media: imx-pxp: add i.MX Pixel Pipeline driver

Add a V4L2 mem-to-mem scaler/CSC driver for the Pixel Pipeline (PXP)
version found on i.MX6ULL SoCs. A similar variant is used on i.MX7D.

Since this driver only uses the legacy pipeline, it should be reasonably
easy to extend it to work with the older PXP versions found on i.MX6UL,
i.MX6SX, i.MX6SL, i.MX28, and i.MX23.

The driver supports scaling and colorspace conversion. There is
currently no support for rotation, alpha-blending, and the LUTs.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>

4 files changed, 3448 insertions, 0 deletions

diff --git a/drivers/media/platform/Kconfig b/drivers/media/platform/Kconfig
index 94c1fe0e9787..f6275874ec9e 100644
--- a/drivers/media/platform/Kconfig
+++ b/drivers/media/platform/Kconfig
@@ -181,6 +181,15 @@ config VIDEO_CODA
 config VIDEO_IMX_VDOA
 	def_tristate VIDEO_CODA if SOC_IMX6Q || COMPILE_TEST
 
+config VIDEO_IMX_PXP
+	tristate "i.MX Pixel Pipeline (PXP)"
+	depends on VIDEO_DEV && VIDEO_V4L2 && (ARCH_MXC || COMPILE_TEST)
+	select VIDEOBUF2_DMA_CONTIG
+	select V4L2_MEM2MEM_DEV
+	help
+	  The i.MX Pixel Pipeline is a memory-to-memory engine for scaling,
+	  color space conversion, and rotation.
+
 config VIDEO_MEDIATEK_JPEG
 	tristate "Mediatek JPEG Codec driver"
 	depends on MTK_IOMMU_V1 || COMPILE_TEST
diff --git a/drivers/media/platform/Makefile b/drivers/media/platform/Makefile
index 41322ab65802..6ab6200dd9c9 100644
--- a/drivers/media/platform/Makefile
+++ b/drivers/media/platform/Makefile
@@ -25,6 +25,8 @@ obj-$(CONFIG_VIDEO_TI_CAL)		+= ti-vpe/
 obj-$(CONFIG_VIDEO_MX2_EMMAPRP)		+= mx2_emmaprp.o
 obj-$(CONFIG_VIDEO_CODA)		+= coda/
 
+obj-$(CONFIG_VIDEO_IMX_PXP)		+= imx-pxp.o
+
 obj-$(CONFIG_VIDEO_SH_VEU)		+= sh_veu.o
 
 obj-$(CONFIG_CEC_GPIO)			+= cec-gpio/
diff --git a/drivers/media/platform/imx-pxp.c b/drivers/media/platform/imx-pxp.c
new file mode 100644
index 000000000000..68ecfed7098b
--- /dev/null
+++ b/drivers/media/platform/imx-pxp.c
@@ -0,0 +1,1752 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * i.MX Pixel Pipeline (PXP) mem-to-mem scaler/CSC/rotator driver
+ *
+ * Copyright (c) 2018 Pengutronix, Philipp Zabel
+ *
+ * based on vim2m
+ *
+ * Copyright (c) 2009-2010 Samsung Electronics Co., Ltd.
+ * Pawel Osciak, <pawel@osciak.com>
+ * Marek Szyprowski, <m.szyprowski@samsung.com>
+ */
+#include <linux/clk.h>
+#include <linux/delay.h>
+#include <linux/dma-mapping.h>
+#include <linux/io.h>
+#include <linux/iopoll.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+
+#include <linux/platform_device.h>
+#include <media/v4l2-mem2mem.h>
+#include <media/v4l2-device.h>
+#include <media/v4l2-ioctl.h>
+#include <media/v4l2-ctrls.h>
+#include <media/v4l2-event.h>
+#include <media/videobuf2-dma-contig.h>
+
+#include "imx-pxp.h"
+
+static unsigned int debug;
+module_param(debug, uint, 0644);
+MODULE_PARM_DESC(debug, "activates debug info");
+
+#define MIN_W 8
+#define MIN_H 8
+#define MAX_W 4096
+#define MAX_H 4096
+#define ALIGN_W 3 /* 8x8 pixel blocks */
+#define ALIGN_H 3
+
+/* Flags that indicate a format can be used for capture/output */
+#define MEM2MEM_CAPTURE	(1 << 0)
+#define MEM2MEM_OUTPUT	(1 << 1)
+
+#define MEM2MEM_NAME		"pxp"
+
+/* Flags that indicate processing mode */
+#define MEM2MEM_HFLIP	(1 << 0)
+#define MEM2MEM_VFLIP	(1 << 1)
+
+#define dprintk(dev, fmt, arg...) \
+	v4l2_dbg(1, debug, &dev->v4l2_dev, "%s: " fmt, __func__, ## arg)
+
+struct pxp_fmt {
+	u32	fourcc;
+	int	depth;
+	/* Types the format can be used for */
+	u32	types;
+};
+
+static struct pxp_fmt formats[] = {
+	{
+		.fourcc	= V4L2_PIX_FMT_XBGR32,
+		.depth	= 32,
+		/* Both capture and output format */
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc	= V4L2_PIX_FMT_ABGR32,
+		.depth	= 32,
+		/* Capture-only format */
+		.types	= MEM2MEM_CAPTURE,
+	}, {
+		.fourcc	= V4L2_PIX_FMT_BGR24,
+		.depth	= 24,
+		.types	= MEM2MEM_CAPTURE,
+	}, {
+		.fourcc	= V4L2_PIX_FMT_RGB565,
+		.depth	= 16,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc	= V4L2_PIX_FMT_RGB555,
+		.depth	= 16,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_RGB444,
+		.depth	= 16,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_YUV32,
+		.depth	= 32,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_UYVY,
+		.depth	= 16,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_YUYV,
+		.depth	= 16,
+		/* Output-only format */
+		.types	= MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_VYUY,
+		.depth	= 16,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_YVYU,
+		.depth	= 16,
+		.types	= MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_GREY,
+		.depth	= 8,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_Y4,
+		.depth	= 4,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_NV16,
+		.depth	= 16,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_NV12,
+		.depth	= 12,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_NV21,
+		.depth	= 12,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_NV61,
+		.depth	= 16,
+		.types	= MEM2MEM_CAPTURE | MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_YUV422P,
+		.depth	= 16,
+		.types	= MEM2MEM_OUTPUT,
+	}, {
+		.fourcc = V4L2_PIX_FMT_YUV420,
+		.depth	= 12,
+		.types	= MEM2MEM_OUTPUT,
+	},
+};
+
+#define NUM_FORMATS ARRAY_SIZE(formats)
+
+/* Per-queue, driver-specific private data */
+struct pxp_q_data {
+	unsigned int		width;
+	unsigned int		height;
+	unsigned int		bytesperline;
+	unsigned int		sizeimage;
+	unsigned int		sequence;
+	struct pxp_fmt		*fmt;
+	enum v4l2_ycbcr_encoding ycbcr_enc;
+	enum v4l2_quantization	quant;
+};
+
+enum {
+	V4L2_M2M_SRC = 0,
+	V4L2_M2M_DST = 1,
+};
+
+static struct pxp_fmt *find_format(struct v4l2_format *f)
+{
+	struct pxp_fmt *fmt;
+	unsigned int k;
+
+	for (k = 0; k < NUM_FORMATS; k++) {
+		fmt = &formats[k];
+		if (fmt->fourcc == f->fmt.pix.pixelformat)
+			break;
+	}
+
+	if (k == NUM_FORMATS)
+		return NULL;
+
+	return &formats[k];
+}
+
+struct pxp_dev {
+	struct v4l2_device	v4l2_dev;
+	struct video_device	vfd;
+
+	struct clk		*clk;
+	void __iomem		*mmio;
+
+	atomic_t		num_inst;
+	struct mutex		dev_mutex;
+	spinlock_t		irqlock;
+
+	struct v4l2_m2m_dev	*m2m_dev;
+};
+
+struct pxp_ctx {
+	struct v4l2_fh		fh;
+	struct pxp_dev	*dev;
+
+	struct v4l2_ctrl_handler hdl;
+
+	/* Abort requested by m2m */
+	int			aborting;
+
+	/* Processing mode */
+	int			mode;
+	u8			alpha_component;
+
+	enum v4l2_colorspace	colorspace;
+	enum v4l2_xfer_func	xfer_func;
+
+	/* Source and destination queue data */
+	struct pxp_q_data   q_data[2];
+};
+
+static inline struct pxp_ctx *file2ctx(struct file *file)
+{
+	return container_of(file->private_data, struct pxp_ctx, fh);
+}
+
+static struct pxp_q_data *get_q_data(struct pxp_ctx *ctx,
+					 enum v4l2_buf_type type)
+{
+	if (type == V4L2_BUF_TYPE_VIDEO_OUTPUT)
+		return &ctx->q_data[V4L2_M2M_SRC];
+	else
+		return &ctx->q_data[V4L2_M2M_DST];
+}
+
+static u32 pxp_v4l2_pix_fmt_to_ps_format(u32 v4l2_pix_fmt)
+{
+	switch (v4l2_pix_fmt) {
+	case V4L2_PIX_FMT_XBGR32:  return BV_PXP_PS_CTRL_FORMAT__RGB888;
+	case V4L2_PIX_FMT_RGB555:  return BV_PXP_PS_CTRL_FORMAT__RGB555;
+	case V4L2_PIX_FMT_RGB444:  return BV_PXP_PS_CTRL_FORMAT__RGB444;
+	case V4L2_PIX_FMT_RGB565:  return BV_PXP_PS_CTRL_FORMAT__RGB565;
+	case V4L2_PIX_FMT_YUV32:   return BV_PXP_PS_CTRL_FORMAT__YUV1P444;
+	case V4L2_PIX_FMT_UYVY:    return BV_PXP_PS_CTRL_FORMAT__UYVY1P422;
+	case V4L2_PIX_FMT_YUYV:    return BM_PXP_PS_CTRL_WB_SWAP |
+					  BV_PXP_PS_CTRL_FORMAT__UYVY1P422;
+	case V4L2_PIX_FMT_VYUY:    return BV_PXP_PS_CTRL_FORMAT__VYUY1P422;
+	case V4L2_PIX_FMT_YVYU:    return BM_PXP_PS_CTRL_WB_SWAP |
+					  BV_PXP_PS_CTRL_FORMAT__VYUY1P422;
+	case V4L2_PIX_FMT_GREY:    return BV_PXP_PS_CTRL_FORMAT__Y8;
+	default:
+	case V4L2_PIX_FMT_Y4:      return BV_PXP_PS_CTRL_FORMAT__Y4;
+	case V4L2_PIX_FMT_NV16:    return BV_PXP_PS_CTRL_FORMAT__YUV2P422;
+	case V4L2_PIX_FMT_NV12:    return BV_PXP_PS_CTRL_FORMAT__YUV2P420;
+	case V4L2_PIX_FMT_NV21:    return BV_PXP_PS_CTRL_FORMAT__YVU2P420;
+	case V4L2_PIX_FMT_NV61:    return BV_PXP_PS_CTRL_FORMAT__YVU2P422;
+	case V4L2_PIX_FMT_YUV422P: return BV_PXP_PS_CTRL_FORMAT__YUV422;
+	case V4L2_PIX_FMT_YUV420:  return BV_PXP_PS_CTRL_FORMAT__YUV420;
+	}
+}
+
+static u32 pxp_v4l2_pix_fmt_to_out_format(u32 v4l2_pix_fmt)
+{
+	switch (v4l2_pix_fmt) {
+	case V4L2_PIX_FMT_XBGR32:   return BV_PXP_OUT_CTRL_FORMAT__RGB888;
+	case V4L2_PIX_FMT_ABGR32:   return BV_PXP_OUT_CTRL_FORMAT__ARGB8888;
+	case V4L2_PIX_FMT_BGR24:    return BV_PXP_OUT_CTRL_FORMAT__RGB888P;
+	/* Missing V4L2 pixel formats for ARGB1555 and ARGB4444 */
+	case V4L2_PIX_FMT_RGB555:   return BV_PXP_OUT_CTRL_FORMAT__RGB555;
+	case V4L2_PIX_FMT_RGB444:   return BV_PXP_OUT_CTRL_FORMAT__RGB444;
+	case V4L2_PIX_FMT_RGB565:   return BV_PXP_OUT_CTRL_FORMAT__RGB565;
+	case V4L2_PIX_FMT_YUV32:    return BV_PXP_OUT_CTRL_FORMAT__YUV1P444;
+	case V4L2_PIX_FMT_UYVY:     return BV_PXP_OUT_CTRL_FORMAT__UYVY1P422;
+	case V4L2_PIX_FMT_VYUY:     return BV_PXP_OUT_CTRL_FORMAT__VYUY1P422;
+	case V4L2_PIX_FMT_GREY:     return BV_PXP_OUT_CTRL_FORMAT__Y8;
+	default:
+	case V4L2_PIX_FMT_Y4:       return BV_PXP_OUT_CTRL_FORMAT__Y4;
+	case V4L2_PIX_FMT_NV16:     return BV_PXP_OUT_CTRL_FORMAT__YUV2P422;
+	case V4L2_PIX_FMT_NV12:     return BV_PXP_OUT_CTRL_FORMAT__YUV2P420;
+	case V4L2_PIX_FMT_NV61:     return BV_PXP_OUT_CTRL_FORMAT__YVU2P422;
+	case V4L2_PIX_FMT_NV21:     return BV_PXP_OUT_CTRL_FORMAT__YVU2P420;
+	}
+}
+
+static bool pxp_v4l2_pix_fmt_is_yuv(u32 v4l2_pix_fmt)
+{
+	switch (v4l2_pix_fmt) {
+	case V4L2_PIX_FMT_YUV32:
+	case V4L2_PIX_FMT_UYVY:
+	case V4L2_PIX_FMT_YUYV:
+	case V4L2_PIX_FMT_VYUY:
+	case V4L2_PIX_FMT_YVYU:
+	case V4L2_PIX_FMT_NV16:
+	case V4L2_PIX_FMT_NV12:
+	case V4L2_PIX_FMT_NV61:
+	case V4L2_PIX_FMT_NV21:
+	case V4L2_PIX_FMT_YUV420:
+	case V4L2_PIX_FMT_YUV422P:
+	case V4L2_PIX_FMT_GREY:
+	case V4L2_PIX_FMT_Y4:
+		return true;
+	default:
+		return false;
+	}
+}
+
+static void pxp_setup_csc(struct pxp_ctx *ctx)
+{
+	struct pxp_dev *dev = ctx->dev;
+	enum v4l2_ycbcr_encoding ycbcr_enc;
+	enum v4l2_quantization quantization;
+
+	if (pxp_v4l2_pix_fmt_is_yuv(ctx->q_data[V4L2_M2M_SRC].fmt->fourcc) &&
+	    !pxp_v4l2_pix_fmt_is_yuv(ctx->q_data[V4L2_M2M_DST].fmt->fourcc)) {
+		/*
+		 * CSC1 YUV/YCbCr to RGB conversion is implemented as follows:
+		 *
+		 * |R|   |C0 0  C1|   |Y  + Yoffset |
+		 * |G| = |C0 C3 C2| * |Cb + UVoffset|
+		 * |B|   |C0 C4 0 |   |Cr + UVoffset|
+		 *
+		 * Results are clamped to 0..255.
+		 *
+		 * BT.601 limited range:
+		 *
+		 * |R|   |1.1644  0.0000  1.5960|   |Y  - 16 |
+		 * |G| = |1.1644 -0.3917 -0.8129| * |Cb - 128|
+		 * |B|   |1.1644  2.0172  0.0000|   |Cr - 128|
+		 */
+		static const u32 csc1_coef_bt601_lim[3] = {
+			BM_PXP_CSC1_COEF0_YCBCR_MODE |
+			BF_PXP_CSC1_COEF0_C0(0x12a) |	/*  1.1641 (-0.03 %) */
+			BF_PXP_CSC1_COEF0_UV_OFFSET(-128) |
+			BF_PXP_CSC1_COEF0_Y_OFFSET(-16),
+			BF_PXP_CSC1_COEF1_C1(0x198) |	/*  1.5938 (-0.23 %) */
+			BF_PXP_CSC1_COEF1_C4(0x204),	/*  2.0156 (-0.16 %) */
+			BF_PXP_CSC1_COEF2_C2(0x730) |	/* -0.8125 (+0.04 %) */
+			BF_PXP_CSC1_COEF2_C3(0x79c),	/* -0.3906 (+0.11 %) */
+		};
+		/*
+		 * BT.601 full range:
+		 *
+		 * |R|   |1.0000  0.0000  1.4020|   |Y  + 0  |
+		 * |G| = |1.0000 -0.3441 -0.7141| * |Cb - 128|
+		 * |B|   |1.0000  1.7720  0.0000|   |Cr - 128|
+		 */
+		static const u32 csc1_coef_bt601_full[3] = {
+			BM_PXP_CSC1_COEF0_YCBCR_MODE |
+			BF_PXP_CSC1_COEF0_C0(0x100) |	/*  1.0000 (+0.00 %) */
+			BF_PXP_CSC1_COEF0_UV_OFFSET(-128) |
+			BF_PXP_CSC1_COEF0_Y_OFFSET(0),
+			BF_PXP_CSC1_COEF1_C1(0x166) |	/*  1.3984 (-0.36 %) */
+			BF_PXP_CSC1_COEF1_C4(0x1c5),	/*  1.7695 (-0.25 %) */
+			BF_PXP_CSC1_COEF2_C2(0x74a) |	/* -0.7109 (+0.32 %) */
+			BF_PXP_CSC1_COEF2_C3(0x7a8),	/* -0.3438 (+0.04 %) */
+		};
+		/*
+		 * Rec.709 limited range:
+		 *
+		 * |R|   |1.1644  0.0000  1.7927|   |Y  - 16 |
+		 * |G| = |1.1644 -0.2132 -0.5329| * |Cb - 128|
+		 * |B|   |1.1644  2.1124  0.0000|   |Cr - 128|
+		 */
+		static const u32 csc1_coef_rec709_lim[3] = {
+			BM_PXP_CSC1_COEF0_YCBCR_MODE |
+			BF_PXP_CSC1_COEF0_C0(0x12a) |	/*  1.1641 (-0.03 %) */
+			BF_PXP_CSC1_COEF0_UV_OFFSET(-128) |
+			BF_PXP_CSC1_COEF0_Y_OFFSET(-16),
+			BF_PXP_CSC1_COEF1_C1(0x1ca) |	/*  1.7891 (-0.37 %) */
+			BF_PXP_CSC1_COEF1_C4(0x21c),	/*  2.1094 (-0.30 %) */
+			BF_PXP_CSC1_COEF2_C2(0x778) |	/* -0.5312 (+0.16 %) */
+			BF_PXP_CSC1_COEF2_C3(0x7ca),	/* -0.2109 (+0.23 %) */
+		};
+		/*
+		 * Rec.709 full range:
+		 *
+		 * |R|   |1.0000  0.0000  1.5748|   |Y  + 0  |
+		 * |G| = |1.0000 -0.1873 -0.4681| * |Cb - 128|
+		 * |B|   |1.0000  1.8556  0.0000|   |Cr - 128|
+		 */
+		static const u32 csc1_coef_rec709_full[3] = {
+			BM_PXP_CSC1_COEF0_YCBCR_MODE |
+			BF_PXP_CSC1_COEF0_C0(0x100) |	/*  1.0000 (+0.00 %) */
+			BF_PXP_CSC1_COEF0_UV_OFFSET(-128) |
+			BF_PXP_CSC1_COEF0_Y_OFFSET(0),
+			BF_PXP_CSC1_COEF1_C1(0x193) |	/*  1.5742 (-0.06 %) */
+			BF_PXP_CSC1_COEF1_C4(0x1db),	/*  1.8555 (-0.01 %) */
+			BF_PXP_CSC1_COEF2_C2(0x789) |	/* -0.4648 (+0.33 %) */
+			BF_PXP_CSC1_COEF2_C3(0x7d1),	/* -0.1836 (+0.37 %) */
+		};
+		/*
+		 * BT.2020 limited range:
+		 *
+		 * |R|   |1.1644  0.0000  1.6787|   |Y  - 16 |
+		 * |G| = |1.1644 -0.1874 -0.6505| * |Cb - 128|
+		 * |B|   |1.1644  2.1418  0.0000|   |Cr - 128|
+		 */
+		static const u32 csc1_coef_bt2020_lim[3] = {
+			BM_PXP_CSC1_COEF0_YCBCR_MODE |
+			BF_PXP_CSC1_COEF0_C0(0x12a) |	/*  1.1641 (-0.03 %) */
+			BF_PXP_CSC1_COEF0_UV_OFFSET(-128) |
+			BF_PXP_CSC1_COEF0_Y_OFFSET(-16),
+			BF_PXP_CSC1_COEF1_C1(0x1ad) |	/*  1.6758 (-0.29 %) */
+			BF_PXP_CSC1_COEF1_C4(0x224),	/*  2.1406 (-0.11 %) */
+			BF_PXP_CSC1_COEF2_C2(0x75a) |	/* -0.6484 (+0.20 %) */
+			BF_PXP_CSC1_COEF2_C3(0x7d1),	/* -0.1836 (+0.38 %) */
+		};
+		/*
+		 * BT.2020 full range:
+		 *
+		 * |R|   |1.0000  0.0000  1.4746|   |Y  + 0  |
+		 * |G| = |1.0000 -0.1646 -0.5714| * |Cb - 128|
+		 * |B|   |1.0000  1.8814  0.0000|   |Cr - 128|
+		 */
+		static const u32 csc1_coef_bt2020_full[3] = {
+			BM_PXP_CSC1_COEF0_YCBCR_MODE |
+			BF_PXP_CSC1_COEF0_C0(0x100) |	/*  1.0000 (+0.00 %) */
+			BF_PXP_CSC1_COEF0_UV_OFFSET(-128) |
+			BF_PXP_CSC1_COEF0_Y_OFFSET(0),
+			BF_PXP_CSC1_COEF1_C1(0x179) |	/*  1.4727 (-0.19 %) */
+			BF_PXP_CSC1_COEF1_C4(0x1e1),	/*  1.8789 (-0.25 %) */
+			BF_PXP_CSC1_COEF2_C2(0x76e) |	/* -0.5703 (+0.11 %) */
+			BF_PXP_CSC1_COEF2_C3(0x7d6),	/* -0.1641 (+0.05 %) */
+		};
+		/*
+		 * SMPTE 240m limited range:
+		 *
+		 * |R|   |1.1644  0.0000  1.7937|   |Y  - 16 |
+		 * |G| = |1.1644 -0.2565 -0.5427| * |Cb - 128|
+		 * |B|   |1.1644  2.0798  0.0000|   |Cr - 128|
+		 */
+		static const u32 csc1_coef_smpte240m_lim[3] = {
+			BM_PXP_CSC1_COEF0_YCBCR_MODE |
+			BF_PXP_CSC1_COEF0_C0(0x12a) |	/*  1.1641 (-0.03 %) */
+			BF_PXP_CSC1_COEF0_UV_OFFSET(-128) |
+			BF_PXP_CSC1_COEF0_Y_OFFSET(-16),
+			BF_PXP_CSC1_COEF1_C1(0x1cb) |	/*  1.7930 (-0.07 %) */
+			BF_PXP_CSC1_COEF1_C4(0x214),	/*  2.0781 (-0.17 %) */
+			BF_PXP_CSC1_COEF2_C2(0x776) |	/* -0.5391 (+0.36 %) */
+			BF_PXP_CSC1_COEF2_C3(0x7bf),	/* -0.2539 (+0.26 %) */
+		};
+		/*
+		 * SMPTE 240m full range:
+		 *
+		 * |R|   |1.0000  0.0000  1.5756|   |Y  + 0  |
+		 * |G| = |1.0000 -0.2253 -0.4767| * |Cb - 128|
+		 * |B|   |1.0000  1.8270  0.0000|   |Cr - 128|
+		 */
+		static const u32 csc1_coef_smpte240m_full[3] = {
+			BM_PXP_CSC1_COEF0_YCBCR_MODE |
+			BF_PXP_CSC1_COEF0_C0(0x100) |	/*  1.0000 (+0.00 %) */
+			BF_PXP_CSC1_COEF0_UV_OFFSET(-128) |
+			BF_PXP_CSC1_COEF0_Y_OFFSET(0),
+			BF_PXP_CSC1_COEF1_C1(0x193) |	/*  1.5742 (-0.14 %) */
+			BF_PXP_CSC1_COEF1_C4(0x1d3),	/*  1.8242 (-0.28 %) */
+			BF_PXP_CSC1_COEF2_C2(0x786) |	/* -0.4766 (+0.01 %) */
+			BF_PXP_CSC1_COEF2_C3(0x7c7),	/* -0.2227 (+0.26 %) */
+		};
+		const u32 *csc1_coef;
+
+		ycbcr_enc = ctx->q_data[V4L2_M2M_SRC].ycbcr_enc;
+		quantization = ctx->q_data[V4L2_M2M_SRC].quant;
+
+		if (ycbcr_enc == V4L2_YCBCR_ENC_601) {
+			if (quantization == V4L2_QUANTIZATION_FULL_RANGE)
+				csc1_coef = csc1_coef_bt601_full;
+			else
+				csc1_coef = csc1_coef_bt601_lim;
+		} else if (ycbcr_enc == V4L2_YCBCR_ENC_709) {
+			if (quantization == V4L2_QUANTIZATION_FULL_RANGE)
+				csc1_coef = csc1_coef_rec709_full;
+			else
+				csc1_coef = csc1_coef_rec709_lim;
+		} else if (ycbcr_enc == V4L2_YCBCR_ENC_BT2020) {
+			if (quantization == V4L2_QUANTIZATION_FULL_RANGE)
+				csc1_coef = csc1_coef_bt2020_full;
+			else
+				csc1_coef = csc1_coef_bt2020_lim;
+		} else {
+			if (quantization == V4L2_QUANTIZATION_FULL_RANGE)
+				csc1_coef = csc1_coef_smpte240m_full;
+			else
+				csc1_coef = csc1_coef_smpte240m_lim;
+		}
+
+		writel(csc1_coef[0], dev->mmio + HW_PXP_CSC1_COEF0);
+		writel(csc1_coef[1], dev->mmio + HW_PXP_CSC1_COEF1);
+		writel(csc1_coef[2], dev->mmio + HW_PXP_CSC1_COEF2);
+	} else {
+		writel(BM_PXP_CSC1_COEF0_BYPASS, dev->mmio + HW_PXP_CSC1_COEF0);
+	}
+
+	if (!pxp_v4l2_pix_fmt_is_yuv(ctx->q_data[V4L2_M2M_SRC].fmt->fourcc) &&
+	    pxp_v4l2_pix_fmt_is_yuv(ctx->q_data[V4L2_M2M_DST].fmt->fourcc)) {
+		/*
+		 * CSC2 RGB to YUV/YCbCr conversion is implemented as follows:
+		 *
+		 * |Y |   |A1 A2 A3|   |R|   |D1|
+		 * |Cb| = |B1 B2 B3| * |G| + |D2|
+		 * |Cr|   |C1 C2 C3|   |B|   |D3|
+		 *
+		 * Results are clamped to 0..255.
+		 *
+		 * BT.601 limited range:
+		 *
+		 * |Y |   | 0.2568  0.5041  0.0979|   |R|   |16 |
+		 * |Cb| = |-0.1482 -0.2910  0.4392| * |G| + |128|
+		 * |Cr|   | 0.4392  0.4392 -0.3678|   |B|   |128|
+		 */
+		static const u32 csc2_coef_bt601_lim[6] = {
+			BF_PXP_CSC2_COEF0_A2(0x081) |	/*  0.5039 (-0.02 %) */
+			BF_PXP_CSC2_COEF0_A1(0x041),	/*  0.2539 (-0.29 %) */
+			BF_PXP_CSC2_COEF1_B1(0x7db) |	/* -0.1445 (+0.37 %) */
+			BF_PXP_CSC2_COEF1_A3(0x019),	/*  0.0977 (-0.02 %) */
+			BF_PXP_CSC2_COEF2_B3(0x070) |	/*  0.4375 (-0.17 %) */
+			BF_PXP_CSC2_COEF2_B2(0x7b6),	/* -0.2891 (+0.20 %) */
+			BF_PXP_CSC2_COEF3_C2(0x7a2) |	/* -0.3672 (+0.06 %) */
+			BF_PXP_CSC2_COEF3_C1(0x070),	/*  0.4375 (-0.17 %) */
+			BF_PXP_CSC2_COEF4_D1(16) |
+			BF_PXP_CSC2_COEF4_C3(0x7ee),	/* -0.0703 (+0.11 %) */
+			BF_PXP_CSC2_COEF5_D3(128) |
+			BF_PXP_CSC2_COEF5_D2(128),
+		};
+		/*
+		 * BT.601 full range:
+		 *
+		 * |Y |   | 0.2990  0.5870  0.1140|   |R|   |0  |
+		 * |Cb| = |-0.1687 -0.3313  0.5000| * |G| + |128|
+		 * |Cr|   | 0.5000  0.5000 -0.4187|   |B|   |128|
+		 */
+		static const u32 csc2_coef_bt601_full[6] = {
+			BF_PXP_CSC2_COEF0_A2(0x096) |	/*  0.5859 (-0.11 %) */
+			BF_PXP_CSC2_COEF0_A1(0x04c),	/*  0.2969 (-0.21 %) */
+			BF_PXP_CSC2_COEF1_B1(0x7d5) |	/* -0.1680 (+0.07 %) */
+			BF_PXP_CSC2_COEF1_A3(0x01d),	/*  0.1133 (-0.07 %) */
+			BF_PXP_CSC2_COEF2_B3(0x080) |	/*  0.5000 (+0.00 %) */
+			BF_PXP_CSC2_COEF2_B2(0x7ac),	/* -0.3281 (+0.32 %) */
+			BF_PXP_CSC2_COEF3_C2(0x795) |	/* -0.4180 (+0.07 %) */
+			BF_PXP_CSC2_COEF3_C1(0x080),	/*  0.5000 (+0.00 %) */
+			BF_PXP_CSC2_COEF4_D1(0) |
+			BF_PXP_CSC2_COEF4_C3(0x7ec),	/* -0.0781 (+0.32 %) */
+			BF_PXP_CSC2_COEF5_D3(128) |
+			BF_PXP_CSC2_COEF5_D2(128),
+		};
+		/*
+		 * Rec.709 limited range:
+		 *
+		 * |Y |   | 0.1826  0.6142  0.0620|   |R|   |16 |
+		 * |Cb| = |-0.1007 -0.3385  0.4392| * |G| + |128|
+		 * |Cr|   | 0.4392  0.4392 -0.3990|   |B|   |128|
+		 */
+		static const u32 csc2_coef_rec709_lim[6] = {
+			BF_PXP_CSC2_COEF0_A2(0x09d) |	/*  0.6133 (-0.09 %) */
+			BF_PXP_CSC2_COEF0_A1(0x02e),	/*  0.1797 (-0.29 %) */
+			BF_PXP_CSC2_COEF1_B1(0x7e7) |	/* -0.0977 (+0.30 %) */
+			BF_PXP_CSC2_COEF1_A3(0x00f),	/*  0.0586 (-0.34 %) */
+			BF_PXP_CSC2_COEF2_B3(0x070) |	/*  0.4375 (-0.17 %) */
+			BF_PXP_CSC2_COEF2_B2(0x7aa),	/* -0.3359 (+0.26 %) */
+			BF_PXP_CSC2_COEF3_C2(0x79a) |	/* -0.3984 (+0.05 %) */
+			BF_PXP_CSC2_COEF3_C1(0x070),	/*  0.4375 (-0.17 %) */
+			BF_PXP_CSC2_COEF4_D1(16) |
+			BF_PXP_CSC2_COEF4_C3(0x7f6),	/* -0.0391 (+0.12 %) */
+			BF_PXP_CSC2_COEF5_D3(128) |
+			BF_PXP_CSC2_COEF5_D2(128),
+		};
+		/*
+		 * Rec.709 full range:
+		 *
+		 * |Y |   | 0.2126  0.7152  0.0722|   |R|   |0  |
+		 * |Cb| = |-0.1146 -0.3854  0.5000| * |G| + |128|
+		 * |Cr|   | 0.5000  0.5000 -0.4542|   |B|   |128|
+		 */
+		static const u32 csc2_coef_rec709_full[6] = {
+			BF_PXP_CSC2_COEF0_A2(0x0b7) |	/*  0.7148 (-0.04 %) */
+			BF_PXP_CSC2_COEF0_A1(0x036),	/*  0.2109 (-0.17 %) */
+			BF_PXP_CSC2_COEF1_B1(0x7e3) |	/* -0.1133 (+0.13 %) */
+			BF_PXP_CSC2_COEF1_A3(0x012),	/*  0.0703 (-0.19 %) */
+			BF_PXP_CSC2_COEF2_B3(0x080) |	/*  0.5000 (+0.00 %) */
+			BF_PXP_CSC2_COEF2_B2(0x79e),	/* -0.3828 (+0.26 %) */
+			BF_PXP_CSC2_COEF3_C2(0x78c) |	/* -0.4531 (+0.11 %) */
+			BF_PXP_CSC2_COEF3_C1(0x080),	/*  0.5000 (+0.00 %) */
+			BF_PXP_CSC2_COEF4_D1(0) |
+			BF_PXP_CSC2_COEF4_C3(0x7f5),	/* -0.0430 (+0.28 %) */
+			BF_PXP_CSC2_COEF5_D3(128) |
+			BF_PXP_CSC2_COEF5_D2(128),
+		};
+		/*
+		 * BT.2020 limited range:
+		 *
+		 * |Y |   | 0.2256  0.5823  0.0509|   |R|   |16 |
+		 * |Cb| = |-0.1226 -0.3166  0.4392| * |G| + |128|
+		 * |Cr|   | 0.4392  0.4392 -0.4039|   |B|   |128|
+		 */
+		static const u32 csc2_coef_bt2020_lim[6] = {
+			BF_PXP_CSC2_COEF0_A2(0x095) |	/*  0.5820 (-0.03 %) */
+			BF_PXP_CSC2_COEF0_A1(0x039),	/*  0.2227 (-0.30 %) */
+			BF_PXP_CSC2_COEF1_B1(0x7e1) |	/* -0.1211 (+0.15 %) */
+			BF_PXP_CSC2_COEF1_A3(0x00d),	/*  0.0508 (-0.01 %) */
+			BF_PXP_CSC2_COEF2_B3(0x070) |	/*  0.4375 (-0.17 %) */
+			BF_PXP_CSC2_COEF2_B2(0x7af),	/* -0.3164 (+0.02 %) */
+			BF_PXP_CSC2_COEF3_C2(0x799) |	/* -0.4023 (+0.16 %) */
+			BF_PXP_CSC2_COEF3_C1(0x070),	/*  0.4375 (-0.17 %) */
+			BF_PXP_CSC2_COEF4_D1(16) |
+			BF_PXP_CSC2_COEF4_C3(0x7f7),	/* -0.0352 (+0.02 %) */
+			BF_PXP_CSC2_COEF5_D3(128) |
+			BF_PXP_CSC2_COEF5_D2(128),
+		};
+		/*
+		 * BT.2020 full range:
+		 *
+		 * |Y |   | 0.2627  0.6780  0.0593|   |R|   |0  |
+		 * |Cb| = |-0.1396 -0.3604  0.5000| * |G| + |128|
+		 * |Cr|   | 0.5000  0.5000 -0.4598|   |B|   |128|
+		 */
+		static const u32 csc2_coef_bt2020_full[6] = {
+			BF_PXP_CSC2_COEF0_A2(0x0ad) |	/*  0.6758 (-0.22 %) */
+			BF_PXP_CSC2_COEF0_A1(0x043),	/*  0.2617 (-0.10 %) */
+			BF_PXP_CSC2_COEF1_B1(0x7dd) |	/* -0.1367 (+0.29 %) */
+			BF_PXP_CSC2_COEF1_A3(0x00f),	/*  0.0586 (-0.07 %) */
+			BF_PXP_CSC2_COEF2_B3(0x080) |	/*  0.5000 (+0.00 %) */
+			BF_PXP_CSC2_COEF2_B2(0x7a4),	/* -0.3594 (+0.10 %) */
+			BF_PXP_CSC2_COEF3_C2(0x78b) |	/* -0.4570 (+0.28 %) */
+			BF_PXP_CSC2_COEF3_C1(0x080),	/*  0.5000 (+0.00 %) */
+			BF_PXP_CSC2_COEF4_D1(0) |
+			BF_PXP_CSC2_COEF4_C3(0x7f6),	/* -0.0391 (+0.11 %) */
+			BF_PXP_CSC2_COEF5_D3(128) |
+			BF_PXP_CSC2_COEF5_D2(128),
+		};
+		/*
+		 * SMPTE 240m limited range:
+		 *
+		 * |Y |   | 0.1821  0.6020  0.0747|   |R|   |16 |
+		 * |Cb| = |-0.1019 -0.3373  0.4392| * |G| + |128|
+		 * |Cr|   | 0.4392  0.4392 -0.3909|   |B|   |128|
+		 */
+		static const u32 csc2_coef_smpte240m_lim[6] = {
+			BF_PXP_CSC2_COEF0_A2(0x09a) |	/*  0.6016 (-0.05 %) */
+			BF_PXP_CSC2_COEF0_A1(0x02e),	/*  0.1797 (-0.24 %) */
+			BF_PXP_CSC2_COEF1_B1(0x7e6) |	/* -0.1016 (+0.03 %) */
+			BF_PXP_CSC2_COEF1_A3(0x013),	/*  0.0742 (-0.05 %) */
+			BF_PXP_CSC2_COEF2_B3(0x070) |	/*  0.4375 (-0.17 %) */
+			BF_PXP_CSC2_COEF2_B2(0x7aa),	/* -0.3359 (+0.14 %) */
+			BF_PXP_CSC2_COEF3_C2(0x79c) |	/* -0.3906 (+0.03 %) */
+			BF_PXP_CSC2_COEF3_C1(0x070),	/*  0.4375 (-0.17 %) */
+			BF_PXP_CSC2_COEF4_D1(16) |
+			BF_PXP_CSC2_COEF4_C3(0x7f4),	/* -0.0469 (+0.14 %) */
+			BF_PXP_CSC2_COEF5_D3(128) |
+			BF_PXP_CSC2_COEF5_D2(128),
+		};
+		/*
+		 * SMPTE 240m full range:
+		 *
+		 * |Y |   | 0.2120  0.7010  0.0870|   |R|   |0  |
+		 * |Cb| = |-0.1160 -0.3840  0.5000| * |G| + |128|
+		 * |Cr|   | 0.5000  0.5000 -0.4450|   |B|   |128|
+		 */
+		static const u32 csc2_coef_smpte240m_full[6] = {
+			BF_PXP_CSC2_COEF0_A2(0x0b3) |	/*  0.6992 (-0.18 %) */
+			BF_PXP_CSC2_COEF0_A1(0x036),	/*  0.2109 (-0.11 %) */
+			BF_PXP_CSC2_COEF1_B1(0x7e3) |	/* -0.1133 (+0.27 %) */
+			BF_PXP_CSC2_COEF1_A3(0x016),	/*  0.0859 (-0.11 %) */
+			BF_PXP_CSC2_COEF2_B3(0x080) |	/*  0.5000 (+0.00 %) */
+			BF_PXP_CSC2_COEF2_B2(0x79e),	/* -0.3828 (+0.12 %) */
+			BF_PXP_CSC2_COEF3_C2(0x78f) |	/* -0.4414 (+0.36 %) */
+			BF_PXP_CSC2_COEF3_C1(0x080),	/*  0.5000 (+0.00 %) */
+			BF_PXP_CSC2_COEF4_D1(0) |
+			BF_PXP_CSC2_COEF4_C3(0x7f2),	/* -0.0547 (+0.03 %) */
+			BF_PXP_CSC2_COEF5_D3(128) |
+			BF_PXP_CSC2_COEF5_D2(128),
+		};
+		const u32 *csc2_coef;
+		u32 csc2_ctrl;
+
+		ycbcr_enc = ctx->q_data[V4L2_M2M_DST].ycbcr_enc;
+		quantization = ctx->q_data[V4L2_M2M_DST].quant;
+
+		if (ycbcr_enc == V4L2_YCBCR_ENC_601) {
+			if (quantization == V4L2_QUANTIZATION_FULL_RANGE)
+				csc2_coef = csc2_coef_bt601_full;
+			else
+				csc2_coef = csc2_coef_bt601_lim;
+		} else if (ycbcr_enc == V4L2_YCBCR_ENC_709) {
+			if (quantization == V4L2_QUANTIZATION_FULL_RANGE)
+				csc2_coef = csc2_coef_rec709_full;
+			else
+				csc2_coef = csc2_coef_rec709_lim;
+		} else if (ycbcr_enc == V4L2_YCBCR_ENC_709) {
+			if (quantization == V4L2_QUANTIZATION_FULL_RANGE)
+				csc2_coef = csc2_coef_bt2020_full;
+			else
+				csc2_coef = csc2_coef_bt2020_lim;
+		} else {
+			if (quantization == V4L2_QUANTIZATION_FULL_RANGE)
+				csc2_coef = csc2_coef_smpte240m_full;
+			else
+				csc2_coef = csc2_coef_smpte240m_lim;
+		}
+		if (quantization == V4L2_QUANTIZATION_FULL_RANGE) {
+			csc2_ctrl = BV_PXP_CSC2_CTRL_CSC_MODE__RGB2YUV <<
+				    BP_PXP_CSC2_CTRL_CSC_MODE;
+		} else {
+			csc2_ctrl = BV_PXP_CSC2_CTRL_CSC_MODE__RGB2YCbCr <<
+				    BP_PXP_CSC2_CTRL_CSC_MODE;
+		}
+
+		writel(csc2_ctrl, dev->mmio + HW_PXP_CSC2_CTRL);
+		writel(csc2_coef[0], dev->mmio + HW_PXP_CSC2_COEF0);
+		writel(csc2_coef[1], dev->mmio + HW_PXP_CSC2_COEF1);
+		writel(csc2_coef[2], dev->mmio + HW_PXP_CSC2_COEF2);
+		writel(csc2_coef[3], dev->mmio + HW_PXP_CSC2_COEF3);
+		writel(csc2_coef[4], dev->mmio + HW_PXP_CSC2_COEF4);
+		writel(csc2_coef[5], dev->mmio + HW_PXP_CSC2_COEF5);
+	} else {
+		writel(BM_PXP_CSC2_CTRL_BYPASS, dev->mmio + HW_PXP_CSC2_CTRL);
+	}
+}
+
+static int pxp_start(struct pxp_ctx *ctx, struct vb2_v4l2_buffer *in_vb,
+		     struct vb2_v4l2_buffer *out_vb)
+{
+	struct pxp_dev *dev = ctx->dev;
+	struct pxp_q_data *q_data;
+	u32 src_width, src_height, src_stride, src_fourcc;
+	u32 dst_width, dst_height, dst_stride, dst_fourcc;
+	dma_addr_t p_in, p_out;
+	u32 ctrl, out_ctrl, out_buf, out_buf2, out_pitch, out_lrc, out_ps_ulc;
+	u32 out_ps_lrc;
+	u32 ps_ctrl, ps_buf, ps_ubuf, ps_vbuf, ps_pitch, ps_scale, ps_offset;
+	u32 as_ulc, as_lrc;
+	u32 y_size;
+	u32 decx, decy, xscale, yscale;
+
+	q_data = get_q_data(ctx, V4L2_BUF_TYPE_VIDEO_OUTPUT);
+
+	src_width = ctx->q_data[V4L2_M2M_SRC].width;
+	dst_width = ctx->q_data[V4L2_M2M_DST].width;
+	src_height = ctx->q_data[V4L2_M2M_SRC].height;
+	dst_height = ctx->q_data[V4L2_M2M_DST].height;
+	src_stride = ctx->q_data[V4L2_M2M_SRC].bytesperline;
+	dst_stride = ctx->q_data[V4L2_M2M_DST].bytesperline;
+	src_fourcc = ctx->q_data[V4L2_M2M_SRC].fmt->fourcc;
+	dst_fourcc = ctx->q_data[V4L2_M2M_DST].fmt->fourcc;
+
+	p_in = vb2_dma_contig_plane_dma_addr(&in_vb->vb2_buf, 0);
+	p_out = vb2_dma_contig_plane_dma_addr(&out_vb->vb2_buf, 0);
+
+	if (!p_in || !p_out) {
+		v4l2_err(&dev->v4l2_dev,
+			 "Acquiring DMA addresses of buffers failed\n");
+		return -EFAULT;
+	}
+
+	out_vb->sequence =
+		get_q_data(ctx, V4L2_BUF_TYPE_VIDEO_CAPTURE)->sequence++;
+	in_vb->sequence = q_data->sequence++;
+	out_vb->vb2_buf.timestamp = in_vb->vb2_buf.timestamp;
+
+	if (in_vb->flags & V4L2_BUF_FLAG_TIMECODE)
+		out_vb->timecode = in_vb->timecode;
+	out_vb->field = in_vb->field;
+	out_vb->flags = in_vb->flags &
+		(V4L2_BUF_FLAG_TIMECODE |
+		 V4L2_BUF_FLAG_KEYFRAME |
+		 V4L2_BUF_FLAG_PFRAME |
+		 V4L2_BUF_FLAG_BFRAME |
+		 V4L2_BUF_FLAG_TSTAMP_SRC_MASK);
+
+	/* Rotation disabled, 8x8 block size */
+	ctrl = BF_PXP_CTRL_VFLIP0(!!(ctx->mode & MEM2MEM_VFLIP)) |
+	       BF_PXP_CTRL_HFLIP0(!!(ctx->mode & MEM2MEM_HFLIP));
+	/* Always write alpha value as V4L2_CID_ALPHA_COMPONENT */
+	out_ctrl = BF_PXP_OUT_CTRL_ALPHA(ctx->alpha_component) |
+		   BF_PXP_OUT_CTRL_ALPHA_OUTPUT(1) |
+		   pxp_v4l2_pix_fmt_to_out_format(dst_fourcc);
+	out_buf = p_out;
+	switch (dst_fourcc) {
+	case V4L2_PIX_FMT_NV12:
+	case V4L2_PIX_FMT_NV21:
+	case V4L2_PIX_FMT_NV16:
+	case V4L2_PIX_FMT_NV61:
+		out_buf2 = out_buf + dst_stride * dst_height;
+		break;
+	default:
+		out_buf2 = 0;
+	}
+
+	out_pitch = BF_PXP_OUT_PITCH_PITCH(dst_stride);
+	out_lrc = BF_PXP_OUT_LRC_X(dst_width - 1) |
+		  BF_PXP_OUT_LRC_Y(dst_height - 1);
+	/* PS covers whole output */
+	out_ps_ulc = BF_PXP_OUT_PS_ULC_X(0) | BF_PXP_OUT_PS_ULC_Y(0);
+	out_ps_lrc = BF_PXP_OUT_PS_LRC_X(dst_width - 1) |
+		     BF_PXP_OUT_PS_LRC_Y(dst_height - 1);
+	/* no AS */
+	as_ulc = BF_PXP_OUT_AS_ULC_X(1) | BF_PXP_OUT_AS_ULC_Y(1);
+	as_lrc = BF_PXP_OUT_AS_LRC_X(0) | BF_PXP_OUT_AS_LRC_Y(0);
+
+	decx = (src_width <= dst_width) ? 0 : ilog2(src_width / dst_width);
+	decy = (src_height <= dst_height) ? 0 : ilog2(src_height / dst_height);
+	ps_ctrl = BF_PXP_PS_CTRL_DECX(decx) | BF_PXP_PS_CTRL_DECY(decy) |
+		  pxp_v4l2_pix_fmt_to_ps_format(src_fourcc);
+	ps_buf = p_in;
+	y_size = src_stride * src_height;
+	switch (src_fourcc) {
+	case V4L2_PIX_FMT_YUV420:
+		ps_ubuf = ps_buf + y_size;
+		ps_vbuf = ps_ubuf + y_size / 4;
+		break;
+	case V4L2_PIX_FMT_YUV422P:
+		ps_ubuf = ps_buf + y_size;
+		ps_vbuf = ps_ubuf + y_size / 2;
+		break;
+	case V4L2_PIX_FMT_NV12:
+	case V4L2_PIX_FMT_NV21:
+	case V4L2_PIX_FMT_NV16:
+	case V4L2_PIX_FMT_NV61:
+		ps_ubuf = ps_buf + y_size;
+		ps_vbuf = 0;
+		break;
+	case V4L2_PIX_FMT_GREY:
+	case V4L2_PIX_FMT_Y4:
+		ps_ubuf = 0;
+		/* In grayscale mode, ps_vbuf contents are reused as CbCr */
+		ps_vbuf = 0x8080;
+		break;
+	default:
+		ps_ubuf = 0;
+		ps_vbuf = 0;
+		break;
+	}
+	ps_pitch = BF_PXP_PS_PITCH_PITCH(src_stride);
+	if (decx) {
+		xscale = (src_width >> decx) * 0x1000 / dst_width;
+	} else {
+		switch (src_fourcc) {
+		case V4L2_PIX_FMT_UYVY:
+		case V4L2_PIX_FMT_YUYV:
+		case V4L2_PIX_FMT_VYUY:
+		case V4L2_PIX_FMT_YVYU:
+		case V4L2_PIX_FMT_NV16:
+		case V4L2_PIX_FMT_NV12:
+		case V4L2_PIX_FMT_NV21:
+		case V4L2_PIX_FMT_NV61:
+		case V4L2_PIX_FMT_YUV422P:
+		case V4L2_PIX_FMT_YUV420:
+			/*
+			 * This avoids sampling past the right edge for
+			 * horizontally chroma subsampled formats.
+			 */
+			xscale = (src_width - 2) * 0x1000 / (dst_width - 1);
+			break;
+		default:
+			xscale = (src_width - 1) * 0x1000 / (dst_width - 1);
+			break;
+		}
+	}
+	if (decy)
+		yscale = (src_height >> decy) * 0x1000 / dst_height;
+	else
+		yscale = (src_height - 1) * 0x1000 / (dst_height - 1);
+	ps_scale = BF_PXP_PS_SCALE_YSCALE(yscale) |
+		   BF_PXP_PS_SCALE_XSCALE(xscale);
+	ps_offset = BF_PXP_PS_OFFSET_YOFFSET(0) | BF_PXP_PS_OFFSET_XOFFSET(0);
+
+	writel(ctrl, dev->mmio + HW_PXP_CTRL);
+	/* skip STAT */
+	writel(out_ctrl, dev->mmio + HW_PXP_OUT_CTRL);
+	writel(out_buf, dev->mmio + HW_PXP_OUT_BUF);
+	writel(out_buf2, dev->mmio + HW_PXP_OUT_BUF2);
+	writel(out_pitch, dev->mmio + HW_PXP_OUT_PITCH);
+	writel(out_lrc, dev->mmio + HW_PXP_OUT_LRC);
+	writel(out_ps_ulc, dev->mmio + HW_PXP_OUT_PS_ULC);
+	writel(out_ps_lrc, dev->mmio + HW_PXP_OUT_PS_LRC);
+	writel(as_ulc, dev->mmio + HW_PXP_OUT_AS_ULC);
+	writel(as_lrc, dev->mmio + HW_PXP_OUT_AS_LRC);
+	writel(ps_ctrl, dev->mmio + HW_PXP_PS_CTRL);
+	writel(ps_buf, dev->mmio + HW_PXP_PS_BUF);
+	writel(ps_ubuf, dev->mmio + HW_PXP_PS_UBUF);
+	writel(ps_vbuf, dev->mmio + HW_PXP_PS_VBUF);
+	writel(ps_pitch, dev->mmio + HW_PXP_PS_PITCH);
+	writel(0x00ffffff, dev->mmio + HW_PXP_PS_BACKGROUND_0);
+	writel(ps_scale, dev->mmio + HW_PXP_PS_SCALE);
+	writel(ps_offset, dev->mmio + HW_PXP_PS_OFFSET);
+	/* disable processed surface color keying */
+	writel(0x00ffffff, dev->mmio + HW_PXP_PS_CLRKEYLOW_0);
+	writel(0x00000000, dev->mmio + HW_PXP_PS_CLRKEYHIGH_0);
+
+	/* disable alpha surface color keying */
+	writel(0x00ffffff, dev->mmio + HW_PXP_AS_CLRKEYLOW_0);
+	writel(0x00000000, dev->mmio + HW_PXP_AS_CLRKEYHIGH_0);
+
+	/* setup CSC */
+	pxp_setup_csc(ctx);
+
+	/* bypass LUT */
+	writel(BM_PXP_LUT_CTRL_BYPASS, dev->mmio + HW_PXP_LUT_CTRL);
+
+	writel(BF_PXP_DATA_PATH_CTRL0_MUX15_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX14_SEL(1)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX13_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX12_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX11_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX10_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX9_SEL(1)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX8_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX7_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX6_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX5_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX4_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX3_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX2_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX1_SEL(0)|
+	       BF_PXP_DATA_PATH_CTRL0_MUX0_SEL(0),
+	       dev->mmio + HW_PXP_DATA_PATH_CTRL0);
+	writel(BF_PXP_DATA_PATH_CTRL1_MUX17_SEL(1) |
+	       BF_PXP_DATA_PATH_CTRL1_MUX16_SEL(1),
+	       dev->mmio + HW_PXP_DATA_PATH_CTRL1);
+
+	writel(0xffff, dev->mmio + HW_PXP_IRQ_MASK);
+
+	/* ungate, enable PS/AS/OUT and PXP operation */
+	writel(BM_PXP_CTRL_IRQ_ENABLE, dev->mmio + HW_PXP_CTRL_SET);
+	writel(BM_PXP_CTRL_ENABLE | BM_PXP_CTRL_ENABLE_CSC2 |
+	       BM_PXP_CTRL_ENABLE_LUT | BM_PXP_CTRL_ENABLE_ROTATE0 |
+	       BM_PXP_CTRL_ENABLE_PS_AS_OUT, dev->mmio + HW_PXP_CTRL_SET);
+
+	return 0;
+}
+
+static void pxp_job_finish(struct pxp_dev *dev)
+{
+	struct pxp_ctx *curr_ctx;
+	struct vb2_v4l2_buffer *src_vb, *dst_vb;
+	unsigned long flags;
+
+	curr_ctx = v4l2_m2m_get_curr_priv(dev->m2m_dev);
+
+	if (curr_ctx == NULL) {
+		pr_err("Instance released before the end of transaction\n");
+		return;
+	}
+
+	src_vb = v4l2_m2m_src_buf_remove(curr_ctx->fh.m2m_ctx);
+	dst_vb = v4l2_m2m_dst_buf_remove(curr_ctx->fh.m2m_ctx);
+
+	spin_lock_irqsave(&dev->irqlock, flags);