @data: ioctl data
> + * @file_priv: DRM file info
> + *
> + * Construct a list of ipp ids.
> + *
> + * Called by the user via ioctl.
> + *
> + * Returns:
> + * Zero on success, negative errno on failure.
> + */
> +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data,
> +				 struct drm_file *file_priv)
> +{
> +	struct drm_exynos_ioctl_ipp_get_res *resp = data;
> +	struct exynos_drm_ipp *ipp;
> +	uint32_t __user *ipp_ptr = (uint32_t __user *)
> +						(unsigned long)resp->ipp_id_ptr;
> +	unsigned int count = num_ipp, copied = 0;
> +
> +	/*
> +	 * This ioctl is called twice, once to determine how much space is
> +	 * needed, and the 2nd time to fill it.
> +	 */
> +	if (count && resp->count_ipps >= count) {
> +		list_for_each_entry(ipp, &ipp_list, head) {
> +			if (put_user(ipp->id, ipp_ptr + copied))
> +				return -EFAULT;
> +			copied++;
> +		}
> +	}
> +	resp->count_ipps = count;
> +
> +	return 0;
> +}
> +
> +static inline struct exynos_drm_ipp *__ipp_get(uint32_t id)
> +{
> +	struct exynos_drm_ipp *ipp;
> +
> +	list_for_each_entry(ipp, &ipp_list, head)
> +		if (ipp->id == id)
> +			return ipp;
> +	return NULL;
> +}
> +
> +/**
> + * exynos_drm_ipp_ioctl_get_caps - get ipp module capabilities and formats
> + * @dev: DRM device
> + * @data: ioctl data
> + * @file_priv: DRM file info
> + *
> + * Construct a structure describing ipp module capabilities.
> + *
> + * Called by the user via ioctl.
> + *
> + * Returns:
> + * Zero on success, negative errno on failure.
> + */
> +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data,
> +				  struct drm_file *file_priv)
> +{
> +	struct drm_exynos_ioctl_ipp_get_caps *resp = data;
> +	void __user *ptr = (void __user *)(unsigned long)resp->formats_ptr;
> +	struct exynos_drm_ipp *ipp;
> +	int i;
> +
> +	ipp = __ipp_get(resp->ipp_id);
> +	if (!ipp)
> +		return -ENOENT;
> +
> +	resp->ipp_id = ipp->id;
> +	resp->capabilities = ipp->capabilities;
> +
> +	/*
> +	 * This ioctl is called twice, once to determine how much space is
> +	 * needed, and the 2nd time to fill it.
> +	 */
> +	if (resp->formats_count >= ipp->num_formats) {
> +		for (i = 0; i < ipp->num_formats; i++) {
> +			struct exynos_drm_ipp_format tmp = {
> +				.fourcc = ipp->formats[i].fourcc,
> +				.type = ipp->formats[i].type,
> +				.modifier = ipp->formats[i].modifier,
> +			};
> +
> +			if (copy_to_user(ptr, &tmp, sizeof(tmp)))
> +				return -EFAULT;
> +			ptr += sizeof(tmp);
> +		}
> +	}
> +	resp->formats_count = ipp->num_formats;
> +
> +	return 0;
> +}
> +
> +static inline const struct exynos_drm_ipp_formats *__ipp_format_get(
> +			uint32_t fourcc, uint64_t mod, unsigned long type,
> +			const struct exynos_drm_ipp_formats *f)
> +{
> +	for (; f->fourcc; f++)
> +		if ((f->type & type) && f->fourcc == fourcc &&
> +		    f->modifier == mod)
> +			return f;
> +	return NULL;
> +}
> +
> +/**
> + * exynos_drm_ipp_get_limits_ioctl - get ipp module limits
> + * @dev: DRM device
> + * @data: ioctl data
> + * @file_priv: DRM file info
> + *
> + * Construct a structure describing ipp module limitations for provided
> + * picture format.
> + *
> + * Called by the user via ioctl.
> + *
> + * Returns:
> + * Zero on success, negative errno on failure.
> + */
> +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data,
> +				    struct drm_file *file_priv)
I'm getting a null-pointer Opps in this ioctl(). I haven't fully debugged this,
but I guess it's because format->limits might be NULL. E.g. the formats for the
FIMC (exynos_drm_ipp_formats fimc_formats[]) don't carry any limits for the
color formats.




> +{
> +	struct drm_exynos_ioctl_ipp_get_limits *resp = data;
> +	void __user *ptr = (void __user *)(unsigned long)resp->limits_ptr;
> +	const struct exynos_drm_ipp_formats *format;
> +	const struct drm_exynos_ipp_limit *limits, *l;
> +	struct exynos_drm_ipp *ipp;
> +	int count = 0;
> +
> +	if (resp->type != DRM_EXYNOS_IPP_FORMAT_SOURCE &&
> +	    resp->type != DRM_EXYNOS_IPP_FORMAT_DESTINATION)
> +		return -EINVAL;
> +
> +	ipp = __ipp_get(resp->ipp_id);
> +	if (!ipp)
> +		return -ENOENT;
> +
> +	format = __ipp_format_get(resp->fourcc, resp->modifier, resp->type,
> +			      ipp->formats);
> +	if (!format)
> +		return -EINVAL;
> +
> +	limits = format->limits;
> +	for (l = limits; l->type; l++)
> +		count++;
> +
> +	/*
> +	 * This ioctl is called twice, once to determine how much space is
> +	 * needed, and the 2nd time to fill it.
> +	 */
> +	if (count && resp->limits_size >= count * sizeof(*l)) {
> +		for (l = limits; l->type; l++, ptr += sizeof(*l))
> +			if (copy_to_user((void __user *)ptr, l, sizeof(*l)))
> +				return -EFAULT;
> +	}
> +	resp->limits_size = count * sizeof(*l);
I wonder why you return the size if bytes here? How about keeping this
consistant with the other ioctls (like exynos_drm_ipp_get_caps_ioctl) and just
return the count (together with renaming limits_size to limits>count).

- Tobias


> +
> +	return 0;
> +}
> +
> +static inline struct exynos_drm_ipp_task *
> +	exynos_drm_ipp_task_alloc(struct exynos_drm_ipp *ipp)
> +{
> +	struct exynos_drm_ipp_task *task;
> +
> +	task = kzalloc(sizeof(*task), GFP_KERNEL);
> +	if (!task)
> +		return NULL;
> +
> +	task->dev = ipp->dev;
> +	task->ipp = ipp;
> +
> +	/* some defaults */
> +	task->src.rect.w = task->dst.rect.w = UINT_MAX;
> +	task->src.rect.h = task->dst.rect.h = UINT_MAX;
> +	task->transform.rotation = DRM_MODE_ROTATE_0;
> +
> +	DRM_DEBUG_DRIVER("Allocated task %pK\n", task);
> +
> +	return task;
> +}
> +
> +struct exynos_drm_param_map {
> +	unsigned int id;
> +	unsigned int size;
> +	unsigned int offset;
> +} exynos_drm_ipp_params_maps[] = {
> +	{
> +		DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
> +		sizeof(struct drm_exynos_ipp_task_buffer),
> +		offsetof(struct exynos_drm_ipp_task, src.buf),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
> +		sizeof(struct drm_exynos_ipp_task_buffer),
> +		offsetof(struct exynos_drm_ipp_task, dst.buf),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
> +		sizeof(struct drm_exynos_ipp_task_rect),
> +		offsetof(struct exynos_drm_ipp_task, src.rect),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
> +		sizeof(struct drm_exynos_ipp_task_rect),
> +		offsetof(struct exynos_drm_ipp_task, dst.rect),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_TRANSFORM,
> +		sizeof(struct drm_exynos_ipp_task_transform),
> +		offsetof(struct exynos_drm_ipp_task, transform),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_ALPHA,
> +		sizeof(struct drm_exynos_ipp_task_alpha),
> +		offsetof(struct exynos_drm_ipp_task, alpha),
> +	},
> +};
> +
> +static int exynos_drm_ipp_task_set(struct exynos_drm_ipp_task *task,
> +				  struct drm_exynos_ioctl_ipp_commit *arg)
> +{
> +	unsigned int size = arg->params_size;
> +	void *p, *params;
> +	int ret = 0;
> +
> +	if (size > PAGE_SIZE)
> +		return -ERANGE;
> +
> +	p = kmalloc(size, GFP_KERNEL);
> +	if (!p)
> +		return -ENOMEM;
> +
> +	if (copy_from_user(p, (void __user *)(unsigned long)arg->params_ptr,
> +			   size)) {
> +		ret = -EFAULT;
> +		DRM_DEBUG_DRIVER("Failed to copy configuration from userspace\n");
> +		goto free;
> +	}
> +
> +	params = p;
> +	while (size) {
> +		struct exynos_drm_param_map *map = exynos_drm_ipp_params_maps;
> +		u32 *id = params;
> +		int i;
> +
> +		for (i = 0; i < ARRAY_SIZE(exynos_drm_ipp_params_maps); i++)
> +			if (map[i].id == *id)
> +				break;
> +
> +		if (i < ARRAY_SIZE(exynos_drm_ipp_params_maps) &&
> +		    map[i].size <= size) {
> +			memcpy((void *)task + map[i].offset, params,
> +			       map[i].size);
> +			params += map[i].size;
> +			size -= map[i].size;
> +		} else {
> +			ret = -EINVAL;
> +			goto free;
> +		}
> +	}
> +
> +	DRM_DEBUG_DRIVER("Got task %pK configuration from userspace\n", task);
> +free:
> +	kfree(p);
> +	return ret;
> +}
> +
> +static int exynos_drm_ipp_task_setup_buffer(struct exynos_drm_ipp_buffer *buf,
> +					    struct drm_file *filp)
> +{
> +	int ret = 0;
> +	int i;
> +
> +	/* basic checks */
> +	if (buf->buf.width == 0 || buf->buf.height == 0)
> +		return -EINVAL;
> +	buf->format = drm_format_info(buf->buf.fourcc);
> +	for (i = 0; i < buf->format->num_planes; i++) {
> +		unsigned int width = (i == 0) ? buf->buf.width :
> +			     DIV_ROUND_UP(buf->buf.width, buf->format->hsub);
> +
> +		if (buf->buf.pitch[i] == 0)
> +			buf->buf.pitch[i] = width * buf->format->cpp[i];
> +		if (buf->buf.pitch[i] < width * buf->format->cpp[i])
> +			return -EINVAL;
> +		if (!buf->buf.gem_id[i])
> +			return -ENOENT;
> +	}
> +
> +	/* get GEM buffers and check their size */
> +	for (i = 0; i < buf->format->num_planes; i++) {
> +		unsigned int height = (i == 0) ? buf->buf.height :
> +			     DIV_ROUND_UP(buf->buf.height, buf->format->vsub);
> +		unsigned long size = height * buf->buf.pitch[i] +
> +				     buf->buf.offset[i];
> +		struct drm_gem_object *obj = drm_gem_object_lookup(filp,
> +							    buf->buf.gem_id[i]);
> +		if (!obj) {
> +			ret = -ENOENT;
> +			goto gem_free;
> +		}
> +		buf->exynos_gem[i] = to_exynos_gem(obj);
> +
> +		if (size > buf->exynos_gem[i]->size) {
> +			i++;
> +			ret = -EINVAL;
> +			goto gem_free;
> +		}
> +		buf->dma_addr[i] = buf->exynos_gem[i]->dma_addr +
> +				   buf->buf.offset[i];
> +	}
> +
> +	return 0;
> +gem_free:
> +	while (i--) {
> +		drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base);
> +		buf->exynos_gem[i] = NULL;
> +	}
> +	return ret;
> +}
> +
> +static void exynos_drm_ipp_task_release_buffer(struct exynos_drm_ipp_buffer *buf)
> +{
> +	int i;
> +
> +	if (!buf->exynos_gem[0])
> +		return;
> +	for (i = 0; i < buf->format->num_planes; i++)
> +		drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base);
> +}
> +
> +static void exynos_drm_ipp_task_free(struct exynos_drm_ipp *ipp,
> +				 struct exynos_drm_ipp_task *task)
> +{
> +	DRM_DEBUG_DRIVER("Freeing task %pK\n", task);
> +
> +	exynos_drm_ipp_task_release_buffer(&task->src);
> +	exynos_drm_ipp_task_release_buffer(&task->dst);
> +	if (task->event)
> +		drm_event_cancel_free(ipp->dev, &task->event->base);
> +	kfree(task);
> +}
> +
> +struct drm_ipp_limit {
> +	struct drm_exynos_ipp_limit_val h;
> +	struct drm_exynos_ipp_limit_val v;
> +};
> +
> +enum drm_ipp_size_id {
> +	IPP_LIMIT_BUFFER, IPP_LIMIT_AREA, IPP_LIMIT_ROTATED, IPP_LIMIT_MAX
> +};
> +
> +static const enum drm_ipp_size_id limit_id_fallback[IPP_LIMIT_MAX][4] = {
> +	[IPP_LIMIT_BUFFER]  = { DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER },
> +	[IPP_LIMIT_AREA]    = { DRM_EXYNOS_IPP_LIMIT_SIZE_AREA,
> +				DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER },
> +	[IPP_LIMIT_ROTATED] = { DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED,
> +				DRM_EXYNOS_IPP_LIMIT_SIZE_AREA,
> +				DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER },
> +};
> +
> +static inline void __limit_set_val(unsigned int *ptr, unsigned int val)
> +{
> +	if (!*ptr)
> +		*ptr = val;
> +}
> +
> +static void __get_size_limit(const struct drm_exynos_ipp_limit *limits,
> +			     enum drm_ipp_size_id id, struct drm_ipp_limit *res)
> +{
> +	const struct drm_exynos_ipp_limit *l = limits;
> +	int i = 0;
> +
> +	memset(res, 0, sizeof(*res));
> +	for (i = 0; limit_id_fallback[id][i]; i++)
> +		for (l = limits; l->type; l++) {
> +			if (((l->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) !=
> +			      DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE) ||
> +			    ((l->type & DRM_EXYNOS_IPP_LIMIT_SIZE_MASK) !=
> +						     limit_id_fallback[id][i]))
> +				continue;
> +			__limit_set_val(&res->h.min, l->h.min);
> +			__limit_set_val(&res->h.max, l->h.max);
> +			__limit_set_val(&res->h.align, l->h.align);
> +			__limit_set_val(&res->v.min, l->v.min);
> +			__limit_set_val(&res->v.max, l->v.max);
> +			__limit_set_val(&res->v.align, l->v.align);
> +		}
> +}
> +
> +static inline bool __align_check(unsigned int val, unsigned int align)
> +{
> +	if (align && (val & (align - 1)))
> +		return false;
> +	return true;
> +}
> +
> +static inline bool __size_limit_check(unsigned int val,
> +				 struct drm_exynos_ipp_limit_val *l)
> +{
> +	if ((l->min && val < l->min) || (l->max && val > l->max) ||
> +	    !__align_check(val, l->align))
> +		return false;
> +	return true;
> +}
> +
> +static int exynos_drm_ipp_check_size_limits(struct exynos_drm_ipp_buffer *buf,
> +	const struct drm_exynos_ipp_limit *limits, bool rotate, bool swap)
> +{
> +	enum drm_ipp_size_id id = rotate ? IPP_LIMIT_ROTATED : IPP_LIMIT_AREA;
> +	struct drm_ipp_limit l;
> +	struct drm_exynos_ipp_limit_val *lh = &l.h, *lv = &l.v;
> +
> +	if (!limits)
> +		return 0;
> +
> +	__get_size_limit(limits, IPP_LIMIT_BUFFER, &l);
> +	if (!__size_limit_check(buf->buf.width, &l.h) ||
> +	    !__size_limit_check(buf->buf.height, &l.v))
> +		return -EINVAL;
> +
> +	if (swap) {
> +		lv = &l.h;
> +		lh = &l.v;
> +	}
> +	__get_size_limit(limits, id, &l);
> +	if (!__size_limit_check(buf->rect.w, lh) ||
> +	    !__align_check(buf->rect.x, lh->align) ||
> +	    !__size_limit_check(buf->rect.h, lv) ||
> +	    !__align_check(buf->rect.y, lv->align))
> +		return -EINVAL;
> +
> +	return 0;
> +}
> +
> +static int exynos_drm_ipp_check_scale_limits(struct drm_exynos_ipp_task_rect *src,
> +	struct drm_exynos_ipp_task_rect *dst,
> +	const struct drm_exynos_ipp_limit *limits, bool swap)
> +{
> +	const struct drm_exynos_ipp_limit_val *lh, *lv;
> +
> +	if (!limits)
> +		return 0;
> +
> +	for (; limits->type; limits++)
> +		if ((limits->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) ==
> +		    DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE)
> +			break;
> +	if (!limits->type)
> +		return 0;
> +
> +	lh = (!swap) ? &limits->h : &limits->v;
> +	lv = (!swap) ? &limits->v : &limits->h;
> +
> +	if ((lh->max && dst->w > src->w * lh->max) ||
> +	    (lh->min && (dst->w << 16) < src->w * lh->min) ||
> +	    (lv->max && dst->h > src->h * lv->max) ||
> +	    (lv->min && (dst->h << 16) < src->h * lv->min))
> +		return -EINVAL;
> +
> +	return 0;
> +}
> +
> +static int exynos_drm_ipp_task_check(struct exynos_drm_ipp_task *task,
> +				    struct drm_file *filp)
> +{
> +	struct exynos_drm_ipp *ipp = task->ipp;
> +	const struct exynos_drm_ipp_formats *src_format, *dst_format;
> +	struct exynos_drm_ipp_buffer *src = &task->src, *dst = &task->dst;
> +	unsigned int rotation = task->transform.rotation;
> +	int ret = 0;
> +	bool scale = false, swap = false;
> +
> +	DRM_DEBUG_DRIVER("Checking %pK\n", task);
> +
> +	if (src->rect.w == UINT_MAX)
> +		src->rect.w = src->buf.width;
> +	if (src->rect.h == UINT_MAX)
> +		src->rect.h = src->buf.height;
> +	if (dst->rect.w == UINT_MAX)
> +		dst->rect.w = dst->buf.width;
> +	if (dst->rect.h == UINT_MAX)
> +		dst->rect.h = dst->buf.height;
> +
> +	if (src->rect.x + src->rect.w > (src->buf.width) ||
> +	    src->rect.y + src->rect.h > (src->buf.height) ||
> +	    dst->rect.x + dst->rect.w > (dst->buf.width) ||
> +	    dst->rect.y + dst->rect.h > (dst->buf.height))
> +		return -EINVAL;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: basic checks done\n", task);
> +
> +	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CROP) &&
> +	    (src->rect.x || src->rect.y || dst->rect.x || dst->rect.y))
> +		return -EINVAL;
> +
> +	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_ROTATE) &&
> +	    rotation != DRM_MODE_ROTATE_0)
> +		return -EINVAL;
> +
> +	if (drm_rotation_90_or_270(rotation))
> +		swap = true;
> +
> +	if ((!swap && (src->rect.w != dst->rect.w || src->rect.h != dst->rect.h)) ||
> +	    (swap && (src->rect.w != dst->rect.h || src->rect.h != dst->rect.w)))
> +		scale = true;
> +
> +	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_SCALE) && scale)
> +		return -EINVAL;
> +
> +	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CONVERT) &&
> +	    src->buf.fourcc != dst->buf.fourcc)
> +		return -EINVAL;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: capability checks done\n", task);
> +
> +	src_format = __ipp_format_get(src->buf.fourcc, src->buf.modifier,
> +				  DRM_EXYNOS_IPP_FORMAT_SOURCE, ipp->formats);
> +	if (!src_format)
> +		return -EINVAL;
> +	ret = exynos_drm_ipp_check_size_limits(src, src_format->limits,
> +					  rotation != DRM_MODE_ROTATE_0, false);
> +	if (ret)
> +		return ret;
> +	ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect,
> +						src_format->limits, swap);
> +	if (ret)
> +		return ret;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: source image checks done\n", task);
> +
> +	dst_format = __ipp_format_get(dst->buf.fourcc, dst->buf.modifier,
> +			       DRM_EXYNOS_IPP_FORMAT_DESTINATION, ipp->formats);
> +	if (!dst_format)
> +		return -EINVAL;
> +	ret = exynos_drm_ipp_check_size_limits(dst, dst_format->limits,
> +					   rotation != DRM_MODE_ROTATE_0, swap);
> +	if (ret)
> +		return ret;
> +	ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect,
> +						dst_format->limits, swap);
> +	if (ret)
> +		return ret;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: destination image checks done\n", task);
> +
> +	ret = exynos_drm_ipp_task_setup_buffer(src, filp);
> +	if (ret)
> +		return ret;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: source buffer checks done\n", task);
> +
> +	ret = exynos_drm_ipp_task_setup_buffer(dst, filp);
> +	if (ret)
> +		return ret;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: destination buffer checks done\n", task);
> +
> +	if (ipp->funcs->check)
> +		ret = ipp->funcs->check(ipp, task);
> +
> +	DRM_DEBUG_DRIVER("Task %pK: all checks done.\n", task);
> +
> +	return ret;
> +}
> +
> +static int exynos_drm_ipp_event_create(struct exynos_drm_ipp_task *task,
> +			struct drm_file *file_priv, uint64_t user_data)
> +{
> +	struct drm_pending_exynos_ipp_event *e = NULL;
> +	int ret;
> +
> +	e = kzalloc(sizeof(*e), GFP_KERNEL);
> +	if (!e)
> +		return -ENOMEM;
> +
> +	e->event.base.type = DRM_EXYNOS_IPP_EVENT;
> +	e->event.base.length = sizeof(e->event);
> +	e->event.user_data = user_data;
> +
> +	ret = drm_event_reserve_init(task->dev, file_priv, &e->base,
> +				     &e->event.base);
> +	if (ret)
> +		goto free;
> +
> +	task->event = e;
> +	return 0;
> +free:
> +	kfree(e);
> +	return ret;
> +}
> +
> +static void exynos_drm_ipp_event_send(struct exynos_drm_ipp_task *task)
> +{
> +	struct timeval now = ktime_to_timeval(ktime_get());
> +
> +	task->event->event.tv_sec = now.tv_sec;
> +	task->event->event.tv_usec = now.tv_usec;
> +	task->event->event.sequence = atomic_inc_return(&task->ipp->sequence);
> +
> +	drm_send_event(task->dev, &task->event->base);
> +}
> +
> +static int exynos_drm_ipp_task_cleanup(struct exynos_drm_ipp_task *task)
> +{
> +	int ret = task->ret;
> +
> +	if (ret == 0 && task->event) {
> +		exynos_drm_ipp_event_send(task);
> +		/* ensure event won't be canceled on task free */
> +		task->event = NULL;
> +	}
> +
> +	exynos_drm_ipp_task_free(task->ipp, task);
> +	return ret;
> +}
> +
> +static void exynos_drm_ipp_cleanup_work(struct work_struct *work)
> +{
> +	struct exynos_drm_ipp_task *task = container_of(work,
> +				struct exynos_drm_ipp_task, cleanup_work);
> +
> +	exynos_drm_ipp_task_cleanup(task);
> +}
> +
> +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp);
> +
> +/**
> + * exynos_drm_ipp_task_done - finish given task and set return code
> + * @task: ipp task to finish
> + * @ret: error code or 0 if operation has been performed successfully
> + */
> +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret)
> +{
> +	struct exynos_drm_ipp *ipp = task->ipp;
> +	unsigned long flags;
> +
> +	DRM_DEBUG_DRIVER("ipp: %d, task %pK done\n", ipp->id, task);
> +
> +	spin_lock_irqsave(&ipp->lock, flags);
> +	if (ipp->task == task)
> +		ipp->task = NULL;
> +	task->flags |= DRM_EXYNOS_IPP_TASK_DONE;
> +	task->ret = ret;
> +	spin_unlock_irqrestore(&ipp->lock, flags);
> +
> +	exynos_drm_ipp_next_task(ipp);
> +	wake_up(&ipp->done_wq);
> +
> +	if (task->flags & DRM_EXYNOS_IPP_TASK_ASYNC) {
> +		INIT_WORK(&task->cleanup_work, exynos_drm_ipp_cleanup_work);
> +		schedule_work(&task->cleanup_work);
> +	}
> +}
> +
> +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp)
> +{
> +	struct exynos_drm_ipp_task *task;
> +	unsigned long flags;
> +	int ret;
> +
> +	DRM_DEBUG_DRIVER("ipp: %d, try to run new task\n", ipp->id);
> +
> +	spin_lock_irqsave(&ipp->lock, flags);
> +
> +	if (ipp->task || list_empty(&ipp->todo_list)) {
> +		spin_unlock_irqrestore(&ipp->lock, flags);
> +		return;
> +	}
> +
> +	task = list_first_entry(&ipp->todo_list, struct exynos_drm_ipp_task,
> +				head);
> +	list_del_init(&task->head);
> +	ipp->task = task;
> +
> +	spin_unlock_irqrestore(&ipp->lock, flags);
> +
> +	DRM_DEBUG_DRIVER("ipp: %d, selected task %pK to run\n", ipp->id, task);
> +
> +	ret = ipp->funcs->commit(ipp, task);
> +	if (ret)
> +		exynos_drm_ipp_task_done(task, ret);
> +}
> +
> +static void exynos_drm_ipp_schedule_task(struct exynos_drm_ipp *ipp,
> +				     struct exynos_drm_ipp_task *task)
> +{
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&ipp->lock, flags);
> +	list_add(&task->head, &ipp->todo_list);
> +	spin_unlock_irqrestore(&ipp->lock, flags);
> +
> +	exynos_drm_ipp_next_task(ipp);
> +}
> +
> +static void exynos_drm_ipp_task_abort(struct exynos_drm_ipp *ipp,
> +				  struct exynos_drm_ipp_task *task)
> +{
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&ipp->lock, flags);
> +	if (task->flags & DRM_EXYNOS_IPP_TASK_DONE) {
> +		/* already completed task */
> +		exynos_drm_ipp_task_cleanup(task);
> +	} else if (ipp->task != task) {
> +		/* task has not been scheduled for execution yet */
> +		list_del_init(&task->head);
> +		exynos_drm_ipp_task_cleanup(task);
> +	} else {
> +		/*
> +		 * currently processed task, call abort() and perform
> +		 * cleanup with async worker
> +		 */
> +		task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC;
> +		spin_unlock_irqrestore(&ipp->lock, flags);
> +		if (ipp->funcs->abort)
> +			ipp->funcs->abort(ipp, task);
> +		return;
> +	}
> +	spin_unlock_irqrestore(&ipp->lock, flags);
> +}
> +
> +/**
> + * exynos_drm_ipp_commit_ioctl - perform image processing operation
> + * @dev: DRM device
> + * @data: ioctl data
> + * @file_priv: DRM file info
> + *
> + * Construct a ipp task from the set of properties provided from the user
> + * and try to schedule it to framebuffer processor hardware.
> + *
> + * Called by the user via ioctl.
> + *
> + * Returns:
> + * Zero on success, negative errno on failure.
> + */
> +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, void *data,
> +			  struct drm_file *file_priv)
> +{
> +	struct drm_exynos_ioctl_ipp_commit *arg = data;
> +	struct exynos_drm_ipp *ipp;
> +	struct exynos_drm_ipp_task *task;
> +	int ret = 0;
> +
> +	if ((arg->flags & ~DRM_EXYNOS_IPP_FLAGS) || arg->reserved)
> +		return -EINVAL;
> +
> +	/* can't test and expect an event at the same time */
> +	if ((arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY) &&
> +			(arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT))
> +		return -EINVAL;
> +
> +	ipp = __ipp_get(arg->ipp_id);
> +	if (!ipp)
> +		return -ENOENT;
> +
> +	task = exynos_drm_ipp_task_alloc(ipp);
> +	if (!task)
> +		return -ENOMEM;
> +
> +	ret = exynos_drm_ipp_task_set(task, arg);
> +	if (ret)
> +		goto free;
> +
> +	ret = exynos_drm_ipp_task_check(task, file_priv);
> +	if (ret || arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY)
> +		goto free;
> +
> +	if (arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT) {
> +		ret = exynos_drm_ipp_event_create(task, file_priv,
> +						 arg->user_data);
> +		if (ret)
> +			goto free;
> +	}
> +
> +	/*
> +	 * Queue task for processing on the hardware. task object will be
> +	 * then freed after exynos_drm_ipp_task_done()
> +	 */
> +	if (arg->flags & DRM_EXYNOS_IPP_FLAG_NONBLOCK) {
> +		DRM_DEBUG_DRIVER("ipp: %d, nonblocking processing task %pK\n",
> +				 ipp->id, task);
> +
> +		task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC;
> +		exynos_drm_ipp_schedule_task(task->ipp, task);
> +		ret = 0;
> +	} else {
> +		DRM_DEBUG_DRIVER("ipp: %d, processing task %pK\n", ipp->id, task);
> +		exynos_drm_ipp_schedule_task(ipp, task);
> +		ret = wait_event_interruptible(ipp->done_wq,
> +					task->flags & DRM_EXYNOS_IPP_TASK_DONE);
> +		if (ret)
> +			exynos_drm_ipp_task_abort(ipp, task);
> +		else
> +			ret = exynos_drm_ipp_task_cleanup(task);
> +	}
> +	return ret;
> +free:
> +	exynos_drm_ipp_task_free(ipp, task);
> +
> +	return ret;
> +}
> 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 000000000000..532a60698a6a
> --- /dev/null
> +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.h
> @@ -0,0 +1,188 @@
> +/*
> + * Copyright (c) 2017 Samsung Electronics Co., Ltd.
> + *
> + * 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.
> + */
> +
> +#ifndef _EXYNOS_DRM_IPP_H_
> +#define _EXYNOS_DRM_IPP_H_
> +
> +#include <drm/drmP.h>
> +
> +struct exynos_drm_ipp;
> +struct exynos_drm_ipp_task;
> +
> +/**
> + * struct exynos_drm_ipp_funcs - exynos_drm_ipp control functions
> + */
> +struct exynos_drm_ipp_funcs {
> +	/**
> +	 * @check:
> +	 *
> +	 * This is the optional hook to validate an ipp task. This function
> +	 * must reject any task which the hardware or driver doesn't support.
> +	 * This includes but is of course not limited to:
> +	 *
> +	 *  - Checking that the buffers, scaling and placement requirements
> +	 *    and so on are within the limits of the hardware not specified by
> +	 *    limits array.
> +	 *
> +	 *  - The driver does not need to repeat basic input validation like
> +	 *    done in the exynos_drm_ipp_task_check() function. The core does
> +	 *    that before calling this hook.
> +	 *
> +	 * RETURNS:
> +	 *
> +	 * 0 on success or one of the below negative error codes:
> +	 *
> +	 *  - -EINVAL, if any of the above constraints are violated.
> +	 */
> +	int (*check)(struct exynos_drm_ipp *ipp,
> +		     struct exynos_drm_ipp_task *task);
> +
> +	/**
> +	 * @commit:
> +	 *
> +	 * This is the main entry point to start framebuffer processing
> +	 * in the hardware. The exynos_drm_ipp_task has been already validated.
> +	 * This function must not wait until the device finishes processing.
> +	 * When the driver finishes processing, it has to call
> +	 * exynos_exynos_drm_ipp_task_done() function.
> +	 *
> +	 * RETURNS:
> +	 *
> +	 * 0 on success or negative error codes in case of failure.
> +	 */
> +	int (*commit)(struct exynos_drm_ipp *ipp,
> +		      struct exynos_drm_ipp_task *task);
> +
> +	/**
> +	 * @abort:
> +	 *
> +	 * Informs the driver that it has to abort the currently running
> +	 * task as soon as possible (i.e. as soon as it can stop the device
> +	 * safely), even if the task would not have been finished by then.
> +	 * After the driver performs the necessary steps, it has to call
> +	 * exynos_drm_ipp_task_done() (as if the task ended normally).
> +	 * This function does not have to (and will usually not) wait
> +	 * until the device enters a state when it can be stopped.
> +	 */
> +	void (*abort)(struct exynos_drm_ipp *ipp,
> +		      struct exynos_drm_ipp_task *task);
> +};
> +
> +/**
> + * struct exynos_drm_ipp - central picture processor module structure
> + */
> +struct exynos_drm_ipp {
> +	struct drm_device *dev;
> +	struct list_head head;
> +	unsigned int id;
> +
> +	const char *name;
> +	const struct exynos_drm_ipp_funcs *funcs;
> +	unsigned int capabilities;
> +	const struct exynos_drm_ipp_formats *formats;
> +	unsigned int num_formats;
> +	atomic_t sequence;
> +
> +	spinlock_t lock;
> +	struct exynos_drm_ipp_task *task;
> +	struct list_head todo_list;
> +	wait_queue_head_t done_wq;
> +};
> +
> +struct exynos_drm_ipp_buffer {
> +	struct drm_exynos_ipp_task_buffer buf;
> +	struct drm_exynos_ipp_task_rect rect;
> +
> +	struct exynos_drm_gem *exynos_gem[MAX_FB_BUFFER];
> +	const struct drm_format_info *format;
> +	dma_addr_t dma_addr[MAX_FB_BUFFER];
> +};
> +
> +/**
> + * struct exynos_drm_ipp_task - a structure describing transformation that
> + * has to be performed by the picture processor hardware module
> + */
> +struct exynos_drm_ipp_task {
> +	struct drm_device *dev;
> +	struct exynos_drm_ipp *ipp;
> +	struct list_head head;
> +
> +	struct exynos_drm_ipp_buffer src;
> +	struct exynos_drm_ipp_buffer dst;
> +
> +	struct drm_exynos_ipp_task_transform transform;
> +	struct drm_exynos_ipp_task_alpha alpha;
> +
> +	struct work_struct cleanup_work;
> +	unsigned int flags;
> +	int ret;
> +
> +	struct drm_pending_exynos_ipp_event *event;
> +};
> +
> +#define DRM_EXYNOS_IPP_TASK_DONE	(1 << 0)
> +#define DRM_EXYNOS_IPP_TASK_ASYNC	(1 << 1)
> +
> +struct exynos_drm_ipp_formats {
> +	u32 fourcc;
> +	u32 type;
> +	u64 modifier;
> +	const struct drm_exynos_ipp_limit *limits;
> +};
> +
> +/* helper macros to set exynos_drm_ipp_formats structure and limits*/
> +#define IPP_SRCDST_MFORMAT(f, m, l) \
> +     .fourcc = DRM_FORMAT_##f, .modifier = m, .limits = l, \
> +     .type = (DRM_EXYNOS_IPP_FORMAT_SOURCE | DRM_EXYNOS_IPP_FORMAT_DESTINATION)
> +
> +#define IPP_SRCDST_FORMAT(f, l) IPP_SRCDST_MFORMAT(f, 0, l)
> +
> +#define IPP_SIZE_LIMIT(l, val...)	\
> +     .type = (DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE | DRM_EXYNOS_IPP_LIMIT_SIZE_##l), \
> +     val
> +
> +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp,
> +		const struct exynos_drm_ipp_funcs *funcs, unsigned int caps,
> +		const struct exynos_drm_ipp_formats *formats, const char *name);
> +void exynos_drm_ipp_unregister(struct drm_device *dev, struct exynos_drm_ipp *ipp);
> +
> +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret);
> +
> +#ifdef CONFIG_DRM_EXYNOS_IPP
> +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data,
> +				 struct drm_file *file_priv);
> +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data,
> +				  struct drm_file *file_priv);
> +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data,
> +				    struct drm_file *file_priv);
> +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev,
> +				void *data, struct drm_file *file_priv);
> +#else
> +static inline int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev,
> +	 void *data, struct drm_file *file_priv)
> +{
> +	return -ENODEV;
> +}
> +static inline int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev,
> +	 void *data, struct drm_file *file_priv)
> +{
> +	return -ENODEV;
> +}
> +static inline int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev,
> +	 void *data, struct drm_file *file_priv)
> +{
> +	return -ENODEV;
> +}
> +static inline int exynos_drm_ipp_commit_ioctl(struct drm_device *dev,
> +	 void *data, struct drm_file *file_priv)
> +{
> +	return -ENODEV;
> +}
> +#endif
> +#endif
> diff --git a/include/uapi/drm/exynos_drm.h b/include/uapi/drm/exynos_drm.h
> index 544db2cecb36..239e09022d75 100644
> --- a/include/uapi/drm/exynos_drm.h
> +++ b/include/uapi/drm/exynos_drm.h
> @@ -134,6 +134,208 @@ struct drm_exynos_g2d_exec {
>  	__u64					async;
>  };
>  
> +/* Exynos DRM IPP v2 API */
> +
> +/**
> + * Enumerate available IPP hardware modules.
> + *
> + * @count_ipps: size of ipp_id array / number of ipp modules (set by driver)
> + * @reserved: padding
> + * @ipp_id_ptr: pointer to ipp_id array or NULL
> + */
> +struct drm_exynos_ioctl_ipp_get_res {
> +	__u32 count_ipps;
> +	__u32 reserved;
> +	__u64 ipp_id_ptr;
> +};
> +
> +enum drm_exynos_ipp_format_type {
> +	DRM_EXYNOS_IPP_FORMAT_SOURCE		= 0x01,
> +	DRM_EXYNOS_IPP_FORMAT_DESTINATION	= 0x02,
> +};
> +
> +struct exynos_drm_ipp_format {
> +	__u32 fourcc;
> +	__u32 type;
> +	__u64 modifier;
> +};
> +
> +enum drm_exynos_ipp_capability {
> +	DRM_EXYNOS_IPP_CAP_CROP		= 0x01,
> +	DRM_EXYNOS_IPP_CAP_ROTATE	= 0x02,
> +	DRM_EXYNOS_IPP_CAP_SCALE	= 0x04,
> +	DRM_EXYNOS_IPP_CAP_CONVERT	= 0x08,
> +};
> +
> +/**
> + * Get IPP hardware capabilities and supported image formats.
> + *
> + * @ipp_id: id of IPP module to query
> + * @capabilities: bitmask of drm_exynos_ipp_capability (set by driver)
> + * @reserved: padding
> + * @formats_count: size of formats array (in entries) / number of filled
> + *		   formats (set by driver)
> + * @formats_ptr: pointer to formats array or NULL
> + */
> +struct drm_exynos_ioctl_ipp_get_caps {
> +	__u32 ipp_id;
> +	__u32 capabilities;
> +	__u32 reserved;
> +	__u32 formats_count;
> +	__u64 formats_ptr;
> +};
> +
> +enum drm_exynos_ipp_limit_type {
> +	/* size (horizontal/vertial) limits, in pixels (min, max, alignment) */
> +	DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE		= 0x0001,
> +	/* scale ratio (horizonta/vertial), 16.16 fixed point (min, max) */
> +	DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE		= 0x0002,
> +
> +	/* image buffer area */
> +	DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER	= 0x0001 << 16,
> +	/* src/dst rectangle area */
> +	DRM_EXYNOS_IPP_LIMIT_SIZE_AREA		= 0x0002 << 16,
> +	/* src/dst rectangle area when rotation enabled */
> +	DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED	= 0x0003 << 16,
> +
> +	DRM_EXYNOS_IPP_LIMIT_TYPE_MASK		= 0x000f,
> +	DRM_EXYNOS_IPP_LIMIT_SIZE_MASK		= 0x000f << 16,
> +};
> +
> +struct drm_exynos_ipp_limit_val {
> +	__u32 min, max, align;
> +};
> +
> +/**
> + * IPP module limitation.
> + *
> + * @type: limit type (see drm_exynos_ipp_limit_type enum)
> + * @h, @v: horizontal and vertical limits
> + */
> +struct drm_exynos_ipp_limit {
> +	__u32 type;
> +	struct drm_exynos_ipp_limit_val h;
> +	struct drm_exynos_ipp_limit_val v;
> +};
> +
> +/**
> + * Get IPP limits for given image format.
> + *
> + * @ipp_id: id of IPP module to query
> + * @fourcc: image format code (see DRM_FORMAT_* in drm_fourcc.h)
> + * @modifier: image format modifier (see DRM_FORMAT_MOD_* in drm_fourcc.h)
> + * @type: source/destination identifier (drm_exynos_ipp_format_flag enum)
> + * @limits_size: size of limits array (in bytes) / size of filled entries
> + *		 (set by driver)
> + * @limits_ptr: pointer to limits array or NULL
> + */
> +struct drm_exynos_ioctl_ipp_get_limits {
> +	__u32 ipp_id;
> +	__u32 fourcc;
> +	__u64 modifier;
> +	__u32 type;
> +	__u32 limits_size;
> +	__u64 limits_ptr;
> +};
> +
> +enum drm_exynos_ipp_task_id {
> +	/* buffer described by struct drm_exynos_ipp_task_buffer */
> +	DRM_EXYNOS_IPP_TASK_BUFFER		= 0x0001,
> +	/* rectangle described by struct drm_exynos_ipp_task_rect */
> +	DRM_EXYNOS_IPP_TASK_RECTANGLE		= 0x0002,
> +	/* transformation described by struct drm_exynos_ipp_task_transform */
> +	DRM_EXYNOS_IPP_TASK_TRANSFORM		= 0x0003,
> +	/* alpha configuration described by struct drm_exynos_ipp_task_alpha */
> +	DRM_EXYNOS_IPP_TASK_ALPHA		= 0x0004,
> +
> +	/* source image data (for buffer and rectangle chunks) */
> +	DRM_EXYNOS_IPP_TASK_TYPE_SOURCE		= 0x0001 << 16,
> +	/* destination image data (for buffer and rectangle chunks) */
> +	DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION	= 0x0002 << 16,
> +};
> +
> +/**
> + * Memory buffer with image data.
> + *
> + * @id: must be DRM_EXYNOS_IPP_TASK_BUFFER
> + * other parameters are same as for AddFB2 generic DRM ioctl
> + */
> +struct drm_exynos_ipp_task_buffer {
> +	__u32	id;
> +	__u32	fourcc;
> +	__u32	width, height;
> +	__u32	gem_id[4];
> +	__u32	offset[4];
> +	__u32	pitch[4];
> +	__u64	modifier;
> +};
> +
> +/**
> + * Rectangle for processing.
> + *
> + * @id: must be DRM_EXYNOS_IPP_TASK_RECTANGLE
> + * @x,@y: left corner in pixels
> + * @w,@h: width/height in pixels
> + */
> +struct drm_exynos_ipp_task_rect {
> +	__u32	id;
> +	__u32	x, y, w, h;
> +};
> +
> +/**
> + * Image tranformation description.
> + *
> + * @id: must be DRM_EXYNOS_IPP_TASK_TRANSFORM
> + * @rotation: DRM_MODE_ROTATE_* and DRM_MODE_REFLECT_* values
> + */
> +struct drm_exynos_ipp_task_transform {
> +	__u32	id;
> +	__u32	rotation;
> +};
> +
> +/**
> + * Image global alpha configuration for formats without alpha values.
> + *
> + * @id: must be DRM_EXYNOS_IPP_TASK_ALPHA
> + * @value: global alpha value (0-255)
> + */
> +struct drm_exynos_ipp_task_alpha {
> +	__u32	id;
> +	__u32	value;
> +};
> +
> +enum drm_exynos_ipp_flag {
> +	/* generate DRM event after processing */
> +	DRM_EXYNOS_IPP_FLAG_EVENT	= 0x01,
> +	/* dry run, only check task parameters */
> +	DRM_EXYNOS_IPP_FLAG_TEST_ONLY	= 0x02,
> +	/* non-blocking processing */
> +	DRM_EXYNOS_IPP_FLAG_NONBLOCK	= 0x04,
> +};
> +
> +#define DRM_EXYNOS_IPP_FLAGS (DRM_EXYNOS_IPP_FLAG_EVENT |\
> +		DRM_EXYNOS_IPP_FLAG_TEST_ONLY | DRM_EXYNOS_IPP_FLAG_NONBLOCK)
> +
> +/**
> + * Perform image processing described by array of drm_exynos_ipp_task_*
> + * structures (parameters array).
> + *
> + * @ipp_id: id of IPP module to run the task
> + * @flags: bitmask of drm_exynos_ipp_flag values
> + * @reserved: padding
> + * @params_size: size of parameters array (in bytes)
> + * @params_ptr: pointer to parameters array or NULL
> + * @user_data: (optional) data for drm event
> + */
> +struct drm_exynos_ioctl_ipp_commit {
> +	__u32 ipp_id;
> +	__u32 flags;
> +	__u32 reserved;
> +	__u32 params_size;
> +	__u64 params_ptr;
> +	__u64 user_data;
> +};
> +
>  #define DRM_EXYNOS_GEM_CREATE		0x00
>  #define DRM_EXYNOS_GEM_MAP		0x01
>  /* Reserved 0x03 ~ 0x05 for exynos specific gem ioctl */
> @@ -146,6 +348,11 @@ struct drm_exynos_g2d_exec {
>  #define DRM_EXYNOS_G2D_EXEC		0x22
>  
>  /* Reserved 0x30 ~ 0x33 for obsolete Exynos IPP ioctls */
> +/* IPP - Image Post Processing */
> +#define DRM_EXYNOS_IPP_GET_RESOURCES	0x40
> +#define DRM_EXYNOS_IPP_GET_CAPS		0x41
> +#define DRM_EXYNOS_IPP_GET_LIMITS	0x42
> +#define DRM_EXYNOS_IPP_COMMIT		0x43
>  
>  #define DRM_IOCTL_EXYNOS_GEM_CREATE		DRM_IOWR(DRM_COMMAND_BASE + \
>  		DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create)
> @@ -164,8 +371,18 @@ 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_RESOURCES	DRM_IOWR(DRM_COMMAND_BASE + \
> +		DRM_EXYNOS_IPP_GET_RESOURCES, struct drm_exynos_ioctl_ipp_get_res)
> +#define DRM_IOCTL_EXYNOS_IPP_GET_CAPS		DRM_IOWR(DRM_COMMAND_BASE + \
> +		DRM_EXYNOS_IPP_GET_CAPS, struct drm_exynos_ioctl_ipp_get_caps)
> +#define DRM_IOCTL_EXYNOS_IPP_GET_LIMITS		DRM_IOWR(DRM_COMMAND_BASE + \
> +		DRM_EXYNOS_IPP_GET_LIMITS, struct drm_exynos_ioctl_ipp_get_limits)
> +#define DRM_IOCTL_EXYNOS_IPP_COMMIT		DRM_IOWR(DRM_COMMAND_BASE + \
> +		DRM_EXYNOS_IPP_COMMIT, struct drm_exynos_ioctl_ipp_commit)
> +
>  /* EXYNOS specific events */
>  #define DRM_EXYNOS_G2D_EVENT		0x80000000
> +#define DRM_EXYNOS_IPP_EVENT		0x80000002
>  
>  struct drm_exynos_g2d_event {
>  	struct drm_event	base;
> @@ -176,6 +393,16 @@ 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			ipp_id;
> +	__u32			sequence;
> +	__u64			reserved;
> +};
> +
>  #if defined(__cplusplus)
>  }
>  #endif
> 

Re: [PATCH 0/6] Exynos DRM: rewrite IPP subsystem and userspace API

[Tobias Jakobi]

Hello everyone,

just finished a first draft of the libdrm API and some test application:
https://github.com/tobiasjakobi/libdrm/blob/ippv2/exynos/exynos_ipp.h
https://github.com/tobiasjakobi/libdrm/blob/ippv2/tests/exynos/exynos_ipp_test.c

Needs a small patch to the kernel to prevent an Oops when a format has no limits.

Does not a lot atm. Just pretty-printing the information from the kernel driver
and testing a simple CSC queue.

Now looking to using this for the mpv vo backend.

With best wishes,
Tobias


P.S.: Output from my system below.

> exynos/ipp: 3 IPP modules found.
> IPP module number 0:
> 	id = 0
> 	supports crop
> 	supports rotate
> 	supports scale
> 	supports convert
> a total of 15 formats are supported:
> 	fourcc: XR24, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: RG24, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: RG16, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: NV12, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: NV16, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: NV21, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: NV61, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: UYVY, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: VYUY, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YUYV, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YVYU, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YU12, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YV12, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YU16, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YU24, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> IPP module number 1:
> 	id = 1
> 	supports crop
> 	supports rotate
> 	supports scale
> 	supports convert
> a total of 15 formats are supported:
> 	fourcc: XR24, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: RG24, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: RG16, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: NV12, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: NV16, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: NV21, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: NV61, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: UYVY, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: VYUY, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YUYV, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YVYU, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YU12, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YV12, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YU16, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> 	fourcc: YU24, type: source+destination, modifier: 0x0
> 		format has no source limits
> 		format has no destination limits
> IPP module number 2:
> 	id = 2
> 	supports crop
> 	supports rotate
> a total of 2 formats are supported:
> 	fourcc: XR24, type: source+destination, modifier: 0x0
> 		source limit: type = size (horizontal/vertical) in pixels, size = image buffer area
> 		h_min = 8, h_max = 8192, h_align = 0
> 		v_min = 8, v_max = 8192, v_align = 0
> 		source limit: type = size (horizontal/vertical) in pixels, size = rectangle area
> 		h_min = 0, h_max = 0, h_align = 4
> 		v_min = 0, v_max = 0, v_align = 4
> 		destination limit: type = size (horizontal/vertical) in pixels, size = image buffer area
> 		h_min = 8, h_max = 8192, h_align = 0
> 		v_min = 8, v_max = 8192, v_align = 0
> 		destination limit: type = size (horizontal/vertical) in pixels, size = rectangle area
> 		h_min = 0, h_max = 0, h_align = 4
> 		v_min = 0, v_max = 0, v_align = 4
> 	fourcc: NV12, type: source+destination, modifier: 0x0
> 		source limit: type = size (horizontal/vertical) in pixels, size = image buffer area
> 		h_min = 32, h_max = 32768, h_align = 0
> 		v_min = 32, v_max = 32768, v_align = 0
> 		source limit: type = size (horizontal/vertical) in pixels, size = rectangle area
> 		h_min = 0, h_max = 0, h_align = 8
> 		v_min = 0, v_max = 0, v_align = 8
> 		destination limit: type = size (horizontal/vertical) in pixels, size = image buffer area
> 		h_min = 32, h_max = 32768, h_align = 0
> 		v_min = 32, v_max = 32768, v_align = 0
> 		destination limit: type = size (horizontal/vertical) in pixels, size = rectangle area
> 		h_min = 0, h_max = 0, h_align = 8
> 		v_min = 0, v_max = 0, v_align = 8
> IPP handler:
> 	fd = 3, ipp_id = 0, seq = 29, tv_sec = 2659, tv_usec = 709341
> 	user_data = 0x0xc47a30
> IPP handler:
> 	fd = 3, ipp_id = 0, seq = 30, tv_sec = 2659, tv_usec = 711837
> 	user_data = 0x0xc47a40
> IPP handler:
> 	fd = 3, ipp_id = 0, seq = 31, tv_sec = 2659, tv_usec = 714027
> 	user_data = 0x0xc47a50




Marek Szyprowski wrote:
> Dear all,
> 
> This patchset performs complete rewrite of Exynos DRM IPP subsystem and
> its userspace API.
> 
> Why such rewrite is needed? Exynos DRM IPP API is over-engineered in
> general, but not really extensible on the other side. It is also buggy,
> with significant design flaws:
> - Userspace API covers memory-2-memory picture operations together with
>   CRTC writeback and duplicating features, which belongs to video plane.
> - Lack of support of the all required image formats (for example NV12
>   Samsung-tiled cannot be used due to lack of pixel format modifier
>   support).
> - Userspace API designed only to mimic hardware behaviour, not easy to
>   understand.
> - Lack of proper input validation in the core, drivers also didn't do that
>   correctly, so it was possible to set incorrect parameters and easil
>   trigger IOMMU fault or memory trash.
> - Drivers were partially disfunctional or supported only a subset of modes.
> 
> Due to the above limitations and issues the Exynos DRM IPP API was not
> used by any of the open-source projects. I assume that it is safe to remove
> this broken API without any damage to open-source community. All remaining
> users (mainly Tizen project related) will be updated to the new version.
> 
> This patchset changes Exynos DRM IPP subsystem to something useful. The
> userspace API is much simpler, state-less and easy to understand. Also
> the code of the core and driver is significantly smaller and easier to
> understand.
> 
> Patches were tested on Exynos4412 based Odroid U3 and Exynos5422
> Odroid XU3 boards, on top of Linux next-20170911 kernel.
> 
> Best regards
> Marek Szyprowski
> Samsung R&D Institute Poland
> 
> 
> My previous works in this area:
> 
> "[RFC v2 0/2] Exynos DRM: add Picture Processor extension"
> https://www.spinics.net/lists/dri-devel/msg140669.html
> - removed usage of DRM objects and properties - replaced them with simple
>   list of parameters with predefined IDs
> 
> "[RFC 0/4] Exynos DRM: add Picture Processor extension"
> https://www.spinics.net/lists/linux-samsung-soc/msg59323.html
> - moved this feature from DRM core to Exynos DRM driver
> - changed name from framebuffer processor to picture processor
> - simplified code to cover only things needed by Exynos drivers
> - implemented simple fifo task scheduler
> - cleaned up rotator driver conversion (removed IPP remainings)
> 
> "[RFC 0/2] New feature: Framebuffer processors"
> https://www.spinics.net/lists/linux-samsung-soc/msg54810.html
> - generic approach implemented in DRM core, rejected
> 
> 
> Patch summary:
> 
> Marek Szyprowski (6):
>   drm/exynos: ipp: Remove Exynos DRM IPP subsystem
>   drm/exynos: ipp: Add IPP v2 framework
>   drm/exynos: rotator: Convert driver to IPP v2 core API
>   drm/exynos: gsc: Convert driver to IPP v2 core API
>   drm/exynos: Add generic support for devices shared with V4L2 subsystem
>   drm/exynos: fimc: Convert driver to IPP v2 core API
> 
>  drivers/gpu/drm/exynos/Kconfig              |   12 +-
>  drivers/gpu/drm/exynos/exynos_drm_drv.c     |   33 +-
>  drivers/gpu/drm/exynos/exynos_drm_drv.h     |    4 +-
>  drivers/gpu/drm/exynos/exynos_drm_fimc.c    |  893 +++--------
>  drivers/gpu/drm/exynos/exynos_drm_fimc.h    |   23 -
>  drivers/gpu/drm/exynos/exynos_drm_gsc.c     |  853 +++-------
>  drivers/gpu/drm/exynos/exynos_drm_gsc.h     |   24 -
>  drivers/gpu/drm/exynos/exynos_drm_ipp.c     | 2239 ++++++++-------------------
>  drivers/gpu/drm/exynos/exynos_drm_ipp.h     |  352 ++---
>  drivers/gpu/drm/exynos/exynos_drm_rotator.c |  735 ++-------
>  drivers/gpu/drm/exynos/exynos_drm_rotator.h |   19 -
>  include/uapi/drm/exynos_drm.h               |  315 ++--
>  12 files changed, 1627 insertions(+), 3875 deletions(-)
>  delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_fimc.h
>  delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_gsc.h
>  delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_rotator.h
> 

Re: [PATCH 1/6] drm/exynos: ipp: Remove Exynos DRM IPP subsystem

[Hoegeun Kwon]

Hi Marek,

Tested-by: Hoegeun Kwon <hoegeun.kwon@samsung.com>

Best regards,
Hoegeun


On 09/12/2017 05:08 PM, Marek Szyprowski wrote:
> Exynos IPP will be rewritten, so remove current IPP core code and mark
> existing drivers as BROKEN.
>
> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
> ---
>   drivers/gpu/drm/exynos/Kconfig          |   11 +-
>   drivers/gpu/drm/exynos/Makefile         |    1 -
>   drivers/gpu/drm/exynos/exynos_drm_drv.c |   12 -
>   drivers/gpu/drm/exynos/exynos_drm_drv.h |    2 -
>   drivers/gpu/drm/exynos/exynos_drm_ipp.c | 1806 -------------------------------
>   drivers/gpu/drm/exynos/exynos_drm_ipp.h |  252 -----
>   include/uapi/drm/exynos_drm.h           |  192 +---
>   7 files changed, 4 insertions(+), 2272 deletions(-)
>   delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.c
>   delete 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 305dc3d4ff77..88cff0e039b6 100644
> --- a/drivers/gpu/drm/exynos/Kconfig
> +++ b/drivers/gpu/drm/exynos/Kconfig
> @@ -94,26 +94,21 @@ config DRM_EXYNOS_G2D
>   	help
>   	  Choose this option if you want to use Exynos G2D for DRM.
>   
> -config DRM_EXYNOS_IPP
> -	bool "Image Post Processor"
> -	help
> -	  Choose this option if you want to use IPP feature for DRM.
> -
>   config DRM_EXYNOS_FIMC
>   	bool "FIMC"
> -	depends on DRM_EXYNOS_IPP && MFD_SYSCON
> +	depends on BROKEN && MFD_SYSCON
>   	help
>   	  Choose this option if you want to use Exynos FIMC for DRM.
>   
>   config DRM_EXYNOS_ROTATOR
>   	bool "Rotator"
> -	depends on DRM_EXYNOS_IPP
> +	depends on BROKEN
>   	help
>   	  Choose this option if you want to use Exynos Rotator for DRM.
>   
>   config DRM_EXYNOS_GSC
>   	bool "GScaler"
> -	depends on DRM_EXYNOS_IPP && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n
> +	depends on BROKEN && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n
>   	help
>   	  Choose this option if you want to use Exynos GSC for DRM.
>   
> diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile
> index f663490e949d..09bb002c9555 100644
> --- a/drivers/gpu/drm/exynos/Makefile
> +++ b/drivers/gpu/drm/exynos/Makefile
> @@ -17,7 +17,6 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER)	+= exynos_mixer.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI)	+= exynos_hdmi.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI)	+= exynos_drm_vidi.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_G2D)	+= exynos_drm_g2d.o
> -exynosdrm-$(CONFIG_DRM_EXYNOS_IPP)	+= exynos_drm_ipp.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC)	+= exynos_drm_fimc.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR)	+= exynos_drm_rotator.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_GSC)	+= exynos_drm_gsc.o
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
> index b1f7299600f0..7f19054ebce3 100644
> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
> @@ -28,7 +28,6 @@
>   #include "exynos_drm_plane.h"
>   #include "exynos_drm_vidi.h"
>   #include "exynos_drm_g2d.h"
> -#include "exynos_drm_ipp.h"
>   #include "exynos_drm_iommu.h"
>   
>   #define DRIVER_NAME	"exynos"
> @@ -115,14 +114,6 @@ static void exynos_drm_lastclose(struct drm_device *dev)
>   			DRM_AUTH | DRM_RENDER_ALLOW),
>   	DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl,
>   			DRM_AUTH | DRM_RENDER_ALLOW),
> -	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_PROPERTY, exynos_drm_ipp_get_property,
> -			DRM_AUTH | DRM_RENDER_ALLOW),
> -	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_SET_PROPERTY, exynos_drm_ipp_set_property,
> -			DRM_AUTH | DRM_RENDER_ALLOW),
> -	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_QUEUE_BUF, exynos_drm_ipp_queue_buf,
> -			DRM_AUTH | DRM_RENDER_ALLOW),
> -	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_CMD_CTRL, exynos_drm_ipp_cmd_ctrl,
> -			DRM_AUTH | DRM_RENDER_ALLOW),
>   };
>   
>   static const struct file_operations exynos_drm_driver_fops = {
> @@ -272,9 +263,6 @@ struct exynos_drm_driver_info {
>   	}, {
>   		DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC),
>   	}, {
> -		DRV_PTR(ipp_driver, CONFIG_DRM_EXYNOS_IPP),
> -		DRM_VIRTUAL_DEVICE
> -	}, {
>   		&exynos_drm_platform_driver,
>   		DRM_VIRTUAL_DEVICE
>   	}
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h b/drivers/gpu/drm/exynos/exynos_drm_drv.h
> index cf131c2aa23e..21f4271c012a 100644
> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.h
> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h
> @@ -185,7 +185,6 @@ struct exynos_drm_g2d_private {
>   
>   struct drm_exynos_file_private {
>   	struct exynos_drm_g2d_private	*g2d_priv;
> -	struct device			*ipp_dev;
>   };
>   
>   /*
> @@ -292,6 +291,5 @@ int exynos_atomic_commit(struct drm_device *dev, struct drm_atomic_state *state,
>   extern struct platform_driver fimc_driver;
>   extern struct platform_driver rotator_driver;
>   extern struct platform_driver gsc_driver;
> -extern struct platform_driver ipp_driver;
>   extern struct platform_driver mic_driver;
>   #endif
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.c b/drivers/gpu/drm/exynos/exynos_drm_ipp.c
> deleted file mode 100644
> index 3edda18cc2d2..000000000000
> --- a/drivers/gpu/drm/exynos/exynos_drm_ipp.c
> +++ /dev/null
> @@ -1,1806 +0,0 @@
> -/*
> - * Copyright (C) 2012 Samsung Electronics Co.Ltd
> - * Authors:
> - *	Eunchul Kim <chulspro.kim@samsung.com>
> - *	Jinyoung Jeon <jy0.jeon@samsung.com>
> - *	Sangmin Lee <lsmin.lee@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/platform_device.h>
> -#include <linux/types.h>
> -#include <linux/clk.h>
> -#include <linux/pm_runtime.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"
> -#include "exynos_drm_iommu.h"
> -
> -/*
> - * IPP stands 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
> - */
> -
> -/*
> - * TODO
> - * 1. expand command control id.
> - * 2. integrate	property and config.
> - * 3. removed send_event id check routine.
> - * 4. compare send_event id if needed.
> - * 5. free subdrv_remove notifier callback list if needed.
> - * 6. need to check subdrv_open about multi-open.
> - * 7. need to power_on implement power and sysmmu ctrl.
> - */
> -
> -#define get_ipp_context(dev)	platform_get_drvdata(to_platform_device(dev))
> -#define ipp_is_m2m_cmd(c)	(c == IPP_CMD_M2M)
> -
> -/*
> - * 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;
> -};
> -
> -/*
> - * 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 DEFINE_MUTEX(exynos_drm_ippdrv_lock);
> -static BLOCKING_NOTIFIER_HEAD(exynos_drm_ippnb_list);
> -
> -int exynos_drm_ippdrv_register(struct exynos_drm_ippdrv *ippdrv)
> -{
> -	mutex_lock(&exynos_drm_ippdrv_lock);
> -	list_add_tail(&ippdrv->drv_list, &exynos_drm_ippdrv_list);
> -	mutex_unlock(&exynos_drm_ippdrv_lock);
> -
> -	return 0;
> -}
> -
> -int exynos_drm_ippdrv_unregister(struct exynos_drm_ippdrv *ippdrv)
> -{
> -	mutex_lock(&exynos_drm_ippdrv_lock);
> -	list_del(&ippdrv->drv_list);
> -	mutex_unlock(&exynos_drm_ippdrv_lock);
> -
> -	return 0;
> -}
> -
> -static int ipp_create_id(struct idr *id_idr, struct mutex *lock, void *obj)
> -{
> -	int ret;
> -
> -	mutex_lock(lock);
> -	ret = idr_alloc(id_idr, obj, 1, 0, GFP_KERNEL);
> -	mutex_unlock(lock);
> -
> -	return ret;
> -}
> -
> -static void ipp_remove_id(struct idr *id_idr, struct mutex *lock, u32 id)
> -{
> -	mutex_lock(lock);
> -	idr_remove(id_idr, id);
> -	mutex_unlock(lock);
> -}
> -
> -static void *ipp_find_obj(struct idr *id_idr, struct mutex *lock, u32 id)
> -{
> -	void *obj;
> -
> -	mutex_lock(lock);
> -	obj = idr_find(id_idr, id);
> -	mutex_unlock(lock);
> -
> -	return obj;
> -}
> -
> -static int ipp_check_driver(struct exynos_drm_ippdrv *ippdrv,
> -			    struct drm_exynos_ipp_property *property)
> -{
> -	if (ippdrv->dedicated || (!ipp_is_m2m_cmd(property->cmd) &&
> -				  !pm_runtime_suspended(ippdrv->dev)))
> -		return -EBUSY;
> -
> -	if (ippdrv->check_property &&
> -	    ippdrv->check_property(ippdrv->dev, property))
> -		return -EINVAL;
> -
> -	return 0;
> -}
> -
> -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;
> -	int ret;
> -
> -	if (ipp_id) {
> -		ippdrv = ipp_find_obj(&ctx->ipp_idr, &ctx->ipp_lock, ipp_id);
> -		if (!ippdrv) {
> -			DRM_DEBUG("ipp%d driver not found\n", ipp_id);
> -			return ERR_PTR(-ENODEV);
> -		}
> -
> -		ret = ipp_check_driver(ippdrv, property);
> -		if (ret < 0) {
> -			DRM_DEBUG("ipp%d driver check error %d\n", ipp_id, ret);
> -			return ERR_PTR(ret);
> -		}
> -
> -		return ippdrv;
> -	} else {
> -		list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
> -			ret = ipp_check_driver(ippdrv, property);
> -			if (ret == 0)
> -				return ippdrv;
> -		}
> -
> -		DRM_DEBUG("cannot find driver suitable for given property.\n");
> -	}
> -
> -	return ERR_PTR(-ENODEV);
> -}
> -
> -static struct exynos_drm_ippdrv *ipp_find_drv_by_handle(u32 prop_id)
> -{
> -	struct exynos_drm_ippdrv *ippdrv;
> -	struct drm_exynos_ipp_cmd_node *c_node;
> -	int count = 0;
> -
> -	DRM_DEBUG_KMS("prop_id[%d]\n", prop_id);
> -
> -	/*
> -	 * This case is search ipp driver by prop_id handle.
> -	 * sometimes, ipp subsystem find driver by prop_id.
> -	 * e.g PAUSE state, queue buf, command control.
> -	 */
> -	list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
> -		DRM_DEBUG_KMS("count[%d]ippdrv[%pK]\n", count++, ippdrv);
> -
> -		mutex_lock(&ippdrv->cmd_lock);
> -		list_for_each_entry(c_node, &ippdrv->cmd_list, list) {
> -			if (c_node->property.prop_id == prop_id) {
> -				mutex_unlock(&ippdrv->cmd_lock);
> -				return ippdrv;
> -			}
> -		}
> -		mutex_unlock(&ippdrv->cmd_lock);
> -	}
> -
> -	return ERR_PTR(-ENODEV);
> -}
> -
> -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 device *dev = file_priv->ipp_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;
> -
> -	if (!ctx) {
> -		DRM_ERROR("invalid context.\n");
> -		return -EINVAL;
> -	}
> -
> -	if (!prop_list) {
> -		DRM_ERROR("invalid property parameter.\n");
> -		return -EINVAL;
> -	}
> -
> -	DRM_DEBUG_KMS("ipp_id[%d]\n", prop_list->ipp_id);
> -
> -	if (!prop_list->ipp_id) {
> -		list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list)
> -			count++;
> -
> -		/*
> -		 * Supports ippdrv list count for user application.
> -		 * First step user application getting ippdrv count.
> -		 * and second step getting ippdrv capability using ipp_id.
> -		 */
> -		prop_list->count = count;
> -	} else {
> -		/*
> -		 * Getting ippdrv capability by ipp_id.
> -		 * some device not supported wb, output interface.
> -		 * so, user application detect correct ipp driver
> -		 * using this ioctl.
> -		 */
> -		ippdrv = ipp_find_obj(&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 -ENODEV;
> -		}
> -
> -		*prop_list = ippdrv->prop_list;
> -	}
> -
> -	return 0;
> -}
> -
> -static void ipp_print_property(struct drm_exynos_ipp_property *property,
> -		int idx)
> -{
> -	struct drm_exynos_ipp_config *config = &property->config[idx];
> -	struct drm_exynos_pos *pos = &config->pos;
> -	struct drm_exynos_sz *sz = &config->sz;
> -
> -	DRM_DEBUG_KMS("prop_id[%d]ops[%s]fmt[0x%x]\n",
> -		property->prop_id, idx ? "dst" : "src", config->fmt);
> -
> -	DRM_DEBUG_KMS("pos[%d %d %d %d]sz[%d %d]f[%d]r[%d]\n",
> -		pos->x, pos->y, pos->w, pos->h,
> -		sz->hsize, sz->vsize, config->flip, config->degree);
> -}
> -
> -static struct drm_exynos_ipp_cmd_work *ipp_create_cmd_work(void)
> -{
> -	struct drm_exynos_ipp_cmd_work *cmd_work;
> -
> -	cmd_work = kzalloc(sizeof(*cmd_work), GFP_KERNEL);
> -	if (!cmd_work)
> -		return ERR_PTR(-ENOMEM);
> -
> -	INIT_WORK((struct work_struct *)cmd_work, ipp_sched_cmd);
> -
> -	return cmd_work;
> -}
> -
> -static struct drm_exynos_ipp_event_work *ipp_create_event_work(void)
> -{
> -	struct drm_exynos_ipp_event_work *event_work;
> -
> -	event_work = kzalloc(sizeof(*event_work), GFP_KERNEL);
> -	if (!event_work)
> -		return ERR_PTR(-ENOMEM);
> -
> -	INIT_WORK(&event_work->work, ipp_sched_event);
> -
> -	return event_work;
> -}
> -
> -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 device *dev = file_priv->ipp_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;
> -	u32 prop_id;
> -	int ret, i;
> -
> -	if (!ctx) {
> -		DRM_ERROR("invalid context.\n");
> -		return -EINVAL;
> -	}
> -
> -	if (!property) {
> -		DRM_ERROR("invalid property parameter.\n");
> -		return -EINVAL;
> -	}
> -
> -	prop_id = property->prop_id;
> -
> -	/*
> -	 * This is log print for user application property.
> -	 * user application set various property.
> -	 */
> -	for_each_ipp_ops(i)
> -		ipp_print_property(property, i);
> -
> -	/*
> -	 * In case prop_id is not zero try to set existing property.
> -	 */
> -	if (prop_id) {
> -		c_node = ipp_find_obj(&ctx->prop_idr, &ctx->prop_lock, prop_id);
> -
> -		if (!c_node || c_node->filp != file) {
> -			DRM_DEBUG_KMS("prop_id[%d] not found\n", prop_id);
> -			return -EINVAL;
> -		}
> -
> -		if (c_node->state != IPP_STATE_STOP) {
> -			DRM_DEBUG_KMS("prop_id[%d] not stopped\n", prop_id);
> -			return -EINVAL;
> -		}
> -
> -		c_node->property = *property;
> -
> -		return 0;
> -	}
> -
> -	/* find ipp driver using ipp id */
> -	ippdrv = ipp_find_driver(ctx, property);
> -	if (IS_ERR(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)
> -		return -ENOMEM;
> -
> -	ret = ipp_create_id(&ctx->prop_idr, &ctx->prop_lock, c_node);
> -	if (ret < 0) {
> -		DRM_ERROR("failed to create id.\n");
> -		goto err_clear;
> -	}
> -	property->prop_id = ret;
> -
> -	DRM_DEBUG_KMS("created prop_id[%d]cmd[%d]ippdrv[%pK]\n",
> -		property->prop_id, property->cmd, ippdrv);
> -
> -	/* stored property information and ippdrv in private data */
> -	c_node->property = *property;
> -	c_node->state = IPP_STATE_IDLE;
> -	c_node->filp = file;
> -
> -	c_node->start_work = ipp_create_cmd_work();
> -	if (IS_ERR(c_node->start_work)) {
> -		DRM_ERROR("failed to create start work.\n");
> -		ret = PTR_ERR(c_node->start_work);
> -		goto err_remove_id;
> -	}
> -
> -	c_node->stop_work = ipp_create_cmd_work();
> -	if (IS_ERR(c_node->stop_work)) {
> -		DRM_ERROR("failed to create stop work.\n");
> -		ret = PTR_ERR(c_node->stop_work);
> -		goto err_free_start;
> -	}
> -
> -	c_node->event_work = ipp_create_event_work();
> -	if (IS_ERR(c_node->event_work)) {
> -		DRM_ERROR("failed to create event work.\n");
> -		ret = PTR_ERR(c_node->event_work);
> -		goto err_free_stop;
> -	}
> -
> -	mutex_init(&c_node->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);
> -	mutex_lock(&ippdrv->cmd_lock);
> -	list_add_tail(&c_node->list, &ippdrv->cmd_list);
> -	mutex_unlock(&ippdrv->cmd_lock);
> -
> -	/* make dedicated state without m2m */
> -	if (!ipp_is_m2m_cmd(property->cmd))
> -		ippdrv->dedicated = true;
> -
> -	return 0;
> -
> -err_free_stop:
> -	kfree(c_node->stop_work);
> -err_free_start:
> -	kfree(c_node->start_work);
> -err_remove_id:
> -	ipp_remove_id(&ctx->prop_idr, &ctx->prop_lock, property->prop_id);
> -err_clear:
> -	kfree(c_node);
> -	return ret;
> -}
> -
> -static int ipp_validate_mem_node(struct drm_device *drm_dev,
> -				 struct drm_exynos_ipp_mem_node *m_node,
> -				 struct drm_exynos_ipp_cmd_node *c_node)
> -{
> -	struct drm_exynos_ipp_config *ipp_cfg;
> -	unsigned int num_plane;
> -	unsigned long size, buf_size = 0, plane_size, img_size = 0;
> -	unsigned int bpp, width, height;
> -	int i;
> -
> -	ipp_cfg = &c_node->property.config[m_node->ops_id];
> -	num_plane = drm_format_num_planes(ipp_cfg->fmt);
> -
> -	/**
> -	 * This is a rather simplified validation of a memory node.
> -	 * It basically verifies provided gem object handles
> -	 * and the buffer sizes with respect to current configuration.
> -	 * This is not the best that can be done
> -	 * but it seems more than enough
> -	 */
> -	for (i = 0; i < num_plane; ++i) {
> -		width = ipp_cfg->sz.hsize;
> -		height = ipp_cfg->sz.vsize;
> -		bpp = drm_format_plane_cpp(ipp_cfg->fmt, i);
> -
> -		/*
> -		 * The result of drm_format_plane_cpp() for chroma planes must
> -		 * be used with drm_format_xxxx_chroma_subsampling() for
> -		 * correct result.
> -		 */
> -		if (i > 0) {
> -			width /= drm_format_horz_chroma_subsampling(
> -								ipp_cfg->fmt);
> -			height /= drm_format_vert_chroma_subsampling(
> -								ipp_cfg->fmt);
> -		}
> -		plane_size = width * height * bpp;
> -		img_size += plane_size;
> -
> -		if (m_node->buf_info.handles[i]) {
> -			size = exynos_drm_gem_get_size(drm_dev,
> -					m_node->buf_info.handles[i],
> -					c_node->filp);
> -			if (plane_size > size) {
> -				DRM_ERROR(
> -					"buffer %d is smaller than required\n",
> -					i);
> -				return -EINVAL;
> -			}
> -
> -			buf_size += size;
> -		}
> -	}
> -
> -	if (buf_size < img_size) {
> -		DRM_ERROR("size of buffers(%lu) is smaller than image(%lu)\n",
> -			buf_size, img_size);
> -		return -EINVAL;
> -	}
> -
> -	return 0;
> -}
> -
> -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;
> -
> -	DRM_DEBUG_KMS("node[%pK]\n", m_node);
> -
> -	if (!m_node) {
> -		DRM_ERROR("invalid dequeue node.\n");
> -		return -EFAULT;
> -	}
> -
> -	DRM_DEBUG_KMS("ops_id[%d]\n", 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,
> -							c_node->filp);
> -	}
> -
> -	list_del(&m_node->list);
> -	kfree(m_node);
> -
> -	return 0;
> -}
> -
> -static struct drm_exynos_ipp_mem_node
> -		*ipp_get_mem_node(struct drm_device *drm_dev,
> -		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;
> -	int i;
> -
> -	m_node = kzalloc(sizeof(*m_node), GFP_KERNEL);
> -	if (!m_node)
> -		return ERR_PTR(-ENOMEM);
> -
> -	buf_info = &m_node->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;
> -	INIT_LIST_HEAD(&m_node->list);
> -
> -	DRM_DEBUG_KMS("m_node[%pK]ops_id[%d]\n", m_node, qbuf->ops_id);
> -	DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]\n", qbuf->prop_id, m_node->buf_id);
> -
> -	for_each_ipp_planar(i) {
> -		DRM_DEBUG_KMS("i[%d]handle[0x%x]\n", i, qbuf->handle[i]);
> -
> -		/* get dma address by handle */
> -		if (qbuf->handle[i]) {
> -			dma_addr_t *addr;
> -
> -			addr = exynos_drm_gem_get_dma_addr(drm_dev,
> -					qbuf->handle[i], c_node->filp);
> -			if (IS_ERR(addr)) {
> -				DRM_ERROR("failed to get addr.\n");
> -				ipp_put_mem_node(drm_dev, c_node, m_node);
> -				return ERR_PTR(-EFAULT);
> -			}
> -
> -			buf_info->handles[i] = qbuf->handle[i];
> -			buf_info->base[i] = *addr;
> -			DRM_DEBUG_KMS("i[%d]base[%pad]hd[0x%lx]\n", i,
> -				      &buf_info->base[i], buf_info->handles[i]);
> -		}
> -	}
> -
> -	mutex_lock(&c_node->mem_lock);
> -	if (ipp_validate_mem_node(drm_dev, m_node, c_node)) {
> -		ipp_put_mem_node(drm_dev, c_node, m_node);
> -		mutex_unlock(&c_node->mem_lock);
> -		return ERR_PTR(-EFAULT);
> -	}
> -	list_add_tail(&m_node->list, &c_node->mem_list[qbuf->ops_id]);
> -	mutex_unlock(&c_node->mem_lock);
> -
> -	return m_node;
> -}
> -
> -static void ipp_clean_mem_nodes(struct drm_device *drm_dev,
> -			       struct drm_exynos_ipp_cmd_node *c_node, int ops)
> -{
> -	struct drm_exynos_ipp_mem_node *m_node, *tm_node;
> -	struct list_head *head = &c_node->mem_list[ops];
> -
> -	mutex_lock(&c_node->mem_lock);
> -
> -	list_for_each_entry_safe(m_node, tm_node, head, list) {
> -		int ret;
> -
> -		ret = ipp_put_mem_node(drm_dev, c_node, m_node);
> -		if (ret)
> -			DRM_ERROR("failed to put m_node.\n");
> -	}
> -
> -	mutex_unlock(&c_node->mem_lock);
> -}
> -
> -static int ipp_get_event(struct drm_device *drm_dev,
> -		struct drm_exynos_ipp_cmd_node *c_node,
> -		struct drm_exynos_ipp_queue_buf *qbuf)
> -{
> -	struct drm_exynos_ipp_send_event *e;
> -	int ret;
> -
> -	DRM_DEBUG_KMS("ops_id[%d]buf_id[%d]\n", qbuf->ops_id, qbuf->buf_id);
> -
> -	e = kzalloc(sizeof(*e), GFP_KERNEL);
> -	if (!e)
> -		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;
> -
> -	ret = drm_event_reserve_init(drm_dev, c_node->filp, &e->base, &e->event.base);
> -	if (ret) {
> -		kfree(e);
> -		return ret;
> -	}
> -
> -	mutex_lock(&c_node->event_lock);
> -	list_add_tail(&e->base.link, &c_node->event_list);
> -	mutex_unlock(&c_node->event_lock);
> -
> -	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;
> -
> -	mutex_lock(&c_node->event_lock);
> -	list_for_each_entry_safe(e, te, &c_node->event_list, base.link) {
> -		DRM_DEBUG_KMS("count[%d]e[%pK]\n", count++, e);
> -
> -		/*
> -		 * qbuf == NULL condition means all event deletion.
> -		 * stop operations want to delete all event list.
> -		 * another case delete only same buf id.
> -		 */
> -		if (!qbuf) {
> -			/* delete list */
> -			list_del(&e->base.link);
> -			kfree(e);
> -		}
> -
> -		/* compare buffer id */
> -		if (qbuf && (qbuf->buf_id ==
> -		    e->event.buf_id[EXYNOS_DRM_OPS_DST])) {
> -			/* delete list */
> -			list_del(&e->base.link);
> -			kfree(e);
> -			goto out_unlock;
> -		}
> -	}
> -
> -out_unlock:
> -	mutex_unlock(&c_node->event_lock);
> -	return;
> -}
> -
> -static void ipp_clean_cmd_node(struct ipp_context *ctx,
> -				struct drm_exynos_ipp_cmd_node *c_node)
> -{
> -	int i;
> -
> -	/* cancel works */
> -	cancel_work_sync(&c_node->start_work->work);
> -	cancel_work_sync(&c_node->stop_work->work);
> -	cancel_work_sync(&c_node->event_work->work);
> -
> -	/* put event */
> -	ipp_put_event(c_node, NULL);
> -
> -	for_each_ipp_ops(i)
> -		ipp_clean_mem_nodes(ctx->subdrv.drm_dev, c_node, i);
> -
> -	/* delete list */
> -	list_del(&c_node->list);
> -
> -	ipp_remove_id(&ctx->prop_idr, &ctx->prop_lock,
> -			c_node->property.prop_id);
> -
> -	/* destroy mutex */
> -	mutex_destroy(&c_node->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 bool ipp_check_mem_list(struct drm_exynos_ipp_cmd_node *c_node)
> -{
> -	switch (c_node->property.cmd) {
> -	case IPP_CMD_WB:
> -		return !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_DST]);
> -	case IPP_CMD_OUTPUT:
> -		return !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_SRC]);
> -	case IPP_CMD_M2M:
> -	default:
> -		return !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_SRC]) &&
> -		       !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_DST]);
> -	}
> -}
> -
> -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("buf_id[%d]\n", qbuf->buf_id);
> -
> -	/* source/destination memory list */
> -	head = &c_node->mem_list[qbuf->ops_id];
> -
> -	/* find memory node from memory list */
> -	list_for_each_entry(m_node, head, list) {
> -		DRM_DEBUG_KMS("count[%d]m_node[%pK]\n", count++, m_node);
> -
> -		/* compare buffer id */
> -		if (m_node->buf_id == qbuf->buf_id)
> -			return m_node;
> -	}
> -
> -	return NULL;
> -}
> -
> -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("node[%pK]\n", m_node);
> -
> -	if (!m_node) {
> -		DRM_ERROR("invalid queue node.\n");
> -		return -EFAULT;
> -	}
> -
> -	DRM_DEBUG_KMS("ops_id[%d]\n", m_node->ops_id);
> -
> -	/* get operations callback */
> -	ops = ippdrv->ops[m_node->ops_id];
> -	if (!ops) {
> -		DRM_ERROR("not support ops.\n");
> -		return -EFAULT;
> -	}
> -
> -	/* 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");
> -			return ret;
> -		}
> -	}
> -
> -	return ret;
> -}
> -
> -static 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, &cmd_work->work);
> -}
> -
> -static int ipp_queue_buf_with_run(struct device *dev,
> -		struct drm_exynos_ipp_cmd_node *c_node,
> -		struct drm_exynos_ipp_mem_node *m_node,
> -		struct drm_exynos_ipp_queue_buf *qbuf)
> -{
> -	struct exynos_drm_ippdrv *ippdrv;
> -	struct drm_exynos_ipp_property *property;
> -	struct exynos_drm_ipp_ops *ops;
> -	int ret;
> -
> -	ippdrv = ipp_find_drv_by_handle(qbuf->prop_id);
> -	if (IS_ERR(ippdrv)) {
> -		DRM_ERROR("failed to get ipp driver.\n");
> -		return -EFAULT;
> -	}
> -
> -	ops = ippdrv->ops[qbuf->ops_id];
> -	if (!ops) {
> -		DRM_ERROR("failed to get ops.\n");
> -		return -EFAULT;
> -	}
> -
> -	property = &c_node->property;
> -
> -	if (c_node->state != IPP_STATE_START) {
> -		DRM_DEBUG_KMS("bypass for invalid state.\n");
> -		return 0;
> -	}
> -
> -	mutex_lock(&c_node->mem_lock);
> -	if (!ipp_check_mem_list(c_node)) {
> -		mutex_unlock(&c_node->mem_lock);
> -		DRM_DEBUG_KMS("empty memory.\n");
> -		return 0;
> -	}
> -
> -	/*
> -	 * If set destination buffer and enabled clock,
> -	 * then m2m operations need start operations at queue_buf
> -	 */
> -	if (ipp_is_m2m_cmd(property->cmd)) {
> -		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 {
> -		ret = ipp_set_mem_node(ippdrv, c_node, m_node);
> -		if (ret) {
> -			mutex_unlock(&c_node->mem_lock);
> -			DRM_ERROR("failed to set m node.\n");
> -			return ret;
> -		}
> -	}
> -	mutex_unlock(&c_node->mem_lock);
> -
> -	return 0;
> -}
> -
> -static void ipp_clean_queue_buf(struct drm_device *drm_dev,
> -		struct drm_exynos_ipp_cmd_node *c_node,
> -		struct drm_exynos_ipp_queue_buf *qbuf)
> -{
> -	struct drm_exynos_ipp_mem_node *m_node, *tm_node;
> -
> -	/* delete list */
> -	mutex_lock(&c_node->mem_lock);
> -	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->mem_lock);
> -}
> -
> -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 device *dev = file_priv->ipp_dev;
> -	struct ipp_context *ctx = get_ipp_context(dev);
> -	struct drm_exynos_ipp_queue_buf *qbuf = data;
> -	struct drm_exynos_ipp_cmd_node *c_node;
> -	struct drm_exynos_ipp_mem_node *m_node;
> -	int ret;
> -
> -	if (!qbuf) {
> -		DRM_ERROR("invalid buf parameter.\n");
> -		return -EINVAL;
> -	}
> -
> -	if (qbuf->ops_id >= EXYNOS_DRM_OPS_MAX) {
> -		DRM_ERROR("invalid ops parameter.\n");
> -		return -EINVAL;
> -	}
> -
> -	DRM_DEBUG_KMS("prop_id[%d]ops_id[%s]buf_id[%d]buf_type[%d]\n",
> -		qbuf->prop_id, qbuf->ops_id ? "dst" : "src",
> -		qbuf->buf_id, qbuf->buf_type);
> -
> -	/* find command node */
> -	c_node = ipp_find_obj(&ctx->prop_idr, &ctx->prop_lock,
> -		qbuf->prop_id);
> -	if (!c_node || c_node->filp != file) {
> -		DRM_ERROR("failed to get command node.\n");
> -		return -ENODEV;
> -	}
> -
> -	/* buffer control */
> -	switch (qbuf->buf_type) {
> -	case IPP_BUF_ENQUEUE:
> -		/* get memory node */
> -		m_node = ipp_get_mem_node(drm_dev, c_node, qbuf);
> -		if (IS_ERR(m_node)) {
> -			DRM_ERROR("failed to get m_node.\n");
> -			return PTR_ERR(m_node);
> -		}
> -
> -		/*
> -		 * first step get event for destination buffer.
> -		 * and second step when M2M case run with destination buffer
> -		 * if needed.
> -		 */
> -		if (qbuf->ops_id == EXYNOS_DRM_OPS_DST) {
> -			/* get event for destination buffer */
> -			ret = ipp_get_event(drm_dev, c_node, qbuf);
> -			if (ret) {
> -				DRM_ERROR("failed to get event.\n");
> -				goto err_clean_node;
> -			}
> -
> -			/*
> -			 * M2M case run play control for streaming feature.
> -			 * other case set address and waiting.
> -			 */
> -			ret = ipp_queue_buf_with_run(dev, c_node, m_node, qbuf);
> -			if (ret) {
> -				DRM_ERROR("failed to run command.\n");
> -				goto err_clean_node;
> -			}
> -		}
> -		break;
> -	case IPP_BUF_DEQUEUE:
> -		mutex_lock(&c_node->lock);
> -
> -		/* put event for destination buffer */
> -		if (qbuf->ops_id == EXYNOS_DRM_OPS_DST)
> -			ipp_put_event(c_node, qbuf);
> -
> -		ipp_clean_queue_buf(drm_dev, c_node, qbuf);
> -
> -		mutex_unlock(&c_node->lock);
> -		break;
> -	default:
> -		DRM_ERROR("invalid buffer control.\n");
> -		return -EINVAL;
> -	}
> -
> -	return 0;
> -
> -err_clean_node:
> -	DRM_ERROR("clean memory nodes.\n");
> -
> -	ipp_clean_queue_buf(drm_dev, c_node, qbuf);
> -	return ret;
> -}
> -
> -static bool exynos_drm_ipp_check_valid(struct device *dev,
> -		enum drm_exynos_ipp_ctrl ctrl, enum drm_exynos_ipp_state state)
> -{
> -	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;
> -	}
> -
> -	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_ippdrv *ippdrv = NULL;
> -	struct device *dev = file_priv->ipp_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;
> -
> -	if (!ctx) {
> -		DRM_ERROR("invalid context.\n");
> -		return -EINVAL;
> -	}
> -
> -	if (!cmd_ctrl) {
> -		DRM_ERROR("invalid control parameter.\n");
> -		return -EINVAL;
> -	}
> -
> -	DRM_DEBUG_KMS("ctrl[%d]prop_id[%d]\n",
> -		cmd_ctrl->ctrl, cmd_ctrl->prop_id);
> -
> -	ippdrv = ipp_find_drv_by_handle(cmd_ctrl->prop_id);
> -	if (IS_ERR(ippdrv)) {
> -		DRM_ERROR("failed to get ipp driver.\n");
> -		return PTR_ERR(ippdrv);
> -	}
> -
> -	c_node = ipp_find_obj(&ctx->prop_idr, &ctx->prop_lock,
> -		cmd_ctrl->prop_id);
> -	if (!c_node || c_node->filp != file) {
> -		DRM_ERROR("invalid command node list.\n");
> -		return -ENODEV;
> -	}
> -
> -	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:
> -		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(300))) {
> -			DRM_ERROR("timeout stop:prop_id[%d]\n",
> -				c_node->property.prop_id);
> -		}
> -
> -		c_node->state = IPP_STATE_STOP;
> -		ippdrv->dedicated = false;
> -		mutex_lock(&ippdrv->cmd_lock);
> -		ipp_clean_cmd_node(ctx, c_node);
> -
> -		if (list_empty(&ippdrv->cmd_list))
> -			pm_runtime_put_sync(ippdrv->dev);
> -		mutex_unlock(&ippdrv->cmd_lock);
> -		break;
> -	case IPP_CTRL_PAUSE:
> -		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);
> -		}
> -
> -		c_node->state = IPP_STATE_STOP;
> -		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:
> -		DRM_ERROR("could not support this state currently.\n");
> -		return -EINVAL;
> -	}
> -
> -	DRM_DEBUG_KMS("done ctrl[%d]prop_id[%d]\n",
> -		cmd_ctrl->ctrl, cmd_ctrl->prop_id);
> -
> -	return 0;
> -}
> -
> -int exynos_drm_ippnb_register(struct notifier_block *nb)
> -{
> -	return blocking_notifier_chain_register(
> -		&exynos_drm_ippnb_list, nb);
> -}
> -
> -int exynos_drm_ippnb_unregister(struct notifier_block *nb)
> -{
> -	return blocking_notifier_chain_unregister(
> -		&exynos_drm_ippnb_list, nb);
> -}
> -
> -int exynos_drm_ippnb_send_event(unsigned long val, void *v)
> -{
> -	return blocking_notifier_call_chain(
> -		&exynos_drm_ippnb_list, val, v);
> -}
> -
> -static int ipp_set_property(struct exynos_drm_ippdrv *ippdrv,
> -		struct drm_exynos_ipp_property *property)
> -{
> -	struct exynos_drm_ipp_ops *ops = NULL;
> -	bool swap = false;
> -	int ret, i;
> -
> -	if (!property) {
> -		DRM_ERROR("invalid property parameter.\n");
> -		return -EINVAL;
> -	}
> -
> -	DRM_DEBUG_KMS("prop_id[%d]\n", property->prop_id);
> -
> -	/* reset h/w block */
> -	if (ippdrv->reset &&
> -	    ippdrv->reset(ippdrv->dev)) {
> -		return -EINVAL;
> -	}
> -
> -	/* set source,destination operations */
> -	for_each_ipp_ops(i) {
> -		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)
> -				return ret;
> -		}
> -
> -		/* set transform for rotation, flip */
> -		if (ops->set_transf) {
> -			ret = ops->set_transf(ippdrv->dev, config->degree,
> -				config->flip, &swap);
> -			if (ret)
> -				return ret;
> -		}
> -
> -		/* set size */
> -		if (ops->set_size) {
> -			ret = ops->set_size(ippdrv->dev, swap, &config->pos,
> -				&config->sz);
> -			if (ret)
> -				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("prop_id[%d]\n", property->prop_id);
> -
> -	/* store command info in ippdrv */
> -	ippdrv->c_node = c_node;
> -
> -	mutex_lock(&c_node->mem_lock);
> -	if (!ipp_check_mem_list(c_node)) {
> -		DRM_DEBUG_KMS("empty memory.\n");
> -		ret = -ENOMEM;
> -		goto err_unlock;
> -	}
> -
> -	/* set current property in ippdrv */
> -	ret = ipp_set_property(ippdrv, property);
> -	if (ret) {
> -		DRM_ERROR("failed to set property.\n");
> -		ippdrv->c_node = NULL;
> -		goto err_unlock;
> -	}
> -
> -	/* 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);
> -
> -			DRM_DEBUG_KMS("m_node[%pK]\n", m_node);
> -
> -			ret = ipp_set_mem_node(ippdrv, c_node, m_node);
> -			if (ret) {
> -				DRM_ERROR("failed to set m node.\n");
> -				goto err_unlock;
> -			}
> -		}
> -		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");
> -				goto err_unlock;
> -			}
> -		}
> -		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");
> -				goto err_unlock;
> -			}
> -		}
> -		break;
> -	default:
> -		DRM_ERROR("invalid operations.\n");
> -		ret = -EINVAL;
> -		goto err_unlock;
> -	}
> -	mutex_unlock(&c_node->mem_lock);
> -
> -	DRM_DEBUG_KMS("cmd[%d]\n", property->cmd);
> -
> -	/* start operations */
> -	if (ippdrv->start) {
> -		ret = ippdrv->start(ippdrv->dev, property->cmd);
> -		if (ret) {
> -			DRM_ERROR("failed to start ops.\n");
> -			ippdrv->c_node = NULL;
> -			return ret;
> -		}
> -	}
> -
> -	return 0;
> -
> -err_unlock:
> -	mutex_unlock(&c_node->mem_lock);
> -	ippdrv->c_node = NULL;
> -	return ret;
> -}
> -
> -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_property *property = &c_node->property;
> -	int i;
> -
> -	DRM_DEBUG_KMS("prop_id[%d]\n", property->prop_id);
> -
> -	/* stop operations */
> -	if (ippdrv->stop)
> -		ippdrv->stop(ippdrv->dev, property->cmd);
> -
> -	/* check command */
> -	switch (property->cmd) {
> -	case IPP_CMD_M2M:
> -		for_each_ipp_ops(i)
> -			ipp_clean_mem_nodes(drm_dev, c_node, i);
> -		break;
> -	case IPP_CMD_WB:
> -		ipp_clean_mem_nodes(drm_dev, c_node, EXYNOS_DRM_OPS_DST);
> -		break;
> -	case IPP_CMD_OUTPUT:
> -		ipp_clean_mem_nodes(drm_dev, c_node, EXYNOS_DRM_OPS_SRC);
> -		break;
> -	default:
> -		DRM_ERROR("invalid operations.\n");
> -		return -EINVAL;
> -	}
> -
> -	return 0;
> -}
> -
> -void ipp_sched_cmd(struct work_struct *work)
> -{
> -	struct drm_exynos_ipp_cmd_work *cmd_work =
> -		container_of(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;
> -
> -	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->lock);
> -
> -	property = &c_node->property;
> -
> -	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 (ipp_is_m2m_cmd(property->cmd)) {
> -			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("ctrl[%d] done.\n", cmd_work->ctrl);
> -
> -err_unlock:
> -	mutex_unlock(&c_node->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;
> -	u32 tbuf_id[EXYNOS_DRM_OPS_MAX] = {0, };
> -	int ret, i;
> -
> -	for_each_ipp_ops(i)
> -		DRM_DEBUG_KMS("%s buf_id[%d]\n", 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;
> -	}
> -
> -	mutex_lock(&c_node->event_lock);
> -	if (list_empty(&c_node->event_list)) {
> -		DRM_DEBUG_KMS("event list is empty.\n");
> -		ret = 0;
> -		goto err_event_unlock;
> -	}
> -
> -	mutex_lock(&c_node->mem_lock);
> -	if (!ipp_check_mem_list(c_node)) {
> -		DRM_DEBUG_KMS("empty memory.\n");
> -		ret = 0;
> -		goto err_mem_unlock;
> -	}
> -
> -	/* 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);
> -
> -			tbuf_id[i] = m_node->buf_id;
> -			DRM_DEBUG_KMS("%s buf_id[%d]\n",
> -				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");
> -			ret = -ENOMEM;
> -			goto err_mem_unlock;
> -		}
> -
> -		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);
> -
> -		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");
> -		ret = -EINVAL;
> -		goto err_mem_unlock;
> -	}
> -	mutex_unlock(&c_node->mem_lock);
> -
> -	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,
> -	 * then we make event and link to event list tail.
> -	 * 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);
> -
> -	do_gettimeofday(&now);
> -	DRM_DEBUG_KMS("tv_sec[%ld]tv_usec[%ld]\n", 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];
> -
> -	drm_send_event(drm_dev, &e->base);
> -	mutex_unlock(&c_node->event_lock);
> -
> -	DRM_DEBUG_KMS("done cmd[%d]prop_id[%d]buf_id[%d]\n",
> -		property->cmd, property->prop_id, tbuf_id[EXYNOS_DRM_OPS_DST]);
> -
> -	return 0;
> -
> -err_mem_unlock:
> -	mutex_unlock(&c_node->mem_lock);
> -err_event_unlock:
> -	mutex_unlock(&c_node->event_lock);
> -	return ret;
> -}
> -
> -void ipp_sched_event(struct work_struct *work)
> -{
> -	struct drm_exynos_ipp_event_work *event_work =
> -		container_of(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("buf_id[%d]\n", 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->c_node;
> -	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, then IPP make
> -	 * synchronization with command thread, so make complete event.
> -	 * or going out operations.
> -	 */
> -	if (c_node->state != IPP_STATE_START) {
> -		DRM_DEBUG_KMS("bypass state[%d]prop_id[%d]\n",
> -			c_node->state, c_node->property.prop_id);
> -		goto err_completion;
> -	}
> -
> -	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 (ipp_is_m2m_cmd(c_node->property.cmd))
> -		complete(&c_node->start_complete);
> -}
> -
> -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;
> -
> -	/* 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);
> -		if (ret < 0) {
> -			DRM_ERROR("failed to create id.\n");
> -			goto err;
> -		}
> -		ippdrv->prop_list.ipp_id = ret;
> -
> -		DRM_DEBUG_KMS("count[%d]ippdrv[%pK]ipp_id[%d]\n",
> -			count++, ippdrv, ret);
> -
> -		/* 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);
> -		mutex_init(&ippdrv->cmd_lock);
> -
> -		ret = drm_iommu_attach_device(drm_dev, ippdrv->dev);
> -		if (ret) {
> -			DRM_ERROR("failed to activate iommu\n");
> -			goto err;
> -		}
> -	}
> -
> -	return 0;
> -
> -err:
> -	/* get ipp driver entry */
> -	list_for_each_entry_continue_reverse(ippdrv, &exynos_drm_ippdrv_list,
> -						drv_list) {
> -		drm_iommu_detach_device(drm_dev, ippdrv->dev);
> -
> -		ipp_remove_id(&ctx->ipp_idr, &ctx->ipp_lock,
> -				ippdrv->prop_list.ipp_id);
> -	}
> -
> -	return ret;
> -}
> -
> -static void ipp_subdrv_remove(struct drm_device *drm_dev, struct device *dev)
> -{
> -	struct exynos_drm_ippdrv *ippdrv, *t;
> -	struct ipp_context *ctx = get_ipp_context(dev);
> -
> -	/* get ipp driver entry */
> -	list_for_each_entry_safe(ippdrv, t, &exynos_drm_ippdrv_list, drv_list) {
> -		drm_iommu_detach_device(drm_dev, ippdrv->dev);
> -
> -		ipp_remove_id(&ctx->ipp_idr, &ctx->ipp_lock,
> -				ippdrv->prop_list.ipp_id);
> -
> -		ippdrv->drm_dev = NULL;
> -		exynos_drm_ippdrv_unregister(ippdrv);
> -	}
> -}
> -
> -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;
> -
> -	file_priv->ipp_dev = dev;
> -
> -	DRM_DEBUG_KMS("done priv[%pK]\n", dev);
> -
> -	return 0;
> -}
> -
> -static void ipp_subdrv_close(struct drm_device *drm_dev, struct device *dev,
> -		struct drm_file *file)
> -{
> -	struct exynos_drm_ippdrv *ippdrv = NULL;
> -	struct ipp_context *ctx = get_ipp_context(dev);
> -	struct drm_exynos_ipp_cmd_node *c_node, *tc_node;
> -	int count = 0;
> -
> -	list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) {
> -		mutex_lock(&ippdrv->cmd_lock);
> -		list_for_each_entry_safe(c_node, tc_node,
> -			&ippdrv->cmd_list, list) {
> -			DRM_DEBUG_KMS("count[%d]ippdrv[%pK]\n",
> -				count++, ippdrv);
> -
> -			if (c_node->filp == file) {
> -				/*
> -				 * 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(ctx, c_node);
> -				if (list_empty(&ippdrv->cmd_list))
> -					pm_runtime_put_sync(ippdrv->dev);
> -			}
> -		}
> -		mutex_unlock(&ippdrv->cmd_lock);
> -	}
> -
> -	return;
> -}
> -
> -static int ipp_probe(struct platform_device *pdev)
> -{
> -	struct device *dev = &pdev->dev;
> -	struct ipp_context *ctx;
> -	struct exynos_drm_subdrv *subdrv;
> -	int ret;
> -
> -	ctx = devm_kzalloc(dev, sizeof(*ctx), GFP_KERNEL);
> -	if (!ctx)
> -		return -ENOMEM;
> -
> -	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");
> -		return -EINVAL;
> -	}
> -
> -	/*
> -	 * 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 then 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(dev, "drm ipp registered successfully.\n");
> -
> -	return 0;
> -
> -err_cmd_workq:
> -	destroy_workqueue(ctx->cmd_workq);
> -err_event_workq:
> -	destroy_workqueue(ctx->event_workq);
> -	return ret;
> -}
> -
> -static int ipp_remove(struct platform_device *pdev)
> -{
> -	struct ipp_context *ctx = platform_get_drvdata(pdev);
> -
> -	/* unregister sub driver */
> -	exynos_drm_subdrv_unregister(&ctx->subdrv);
> -
> -	/* remove,destroy ipp idr */
> -	idr_destroy(&ctx->ipp_idr);
> -	idr_destroy(&ctx->prop_idr);
> -
> -	mutex_destroy(&ctx->ipp_lock);
> -	mutex_destroy(&ctx->prop_lock);
> -
> -	/* destroy command, event work queue */
> -	destroy_workqueue(ctx->cmd_workq);
> -	destroy_workqueue(ctx->event_workq);
> -
> -	return 0;
> -}
> -
> -struct platform_driver ipp_driver = {
> -	.probe		= ipp_probe,
> -	.remove		= ipp_remove,
> -	.driver		= {
> -		.name	= "exynos-drm-ipp",
> -		.owner	= THIS_MODULE,
> -	},
> -};
> -
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.h b/drivers/gpu/drm/exynos/exynos_drm_ipp.h
> deleted file mode 100644
> index 2a61547a39d0..000000000000
> --- a/drivers/gpu/drm/exynos/exynos_drm_ipp.h
> +++ /dev/null
> @@ -1,252 +0,0 @@
> -/*
> - * Copyright (c) 2012 Samsung Electronics Co., Ltd.
> - *
> - * Authors:
> - *	Eunchul Kim <chulspro.kim@samsung.com>
> - *	Jinyoung Jeon <jy0.jeon@samsung.com>
> - *	Sangmin Lee <lsmin.lee@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.
> - */
> -
> -#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.
> - *
> - * @list: list head to command queue information.
> - * @event_list: list head of event.
> - * @mem_list: list head to source,destination memory queue information.
> - * @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.
> - * @filp: associated file pointer.
> - */
> -struct drm_exynos_ipp_cmd_node {
> -	struct list_head	list;
> -	struct list_head	event_list;
> -	struct list_head	mem_list[EXYNOS_DRM_OPS_MAX];
> -	struct mutex	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;
> -	struct drm_file	*filp;
> -};
> -
> -/*
> - * A structure of buffer information.
> - *
> - * @handles: Y, Cb, Cr each gem object handle.
> - * @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 information.
> - *
> - * @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, bool *swap);
> -	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.
> - * @dedicated: dedicated ipp device.
> - * @ops: source, destination operations.
> - * @event_workq: event work queue.
> - * @c_node: current command information.
> - * @cmd_list: list head for command information.
> - * @cmd_lock: lock for synchronization of access to cmd_list.
> - * @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;
> -	bool	dedicated;
> -	struct exynos_drm_ipp_ops	*ops[EXYNOS_DRM_OPS_MAX];
> -	struct workqueue_struct	*event_workq;
> -	struct drm_exynos_ipp_cmd_node *c_node;
> -	struct list_head	cmd_list;
> -	struct mutex	cmd_lock;
> -	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 cb3e9f9d029f..544db2cecb36 100644
> --- a/include/uapi/drm/exynos_drm.h
> +++ b/include/uapi/drm/exynos_drm.h
> @@ -134,172 +134,6 @@ 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),
> -	EXYNOS_DRM_FLIP_VERTICAL = (1 << 0),
> -	EXYNOS_DRM_FLIP_HORIZONTAL = (1 << 1),
> -	EXYNOS_DRM_FLIP_BOTH = EXYNOS_DRM_FLIP_VERTICAL |
> -			EXYNOS_DRM_FLIP_HORIZONTAL,
> -};
> -
> -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 {
> -	__u32 ops_id;
> -	__u32 flip;
> -	__u32 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.
> - * @refresh_rate: refresh rate.
> - */
> -struct drm_exynos_ipp_property {
> -	struct drm_exynos_ipp_config config[EXYNOS_DRM_OPS_MAX];
> -	__u32	cmd;
> -	__u32	ipp_id;
> -	__u32	prop_id;
> -	__u32	refresh_rate;
> -};
> -
> -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 {
> -	__u32	ops_id;
> -	__u32	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;
> -	__u32	ctrl;
> -};
> -
>   #define DRM_EXYNOS_GEM_CREATE		0x00
>   #define DRM_EXYNOS_GEM_MAP		0x01
>   /* Reserved 0x03 ~ 0x05 for exynos specific gem ioctl */
> @@ -311,11 +145,7 @@ struct drm_exynos_ipp_cmd_ctrl {
>   #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
> +/* Reserved 0x30 ~ 0x33 for obsolete Exynos IPP ioctls */
>   
>   #define DRM_IOCTL_EXYNOS_GEM_CREATE		DRM_IOWR(DRM_COMMAND_BASE + \
>   		DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create)
> @@ -334,18 +164,8 @@ struct drm_exynos_ipp_cmd_ctrl {
>   #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;
> @@ -356,16 +176,6 @@ 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];
> -};
> -
>   #if defined(__cplusplus)
>   }
>   #endif

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 2/6] drm/exynos: ipp: Add IPP v2 framework

[Hoegeun Kwon]

Hi Marek,

Tested-by: Hoegeun Kwon <hoegeun.kwon@samsung.com>

Best regards,
Hoegeun



On 09/12/2017 05:08 PM, Marek Szyprowski wrote:
> This patch adds Exynos IPP v2 subsystem and userspace API.
>
> New userspace API is focused ONLY on memory-to-memory image processing.
> The two remainging IPP operation modes (framebuffer writeback and
> local-path output with image processing) can be implemented using
> standard DRM features: writeback connectors and additional DRM planes with
> scaling features.
>
> V2 IPP userspace API is not compatible with old IPP ioctls. This is a
> significant change, but the old IPP subsystem in mainline Linux kernel was
> partially disfunctional anyway and not used in any open-source project.
>
> V2 IPP userspace API is based on stateless approach, which much better fits
> to memory-to-memory image processing model. It also provides support for
> all image formats, which are both already defined in DRM API and supported
> by the existing IPP hardware modules.
>
> The API consists of the following ioctls:
> - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image
>    processing modules,
> - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image
>    formats of given IPP module,
> - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for
>    selected image format of given IPP module,
> - DRM_IOCTL_EXYNOS_IPP_COMMIT: to perform operation described by the
>    provided structures (source and destination buffers, operation rectangle,
>    transformation, etc).
>
> The proposed userspace API is extensible. In the future more advanced image
> processing operations can be defined to support for example blending.
>
> Userspace API is fully functional also on DRM render nodes, so it is not
> limited to the root/privileged client.
>
> Internal driver API also has been completely rewritten. New IPP core
> performs all possible input validation, checks and object life-time
> control. The drivers can focus only on writing configuration to hardware
> registers. Stateless nature of DRM_IOCTL_EXYNOS_IPP_COMMIT ioctl simplifies
> the driver API. Minimal driver needs to provide a single callback for
> starting processing and an array with supported image formats.
>
> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
> ---
>   drivers/gpu/drm/exynos/Kconfig          |   3 +
>   drivers/gpu/drm/exynos/Makefile         |   1 +
>   drivers/gpu/drm/exynos/exynos_drm_drv.c |   9 +
>   drivers/gpu/drm/exynos/exynos_drm_ipp.c | 893 ++++++++++++++++++++++++++++++++
>   drivers/gpu/drm/exynos/exynos_drm_ipp.h | 188 +++++++
>   include/uapi/drm/exynos_drm.h           | 227 ++++++++
>   6 files changed, 1321 insertions(+)
>   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 88cff0e039b6..2e097a81df12 100644
> --- a/drivers/gpu/drm/exynos/Kconfig
> +++ b/drivers/gpu/drm/exynos/Kconfig
> @@ -94,6 +94,9 @@ config DRM_EXYNOS_G2D
>   	help
>   	  Choose this option if you want to use Exynos G2D for DRM.
>   
> +config DRM_EXYNOS_IPP
> +	bool
> +
>   config DRM_EXYNOS_FIMC
>   	bool "FIMC"
>   	depends on BROKEN && MFD_SYSCON
> diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile
> index 09bb002c9555..f663490e949d 100644
> --- a/drivers/gpu/drm/exynos/Makefile
> +++ b/drivers/gpu/drm/exynos/Makefile
> @@ -17,6 +17,7 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER)	+= exynos_mixer.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI)	+= exynos_hdmi.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI)	+= exynos_drm_vidi.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_G2D)	+= exynos_drm_g2d.o
> +exynosdrm-$(CONFIG_DRM_EXYNOS_IPP)	+= exynos_drm_ipp.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC)	+= exynos_drm_fimc.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR)	+= exynos_drm_rotator.o
>   exynosdrm-$(CONFIG_DRM_EXYNOS_GSC)	+= exynos_drm_gsc.o
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
> index 7f19054ebce3..645046c5943a 100644
> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
> @@ -26,6 +26,7 @@
>   #include "exynos_drm_fb.h"
>   #include "exynos_drm_gem.h"
>   #include "exynos_drm_plane.h"
> +#include "exynos_drm_ipp.h"
>   #include "exynos_drm_vidi.h"
>   #include "exynos_drm_g2d.h"
>   #include "exynos_drm_iommu.h"
> @@ -114,6 +115,14 @@ static void exynos_drm_lastclose(struct drm_device *dev)
>   			DRM_AUTH | DRM_RENDER_ALLOW),
>   	DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl,
>   			DRM_AUTH | DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_RESOURCES, exynos_drm_ipp_get_res_ioctl,
> +			DRM_AUTH | DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_CAPS, exynos_drm_ipp_get_caps_ioctl,
> +			DRM_AUTH | DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_LIMITS, exynos_drm_ipp_get_limits_ioctl,
> +			DRM_AUTH | DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_COMMIT, exynos_drm_ipp_commit_ioctl,
> +			DRM_AUTH | DRM_RENDER_ALLOW),
>   };
>   
>   static const struct file_operations exynos_drm_driver_fops = {
> 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 000000000000..11be6a1bf839
> --- /dev/null
> +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.c
> @@ -0,0 +1,892 @@
> +/*
> + * Copyright (C) 2017 Samsung Electronics Co.Ltd
> + * Authors:
> + *	Marek Szyprowski <m.szyprowski@samsung.com>
> + *
> + * Exynos DRM Image Post Processing (IPP) related functions
> + *
> + * 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 shall be included in
> + * all copies or substantial portions of the Software.
> + */
> +
> +
> +#include <drm/drmP.h>
> +#include <drm/drm_mode.h>
> +#include <uapi/drm/exynos_drm.h>
> +
> +#include "exynos_drm_drv.h"
> +#include "exynos_drm_gem.h"
> +#include "exynos_drm_ipp.h"
> +
> +static int num_ipp;
> +static LIST_HEAD(ipp_list);
> +
> +struct drm_pending_exynos_ipp_event {
> +	struct drm_pending_event base;
> +	struct drm_exynos_ipp_event event;
> +};
> +
> +/**
> + * exynos_drm_ipp_register - Register a new picture processor hardware module
> + * @dev: DRM device
> + * @ipp: ipp module to init
> + * @funcs: callbacks for the new ipp object
> + * @caps: bitmask of ipp capabilities (%DRM_EXYNOS_IPP_CAP_*)
> + * @formats: array of supported formats (%DRM_FORMAT_*)
> + * @name: name (for debugging purposes)
> + *
> + * Initializes a ipp module.
> + *
> + * Returns:
> + * Zero on success, error code on failure.
> + */
> +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp,
> +		const struct exynos_drm_ipp_funcs *funcs, unsigned int caps,
> +		const struct exynos_drm_ipp_formats *formats, const char *name)
> +{
> +	WARN_ON(!ipp);
> +	WARN_ON(!funcs);
> +	WARN_ON(!formats);
> +
> +	spin_lock_init(&ipp->lock);
> +	INIT_LIST_HEAD(&ipp->todo_list);
> +	init_waitqueue_head(&ipp->done_wq);
> +	ipp->dev = dev;
> +	ipp->funcs = funcs;
> +	ipp->capabilities = caps;
> +	ipp->name = name;
> +	ipp->formats = formats;
> +	while (formats++->fourcc)
> +		ipp->num_formats++;
> +
> +	list_add_tail(&ipp->head, &ipp_list);
> +	ipp->id = num_ipp++;
> +
> +	DRM_DEBUG_DRIVER("Registered ipp %d\n", ipp->id);
> +
> +	return 0;
> +}
> +
> +/**
> + * exynos_drm_ipp_unregister - Unregister the picture processor module
> + * @dev: DRM device
> + * @ipp: ipp module
> + */
> +void exynos_drm_ipp_unregister(struct drm_device *dev,
> +			       struct exynos_drm_ipp *ipp)
> +{
> +	BUG_ON(ipp->task);
> +	BUG_ON(!list_empty(&ipp->todo_list));
> +	list_del(&ipp->head);
> +}
> +
> +/**
> + * exynos_drm_ipp_ioctl_get_res_ioctl - enumerate all ipp modules
> + * @dev: DRM device
> + * @data: ioctl data
> + * @file_priv: DRM file info
> + *
> + * Construct a list of ipp ids.
> + *
> + * Called by the user via ioctl.
> + *
> + * Returns:
> + * Zero on success, negative errno on failure.
> + */
> +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data,
> +				 struct drm_file *file_priv)
> +{
> +	struct drm_exynos_ioctl_ipp_get_res *resp = data;
> +	struct exynos_drm_ipp *ipp;
> +	uint32_t __user *ipp_ptr = (uint32_t __user *)
> +						(unsigned long)resp->ipp_id_ptr;
> +	unsigned int count = num_ipp, copied = 0;
> +
> +	/*
> +	 * This ioctl is called twice, once to determine how much space is
> +	 * needed, and the 2nd time to fill it.
> +	 */
> +	if (count && resp->count_ipps >= count) {
> +		list_for_each_entry(ipp, &ipp_list, head) {
> +			if (put_user(ipp->id, ipp_ptr + copied))
> +				return -EFAULT;
> +			copied++;
> +		}
> +	}
> +	resp->count_ipps = count;
> +
> +	return 0;
> +}
> +
> +static inline struct exynos_drm_ipp *__ipp_get(uint32_t id)
> +{
> +	struct exynos_drm_ipp *ipp;
> +
> +	list_for_each_entry(ipp, &ipp_list, head)
> +		if (ipp->id == id)
> +			return ipp;
> +	return NULL;
> +}
> +
> +/**
> + * exynos_drm_ipp_ioctl_get_caps - get ipp module capabilities and formats
> + * @dev: DRM device
> + * @data: ioctl data
> + * @file_priv: DRM file info
> + *
> + * Construct a structure describing ipp module capabilities.
> + *
> + * Called by the user via ioctl.
> + *
> + * Returns:
> + * Zero on success, negative errno on failure.
> + */
> +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data,
> +				  struct drm_file *file_priv)
> +{
> +	struct drm_exynos_ioctl_ipp_get_caps *resp = data;
> +	void __user *ptr = (void __user *)(unsigned long)resp->formats_ptr;
> +	struct exynos_drm_ipp *ipp;
> +	int i;
> +
> +	ipp = __ipp_get(resp->ipp_id);
> +	if (!ipp)
> +		return -ENOENT;
> +
> +	resp->ipp_id = ipp->id;
> +	resp->capabilities = ipp->capabilities;
> +
> +	/*
> +	 * This ioctl is called twice, once to determine how much space is
> +	 * needed, and the 2nd time to fill it.
> +	 */
> +	if (resp->formats_count >= ipp->num_formats) {
> +		for (i = 0; i < ipp->num_formats; i++) {
> +			struct exynos_drm_ipp_format tmp = {
> +				.fourcc = ipp->formats[i].fourcc,
> +				.type = ipp->formats[i].type,
> +				.modifier = ipp->formats[i].modifier,
> +			};
> +
> +			if (copy_to_user(ptr, &tmp, sizeof(tmp)))
> +				return -EFAULT;
> +			ptr += sizeof(tmp);
> +		}
> +	}
> +	resp->formats_count = ipp->num_formats;
> +
> +	return 0;
> +}
> +
> +static inline const struct exynos_drm_ipp_formats *__ipp_format_get(
> +			uint32_t fourcc, uint64_t mod, unsigned long type,
> +			const struct exynos_drm_ipp_formats *f)
> +{
> +	for (; f->fourcc; f++)
> +		if ((f->type & type) && f->fourcc == fourcc &&
> +		    f->modifier == mod)
> +			return f;
> +	return NULL;
> +}
> +
> +/**
> + * exynos_drm_ipp_get_limits_ioctl - get ipp module limits
> + * @dev: DRM device
> + * @data: ioctl data
> + * @file_priv: DRM file info
> + *
> + * Construct a structure describing ipp module limitations for provided
> + * picture format.
> + *
> + * Called by the user via ioctl.
> + *
> + * Returns:
> + * Zero on success, negative errno on failure.
> + */
> +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data,
> +				    struct drm_file *file_priv)
> +{
> +	struct drm_exynos_ioctl_ipp_get_limits *resp = data;
> +	void __user *ptr = (void __user *)(unsigned long)resp->limits_ptr;
> +	const struct exynos_drm_ipp_formats *format;
> +	const struct drm_exynos_ipp_limit *limits, *l;
> +	struct exynos_drm_ipp *ipp;
> +	int count = 0;
> +
> +	if (resp->type != DRM_EXYNOS_IPP_FORMAT_SOURCE &&
> +	    resp->type != DRM_EXYNOS_IPP_FORMAT_DESTINATION)
> +		return -EINVAL;
> +
> +	ipp = __ipp_get(resp->ipp_id);
> +	if (!ipp)
> +		return -ENOENT;
> +
> +	format = __ipp_format_get(resp->fourcc, resp->modifier, resp->type,
> +			      ipp->formats);
> +	if (!format)
> +		return -EINVAL;
> +
> +	limits = format->limits;
> +	for (l = limits; l->type; l++)
> +		count++;
> +
> +	/*
> +	 * This ioctl is called twice, once to determine how much space is
> +	 * needed, and the 2nd time to fill it.
> +	 */
> +	if (count && resp->limits_size >= count * sizeof(*l)) {
> +		for (l = limits; l->type; l++, ptr += sizeof(*l))
> +			if (copy_to_user((void __user *)ptr, l, sizeof(*l)))
> +				return -EFAULT;
> +	}
> +	resp->limits_size = count * sizeof(*l);
> +
> +	return 0;
> +}
> +
> +static inline struct exynos_drm_ipp_task *
> +	exynos_drm_ipp_task_alloc(struct exynos_drm_ipp *ipp)
> +{
> +	struct exynos_drm_ipp_task *task;
> +
> +	task = kzalloc(sizeof(*task), GFP_KERNEL);
> +	if (!task)
> +		return NULL;
> +
> +	task->dev = ipp->dev;
> +	task->ipp = ipp;
> +
> +	/* some defaults */
> +	task->src.rect.w = task->dst.rect.w = UINT_MAX;
> +	task->src.rect.h = task->dst.rect.h = UINT_MAX;
> +	task->transform.rotation = DRM_MODE_ROTATE_0;
> +
> +	DRM_DEBUG_DRIVER("Allocated task %pK\n", task);
> +
> +	return task;
> +}
> +
> +struct exynos_drm_param_map {
> +	unsigned int id;
> +	unsigned int size;
> +	unsigned int offset;
> +} exynos_drm_ipp_params_maps[] = {
> +	{
> +		DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
> +		sizeof(struct drm_exynos_ipp_task_buffer),
> +		offsetof(struct exynos_drm_ipp_task, src.buf),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
> +		sizeof(struct drm_exynos_ipp_task_buffer),
> +		offsetof(struct exynos_drm_ipp_task, dst.buf),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE,
> +		sizeof(struct drm_exynos_ipp_task_rect),
> +		offsetof(struct exynos_drm_ipp_task, src.rect),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION,
> +		sizeof(struct drm_exynos_ipp_task_rect),
> +		offsetof(struct exynos_drm_ipp_task, dst.rect),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_TRANSFORM,
> +		sizeof(struct drm_exynos_ipp_task_transform),
> +		offsetof(struct exynos_drm_ipp_task, transform),
> +	}, {
> +		DRM_EXYNOS_IPP_TASK_ALPHA,
> +		sizeof(struct drm_exynos_ipp_task_alpha),
> +		offsetof(struct exynos_drm_ipp_task, alpha),
> +	},
> +};
> +
> +static int exynos_drm_ipp_task_set(struct exynos_drm_ipp_task *task,
> +				  struct drm_exynos_ioctl_ipp_commit *arg)
> +{
> +	unsigned int size = arg->params_size;
> +	void *p, *params;
> +	int ret = 0;
> +
> +	if (size > PAGE_SIZE)
> +		return -ERANGE;
> +
> +	p = kmalloc(size, GFP_KERNEL);
> +	if (!p)
> +		return -ENOMEM;
> +
> +	if (copy_from_user(p, (void __user *)(unsigned long)arg->params_ptr,
> +			   size)) {
> +		ret = -EFAULT;
> +		DRM_DEBUG_DRIVER("Failed to copy configuration from userspace\n");
> +		goto free;
> +	}
> +
> +	params = p;
> +	while (size) {
> +		struct exynos_drm_param_map *map = exynos_drm_ipp_params_maps;
> +		u32 *id = params;
> +		int i;
> +
> +		for (i = 0; i < ARRAY_SIZE(exynos_drm_ipp_params_maps); i++)
> +			if (map[i].id == *id)
> +				break;
> +
> +		if (i < ARRAY_SIZE(exynos_drm_ipp_params_maps) &&
> +		    map[i].size <= size) {
> +			memcpy((void *)task + map[i].offset, params,
> +			       map[i].size);
> +			params += map[i].size;
> +			size -= map[i].size;
> +		} else {
> +			ret = -EINVAL;
> +			goto free;
> +		}
> +	}
> +
> +	DRM_DEBUG_DRIVER("Got task %pK configuration from userspace\n", task);
> +free:
> +	kfree(p);
> +	return ret;
> +}
> +
> +static int exynos_drm_ipp_task_setup_buffer(struct exynos_drm_ipp_buffer *buf,
> +					    struct drm_file *filp)
> +{
> +	int ret = 0;
> +	int i;
> +
> +	/* basic checks */
> +	if (buf->buf.width == 0 || buf->buf.height == 0)
> +		return -EINVAL;
> +	buf->format = drm_format_info(buf->buf.fourcc);
> +	for (i = 0; i < buf->format->num_planes; i++) {
> +		unsigned int width = (i == 0) ? buf->buf.width :
> +			     DIV_ROUND_UP(buf->buf.width, buf->format->hsub);
> +
> +		if (buf->buf.pitch[i] == 0)
> +			buf->buf.pitch[i] = width * buf->format->cpp[i];
> +		if (buf->buf.pitch[i] < width * buf->format->cpp[i])
> +			return -EINVAL;
> +		if (!buf->buf.gem_id[i])
> +			return -ENOENT;
> +	}
> +
> +	/* get GEM buffers and check their size */
> +	for (i = 0; i < buf->format->num_planes; i++) {
> +		unsigned int height = (i == 0) ? buf->buf.height :
> +			     DIV_ROUND_UP(buf->buf.height, buf->format->vsub);
> +		unsigned long size = height * buf->buf.pitch[i] +
> +				     buf->buf.offset[i];
> +		struct drm_gem_object *obj = drm_gem_object_lookup(filp,
> +							    buf->buf.gem_id[i]);
> +		if (!obj) {
> +			ret = -ENOENT;
> +			goto gem_free;
> +		}
> +		buf->exynos_gem[i] = to_exynos_gem(obj);
> +
> +		if (size > buf->exynos_gem[i]->size) {
> +			i++;
> +			ret = -EINVAL;
> +			goto gem_free;
> +		}
> +		buf->dma_addr[i] = buf->exynos_gem[i]->dma_addr +
> +				   buf->buf.offset[i];
> +	}
> +
> +	return 0;
> +gem_free:
> +	while (i--) {
> +		drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base);
> +		buf->exynos_gem[i] = NULL;
> +	}
> +	return ret;
> +}
> +
> +static void exynos_drm_ipp_task_release_buffer(struct exynos_drm_ipp_buffer *buf)
> +{
> +	int i;
> +
> +	if (!buf->exynos_gem[0])
> +		return;
> +	for (i = 0; i < buf->format->num_planes; i++)
> +		drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base);
> +}
> +
> +static void exynos_drm_ipp_task_free(struct exynos_drm_ipp *ipp,
> +				 struct exynos_drm_ipp_task *task)
> +{
> +	DRM_DEBUG_DRIVER("Freeing task %pK\n", task);
> +
> +	exynos_drm_ipp_task_release_buffer(&task->src);
> +	exynos_drm_ipp_task_release_buffer(&task->dst);
> +	if (task->event)
> +		drm_event_cancel_free(ipp->dev, &task->event->base);
> +	kfree(task);
> +}
> +
> +struct drm_ipp_limit {
> +	struct drm_exynos_ipp_limit_val h;
> +	struct drm_exynos_ipp_limit_val v;
> +};
> +
> +enum drm_ipp_size_id {
> +	IPP_LIMIT_BUFFER, IPP_LIMIT_AREA, IPP_LIMIT_ROTATED, IPP_LIMIT_MAX
> +};
> +
> +static const enum drm_ipp_size_id limit_id_fallback[IPP_LIMIT_MAX][4] = {
> +	[IPP_LIMIT_BUFFER]  = { DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER },
> +	[IPP_LIMIT_AREA]    = { DRM_EXYNOS_IPP_LIMIT_SIZE_AREA,
> +				DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER },
> +	[IPP_LIMIT_ROTATED] = { DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED,
> +				DRM_EXYNOS_IPP_LIMIT_SIZE_AREA,
> +				DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER },
> +};
> +
> +static inline void __limit_set_val(unsigned int *ptr, unsigned int val)
> +{
> +	if (!*ptr)
> +		*ptr = val;
> +}
> +
> +static void __get_size_limit(const struct drm_exynos_ipp_limit *limits,
> +			     enum drm_ipp_size_id id, struct drm_ipp_limit *res)
> +{
> +	const struct drm_exynos_ipp_limit *l = limits;
> +	int i = 0;
> +
> +	memset(res, 0, sizeof(*res));
> +	for (i = 0; limit_id_fallback[id][i]; i++)
> +		for (l = limits; l->type; l++) {
> +			if (((l->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) !=
> +			      DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE) ||
> +			    ((l->type & DRM_EXYNOS_IPP_LIMIT_SIZE_MASK) !=
> +						     limit_id_fallback[id][i]))
> +				continue;
> +			__limit_set_val(&res->h.min, l->h.min);
> +			__limit_set_val(&res->h.max, l->h.max);
> +			__limit_set_val(&res->h.align, l->h.align);
> +			__limit_set_val(&res->v.min, l->v.min);
> +			__limit_set_val(&res->v.max, l->v.max);
> +			__limit_set_val(&res->v.align, l->v.align);
> +		}
> +}
> +
> +static inline bool __align_check(unsigned int val, unsigned int align)
> +{
> +	if (align && (val & (align - 1)))
> +		return false;
> +	return true;
> +}
> +
> +static inline bool __size_limit_check(unsigned int val,
> +				 struct drm_exynos_ipp_limit_val *l)
> +{
> +	if ((l->min && val < l->min) || (l->max && val > l->max) ||
> +	    !__align_check(val, l->align))
> +		return false;
> +	return true;
> +}
> +
> +static int exynos_drm_ipp_check_size_limits(struct exynos_drm_ipp_buffer *buf,
> +	const struct drm_exynos_ipp_limit *limits, bool rotate, bool swap)
> +{
> +	enum drm_ipp_size_id id = rotate ? IPP_LIMIT_ROTATED : IPP_LIMIT_AREA;
> +	struct drm_ipp_limit l;
> +	struct drm_exynos_ipp_limit_val *lh = &l.h, *lv = &l.v;
> +
> +	if (!limits)
> +		return 0;
> +
> +	__get_size_limit(limits, IPP_LIMIT_BUFFER, &l);
> +	if (!__size_limit_check(buf->buf.width, &l.h) ||
> +	    !__size_limit_check(buf->buf.height, &l.v))
> +		return -EINVAL;
> +
> +	if (swap) {
> +		lv = &l.h;
> +		lh = &l.v;
> +	}
> +	__get_size_limit(limits, id, &l);
> +	if (!__size_limit_check(buf->rect.w, lh) ||
> +	    !__align_check(buf->rect.x, lh->align) ||
> +	    !__size_limit_check(buf->rect.h, lv) ||
> +	    !__align_check(buf->rect.y, lv->align))
> +		return -EINVAL;
> +
> +	return 0;
> +}
> +
> +static int exynos_drm_ipp_check_scale_limits(struct drm_exynos_ipp_task_rect *src,
> +	struct drm_exynos_ipp_task_rect *dst,
> +	const struct drm_exynos_ipp_limit *limits, bool swap)
> +{
> +	const struct drm_exynos_ipp_limit_val *lh, *lv;
> +
> +	if (!limits)
> +		return 0;
> +
> +	for (; limits->type; limits++)
> +		if ((limits->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) ==
> +		    DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE)
> +			break;
> +	if (!limits->type)
> +		return 0;
> +
> +	lh = (!swap) ? &limits->h : &limits->v;
> +	lv = (!swap) ? &limits->v : &limits->h;
> +
> +	if ((lh->max && dst->w > src->w * lh->max) ||
> +	    (lh->min && (dst->w << 16) < src->w * lh->min) ||
> +	    (lv->max && dst->h > src->h * lv->max) ||
> +	    (lv->min && (dst->h << 16) < src->h * lv->min))
> +		return -EINVAL;
> +
> +	return 0;
> +}
> +
> +static int exynos_drm_ipp_task_check(struct exynos_drm_ipp_task *task,
> +				    struct drm_file *filp)
> +{
> +	struct exynos_drm_ipp *ipp = task->ipp;
> +	const struct exynos_drm_ipp_formats *src_format, *dst_format;
> +	struct exynos_drm_ipp_buffer *src = &task->src, *dst = &task->dst;
> +	unsigned int rotation = task->transform.rotation;
> +	int ret = 0;
> +	bool scale = false, swap = false;
> +
> +	DRM_DEBUG_DRIVER("Checking %pK\n", task);
> +
> +	if (src->rect.w == UINT_MAX)
> +		src->rect.w = src->buf.width;
> +	if (src->rect.h == UINT_MAX)
> +		src->rect.h = src->buf.height;
> +	if (dst->rect.w == UINT_MAX)
> +		dst->rect.w = dst->buf.width;
> +	if (dst->rect.h == UINT_MAX)
> +		dst->rect.h = dst->buf.height;
> +
> +	if (src->rect.x + src->rect.w > (src->buf.width) ||
> +	    src->rect.y + src->rect.h > (src->buf.height) ||
> +	    dst->rect.x + dst->rect.w > (dst->buf.width) ||
> +	    dst->rect.y + dst->rect.h > (dst->buf.height))
> +		return -EINVAL;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: basic checks done\n", task);
> +
> +	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CROP) &&
> +	    (src->rect.x || src->rect.y || dst->rect.x || dst->rect.y))
> +		return -EINVAL;
> +
> +	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_ROTATE) &&
> +	    rotation != DRM_MODE_ROTATE_0)
> +		return -EINVAL;
> +
> +	if (drm_rotation_90_or_270(rotation))
> +		swap = true;
> +
> +	if ((!swap && (src->rect.w != dst->rect.w || src->rect.h != dst->rect.h)) ||
> +	    (swap && (src->rect.w != dst->rect.h || src->rect.h != dst->rect.w)))
> +		scale = true;
> +
> +	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_SCALE) && scale)
> +		return -EINVAL;
> +
> +	if (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CONVERT) &&
> +	    src->buf.fourcc != dst->buf.fourcc)
> +		return -EINVAL;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: capability checks done\n", task);
> +
> +	src_format = __ipp_format_get(src->buf.fourcc, src->buf.modifier,
> +				  DRM_EXYNOS_IPP_FORMAT_SOURCE, ipp->formats);
> +	if (!src_format)
> +		return -EINVAL;
> +	ret = exynos_drm_ipp_check_size_limits(src, src_format->limits,
> +					  rotation != DRM_MODE_ROTATE_0, false);
> +	if (ret)
> +		return ret;
> +	ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect,
> +						src_format->limits, swap);
> +	if (ret)
> +		return ret;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: source image checks done\n", task);
> +
> +	dst_format = __ipp_format_get(dst->buf.fourcc, dst->buf.modifier,
> +			       DRM_EXYNOS_IPP_FORMAT_DESTINATION, ipp->formats);
> +	if (!dst_format)
> +		return -EINVAL;
> +	ret = exynos_drm_ipp_check_size_limits(dst, dst_format->limits,
> +					   rotation != DRM_MODE_ROTATE_0, swap);
> +	if (ret)
> +		return ret;
> +	ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect,
> +						dst_format->limits, swap);
> +	if (ret)
> +		return ret;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: destination image checks done\n", task);
> +
> +	ret = exynos_drm_ipp_task_setup_buffer(src, filp);
> +	if (ret)
> +		return ret;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: source buffer checks done\n", task);
> +
> +	ret = exynos_drm_ipp_task_setup_buffer(dst, filp);
> +	if (ret)
> +		return ret;
> +
> +	DRM_DEBUG_DRIVER("Task %pK: destination buffer checks done\n", task);
> +
> +	if (ipp->funcs->check)
> +		ret = ipp->funcs->check(ipp, task);
> +
> +	DRM_DEBUG_DRIVER("Task %pK: all checks done.\n", task);
> +
> +	return ret;
> +}
> +
> +static int exynos_drm_ipp_event_create(struct exynos_drm_ipp_task *task,
> +			struct drm_file *file_priv, uint64_t user_data)
> +{
> +	struct drm_pending_exynos_ipp_event *e = NULL;
> +	int ret;
> +
> +	e = kzalloc(sizeof(*e), GFP_KERNEL);
> +	if (!e)
> +		return -ENOMEM;
> +
> +	e->event.base.type = DRM_EXYNOS_IPP_EVENT;
> +	e->event.base.length = sizeof(e->event);
> +	e->event.user_data = user_data;
> +
> +	ret = drm_event_reserve_init(task->dev, file_priv, &e->base,
> +				     &e->event.base);
> +	if (ret)
> +		goto free;
> +
> +	task->event = e;
> +	return 0;
> +free:
> +	kfree(e);
> +	return ret;
> +}
> +
> +static void exynos_drm_ipp_event_send(struct exynos_drm_ipp_task *task)
> +{
> +	struct timeval now = ktime_to_timeval(ktime_get());
> +
> +	task->event->event.tv_sec = now.tv_sec;
> +	task->event->event.tv_usec = now.tv_usec;
> +	task->event->event.sequence = atomic_inc_return(&task->ipp->sequence);
> +
> +	drm_send_event(task->dev, &task->event->base);
> +}
> +
> +static int exynos_drm_ipp_task_cleanup(struct exynos_drm_ipp_task *task)
> +{
> +	int ret = task->ret;
> +
> +	if (ret == 0 && task->event) {
> +		exynos_drm_ipp_event_send(task);
> +		/* ensure event won't be canceled on task free */
> +		task->event = NULL;
> +	}
> +
> +	exynos_drm_ipp_task_free(task->ipp, task);
> +	return ret;
> +}
> +
> +static void exynos_drm_ipp_cleanup_work(struct work_struct *work)
> +{
> +	struct exynos_drm_ipp_task *task = container_of(work,
> +				struct exynos_drm_ipp_task, cleanup_work);
> +
> +	exynos_drm_ipp_task_cleanup(task);
> +}
> +
> +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp);
> +
> +/**
> + * exynos_drm_ipp_task_done - finish given task and set return code
> + * @task: ipp task to finish
> + * @ret: error code or 0 if operation has been performed successfully
> + */
> +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret)
> +{
> +	struct exynos_drm_ipp *ipp = task->ipp;
> +	unsigned long flags;
> +
> +	DRM_DEBUG_DRIVER("ipp: %d, task %pK done\n", ipp->id, task);
> +
> +	spin_lock_irqsave(&ipp->lock, flags);
> +	if (ipp->task == task)
> +		ipp->task = NULL;
> +	task->flags |= DRM_EXYNOS_IPP_TASK_DONE;
> +	task->ret = ret;
> +	spin_unlock_irqrestore(&ipp->lock, flags);
> +
> +	exynos_drm_ipp_next_task(ipp);
> +	wake_up(&ipp->done_wq);
> +
> +	if (task->flags & DRM_EXYNOS_IPP_TASK_ASYNC) {
> +		INIT_WORK(&task->cleanup_work, exynos_drm_ipp_cleanup_work);
> +		schedule_work(&task->cleanup_work);
> +	}
> +}
> +
> +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp)
> +{
> +	struct exynos_drm_ipp_task *task;
> +	unsigned long flags;
> +	int ret;
> +
> +	DRM_DEBUG_DRIVER("ipp: %d, try to run new task\n", ipp->id);
> +
> +	spin_lock_irqsave(&ipp->lock, flags);
> +
> +	if (ipp->task || list_empty(&ipp->todo_list)) {
> +		spin_unlock_irqrestore(&ipp->lock, flags);
> +		return;
> +	}
> +
> +	task = list_first_entry(&ipp->todo_list, struct exynos_drm_ipp_task,
> +				head);
> +	list_del_init(&task->head);
> +	ipp->task = task;
> +
> +	spin_unlock_irqrestore(&ipp->lock, flags);
> +
> +	DRM_DEBUG_DRIVER("ipp: %d, selected task %pK to run\n", ipp->id, task);
> +
> +	ret = ipp->funcs->commit(ipp, task);
> +	if (ret)
> +		exynos_drm_ipp_task_done(task, ret);
> +}
> +
> +static void exynos_drm_ipp_schedule_task(struct exynos_drm_ipp *ipp,
> +				     struct exynos_drm_ipp_task *task)
> +{
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&ipp->lock, flags);
> +	list_add(&task->head, &ipp->todo_list);
> +	spin_unlock_irqrestore(&ipp->lock, flags);
> +
> +	exynos_drm_ipp_next_task(ipp);
> +}
> +
> +static void exynos_drm_ipp_task_abort(struct exynos_drm_ipp *ipp,
> +				  struct exynos_drm_ipp_task *task)
> +{
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&ipp->lock, flags);
> +	if (task->flags & DRM_EXYNOS_IPP_TASK_DONE) {
> +		/* already completed task */
> +		exynos_drm_ipp_task_cleanup(task);
> +	} else if (ipp->task != task) {
> +		/* task has not been scheduled for execution yet */
> +		list_del_init(&task->head);
> +		exynos_drm_ipp_task_cleanup(task);
> +	} else {
> +		/*
> +		 * currently processed task, call abort() and perform
> +		 * cleanup with async worker
> +		 */
> +		task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC;
> +		spin_unlock_irqrestore(&ipp->lock, flags);
> +		if (ipp->funcs->abort)
> +			ipp->funcs->abort(ipp, task);
> +		return;
> +	}
> +	spin_unlock_irqrestore(&ipp->lock, flags);
> +}
> +
> +/**
> + * exynos_drm_ipp_commit_ioctl - perform image processing operation
> + * @dev: DRM device
> + * @data: ioctl data
> + * @file_priv: DRM file info
> + *
> + * Construct a ipp task from the set of properties provided from the user
> + * and try to schedule it to framebuffer processor hardware.
> + *
> + * Called by the user via ioctl.
> + *
> + * Returns:
> + * Zero on success, negative errno on failure.
> + */
> +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, void *data,
> +			  struct drm_file *file_priv)
> +{
> +	struct drm_exynos_ioctl_ipp_commit *arg = data;
> +	struct exynos_drm_ipp *ipp;
> +	struct exynos_drm_ipp_task *task;
> +	int ret = 0;
> +
> +	if ((arg->flags & ~DRM_EXYNOS_IPP_FLAGS) || arg->reserved)
> +		return -EINVAL;
> +
> +	/* can't test and expect an event at the same time */
> +	if ((arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY) &&
> +			(arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT))
> +		return -EINVAL;
> +
> +	ipp = __ipp_get(arg->ipp_id);
> +	if (!ipp)
> +		return -ENOENT;
> +
> +	task = exynos_drm_ipp_task_alloc(ipp);
> +	if (!task)
> +		return -ENOMEM;
> +
> +	ret = exynos_drm_ipp_task_set(task, arg);
> +	if (ret)
> +		goto free;
> +
> +	ret = exynos_drm_ipp_task_check(task, file_priv);
> +	if (ret || arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY)
> +		goto free;
> +
> +	if (arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT) {
> +		ret = exynos_drm_ipp_event_create(task, file_priv,
> +						 arg->user_data);
> +		if (ret)
> +			goto free;
> +	}
> +
> +	/*
> +	 * Queue task for processing on the hardware. task object will be
> +	 * then freed after exynos_drm_ipp_task_done()
> +	 */
> +	if (arg->flags & DRM_EXYNOS_IPP_FLAG_NONBLOCK) {
> +		DRM_DEBUG_DRIVER("ipp: %d, nonblocking processing task %pK\n",
> +				 ipp->id, task);
> +
> +		task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC;
> +		exynos_drm_ipp_schedule_task(task->ipp, task);
> +		ret = 0;
> +	} else {
> +		DRM_DEBUG_DRIVER("ipp: %d, processing task %pK\n", ipp->id, task);
> +		exynos_drm_ipp_schedule_task(ipp, task);
> +		ret = wait_event_interruptible(ipp->done_wq,
> +					task->flags & DRM_EXYNOS_IPP_TASK_DONE);
> +		if (ret)
> +			exynos_drm_ipp_task_abort(ipp, task);
> +		else
> +			ret = exynos_drm_ipp_task_cleanup(task);
> +	}
> +	return ret;
> +free:
> +	exynos_drm_ipp_task_free(ipp, task);
> +
> +	return ret;
> +}
> 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 000000000000..532a60698a6a
> --- /dev/null
> +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.h
> @@ -0,0 +1,188 @@
> +/*
> + * Copyright (c) 2017 Samsung Electronics Co., Ltd.
> + *
> + * 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.
> + */
> +
> +#ifndef _EXYNOS_DRM_IPP_H_
> +#define _EXYNOS_DRM_IPP_H_
> +
> +#include <drm/drmP.h>
> +
> +struct exynos_drm_ipp;
> +struct exynos_drm_ipp_task;
> +
> +/**
> + * struct exynos_drm_ipp_funcs - exynos_drm_ipp control functions
> + */
> +struct exynos_drm_ipp_funcs {
> +	/**
> +	 * @check:
> +	 *
> +	 * This is the optional hook to validate an ipp task. This function
> +	 * must reject any task which the hardware or driver doesn't support.
> +	 * This includes but is of course not limited to:
> +	 *
> +	 *  - Checking that the buffers, scaling and placement requirements
> +	 *    and so on are within the limits of the hardware not specified by
> +	 *    limits array.
> +	 *
> +	 *  - The driver does not need to repeat basic input validation like
> +	 *    done in the exynos_drm_ipp_task_check() function. The core does
> +	 *    that before calling this hook.
> +	 *
> +	 * RETURNS:
> +	 *
> +	 * 0 on success or one of the below negative error codes:
> +	 *
> +	 *  - -EINVAL, if any of the above constraints are violated.
> +	 */
> +	int (*check)(struct exynos_drm_ipp *ipp,
> +		     struct exynos_drm_ipp_task *task);
> +
> +	/**
> +	 * @commit:
> +	 *
> +	 * This is the main entry point to start framebuffer processing
> +	 * in the hardware. The exynos_drm_ipp_task has been already validated.
> +	 * This function must not wait until the device finishes processing.
> +	 * When the driver finishes processing, it has to call
> +	 * exynos_exynos_drm_ipp_task_done() function.
> +	 *
> +	 * RETURNS:
> +	 *
> +	 * 0 on success or negative error codes in case of failure.
> +	 */
> +	int (*commit)(struct exynos_drm_ipp *ipp,
> +		      struct exynos_drm_ipp_task *task);
> +
> +	/**
> +	 * @abort:
> +	 *
> +	 * Informs the driver that it has to abort the currently running
> +	 * task as soon as possible (i.e. as soon as it can stop the device
> +	 * safely), even if the task would not have been finished by then.
> +	 * After the driver performs the necessary steps, it has to call
> +	 * exynos_drm_ipp_task_done() (as if the task ended normally).
> +	 * This function does not have to (and will usually not) wait
> +	 * until the device enters a state when it can be stopped.
> +	 */
> +	void (*abort)(struct exynos_drm_ipp *ipp,
> +		      struct exynos_drm_ipp_task *task);
> +};
> +
> +/**
> + * struct exynos_drm_ipp - central picture processor module structure
> + */
> +struct exynos_drm_ipp {
> +	struct drm_device *dev;
> +	struct list_head head;
> +	unsigned int id;
> +
> +	const char *name;
> +	const struct exynos_drm_ipp_funcs *funcs;
> +	unsigned int capabilities;
> +	const struct exynos_drm_ipp_formats *formats;
> +	unsigned int num_formats;
> +	atomic_t sequence;
> +
> +	spinlock_t lock;
> +	struct exynos_drm_ipp_task *task;
> +	struct list_head todo_list;
> +	wait_queue_head_t done_wq;
> +};
> +
> +struct exynos_drm_ipp_buffer {
> +	struct drm_exynos_ipp_task_buffer buf;
> +	struct drm_exynos_ipp_task_rect rect;
> +
> +	struct exynos_drm_gem *exynos_gem[MAX_FB_BUFFER];
> +	const struct drm_format_info *format;
> +	dma_addr_t dma_addr[MAX_FB_BUFFER];
> +};
> +
> +/**
> + * struct exynos_drm_ipp_task - a structure describing transformation that
> + * has to be performed by the picture processor hardware module
> + */
> +struct exynos_drm_ipp_task {
> +	struct drm_device *dev;
> +	struct exynos_drm_ipp *ipp;
> +	struct list_head head;
> +
> +	struct exynos_drm_ipp_buffer src;
> +	struct exynos_drm_ipp_buffer dst;
> +
> +	struct drm_exynos_ipp_task_transform transform;
> +	struct drm_exynos_ipp_task_alpha alpha;
> +
> +	struct work_struct cleanup_work;
> +	unsigned int flags;
> +	int ret;
> +
> +	struct drm_pending_exynos_ipp_event *event;
> +};
> +
> +#define DRM_EXYNOS_IPP_TASK_DONE	(1 << 0)
> +#define DRM_EXYNOS_IPP_TASK_ASYNC	(1 << 1)
> +
> +struct exynos_drm_ipp_formats {
> +	u32 fourcc;
> +	u32 type;
> +	u64 modifier;
> +	const struct drm_exynos_ipp_limit *limits;
> +};
> +
> +/* helper macros to set exynos_drm_ipp_formats structure and limits*/
> +#define IPP_SRCDST_MFORMAT(f, m, l) \
> +     .fourcc = DRM_FORMAT_##f, .modifier = m, .limits = l, \
> +     .type = (DRM_EXYNOS_IPP_FORMAT_SOURCE | DRM_EXYNOS_IPP_FORMAT_DESTINATION)
> +
> +#define IPP_SRCDST_FORMAT(f, l) IPP_SRCDST_MFORMAT(f, 0, l)
> +
> +#define IPP_SIZE_LIMIT(l, val...)	\
> +     .type = (DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE | DRM_EXYNOS_IPP_LIMIT_SIZE_##l), \
> +     val
> +
> +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp,
> +		const struct exynos_drm_ipp_funcs *funcs, unsigned int caps,
> +		const struct exynos_drm_ipp_formats *formats, const char *name);
> +void exynos_drm_ipp_unregister(struct drm_device *dev, struct exynos_drm_ipp *ipp);
> +
> +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret);
> +
> +#ifdef CONFIG_DRM_EXYNOS_IPP
> +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data,
> +				 struct drm_file *file_priv);
> +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data,
> +				  struct drm_file *file_priv);
> +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data,
> +				    struct drm_file *file_priv);
> +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev,
> +				void *data, struct drm_file *file_priv);
> +#else
> +static inline int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev,
> +	 void *data, struct drm_file *file_priv)
> +{
> +	return -ENODEV;
> +}
> +static inline int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev,
> +	 void *data, struct drm_file *file_priv)
> +{
> +	return -ENODEV;
> +}
> +static inline int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev,
> +	 void *data, struct drm_file *file_priv)
> +{
> +	return -ENODEV;
> +}
> +static inline int exynos_drm_ipp_commit_ioctl(struct drm_device *dev,
> +	 void *data, struct drm_file *file_priv)
> +{
> +	return -ENODEV;
> +}
> +#endif
> +#endif
> diff --git a/include/uapi/drm/exynos_drm.h b/include/uapi/drm/exynos_drm.h
> index 544db2cecb36..239e09022d75 100644
> --- a/include/uapi/drm/exynos_drm.h
> +++ b/include/uapi/drm/exynos_drm.h
> @@ -134,6 +134,208 @@ struct drm_exynos_g2d_exec {
>   	__u64					async;
>   };
>   
> +/* Exynos DRM IPP v2 API */
> +
> +/**
> + * Enumerate available IPP hardware modules.
> + *
> + * @count_ipps: size of ipp_id array / number of ipp modules (set by driver)
> + * @reserved: padding
> + * @ipp_id_ptr: pointer to ipp_id array or NULL
> + */
> +struct drm_exynos_ioctl_ipp_get_res {
> +	__u32 count_ipps;
> +	__u32 reserved;
> +	__u64 ipp_id_ptr;
> +};
> +
> +enum drm_exynos_ipp_format_type {
> +	DRM_EXYNOS_IPP_FORMAT_SOURCE		= 0x01,
> +	DRM_EXYNOS_IPP_FORMAT_DESTINATION	= 0x02,
> +};
> +
> +struct exynos_drm_ipp_format {
> +	__u32 fourcc;
> +	__u32 type;
> +	__u64 modifier;
> +};
> +
> +enum drm_exynos_ipp_capability {
> +	DRM_EXYNOS_IPP_CAP_CROP		= 0x01,
> +	DRM_EXYNOS_IPP_CAP_ROTATE	= 0x02,
> +	DRM_EXYNOS_IPP_CAP_SCALE	= 0x04,
> +	DRM_EXYNOS_IPP_CAP_CONVERT	= 0x08,
> +};
> +
> +/**
> + * Get IPP hardware capabilities and supported image formats.
> + *
> + * @ipp_id: id of IPP module to query
> + * @capabilities: bitmask of drm_exynos_ipp_capability (set by driver)
> + * @reserved: padding
> + * @formats_count: size of formats array (in entries) / number of filled
> + *		   formats (set by driver)
> + * @formats_ptr: pointer to formats array or NULL
> + */
> +struct drm_exynos_ioctl_ipp_get_caps {
> +	__u32 ipp_id;
> +	__u32 capabilities;
> +	__u32 reserved;
> +	__u32 formats_count;
> +	__u64 formats_ptr;
> +};
> +
> +enum drm_exynos_ipp_limit_type {
> +	/* size (horizontal/vertial) limits, in pixels (min, max, alignment) */
> +	DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE		= 0x0001,
> +	/* scale ratio (horizonta/vertial), 16.16 fixed point (min, max) */
> +	DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE		= 0x0002,
> +
> +	/* image buffer area */
> +	DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER	= 0x0001 << 16,
> +	/* src/dst rectangle area */
> +	DRM_EXYNOS_IPP_LIMIT_SIZE_AREA		= 0x0002 << 16,
> +	/* src/dst rectangle area when rotation enabled */
> +	DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED	= 0x0003 << 16,
> +
> +	DRM_EXYNOS_IPP_LIMIT_TYPE_MASK		= 0x000f,
> +	DRM_EXYNOS_IPP_LIMIT_SIZE_MASK		= 0x000f << 16,
> +};
> +
> +struct drm_exynos_ipp_limit_val {
> +	__u32 min, max, align;
> +};
> +
> +/**
> + * IPP module limitation.
> + *
> + * @type: limit type (see drm_exynos_ipp_limit_type enum)
> + * @h, @v: horizontal and vertical limits
> + */
> +struct drm_exynos_ipp_limit {
> +	__u32 type;
> +	struct drm_exynos_ipp_limit_val h;
> +	struct drm_exynos_ipp_limit_val v;
> +};
> +
> +/**
> + * Get IPP limits for given image format.
> + *
> + * @ipp_id: id of IPP module to query
> + * @fourcc: image format code (see DRM_FORMAT_* in drm_fourcc.h)
> + * @modifier: image format modifier (see DRM_FORMAT_MOD_* in drm_fourcc.h)
> + * @type: source/destination identifier (drm_exynos_ipp_format_flag enum)
> + * @limits_size: size of limits array (in bytes) / size of filled entries
> + *		 (set by driver)
> + * @limits_ptr: pointer to limits array or NULL
> + */
> +struct drm_exynos_ioctl_ipp_get_limits {
> +	__u32 ipp_id;
> +	__u32 fourcc;
> +	__u64 modifier;
> +	__u32 type;
> +	__u32 limits_size;
> +	__u64 limits_ptr;
> +};
> +
> +enum drm_exynos_ipp_task_id {
> +	/* buffer described by struct drm_exynos_ipp_task_buffer */
> +	DRM_EXYNOS_IPP_TASK_BUFFER		= 0x0001,
> +	/* rectangle described by struct drm_exynos_ipp_task_rect */
> +	DRM_EXYNOS_IPP_TASK_RECTANGLE		= 0x0002,
> +	/* transformation described by struct drm_exynos_ipp_task_transform */
> +	DRM_EXYNOS_IPP_TASK_TRANSFORM		= 0x0003,
> +	/* alpha configuration described by struct drm_exynos_ipp_task_alpha */
> +	DRM_EXYNOS_IPP_TASK_ALPHA		= 0x0004,
> +
> +	/* source image data (for buffer and rectangle chunks) */
> +	DRM_EXYNOS_IPP_TASK_TYPE_SOURCE		= 0x0001 << 16,
> +	/* destination image data (for buffer and rectangle chunks) */
> +	DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION	= 0x0002 << 16,
> +};
> +
> +/**
> + * Memory buffer with image data.
> + *
> + * @id: must be DRM_EXYNOS_IPP_TASK_BUFFER
> + * other parameters are same as for AddFB2 generic DRM ioctl
> + */
> +struct drm_exynos_ipp_task_buffer {
> +	__u32	id;
> +	__u32	fourcc;
> +	__u32	width, height;
> +	__u32	gem_id[4];
> +	__u32	offset[4];
> +	__u32	pitch[4];
> +	__u64	modifier;
> +};
> +
> +/**
> + * Rectangle for processing.
> + *
> + * @id: must be DRM_EXYNOS_IPP_TASK_RECTANGLE
> + * @x,@y: left corner in pixels
> + * @w,@h: width/height in pixels
> + */
> +struct drm_exynos_ipp_task_rect {
> +	__u32	id;
> +	__u32	x, y, w, h;
> +};
> +
> +/**
> + * Image tranformation description.
> + *
> + * @id: must be DRM_EXYNOS_IPP_TASK_TRANSFORM
> + * @rotation: DRM_MODE_ROTATE_* and DRM_MODE_REFLECT_* values
> + */
> +struct drm_exynos_ipp_task_transform {
> +	__u32	id;
> +	__u32	rotation;
> +};
> +
> +/**
> + * Image global alpha configuration for formats without alpha values.
> + *
> + * @id: must be DRM_EXYNOS_IPP_TASK_ALPHA
> + * @value: global alpha value (0-255)
> + */
> +struct drm_exynos_ipp_task_alpha {
> +	__u32	id;
> +	__u32	value;
> +};
> +
> +enum drm_exynos_ipp_flag {
> +	/* generate DRM event after processing */
> +	DRM_EXYNOS_IPP_FLAG_EVENT	= 0x01,
> +	/* dry run, only check task parameters */
> +	DRM_EXYNOS_IPP_FLAG_TEST_ONLY	= 0x02,
> +	/* non-blocking processing */
> +	DRM_EXYNOS_IPP_FLAG_NONBLOCK	= 0x04,
> +};
> +
> +#define DRM_EXYNOS_IPP_FLAGS (DRM_EXYNOS_IPP_FLAG_EVENT |\
> +		DRM_EXYNOS_IPP_FLAG_TEST_ONLY | DRM_EXYNOS_IPP_FLAG_NONBLOCK)
> +
> +/**
> + * Perform image processing described by array of drm_exynos_ipp_task_*
> + * structures (parameters array).
> + *
> + * @ipp_id: id of IPP module to run the task
> + * @flags: bitmask of drm_exynos_ipp_flag values
> + * @reserved: padding
> + * @params_size: size of parameters array (in bytes)
> + * @params_ptr: pointer to parameters array or NULL
> + * @user_data: (optional) data for drm event
> + */
> +struct drm_exynos_ioctl_ipp_commit {
> +	__u32 ipp_id;
> +	__u32 flags;
> +	__u32 reserved;
> +	__u32 params_size;
> +	__u64 params_ptr;
> +	__u64 user_data;
> +};
> +
>   #define DRM_EXYNOS_GEM_CREATE		0x00
>   #define DRM_EXYNOS_GEM_MAP		0x01
>   /* Reserved 0x03 ~ 0x05 for exynos specific gem ioctl */
> @@ -146,6 +348,11 @@ struct drm_exynos_g2d_exec {
>   #define DRM_EXYNOS_G2D_EXEC		0x22
>   
>   /* Reserved 0x30 ~ 0x33 for obsolete Exynos IPP ioctls */
> +/* IPP - Image Post Processing */
> +#define DRM_EXYNOS_IPP_GET_RESOURCES	0x40
> +#define DRM_EXYNOS_IPP_GET_CAPS		0x41
> +#define DRM_EXYNOS_IPP_GET_LIMITS	0x42
> +#define DRM_EXYNOS_IPP_COMMIT		0x43
>   
>   #define DRM_IOCTL_EXYNOS_GEM_CREATE		DRM_IOWR(DRM_COMMAND_BASE + \
>   		DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create)
> @@ -164,8 +371,18 @@ 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_RESOURCES	DRM_IOWR(DRM_COMMAND_BASE + \
> +		DRM_EXYNOS_IPP_GET_RESOURCES, struct drm_exynos_ioctl_ipp_get_res)
> +#define DRM_IOCTL_EXYNOS_IPP_GET_CAPS		DRM_IOWR(DRM_COMMAND_BASE + \
> +		DRM_EXYNOS_IPP_GET_CAPS, struct drm_exynos_ioctl_ipp_get_caps)
> +#define DRM_IOCTL_EXYNOS_IPP_GET_LIMITS		DRM_IOWR(DRM_COMMAND_BASE + \
> +		DRM_EXYNOS_IPP_GET_LIMITS, struct drm_exynos_ioctl_ipp_get_limits)
> +#define DRM_IOCTL_EXYNOS_IPP_COMMIT		DRM_IOWR(DRM_COMMAND_BASE + \
> +		DRM_EXYNOS_IPP_COMMIT, struct drm_exynos_ioctl_ipp_commit)
> +
>   /* EXYNOS specific events */
>   #define DRM_EXYNOS_G2D_EVENT		0x80000000
> +#define DRM_EXYNOS_IPP_EVENT		0x80000002
>   
>   struct drm_exynos_g2d_event {
>   	struct drm_event	base;
> @@ -176,6 +393,16 @@ 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			ipp_id;
> +	__u32			sequence;
> +	__u64			reserved;
> +};
> +
>   #if defined(__cplusplus)
>   }
>   #endif

Re: [PATCH 4/6] drm/exynos: gsc: Convert driver to IPP v2 core API

[Hoegeun Kwon]

Hi Marek,

Tested-by: Hoegeun Kwon <hoegeun.kwon@samsung.com>

Best regards,
Hoegeun



On 09/12/2017 05:08 PM, Marek Szyprowski wrote:
> This patch adapts Exynos DRM rotator driver to new IPP v2 core API.
> The side effect of this conversion is a switch to driver component API
> to register properly in the Exynos DRM core.
>
> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
> ---
>   drivers/gpu/drm/exynos/Kconfig          |   3 +-
>   drivers/gpu/drm/exynos/exynos_drm_drv.c |   1 +
>   drivers/gpu/drm/exynos/exynos_drm_gsc.c | 853 ++++++++------------------------
>   drivers/gpu/drm/exynos/exynos_drm_gsc.h |  24 -
>   4 files changed, 198 insertions(+), 683 deletions(-)
>   delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_gsc.h
>
> diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig
> index c10c9ca0d8b4..4bb9edb00601 100644
> --- a/drivers/gpu/drm/exynos/Kconfig
> +++ b/drivers/gpu/drm/exynos/Kconfig
> @@ -111,7 +111,8 @@ config DRM_EXYNOS_ROTATOR
>   
>   config DRM_EXYNOS_GSC
>   	bool "GScaler"
> -	depends on BROKEN && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n
> +	depends on ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n
> +	select DRM_EXYNOS_IPP
>   	help
>   	  Choose this option if you want to use Exynos GSC for DRM.
>   
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
> index 277e444b0be6..27242af103ea 100644
> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
> @@ -272,6 +272,7 @@ struct exynos_drm_driver_info {
>   		DRM_COMPONENT_DRIVER
>   	}, {
>   		DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC),
> +		DRM_COMPONENT_DRIVER
>   	}, {
>   		&exynos_drm_platform_driver,
>   		DRM_VIRTUAL_DEVICE
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_gsc.c b/drivers/gpu/drm/exynos/exynos_drm_gsc.c
> index 0506b2b17ac1..1293441a4212 100644
> --- a/drivers/gpu/drm/exynos/exynos_drm_gsc.c
> +++ b/drivers/gpu/drm/exynos/exynos_drm_gsc.c
> @@ -12,6 +12,7 @@
>    *
>    */
>   #include <linux/kernel.h>
> +#include <linux/component.h>
>   #include <linux/platform_device.h>
>   #include <linux/clk.h>
>   #include <linux/pm_runtime.h>
> @@ -22,8 +23,8 @@
>   #include <drm/exynos_drm.h>
>   #include "regs-gsc.h"
>   #include "exynos_drm_drv.h"
> +#include "exynos_drm_iommu.h"
>   #include "exynos_drm_ipp.h"
> -#include "exynos_drm_gsc.h"
>   
>   /*
>    * GSC stands for General SCaler and
> @@ -31,26 +32,9 @@
>    * input DMA reads image data from the memory.
>    * output DMA writes image data to memory.
>    * GSC supports image rotation and image effect functions.
> - *
> - * M2M operation : supports crop/scale/rotation/csc so on.
> - * Memory ----> GSC H/W ----> Memory.
> - * Writeback operation : supports cloned screen with FIMD.
> - * FIMD ----> GSC H/W ----> Memory.
> - * Output operation : supports direct display using local path.
> - * Memory ----> GSC H/W ----> FIMD, Mixer.
>    */
>   
> -/*
> - * TODO
> - * 1. check suspend/resume api if needed.
> - * 2. need to check use case platform_device_id.
> - * 3. check src/dst size with, height.
> - * 4. added check_prepare api for right register.
> - * 5. need to add supported list in prop_list.
> - * 6. check prescaler/scaler optimization.
> - */
>   
> -#define GSC_MAX_DEVS	4
>   #define GSC_MAX_SRC		4
>   #define GSC_MAX_DST		16
>   #define GSC_RESET_TIMEOUT	50
> @@ -65,8 +49,6 @@
>   #define GSC_SC_DOWN_RATIO_4_8		131072
>   #define GSC_SC_DOWN_RATIO_3_8		174762
>   #define GSC_SC_DOWN_RATIO_2_8		262144
> -#define GSC_REFRESH_MIN	12
> -#define GSC_REFRESH_MAX	60
>   #define GSC_CROP_MAX	8192
>   #define GSC_CROP_MIN	32
>   #define GSC_SCALE_MAX	4224
> @@ -79,8 +61,6 @@
>   #define GSC_COEF_DEPTH	3
>   
>   #define get_gsc_context(dev)	platform_get_drvdata(to_platform_device(dev))
> -#define get_ctx_from_ippdrv(ippdrv)	container_of(ippdrv,\
> -					struct gsc_context, ippdrv);
>   #define gsc_read(offset)		readl(ctx->regs + (offset))
>   #define gsc_write(cfg, offset)	writel(cfg, ctx->regs + (offset))
>   
> @@ -124,7 +104,6 @@ struct gsc_capability {
>   /*
>    * A structure of gsc context.
>    *
> - * @ippdrv: prepare initialization using ippdrv.
>    * @regs_res: register resources.
>    * @regs: memory mapped io registers.
>    * @sysreg: handle to SYSREG block regmap.
> @@ -137,11 +116,14 @@ struct gsc_capability {
>    * @suspended: qos operations.
>    */
>   struct gsc_context {
> -	struct exynos_drm_ippdrv	ippdrv;
> +	struct exynos_drm_ipp ipp;
> +	struct drm_device *drm_dev;
> +	struct device	*dev;
> +	struct exynos_drm_ipp_task	*task;
> +
>   	struct resource	*regs_res;
>   	void __iomem	*regs;
>   	struct regmap	*sysreg;
> -	struct mutex	lock;
>   	struct clk	*gsc_clk;
>   	struct gsc_scaler	sc;
>   	int	id;
> @@ -438,25 +420,6 @@ static int gsc_sw_reset(struct gsc_context *ctx)
>   	return 0;
>   }
>   
> -static void gsc_set_gscblk_fimd_wb(struct gsc_context *ctx, bool enable)
> -{
> -	unsigned int gscblk_cfg;
> -
> -	if (!ctx->sysreg)
> -		return;
> -
> -	regmap_read(ctx->sysreg, SYSREG_GSCBLK_CFG1, &gscblk_cfg);
> -
> -	if (enable)
> -		gscblk_cfg |= GSC_BLK_DISP1WB_DEST(ctx->id) |
> -				GSC_BLK_GSCL_WB_IN_SRC_SEL(ctx->id) |
> -				GSC_BLK_SW_RESET_WB_DEST(ctx->id);
> -	else
> -		gscblk_cfg |= GSC_BLK_PXLASYNC_LO_MASK_WB(ctx->id);
> -
> -	regmap_write(ctx->sysreg, SYSREG_GSCBLK_CFG1, gscblk_cfg);
> -}
> -
>   static void gsc_handle_irq(struct gsc_context *ctx, bool enable,
>   		bool overflow, bool done)
>   {
> @@ -487,10 +450,8 @@ static void gsc_handle_irq(struct gsc_context *ctx, bool enable,
>   }
>   
>   
> -static int gsc_src_set_fmt(struct device *dev, u32 fmt)
> +static int gsc_src_set_fmt(struct gsc_context *ctx, u32 fmt)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
>   	u32 cfg;
>   
>   	DRM_DEBUG_KMS("fmt[0x%x]\n", fmt);
> @@ -549,7 +510,7 @@ static int gsc_src_set_fmt(struct device *dev, u32 fmt)
>   			GSC_IN_YUV420_2P);
>   		break;
>   	default:
> -		dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt);
> +		dev_err(ctx->dev, "invalid target yuv order 0x%x.\n", fmt);
>   		return -EINVAL;
>   	}
>   
> @@ -558,105 +519,81 @@ static int gsc_src_set_fmt(struct device *dev, u32 fmt)
>   	return 0;
>   }
>   
> -static int gsc_src_set_transf(struct device *dev,
> -		enum drm_exynos_degree degree,
> -		enum drm_exynos_flip flip, bool *swap)
> +static int gsc_src_set_transf(struct gsc_context *ctx, unsigned int rotation)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> +	unsigned int degree = rotation & DRM_MODE_ROTATE_MASK;
>   	u32 cfg;
>   
> -	DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip);
> -
>   	cfg = gsc_read(GSC_IN_CON);
>   	cfg &= ~GSC_IN_ROT_MASK;
>   
>   	switch (degree) {
> -	case EXYNOS_DRM_DEGREE_0:
> -		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
> +	case DRM_MODE_ROTATE_0:
> +		if (rotation & DRM_MODE_REFLECT_Y)
>   			cfg |= GSC_IN_ROT_XFLIP;
> -		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
> +		if (rotation & DRM_MODE_REFLECT_X)
>   			cfg |= GSC_IN_ROT_YFLIP;
>   		break;
> -	case EXYNOS_DRM_DEGREE_90:
> -		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
> +	case DRM_MODE_ROTATE_90:
> +		if (rotation & DRM_MODE_REFLECT_Y)
>   			cfg |= GSC_IN_ROT_90_XFLIP;
> -		else if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
> +		else if (rotation & DRM_MODE_REFLECT_X)
>   			cfg |= GSC_IN_ROT_90_YFLIP;
>   		else
>   			cfg |= GSC_IN_ROT_90;
>   		break;
> -	case EXYNOS_DRM_DEGREE_180:
> +	case DRM_MODE_ROTATE_180:
>   		cfg |= GSC_IN_ROT_180;
> -		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
> +		if (rotation & DRM_MODE_REFLECT_Y)
>   			cfg &= ~GSC_IN_ROT_XFLIP;
> -		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
> +		if (rotation & DRM_MODE_REFLECT_X)
>   			cfg &= ~GSC_IN_ROT_YFLIP;
>   		break;
> -	case EXYNOS_DRM_DEGREE_270:
> +	case DRM_MODE_ROTATE_270:
>   		cfg |= GSC_IN_ROT_270;
> -		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
> +		if (rotation & DRM_MODE_REFLECT_Y)
>   			cfg &= ~GSC_IN_ROT_XFLIP;
> -		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
> +		if (rotation & DRM_MODE_REFLECT_X)
>   			cfg &= ~GSC_IN_ROT_YFLIP;
>   		break;
> -	default:
> -		dev_err(ippdrv->dev, "invalid degree value %d.\n", degree);
> -		return -EINVAL;
>   	}
>   
>   	gsc_write(cfg, GSC_IN_CON);
>   
>   	ctx->rotation = (cfg & GSC_IN_ROT_90) ? 1 : 0;
> -	*swap = ctx->rotation;
> -
>   	return 0;
>   }
>   
> -static int gsc_src_set_size(struct device *dev, int swap,
> -		struct drm_exynos_pos *pos, struct drm_exynos_sz *sz)
> +static int gsc_src_set_size(struct gsc_context *ctx,
> +		struct exynos_drm_ipp_buffer *buf)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct drm_exynos_pos img_pos = *pos;
>   	struct gsc_scaler *sc = &ctx->sc;
>   	u32 cfg;
>   
> -	DRM_DEBUG_KMS("swap[%d]x[%d]y[%d]w[%d]h[%d]\n",
> -		swap, pos->x, pos->y, pos->w, pos->h);
> -
> -	if (swap) {
> -		img_pos.w = pos->h;
> -		img_pos.h = pos->w;
> -	}
> -
>   	/* pixel offset */
> -	cfg = (GSC_SRCIMG_OFFSET_X(img_pos.x) |
> -		GSC_SRCIMG_OFFSET_Y(img_pos.y));
> +	cfg = (GSC_SRCIMG_OFFSET_X(buf->rect.x) |
> +		GSC_SRCIMG_OFFSET_Y(buf->rect.y));
>   	gsc_write(cfg, GSC_SRCIMG_OFFSET);
>   
>   	/* cropped size */
> -	cfg = (GSC_CROPPED_WIDTH(img_pos.w) |
> -		GSC_CROPPED_HEIGHT(img_pos.h));
> +	cfg = (GSC_CROPPED_WIDTH(buf->rect.w) |
> +		GSC_CROPPED_HEIGHT(buf->rect.h));
>   	gsc_write(cfg, GSC_CROPPED_SIZE);
>   
> -	DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", sz->hsize, sz->vsize);
> -
>   	/* original size */
>   	cfg = gsc_read(GSC_SRCIMG_SIZE);
>   	cfg &= ~(GSC_SRCIMG_HEIGHT_MASK |
>   		GSC_SRCIMG_WIDTH_MASK);
>   
> -	cfg |= (GSC_SRCIMG_WIDTH(sz->hsize) |
> -		GSC_SRCIMG_HEIGHT(sz->vsize));
> +	cfg |= (GSC_SRCIMG_WIDTH(buf->buf.width) |
> +		GSC_SRCIMG_HEIGHT(buf->buf.height));
>   
>   	gsc_write(cfg, GSC_SRCIMG_SIZE);
>   
>   	cfg = gsc_read(GSC_IN_CON);
>   	cfg &= ~GSC_IN_RGB_TYPE_MASK;
>   
> -	DRM_DEBUG_KMS("width[%d]range[%d]\n", pos->w, sc->range);
> -
> -	if (pos->w >= GSC_WIDTH_ITU_709)
> +	if (buf->rect.w >= GSC_WIDTH_ITU_709)
>   		if (sc->range)
>   			cfg |= GSC_IN_RGB_HD_WIDE;
>   		else
> @@ -673,30 +610,15 @@ static int gsc_src_set_size(struct device *dev, int swap,
>   }
>   
>   static int gsc_src_set_buf_seq(struct gsc_context *ctx, u32 buf_id,
> -		enum drm_exynos_ipp_buf_type buf_type)
> +			       bool enqueue)
>   {
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> -	bool masked;
> +	bool masked = !enqueue;
>   	u32 cfg;
>   	u32 mask = 0x00000001 << buf_id;
>   
> -	DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type);
> -
>   	/* mask register set */
>   	cfg = gsc_read(GSC_IN_BASE_ADDR_Y_MASK);
>   
> -	switch (buf_type) {
> -	case IPP_BUF_ENQUEUE:
> -		masked = false;
> -		break;
> -	case IPP_BUF_DEQUEUE:
> -		masked = true;
> -		break;
> -	default:
> -		dev_err(ippdrv->dev, "invalid buf ctrl parameter.\n");
> -		return -EINVAL;
> -	}
> -
>   	/* sequence id */
>   	cfg &= ~mask;
>   	cfg |= masked << buf_id;
> @@ -707,64 +629,19 @@ static int gsc_src_set_buf_seq(struct gsc_context *ctx, u32 buf_id,
>   	return 0;
>   }
>   
> -static int gsc_src_set_addr(struct device *dev,
> -		struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id,
> -		enum drm_exynos_ipp_buf_type buf_type)
> +static int gsc_src_set_addr(struct gsc_context *ctx, u32 buf_id,
> +			    struct exynos_drm_ipp_buffer *buf)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> -	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
> -	struct drm_exynos_ipp_property *property;
> -
> -	if (!c_node) {
> -		DRM_ERROR("failed to get c_node.\n");
> -		return -EFAULT;
> -	}
> -
> -	property = &c_node->property;
> -
> -	DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n",
> -		property->prop_id, buf_id, buf_type);
> -
> -	if (buf_id > GSC_MAX_SRC) {
> -		dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id);
> -		return -EINVAL;
> -	}
> -
>   	/* address register set */
> -	switch (buf_type) {
> -	case IPP_BUF_ENQUEUE:
> -		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_Y],
> -			GSC_IN_BASE_ADDR_Y(buf_id));
> -		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CB],
> -			GSC_IN_BASE_ADDR_CB(buf_id));
> -		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CR],
> -			GSC_IN_BASE_ADDR_CR(buf_id));
> -		break;
> -	case IPP_BUF_DEQUEUE:
> -		gsc_write(0x0, GSC_IN_BASE_ADDR_Y(buf_id));
> -		gsc_write(0x0, GSC_IN_BASE_ADDR_CB(buf_id));
> -		gsc_write(0x0, GSC_IN_BASE_ADDR_CR(buf_id));
> -		break;
> -	default:
> -		/* bypass */
> -		break;
> -	}
> +	gsc_write(buf->dma_addr[0], GSC_IN_BASE_ADDR_Y(buf_id));
> +	gsc_write(buf->dma_addr[1], GSC_IN_BASE_ADDR_CB(buf_id));
> +	gsc_write(buf->dma_addr[2], GSC_IN_BASE_ADDR_CR(buf_id));
>   
> -	return gsc_src_set_buf_seq(ctx, buf_id, buf_type);
> +	return gsc_src_set_buf_seq(ctx, buf_id, true);
>   }
>   
> -static struct exynos_drm_ipp_ops gsc_src_ops = {
> -	.set_fmt = gsc_src_set_fmt,
> -	.set_transf = gsc_src_set_transf,
> -	.set_size = gsc_src_set_size,
> -	.set_addr = gsc_src_set_addr,
> -};
> -
> -static int gsc_dst_set_fmt(struct device *dev, u32 fmt)
> +static int gsc_dst_set_fmt(struct gsc_context *ctx, u32 fmt)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
>   	u32 cfg;
>   
>   	DRM_DEBUG_KMS("fmt[0x%x]\n", fmt);
> @@ -819,9 +696,6 @@ static int gsc_dst_set_fmt(struct device *dev, u32 fmt)
>   		cfg |= (GSC_OUT_CHROMA_ORDER_CBCR |
>   			GSC_OUT_YUV420_2P);
>   		break;
> -	default:
> -		dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt);
> -		return -EINVAL;
>   	}
>   
>   	gsc_write(cfg, GSC_OUT_CON);
> @@ -829,61 +703,6 @@ static int gsc_dst_set_fmt(struct device *dev, u32 fmt)
>   	return 0;
>   }
>   
> -static int gsc_dst_set_transf(struct device *dev,
> -		enum drm_exynos_degree degree,
> -		enum drm_exynos_flip flip, bool *swap)
> -{
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> -	u32 cfg;
> -
> -	DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip);
> -
> -	cfg = gsc_read(GSC_IN_CON);
> -	cfg &= ~GSC_IN_ROT_MASK;
> -
> -	switch (degree) {
> -	case EXYNOS_DRM_DEGREE_0:
> -		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
> -			cfg |= GSC_IN_ROT_XFLIP;
> -		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
> -			cfg |= GSC_IN_ROT_YFLIP;
> -		break;
> -	case EXYNOS_DRM_DEGREE_90:
> -		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
> -			cfg |= GSC_IN_ROT_90_XFLIP;
> -		else if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
> -			cfg |= GSC_IN_ROT_90_YFLIP;
> -		else
> -			cfg |= GSC_IN_ROT_90;
> -		break;
> -	case EXYNOS_DRM_DEGREE_180:
> -		cfg |= GSC_IN_ROT_180;
> -		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
> -			cfg &= ~GSC_IN_ROT_XFLIP;
> -		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
> -			cfg &= ~GSC_IN_ROT_YFLIP;
> -		break;
> -	case EXYNOS_DRM_DEGREE_270:
> -		cfg |= GSC_IN_ROT_270;
> -		if (flip & EXYNOS_DRM_FLIP_VERTICAL)
> -			cfg &= ~GSC_IN_ROT_XFLIP;
> -		if (flip & EXYNOS_DRM_FLIP_HORIZONTAL)
> -			cfg &= ~GSC_IN_ROT_YFLIP;
> -		break;
> -	default:
> -		dev_err(ippdrv->dev, "invalid degree value %d.\n", degree);
> -		return -EINVAL;
> -	}
> -
> -	gsc_write(cfg, GSC_IN_CON);
> -
> -	ctx->rotation = (cfg & GSC_IN_ROT_90) ? 1 : 0;
> -	*swap = ctx->rotation;
> -
> -	return 0;
> -}
> -
>   static int gsc_get_ratio_shift(u32 src, u32 dst, u32 *ratio)
>   {
>   	DRM_DEBUG_KMS("src[%d]dst[%d]\n", src, dst);
> @@ -919,9 +738,8 @@ static void gsc_get_prescaler_shfactor(u32 hratio, u32 vratio, u32 *shfactor)
>   }
>   
>   static int gsc_set_prescaler(struct gsc_context *ctx, struct gsc_scaler *sc,
> -		struct drm_exynos_pos *src, struct drm_exynos_pos *dst)
> +     struct drm_exynos_ipp_task_rect *src, struct drm_exynos_ipp_task_rect *dst)
>   {
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
>   	u32 cfg;
>   	u32 src_w, src_h, dst_w, dst_h;
>   	int ret = 0;
> @@ -939,13 +757,13 @@ static int gsc_set_prescaler(struct gsc_context *ctx, struct gsc_scaler *sc,
>   
>   	ret = gsc_get_ratio_shift(src_w, dst_w, &sc->pre_hratio);
>   	if (ret) {
> -		dev_err(ippdrv->dev, "failed to get ratio horizontal.\n");
> +		dev_err(ctx->dev, "failed to get ratio horizontal.\n");
>   		return ret;
>   	}
>   
>   	ret = gsc_get_ratio_shift(src_h, dst_h, &sc->pre_vratio);
>   	if (ret) {
> -		dev_err(ippdrv->dev, "failed to get ratio vertical.\n");
> +		dev_err(ctx->dev, "failed to get ratio vertical.\n");
>   		return ret;
>   	}
>   
> @@ -1039,47 +857,31 @@ static void gsc_set_scaler(struct gsc_context *ctx, struct gsc_scaler *sc)
>   	gsc_write(cfg, GSC_MAIN_V_RATIO);
>   }
>   
> -static int gsc_dst_set_size(struct device *dev, int swap,
> -		struct drm_exynos_pos *pos, struct drm_exynos_sz *sz)
> +static int gsc_dst_set_size(struct gsc_context *ctx,
> +			    struct exynos_drm_ipp_buffer *buf)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct drm_exynos_pos img_pos = *pos;
>   	struct gsc_scaler *sc = &ctx->sc;
>   	u32 cfg;
>   
> -	DRM_DEBUG_KMS("swap[%d]x[%d]y[%d]w[%d]h[%d]\n",
> -		swap, pos->x, pos->y, pos->w, pos->h);
> -
> -	if (swap) {
> -		img_pos.w = pos->h;
> -		img_pos.h = pos->w;
> -	}
> -
>   	/* pixel offset */
> -	cfg = (GSC_DSTIMG_OFFSET_X(pos->x) |
> -		GSC_DSTIMG_OFFSET_Y(pos->y));
> +	cfg = (GSC_DSTIMG_OFFSET_X(buf->rect.x) |
> +		GSC_DSTIMG_OFFSET_Y(buf->rect.y));
>   	gsc_write(cfg, GSC_DSTIMG_OFFSET);
>   
>   	/* scaled size */
> -	cfg = (GSC_SCALED_WIDTH(img_pos.w) | GSC_SCALED_HEIGHT(img_pos.h));
> +	cfg = (GSC_SCALED_WIDTH(buf->rect.w) | GSC_SCALED_HEIGHT(buf->rect.h));
>   	gsc_write(cfg, GSC_SCALED_SIZE);
>   
> -	DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", sz->hsize, sz->vsize);
> -
>   	/* original size */
>   	cfg = gsc_read(GSC_DSTIMG_SIZE);
> -	cfg &= ~(GSC_DSTIMG_HEIGHT_MASK |
> -		GSC_DSTIMG_WIDTH_MASK);
> -	cfg |= (GSC_DSTIMG_WIDTH(sz->hsize) |
> -		GSC_DSTIMG_HEIGHT(sz->vsize));
> +	cfg &= ~(GSC_DSTIMG_HEIGHT_MASK | GSC_DSTIMG_WIDTH_MASK);
> +	cfg |= (GSC_DSTIMG_WIDTH(buf->buf.width) | GSC_DSTIMG_HEIGHT(buf->buf.height));
>   	gsc_write(cfg, GSC_DSTIMG_SIZE);
>   
>   	cfg = gsc_read(GSC_OUT_CON);
>   	cfg &= ~GSC_OUT_RGB_TYPE_MASK;
>   
> -	DRM_DEBUG_KMS("width[%d]range[%d]\n", pos->w, sc->range);
> -
> -	if (pos->w >= GSC_WIDTH_ITU_709)
> +	if (buf->rect.w >= GSC_WIDTH_ITU_709)
>   		if (sc->range)
>   			cfg |= GSC_OUT_RGB_HD_WIDE;
>   		else
> @@ -1112,34 +914,16 @@ static int gsc_dst_get_buf_seq(struct gsc_context *ctx)
>   }
>   
>   static int gsc_dst_set_buf_seq(struct gsc_context *ctx, u32 buf_id,
> -		enum drm_exynos_ipp_buf_type buf_type)
> +			       bool enqueue)
>   {
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> -	bool masked;
> +	bool masked = !enqueue;
>   	u32 cfg;
>   	u32 mask = 0x00000001 << buf_id;
>   	int ret = 0;
>   
> -	DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type);
> -
> -	mutex_lock(&ctx->lock);
> -
>   	/* mask register set */
>   	cfg = gsc_read(GSC_OUT_BASE_ADDR_Y_MASK);
>   
> -	switch (buf_type) {
> -	case IPP_BUF_ENQUEUE:
> -		masked = false;
> -		break;
> -	case IPP_BUF_DEQUEUE:
> -		masked = true;
> -		break;
> -	default:
> -		dev_err(ippdrv->dev, "invalid buf ctrl parameter.\n");
> -		ret =  -EINVAL;
> -		goto err_unlock;
> -	}
> -
>   	/* sequence id */
>   	cfg &= ~mask;
>   	cfg |= masked << buf_id;
> @@ -1148,74 +932,27 @@ static int gsc_dst_set_buf_seq(struct gsc_context *ctx, u32 buf_id,
>   	gsc_write(cfg, GSC_OUT_BASE_ADDR_CR_MASK);
>   
>   	/* interrupt enable */
> -	if (buf_type == IPP_BUF_ENQUEUE &&
> -	    gsc_dst_get_buf_seq(ctx) >= GSC_BUF_START)
> +	if (enqueue && gsc_dst_get_buf_seq(ctx) >= GSC_BUF_START)
>   		gsc_handle_irq(ctx, true, false, true);
>   
>   	/* interrupt disable */
> -	if (buf_type == IPP_BUF_DEQUEUE &&
> -	    gsc_dst_get_buf_seq(ctx) <= GSC_BUF_STOP)
> +	if (!enqueue && gsc_dst_get_buf_seq(ctx) <= GSC_BUF_STOP)
>   		gsc_handle_irq(ctx, false, false, true);
>   
> -err_unlock:
> -	mutex_unlock(&ctx->lock);
>   	return ret;
>   }
>   
> -static int gsc_dst_set_addr(struct device *dev,
> -		struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id,
> -		enum drm_exynos_ipp_buf_type buf_type)
> +static int gsc_dst_set_addr(struct gsc_context *ctx,
> +		u32 buf_id, struct exynos_drm_ipp_buffer *buf)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> -	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
> -	struct drm_exynos_ipp_property *property;
> -
> -	if (!c_node) {
> -		DRM_ERROR("failed to get c_node.\n");
> -		return -EFAULT;
> -	}
> -
> -	property = &c_node->property;
> -
> -	DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n",
> -		property->prop_id, buf_id, buf_type);
> -
> -	if (buf_id > GSC_MAX_DST) {
> -		dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id);
> -		return -EINVAL;
> -	}
> -
>   	/* address register set */
> -	switch (buf_type) {
> -	case IPP_BUF_ENQUEUE:
> -		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_Y],
> -			GSC_OUT_BASE_ADDR_Y(buf_id));
> -		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CB],
> -			GSC_OUT_BASE_ADDR_CB(buf_id));
> -		gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CR],
> -			GSC_OUT_BASE_ADDR_CR(buf_id));
> -		break;
> -	case IPP_BUF_DEQUEUE:
> -		gsc_write(0x0, GSC_OUT_BASE_ADDR_Y(buf_id));
> -		gsc_write(0x0, GSC_OUT_BASE_ADDR_CB(buf_id));
> -		gsc_write(0x0, GSC_OUT_BASE_ADDR_CR(buf_id));
> -		break;
> -	default:
> -		/* bypass */
> -		break;
> -	}
> +	gsc_write(buf->dma_addr[0], GSC_OUT_BASE_ADDR_Y(buf_id));
> +	gsc_write(buf->dma_addr[1], GSC_OUT_BASE_ADDR_CB(buf_id));
> +	gsc_write(buf->dma_addr[2], GSC_OUT_BASE_ADDR_CR(buf_id));
>   
> -	return gsc_dst_set_buf_seq(ctx, buf_id, buf_type);
> +	return gsc_dst_set_buf_seq(ctx, buf_id, true);
>   }
>   
> -static struct exynos_drm_ipp_ops gsc_dst_ops = {
> -	.set_fmt = gsc_dst_set_fmt,
> -	.set_transf = gsc_dst_set_transf,
> -	.set_size = gsc_dst_set_size,
> -	.set_addr = gsc_dst_set_addr,
> -};
> -
>   static int gsc_clk_ctrl(struct gsc_context *ctx, bool enable)
>   {
>   	DRM_DEBUG_KMS("enable[%d]\n", enable);
> @@ -1249,20 +986,20 @@ static int gsc_get_src_buf_index(struct gsc_context *ctx)
>   		}
>   	}
>   
> +	DRM_DEBUG_KMS("cfg[0x%x]curr_index[%d]buf_id[%d]\n", cfg,
> +		curr_index, buf_id);
> +
>   	if (buf_id == GSC_MAX_SRC) {
>   		DRM_ERROR("failed to get in buffer index.\n");
>   		return -EINVAL;
>   	}
>   
> -	ret = gsc_src_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE);
> +	ret = gsc_src_set_buf_seq(ctx, buf_id, false);
>   	if (ret < 0) {
>   		DRM_ERROR("failed to dequeue.\n");
>   		return ret;
>   	}
>   
> -	DRM_DEBUG_KMS("cfg[0x%x]curr_index[%d]buf_id[%d]\n", cfg,
> -		curr_index, buf_id);
> -
>   	return buf_id;
>   }
>   
> @@ -1289,7 +1026,7 @@ static int gsc_get_dst_buf_index(struct gsc_context *ctx)
>   		return -EINVAL;
>   	}
>   
> -	ret = gsc_dst_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE);
> +	ret = gsc_dst_set_buf_seq(ctx, buf_id, false);
>   	if (ret < 0) {
>   		DRM_ERROR("failed to dequeue.\n");
>   		return ret;
> @@ -1304,215 +1041,54 @@ static int gsc_get_dst_buf_index(struct gsc_context *ctx)
>   static irqreturn_t gsc_irq_handler(int irq, void *dev_id)
>   {
>   	struct gsc_context *ctx = dev_id;
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> -	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
> -	struct drm_exynos_ipp_event_work *event_work =
> -		c_node->event_work;
>   	u32 status;
> -	int buf_id[EXYNOS_DRM_OPS_MAX];
> +	int err = 0;
>   
>   	DRM_DEBUG_KMS("gsc id[%d]\n", ctx->id);
>   
>   	status = gsc_read(GSC_IRQ);
>   	if (status & GSC_IRQ_STATUS_OR_IRQ) {
> -		dev_err(ippdrv->dev, "occurred overflow at %d, status 0x%x.\n",
> +		dev_err(ctx->dev, "occurred overflow at %d, status 0x%x.\n",
>   			ctx->id, status);
> -		return IRQ_NONE;
> +		err = -EINVAL;
>   	}
>   
>   	if (status & GSC_IRQ_STATUS_OR_FRM_DONE) {
> -		dev_dbg(ippdrv->dev, "occurred frame done at %d, status 0x%x.\n",
> -			ctx->id, status);
> -
> -		buf_id[EXYNOS_DRM_OPS_SRC] = gsc_get_src_buf_index(ctx);
> -		if (buf_id[EXYNOS_DRM_OPS_SRC] < 0)
> -			return IRQ_HANDLED;
> -
> -		buf_id[EXYNOS_DRM_OPS_DST] = gsc_get_dst_buf_index(ctx);
> -		if (buf_id[EXYNOS_DRM_OPS_DST] < 0)
> -			return IRQ_HANDLED;
> +		int src_buf_id, dst_buf_id;
>   
> -		DRM_DEBUG_KMS("buf_id_src[%d]buf_id_dst[%d]\n",
> -			buf_id[EXYNOS_DRM_OPS_SRC], buf_id[EXYNOS_DRM_OPS_DST]);
> -
> -		event_work->ippdrv = ippdrv;
> -		event_work->buf_id[EXYNOS_DRM_OPS_SRC] =
> -			buf_id[EXYNOS_DRM_OPS_SRC];
> -		event_work->buf_id[EXYNOS_DRM_OPS_DST] =
> -			buf_id[EXYNOS_DRM_OPS_DST];
> -		queue_work(ippdrv->event_workq, &event_work->work);
> -	}
> -
> -	return IRQ_HANDLED;
> -}
> -
> -static int gsc_init_prop_list(struct exynos_drm_ippdrv *ippdrv)
> -{
> -	struct drm_exynos_ipp_prop_list *prop_list = &ippdrv->prop_list;
> -
> -	prop_list->version = 1;
> -	prop_list->writeback = 1;
> -	prop_list->refresh_min = GSC_REFRESH_MIN;
> -	prop_list->refresh_max = GSC_REFRESH_MAX;
> -	prop_list->flip = (1 << EXYNOS_DRM_FLIP_VERTICAL) |
> -				(1 << EXYNOS_DRM_FLIP_HORIZONTAL);
> -	prop_list->degree = (1 << EXYNOS_DRM_DEGREE_0) |
> -				(1 << EXYNOS_DRM_DEGREE_90) |
> -				(1 << EXYNOS_DRM_DEGREE_180) |
> -				(1 << EXYNOS_DRM_DEGREE_270);
> -	prop_list->csc = 1;
> -	prop_list->crop = 1;
> -	prop_list->crop_max.hsize = GSC_CROP_MAX;
> -	prop_list->crop_max.vsize = GSC_CROP_MAX;
> -	prop_list->crop_min.hsize = GSC_CROP_MIN;
> -	prop_list->crop_min.vsize = GSC_CROP_MIN;
> -	prop_list->scale = 1;
> -	prop_list->scale_max.hsize = GSC_SCALE_MAX;
> -	prop_list->scale_max.vsize = GSC_SCALE_MAX;
> -	prop_list->scale_min.hsize = GSC_SCALE_MIN;
> -	prop_list->scale_min.vsize = GSC_SCALE_MIN;
> -
> -	return 0;
> -}
> -
> -static inline bool gsc_check_drm_flip(enum drm_exynos_flip flip)
> -{
> -	switch (flip) {
> -	case EXYNOS_DRM_FLIP_NONE:
> -	case EXYNOS_DRM_FLIP_VERTICAL:
> -	case EXYNOS_DRM_FLIP_HORIZONTAL:
> -	case EXYNOS_DRM_FLIP_BOTH:
> -		return true;
> -	default:
> -		DRM_DEBUG_KMS("invalid flip\n");
> -		return false;
> -	}
> -}
> -
> -static int gsc_ippdrv_check_property(struct device *dev,
> -		struct drm_exynos_ipp_property *property)
> -{
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> -	struct drm_exynos_ipp_prop_list *pp = &ippdrv->prop_list;
> -	struct drm_exynos_ipp_config *config;
> -	struct drm_exynos_pos *pos;
> -	struct drm_exynos_sz *sz;
> -	bool swap;
> -	int i;
> -
> -	for_each_ipp_ops(i) {
> -		if ((i == EXYNOS_DRM_OPS_SRC) &&
> -			(property->cmd == IPP_CMD_WB))
> -			continue;
> -
> -		config = &property->config[i];
> -		pos = &config->pos;
> -		sz = &config->sz;
> -
> -		/* check for flip */
> -		if (!gsc_check_drm_flip(config->flip)) {
> -			DRM_ERROR("invalid flip.\n");
> -			goto err_property;
> -		}
> -
> -		/* check for degree */
> -		switch (config->degree) {
> -		case EXYNOS_DRM_DEGREE_90:
> -		case EXYNOS_DRM_DEGREE_270:
> -			swap = true;
> -			break;
> -		case EXYNOS_DRM_DEGREE_0:
> -		case EXYNOS_DRM_DEGREE_180:
> -			swap = false;
> -			break;
> -		default:
> -			DRM_ERROR("invalid degree.\n");
> -			goto err_property;
> -		}
> +		dev_dbg(ctx->dev, "occurred frame done at %d, status 0x%x.\n",
> +			ctx->id, status);
>   
> -		/* check for buffer bound */
> -		if ((pos->x + pos->w > sz->hsize) ||
> -			(pos->y + pos->h > sz->vsize)) {
> -			DRM_ERROR("out of buf bound.\n");
> -			goto err_property;
> -		}
> +		src_buf_id = gsc_get_src_buf_index(ctx);
> +		dst_buf_id = gsc_get_dst_buf_index(ctx);
>   
> -		/* check for crop */
> -		if ((i == EXYNOS_DRM_OPS_SRC) && (pp->crop)) {
> -			if (swap) {
> -				if ((pos->h < pp->crop_min.hsize) ||
> -					(sz->vsize > pp->crop_max.hsize) ||
> -					(pos->w < pp->crop_min.vsize) ||
> -					(sz->hsize > pp->crop_max.vsize)) {
> -					DRM_ERROR("out of crop size.\n");
> -					goto err_property;
> -				}
> -			} else {
> -				if ((pos->w < pp->crop_min.hsize) ||
> -					(sz->hsize > pp->crop_max.hsize) ||
> -					(pos->h < pp->crop_min.vsize) ||
> -					(sz->vsize > pp->crop_max.vsize)) {
> -					DRM_ERROR("out of crop size.\n");
> -					goto err_property;
> -				}
> -			}
> -		}
> +		DRM_DEBUG_KMS("buf_id_src[%d]buf_id_dst[%d]\n",	src_buf_id,
> +			      dst_buf_id);
>   
> -		/* check for scale */
> -		if ((i == EXYNOS_DRM_OPS_DST) && (pp->scale)) {
> -			if (swap) {
> -				if ((pos->h < pp->scale_min.hsize) ||
> -					(sz->vsize > pp->scale_max.hsize) ||
> -					(pos->w < pp->scale_min.vsize) ||
> -					(sz->hsize > pp->scale_max.vsize)) {
> -					DRM_ERROR("out of scale size.\n");
> -					goto err_property;
> -				}
> -			} else {
> -				if ((pos->w < pp->scale_min.hsize) ||
> -					(sz->hsize > pp->scale_max.hsize) ||
> -					(pos->h < pp->scale_min.vsize) ||
> -					(sz->vsize > pp->scale_max.vsize)) {
> -					DRM_ERROR("out of scale size.\n");
> -					goto err_property;
> -				}
> -			}
> -		}
> +		if (src_buf_id < 0 || dst_buf_id < 0)
> +			err = -EINVAL;
>   	}
>   
> -	return 0;
> +	if (ctx->task) {
> +		struct exynos_drm_ipp_task *task = ctx->task;
>   
> -err_property:
> -	for_each_ipp_ops(i) {
> -		if ((i == EXYNOS_DRM_OPS_SRC) &&
> -			(property->cmd == IPP_CMD_WB))
> -			continue;
> -
> -		config = &property->config[i];
> -		pos = &config->pos;
> -		sz = &config->sz;
> -
> -		DRM_ERROR("[%s]f[%d]r[%d]pos[%d %d %d %d]sz[%d %d]\n",
> -			i ? "dst" : "src", config->flip, config->degree,
> -			pos->x, pos->y, pos->w, pos->h,
> -			sz->hsize, sz->vsize);
> +		ctx->task = NULL;
> +		pm_runtime_put(ctx->dev);
> +		exynos_drm_ipp_task_done(task, err);
>   	}
>   
> -	return -EINVAL;
> +	return IRQ_HANDLED;
>   }
>   
> -
> -static int gsc_ippdrv_reset(struct device *dev)
> +static int gsc_reset(struct gsc_context *ctx)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
>   	struct gsc_scaler *sc = &ctx->sc;
>   	int ret;
>   
>   	/* reset h/w block */
>   	ret = gsc_sw_reset(ctx);
>   	if (ret < 0) {
> -		dev_err(dev, "failed to reset hardware.\n");
> +		dev_err(ctx->dev, "failed to reset hardware.\n");
>   		return ret;
>   	}
>   
> @@ -1523,96 +1099,29 @@ static int gsc_ippdrv_reset(struct device *dev)
>   	return 0;
>   }
>   
> -static int gsc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd)
> +static int gsc_start(struct gsc_context *ctx)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> -	struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node;
> -	struct drm_exynos_ipp_property *property;
> -	struct drm_exynos_ipp_config *config;
> -	struct drm_exynos_pos	img_pos[EXYNOS_DRM_OPS_MAX];
> -	struct drm_exynos_ipp_set_wb set_wb;
>   	u32 cfg;
> -	int ret, i;
> -
> -	DRM_DEBUG_KMS("cmd[%d]\n", cmd);
> -
> -	if (!c_node) {
> -		DRM_ERROR("failed to get c_node.\n");
> -		return -EINVAL;
> -	}
> -
> -	property = &c_node->property;
>   
>   	gsc_handle_irq(ctx, true, false, true);
>   
> -	for_each_ipp_ops(i) {
> -		config = &property->config[i];
> -		img_pos[i] = config->pos;
> -	}
> +	/* enable one shot */
> +	cfg = gsc_read(GSC_ENABLE);
> +	cfg &= ~(GSC_ENABLE_ON_CLEAR_MASK |
> +		GSC_ENABLE_CLK_GATE_MODE_MASK);
> +	cfg |= GSC_ENABLE_ON_CLEAR_ONESHOT;
> +	gsc_write(cfg, GSC_ENABLE);
>   
> -	switch (cmd) {
> -	case IPP_CMD_M2M:
> -		/* enable one shot */
> -		cfg = gsc_read(GSC_ENABLE);
> -		cfg &= ~(GSC_ENABLE_ON_CLEAR_MASK |
> -			GSC_ENABLE_CLK_GATE_MODE_MASK);
> -		cfg |= GSC_ENABLE_ON_CLEAR_ONESHOT;
> -		gsc_write(cfg, GSC_ENABLE);
> -
> -		/* src dma memory */
> -		cfg = gsc_read(GSC_IN_CON);
> -		cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK);
> -		cfg |= GSC_IN_PATH_MEMORY;
> -		gsc_write(cfg, GSC_IN_CON);
> -
> -		/* dst dma memory */
> -		cfg = gsc_read(GSC_OUT_CON);
> -		cfg |= GSC_OUT_PATH_MEMORY;
> -		gsc_write(cfg, GSC_OUT_CON);
> -		break;
> -	case IPP_CMD_WB:
> -		set_wb.enable = 1;
> -		set_wb.refresh = property->refresh_rate;
> -		gsc_set_gscblk_fimd_wb(ctx, set_wb.enable);
> -		exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb);
> -
> -		/* src local path */
> -		cfg = gsc_read(GSC_IN_CON);
> -		cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK);
> -		cfg |= (GSC_IN_PATH_LOCAL | GSC_IN_LOCAL_FIMD_WB);
> -		gsc_write(cfg, GSC_IN_CON);
> -
> -		/* dst dma memory */
> -		cfg = gsc_read(GSC_OUT_CON);
> -		cfg |= GSC_OUT_PATH_MEMORY;
> -		gsc_write(cfg, GSC_OUT_CON);
> -		break;
> -	case IPP_CMD_OUTPUT:
> -		/* src dma memory */
> -		cfg = gsc_read(GSC_IN_CON);
> -		cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK);
> -		cfg |= GSC_IN_PATH_MEMORY;
> -		gsc_write(cfg, GSC_IN_CON);
> -
> -		/* dst local path */
> -		cfg = gsc_read(GSC_OUT_CON);
> -		cfg |= GSC_OUT_PATH_MEMORY;
> -		gsc_write(cfg, GSC_OUT_CON);
> -		break;
> -	default:
> -		ret = -EINVAL;
> -		dev_err(dev, "invalid operations.\n");
> -		return ret;
> -	}
> +	/* src dma memory */
> +	cfg = gsc_read(GSC_IN_CON);
> +	cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK);
> +	cfg |= GSC_IN_PATH_MEMORY;
> +	gsc_write(cfg, GSC_IN_CON);
>   
> -	ret = gsc_set_prescaler(ctx, &ctx->sc,
> -		&img_pos[EXYNOS_DRM_OPS_SRC],
> -		&img_pos[EXYNOS_DRM_OPS_DST]);
> -	if (ret) {
> -		dev_err(dev, "failed to set prescaler.\n");
> -		return ret;
> -	}
> +	/* dst dma memory */
> +	cfg = gsc_read(GSC_OUT_CON);
> +	cfg |= GSC_OUT_PATH_MEMORY;
> +	gsc_write(cfg, GSC_OUT_CON);
>   
>   	gsc_set_scaler(ctx, &ctx->sc);
>   
> @@ -1623,60 +1132,112 @@ static int gsc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd)
>   	return 0;
>   }
>   
> -static void gsc_ippdrv_stop(struct device *dev, enum drm_exynos_ipp_cmd cmd)
> +static int gsc_commit(struct exynos_drm_ipp *ipp,
> +			  struct exynos_drm_ipp_task *task)
>   {
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct drm_exynos_ipp_set_wb set_wb = {0, 0};
> -	u32 cfg;
> +	struct gsc_context *ctx = container_of(ipp, struct gsc_context, ipp);
> +
> +	pm_runtime_get_sync(ctx->dev);
> +	ctx->task = task;
> +
> +	gsc_reset(ctx);
> +	gsc_src_set_fmt(ctx, task->src.buf.fourcc);
> +	gsc_src_set_transf(ctx, task->transform.rotation);
> +	gsc_src_set_size(ctx, &task->src);
> +	gsc_src_set_addr(ctx, 0, &task->src);
> +	gsc_dst_set_fmt(ctx, task->dst.buf.fourcc);
> +	gsc_dst_set_size(ctx, &task->dst);
> +	gsc_dst_set_addr(ctx, 0, &task->dst);
> +	gsc_set_prescaler(ctx, &ctx->sc, &task->src.rect, &task->dst.rect);
> +	gsc_start(ctx);
>   
> -	DRM_DEBUG_KMS("cmd[%d]\n", cmd);
> +	return 0;
> +}
>   
> -	switch (cmd) {
> -	case IPP_CMD_M2M:
> -		/* bypass */
> -		break;
> -	case IPP_CMD_WB:
> -		gsc_set_gscblk_fimd_wb(ctx, set_wb.enable);
> -		exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb);
> -		break;
> -	case IPP_CMD_OUTPUT:
> -	default:
> -		dev_err(dev, "invalid operations.\n");
> -		break;
> +static void gsc_abort(struct exynos_drm_ipp *ipp,
> +			  struct exynos_drm_ipp_task *task)
> +{
> +	struct gsc_context *ctx =
> +			container_of(ipp, struct gsc_context, ipp);
> +
> +	gsc_reset(ctx);
> +	if (ctx->task) {
> +		struct exynos_drm_ipp_task *task = ctx->task;
> +
> +		ctx->task = NULL;
> +		pm_runtime_put(ctx->dev);
> +		exynos_drm_ipp_task_done(task, -EIO);
>   	}
> +}
>   
> -	gsc_handle_irq(ctx, false, false, true);
> +static struct exynos_drm_ipp_funcs ipp_funcs = {
> +	.commit = gsc_commit,
> +	.abort = gsc_abort,
> +};
>   
> -	/* reset sequence */
> -	gsc_write(0xff, GSC_OUT_BASE_ADDR_Y_MASK);
> -	gsc_write(0xff, GSC_OUT_BASE_ADDR_CB_MASK);
> -	gsc_write(0xff, GSC_OUT_BASE_ADDR_CR_MASK);
> +static const struct exynos_drm_ipp_formats gsc_formats[] = {
> +	{ IPP_SRCDST_FORMAT(XRGB8888, NULL) },
> +	{ IPP_SRCDST_FORMAT(RGB565, NULL) },
> +	{ IPP_SRCDST_FORMAT(BGRX8888, NULL) },
> +	{ IPP_SRCDST_FORMAT(NV12, NULL) },
> +	{ IPP_SRCDST_FORMAT(NV16, NULL) },
> +	{ IPP_SRCDST_FORMAT(NV21, NULL) },
> +	{ IPP_SRCDST_FORMAT(NV61, NULL) },
> +	{ IPP_SRCDST_FORMAT(UYVY, NULL) },
> +	{ IPP_SRCDST_FORMAT(VYUY, NULL) },
> +	{ IPP_SRCDST_FORMAT(YUYV, NULL) },
> +	{ IPP_SRCDST_FORMAT(YVYU, NULL) },
> +	{ IPP_SRCDST_FORMAT(YUV420, NULL) },
> +	{ IPP_SRCDST_FORMAT(YVU420, NULL) },
> +	{ IPP_SRCDST_FORMAT(YUV422, NULL) },
> +	{ }
> +};
>   
> -	cfg = gsc_read(GSC_ENABLE);
> -	cfg &= ~GSC_ENABLE_ON;
> -	gsc_write(cfg, GSC_ENABLE);
> +static int gsc_bind(struct device *dev, struct device *master, void *data)
> +{
> +	struct gsc_context *ctx = dev_get_drvdata(dev);
> +	struct drm_device *drm_dev = data;
> +	struct exynos_drm_ipp *ipp = &ctx->ipp;
> +
> +	ctx->drm_dev = drm_dev;
> +	drm_iommu_attach_device(drm_dev, dev);
> +
> +	exynos_drm_ipp_register(drm_dev, ipp, &ipp_funcs,
> +			   DRM_EXYNOS_IPP_CAP_CROP | DRM_EXYNOS_IPP_CAP_ROTATE |
> +			   DRM_EXYNOS_IPP_CAP_SCALE | DRM_EXYNOS_IPP_CAP_CONVERT,
> +			   gsc_formats, "gsc");
> +
> +	dev_info(dev, "The exynos gscaler is probed successfully\n");
> +
> +	return 0;
>   }
>   
> +static void gsc_unbind(struct device *dev, struct device *master,
> +			void *data)
> +{
> +	struct gsc_context *ctx = dev_get_drvdata(dev);
> +
> +	drm_iommu_detach_device(ctx->drm_dev, ctx->dev);
> +}
> +
> +static const struct component_ops gsc_component_ops = {
> +	.bind	= gsc_bind,
> +	.unbind = gsc_unbind,
> +};
> +
> +
>   static int gsc_probe(struct platform_device *pdev)
>   {
>   	struct device *dev = &pdev->dev;
>   	struct gsc_context *ctx;
>   	struct resource *res;
> -	struct exynos_drm_ippdrv *ippdrv;
>   	int ret;
>   
>   	ctx = devm_kzalloc(dev, sizeof(*ctx), GFP_KERNEL);
>   	if (!ctx)
>   		return -ENOMEM;
>   
> -	if (dev->of_node) {
> -		ctx->sysreg = syscon_regmap_lookup_by_phandle(dev->of_node,
> -							"samsung,sysreg");
> -		if (IS_ERR(ctx->sysreg)) {
> -			dev_warn(dev, "failed to get system register.\n");
> -			ctx->sysreg = NULL;
> -		}
> -	}
> +	ctx->dev = dev;
>   
>   	/* clock control */
>   	ctx->gsc_clk = devm_clk_get(dev, "gscl");
> @@ -1699,8 +1260,8 @@ static int gsc_probe(struct platform_device *pdev)
>   	}
>   
>   	ctx->irq = res->start;
> -	ret = devm_request_threaded_irq(dev, ctx->irq, NULL, gsc_irq_handler,
> -		IRQF_ONESHOT, "drm_gsc", ctx);
> +	ret = devm_request_irq(dev, ctx->irq, gsc_irq_handler, 0,
> +			       dev_name(dev), ctx);
>   	if (ret < 0) {
>   		dev_err(dev, "failed to request irq.\n");
>   		return ret;
> @@ -1709,38 +1270,19 @@ static int gsc_probe(struct platform_device *pdev)
>   	/* context initailization */
>   	ctx->id = pdev->id;
>   
> -	ippdrv = &ctx->ippdrv;
> -	ippdrv->dev = dev;
> -	ippdrv->ops[EXYNOS_DRM_OPS_SRC] = &gsc_src_ops;
> -	ippdrv->ops[EXYNOS_DRM_OPS_DST] = &gsc_dst_ops;
> -	ippdrv->check_property = gsc_ippdrv_check_property;
> -	ippdrv->reset = gsc_ippdrv_reset;
> -	ippdrv->start = gsc_ippdrv_start;
> -	ippdrv->stop = gsc_ippdrv_stop;
> -	ret = gsc_init_prop_list(ippdrv);
> -	if (ret < 0) {
> -		dev_err(dev, "failed to init property list.\n");
> -		return ret;
> -	}
> -
> -	DRM_DEBUG_KMS("id[%d]ippdrv[%pK]\n", ctx->id, ippdrv);
> -
> -	mutex_init(&ctx->lock);
>   	platform_set_drvdata(pdev, ctx);
>   
>   	pm_runtime_enable(dev);
>   
> -	ret = exynos_drm_ippdrv_register(ippdrv);
> -	if (ret < 0) {
> -		dev_err(dev, "failed to register drm gsc device.\n");
> -		goto err_ippdrv_register;
> -	}
> +	ret = component_add(dev, &gsc_component_ops);
> +	if (ret)
> +		goto err_pm_dis;
>   
>   	dev_info(dev, "drm gsc registered successfully.\n");
>   
>   	return 0;
>   
> -err_ippdrv_register:
> +err_pm_dis:
>   	pm_runtime_disable(dev);
>   	return ret;
>   }
> @@ -1748,11 +1290,6 @@ static int gsc_probe(struct platform_device *pdev)
>   static int gsc_remove(struct platform_device *pdev)
>   {
>   	struct device *dev = &pdev->dev;
> -	struct gsc_context *ctx = get_gsc_context(dev);
> -	struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv;
> -
> -	exynos_drm_ippdrv_unregister(ippdrv);
> -	mutex_destroy(&ctx->lock);
>   
>   	pm_runtime_set_suspended(dev);
>   	pm_runtime_disable(dev);
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_gsc.h b/drivers/gpu/drm/exynos/exynos_drm_gsc.h
> deleted file mode 100644
> index 29ec1c5efcf2..000000000000
> --- a/drivers/gpu/drm/exynos/exynos_drm_gsc.h
> +++ /dev/null
> @@ -1,24 +0,0 @@
> -/*
> - * Copyright (c) 2012 Samsung Electronics Co., Ltd.
> - *
> - * Authors:
> - *	Eunchul Kim <chulspro.kim@samsung.com>
> - *	Jinyoung Jeon <jy0.jeon@samsung.com>
> - *	Sangmin Lee <lsmin.lee@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.
> - */
> -
> -#ifndef _EXYNOS_DRM_GSC_H_
> -#define _EXYNOS_DRM_GSC_H_
> -
> -/*
> - * TODO
> - * FIMD output interface notifier callback.
> - * Mixer output interface notifier callback.
> - */
> -
> -#endif /* _EXYNOS_DRM_GSC_H_ */

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 2/6] drm/exynos: ipp: Add IPP v2 framework

[Marek Szyprowski]

Hi Tobias,

Thanks for testing!

On 2017-09-15 19:18, Tobias Jakobi wrote:
> Hello Marek,
>
> Marek Szyprowski wrote:
>> This patch adds Exynos IPP v2 subsystem and userspace API.
>>
>> New userspace API is focused ONLY on memory-to-memory image processing.
>> The two remainging IPP operation modes (framebuffer writeback and
>> local-path output with image processing) can be implemented using
>> standard DRM features: writeback connectors and additional DRM planes with
>> scaling features.
>>
>> V2 IPP userspace API is not compatible with old IPP ioctls. This is a
>> significant change, but the old IPP subsystem in mainline Linux kernel was
>> partially disfunctional anyway and not used in any open-source project.
>>
>> V2 IPP userspace API is based on stateless approach, which much better fits
>> to memory-to-memory image processing model. It also provides support for
>> all image formats, which are both already defined in DRM API and supported
>> by the existing IPP hardware modules.
>>
>> The API consists of the following ioctls:
>> - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image
>>    processing modules,
>> - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image
>>    formats of given IPP module,
>> - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for
>>    selected image format of given IPP module,
>> - DRM_IOCTL_EXYNOS_IPP_COMMIT: to perform operation described by the
>>    provided structures (source and destination buffers, operation rectangle,
>>    transformation, etc).
>>
>> The proposed userspace API is extensible. In the future more advanced image
>> processing operations can be defined to support for example blending.
>>
>> Userspace API is fully functional also on DRM render nodes, so it is not
>> limited to the root/privileged client.
>>
>> Internal driver API also has been completely rewritten. New IPP core
>> performs all possible input validation, checks and object life-time
>> control. The drivers can focus only on writing configuration to hardware
>> registers. Stateless nature of DRM_IOCTL_EXYNOS_IPP_COMMIT ioctl simplifies
>> the driver API. Minimal driver needs to provide a single callback for
>> starting processing and an array with supported image formats.
>>
>> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
>> ---
>>   drivers/gpu/drm/exynos/Kconfig          |   3 +
>>   drivers/gpu/drm/exynos/Makefile         |   1 +
>>   drivers/gpu/drm/exynos/exynos_drm_drv.c |   9 +
>>   drivers/gpu/drm/exynos/exynos_drm_ipp.c | 893 ++++++++++++++++++++++++++++++++
>>   drivers/gpu/drm/exynos/exynos_drm_ipp.h | 188 +++++++
>>   include/uapi/drm/exynos_drm.h           | 227 ++++++++
>>   6 files changed, 1321 insertions(+)
>>   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 88cff0e039b6..2e097a81df12 100644
>> --- a/drivers/gpu/drm/exynos/Kconfig
>> +++ b/drivers/gpu/drm/exynos/Kconfig
>> @@ -94,6 +94,9 @@ config DRM_EXYNOS_G2D
>>   	help
>>   	  Choose this option if you want to use Exynos G2D for DRM.
>>   
>> +config DRM_EXYNOS_IPP
>> +	bool
>> +
>>   config DRM_EXYNOS_FIMC
>>   	bool "FIMC"
>>   	depends on BROKEN && MFD_SYSCON
>> diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile
>> index 09bb002c9555..f663490e949d 100644
>> --- a/drivers/gpu/drm/exynos/Makefile
>> +++ b/drivers/gpu/drm/exynos/Makefile
>> @@ -17,6 +17,7 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER)	+= exynos_mixer.o
>>   exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI)	+= exynos_hdmi.o
>>   exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI)	+= exynos_drm_vidi.o
>>   exynosdrm-$(CONFIG_DRM_EXYNOS_G2D)	+= exynos_drm_g2d.o
>> +exynosdrm-$(CONFIG_DRM_EXYNOS_IPP)	+= exynos_drm_ipp.o
>>   exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC)	+= exynos_drm_fimc.o
>>   exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR)	+= exynos_drm_rotator.o
>>   exynosdrm-$(CONFIG_DRM_EXYNOS_GSC)	+= exynos_drm_gsc.o
>> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c
>> index 7f19054ebce3..645046c5943a 100644
>> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c
>> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c
>> @@ -26,6 +26,7 @@
>>   #include "exynos_drm_fb.h"
>>   #include "exynos_drm_gem.h"
>>   #include "exynos_drm_plane.h"
>> +#include "exynos_drm_ipp.h"
>>   #include "exynos_drm_vidi.h"
>>   #include "exynos_drm_g2d.h"
>>   #include "exynos_drm_iommu.h"
>> @@ -114,6 +115,14 @@ static void exynos_drm_lastclose(struct drm_device *dev)
>>   			DRM_AUTH | DRM_RENDER_ALLOW),
>>   	DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl,
>>   			DRM_AUTH | DRM_RENDER_ALLOW),
>> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_RESOURCES, exynos_drm_ipp_get_res_ioctl,
>> +			DRM_AUTH | DRM_RENDER_ALLOW),
>> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_CAPS, exynos_drm_ipp_get_caps_ioctl,
>> +			DRM_AUTH | DRM_RENDER_ALLOW),
>> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_LIMITS, exynos_drm_ipp_get_limits_ioctl,
>> +			DRM_AUTH | DRM_RENDER_ALLOW),
>> +	DRM_IOCTL_DEF_DRV(EXYNOS_IPP_COMMIT, exynos_drm_ipp_commit_ioctl,
>> +			DRM_AUTH | DRM_RENDER_ALLOW),
>>   };
>>   
>>   static const struct file_operations exynos_drm_driver_fops = {
>> 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 000000000000..11be6a1bf839
>> --- /dev/null
>> +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.c
>> @@ -0,0 +1,892 @@
>> +/*
>> + * Copyright (C) 2017 Samsung Electronics Co.Ltd
>> + * Authors:
>> + *	Marek Szyprowski <m.szyprowski@samsung.com>
>> + *
>> + * Exynos DRM Image Post Processing (IPP) related functions
>> + *
>> + * 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 shall be included in
>> + * all copies or substantial portions of the Software.
>> + */
>> +
>> +
>> +#include <drm/drmP.h>
>> +#include <drm/drm_mode.h>
>> +#include <uapi/drm/exynos_drm.h>
>> +
>> +#include "exynos_drm_drv.h"
>> +#include "exynos_drm_gem.h"
>> +#include "exynos_drm_ipp.h"
>> +
>> +static int num_ipp;
>> +static LIST_HEAD(ipp_list);
>> +
>> +struct drm_pending_exynos_ipp_event {
>> +	struct drm_pending_event base;
>> +	struct drm_exynos_ipp_event event;
>> +};
>> +
>> +/**
>> + * exynos_drm_ipp_register - Register a new picture processor hardware module
>> + * @dev: DRM device
>> + * @ipp: ipp module to init
>> + * @funcs: callbacks for the new ipp object
>> + * @caps: bitmask of ipp capabilities (%DRM_EXYNOS_IPP_CAP_*)
>> + * @formats: array of supported formats (%DRM_FORMAT_*)
>> + * @name: name (for debugging purposes)
>> + *
>> + * Initializes a ipp module.
>> + *
>> + * Returns:
>> + * Zero on success, error code on failure.
>> + */
>> +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp,
>> +		const struct exynos_drm_ipp_funcs *funcs, unsigned int caps,
>> +		const struct exynos_drm_ipp_formats *formats, const char *name)
>> +{
>> +	WARN_ON(!ipp);
>> +	WARN_ON(!funcs);
>> +	WARN_ON(!formats);
>> +
>> +	spin_lock_init(&ipp->lock);
>> +	INIT_LIST_HEAD(&ipp->todo_list);
>> +	init_waitqueue_head(&ipp->done_wq);
>> +	ipp->dev = dev;
>> +	ipp->funcs = funcs;
>> +	ipp->capabilities = caps;
>> +	ipp->name = name;
>> +	ipp->formats = formats;
>> +	while (formats++->fourcc)
>> +		ipp->num_formats++;
>> +
>> +	list_add_tail(&ipp->head, &ipp_list);
>> +	ipp->id = num_ipp++;
>> +
>> +	DRM_DEBUG_DRIVER("Registered ipp %d\n", ipp->id);
>> +
>> +	return 0;
>> +}
>> +
>> +/**
>> + * exynos_drm_ipp_unregister - Unregister the picture processor module
>> + * @dev: DRM device
>> + * @ipp: ipp module
>> + */
>> +void exynos_drm_ipp_unregister(struct drm_device *dev,
>> +			       struct exynos_drm_ipp *ipp)
>> +{
>> +	BUG_ON(ipp->task);
>> +	BUG_ON(!list_empty(&ipp->todo_list));
>> +	list_del(&ipp->head);
>> +}
>> +
>> +/**
>> + * exynos_drm_ipp_ioctl_get_res_ioctl - enumerate all ipp modules
>> + * @dev: DRM device
>> + * @data: ioctl data
>> + * @file_priv: DRM file info
>> + *
>> + * Construct a list of ipp ids.
>> + *
>> + * Called by the user via ioctl.
>> + *
>> + * Returns:
>> + * Zero on success, negative errno on failure.
>> + */
>> +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data,
>> +				 struct drm_file *file_priv)
>> +{
>> +	struct drm_exynos_ioctl_ipp_get_res *resp = data;
>> +	struct exynos_drm_ipp *ipp;
>> +	uint32_t __user *ipp_ptr = (uint32_t __user *)
>> +						(unsigned long)resp->ipp_id_ptr;
>> +	unsigned int count = num_ipp, copied = 0;
>> +
>> +	/*
>> +	 * This ioctl is called twice, once to determine how much space is
>> +	 * needed, and the 2nd time to fill it.
>> +	 */
>> +	if (count && resp->count_ipps >= count) {
>> +		list_for_each_entry(ipp, &ipp_list, head) {
>> +			if (put_user(ipp->id, ipp_ptr + copied))
>> +				return -EFAULT;
>> +			copied++;
>> +		}
>> +	}
>> +	resp->count_ipps = count;
>> +
>> +	return 0;
>> +}
>> +
>> +static inline struct exynos_drm_ipp *__ipp_get(uint32_t id)
>> +{
>> +	struct exynos_drm_ipp *ipp;
>> +
>> +	list_for_each_entry(ipp, &ipp_list, head)
>> +		if (ipp->id == id)
>> +			return ipp;
>> +	return NULL;
>> +}
>> +
>> +/**
>> + * exynos_drm_ipp_ioctl_get_caps - get ipp module capabilities and formats
>> + * @dev: DRM device
>> + * @data: ioctl data
>> + * @file_priv: DRM file info
>> + *
>> + * Construct a structure describing ipp module capabilities.
>> + *
>> + * Called by the user via ioctl.
>> + *
>> + * Returns:
>> + * Zero on success, negative errno on failure.
>> + */
>> +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data,
>> +				  struct drm_file *file_priv)
>> +{
>> +	struct drm_exynos_ioctl_ipp_get_caps *resp = data;
>> +	void __user *ptr = (void __user *)(unsigned long)resp->formats_ptr;
>> +	struct exynos_drm_ipp *ipp;
>> +	int i;
>> +
>> +	ipp = __ipp_get(resp->ipp_id);
>> +	if (!ipp)
>> +		return -ENOENT;
>> +
>> +	resp->ipp_id = ipp->id;
>> +	resp->capabilities = ipp->capabilities;
>> +
>> +	/*
>> +	 * This ioctl is called twice, once to determine how much space is
>> +	 * needed, and the 2nd time to fill it.
>> +	 */
>> +	if (resp->formats_count >= ipp->num_formats) {
>> +		for (i = 0; i < ipp->num_formats; i++) {
>> +			struct exynos_drm_ipp_format tmp = {
>> +				.fourcc = ipp->formats[i].fourcc,
>> +				.type = ipp->formats[i].type,
>> +				.modifier = ipp->formats[i].modifier,
>> +			};
>> +
>> +			if (copy_to_user(ptr, &tmp, sizeof(tmp)))
>> +				return -EFAULT;
>> +			ptr += sizeof(tmp);
>> +		}
>> +	}
>> +	resp->formats_count = ipp->num_formats;
>> +
>> +	return 0;
>> +}
>> +
>> +static inline const struct exynos_drm_ipp_formats *__ipp_format_get(
>> +			uint32_t fourcc, uint64_t mod, unsigned long type,
>> +			const struct exynos_drm_ipp_formats *f)
>> +{
>> +	for (; f->fourcc; f++)
>> +		if ((f->type & type) && f->fourcc == fourcc &&
>> +		    f->modifier == mod)
>> +			return f;
>> +	return NULL;
>> +}
>> +
>> +/**
>> + * exynos_drm_ipp_get_limits_ioctl - get ipp module limits
>> + * @dev: DRM device
>> + * @data: ioctl data
>> + * @file_priv: DRM file info
>> + *
>> + * Construct a structure describing ipp module limitations for provided
>> + * picture format.
>> + *
>> + * Called by the user via ioctl.
>> + *
>> + * Returns:
>> + * Zero on success, negative errno on failure.
>> + */
>> +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data,
>> +				    struct drm_file *file_priv)
> I'm getting a null-pointer Opps in this ioctl(). I haven't fully debugged this,
> but I guess it's because format->limits might be NULL. E.g. the formats for the
> FIMC (exynos_drm_ipp_formats fimc_formats[]) don't carry any limits for the
> color formats.

Right, my fault. There should be a check against NULL. On the other 
hand, drivers
should provide some limits, but those were not added to FIMC and GSC 
drivers yet.

>> +{
>> +	struct drm_exynos_ioctl_ipp_get_limits *resp = data;
>> +	void __user *ptr = (void __user *)(unsigned long)resp->limits_ptr;
>> +	const struct exynos_drm_ipp_formats *format;
>> +	const struct drm_exynos_ipp_limit *limits, *l;
>> +	struct exynos_drm_ipp *ipp;
>> +	int count = 0;
>> +
>> +	if (resp->type != DRM_EXYNOS_IPP_FORMAT_SOURCE &&
>> +	    resp->type != DRM_EXYNOS_IPP_FORMAT_DESTINATION)
>> +		return -EINVAL;
>> +
>> +	ipp = __ipp_get(resp->ipp_id);
>> +	if (!ipp)
>> +		return -ENOENT;
>> +
>> +	format = __ipp_format_get(resp->fourcc, resp->modifier, resp->type,
>> +			      ipp->formats);
>> +	if (!format)
>> +		return -EINVAL;
>> +
>> +	limits = format->limits;
>> +	for (l = limits; l->type; l++)
>> +		count++;
>> +
>> +	/*
>> +	 * This ioctl is called twice, once to determine how much space is
>> +	 * needed, and the 2nd time to fill it.
>> +	 */
>> +	if (count && resp->limits_size >= count * sizeof(*l)) {
>> +		for (l = limits; l->type; l++, ptr += sizeof(*l))
>> +			if (copy_to_user((void __user *)ptr, l, sizeof(*l)))
>> +				return -EFAULT;
>> +	}
>> +	resp->limits_size = count * sizeof(*l);
> I wonder why you return the size if bytes here? How about keeping this
> consistant with the other ioctls (like exynos_drm_ipp_get_caps_ioctl) and just
> return the count (together with renaming limits_size to limits>count).

Frankly, this is a leftover from my previous approach to limit handling, 
where
limits were described by different structures. I will change this to 
limit count
to make it consistent.

 >> ...

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland