[media] rename drivers/media/video as .../platform

The remaining drivers are mostly platform drivers. Name the
dir to reflect it.

It makes sense to latter break it into a few other dirs.

Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

1 files changed, 1778 insertions, 0 deletions

diff --git a/drivers/media/platform/omap3isp/ispresizer.c b/drivers/media/platform/omap3isp/ispresizer.c
new file mode 100644
index 000000000000..d11fb261d530
--- /dev/null
+++ b/drivers/media/platform/omap3isp/ispresizer.c
@@ -0,0 +1,1778 @@
+/*
+ * ispresizer.c
+ *
+ * TI OMAP3 ISP - Resizer module
+ *
+ * Copyright (C) 2010 Nokia Corporation
+ * Copyright (C) 2009 Texas Instruments, Inc
+ *
+ * Contacts: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
+ *	     Sakari Ailus <sakari.ailus@iki.fi>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA
+ * 02110-1301 USA
+ */
+
+#include <linux/device.h>
+#include <linux/mm.h>
+#include <linux/module.h>
+
+#include "isp.h"
+#include "ispreg.h"
+#include "ispresizer.h"
+
+/*
+ * Resizer Constants
+ */
+#define MIN_RESIZE_VALUE		64
+#define MID_RESIZE_VALUE		512
+#define MAX_RESIZE_VALUE		1024
+
+#define MIN_IN_WIDTH			32
+#define MIN_IN_HEIGHT			32
+#define MAX_IN_WIDTH_MEMORY_MODE	4095
+#define MAX_IN_WIDTH_ONTHEFLY_MODE_ES1	1280
+#define MAX_IN_WIDTH_ONTHEFLY_MODE_ES2	4095
+#define MAX_IN_HEIGHT			4095
+
+#define MIN_OUT_WIDTH			16
+#define MIN_OUT_HEIGHT			2
+#define MAX_OUT_HEIGHT			4095
+
+/*
+ * Resizer Use Constraints
+ * "TRM ES3.1, table 12-46"
+ */
+#define MAX_4TAP_OUT_WIDTH_ES1		1280
+#define MAX_7TAP_OUT_WIDTH_ES1		640
+#define MAX_4TAP_OUT_WIDTH_ES2		3312
+#define MAX_7TAP_OUT_WIDTH_ES2		1650
+#define MAX_4TAP_OUT_WIDTH_3630		4096
+#define MAX_7TAP_OUT_WIDTH_3630		2048
+
+/*
+ * Constants for ratio calculation
+ */
+#define RESIZE_DIVISOR			256
+#define DEFAULT_PHASE			1
+
+/*
+ * Default (and only) configuration of filter coefficients.
+ * 7-tap mode is for scale factors 0.25x to 0.5x.
+ * 4-tap mode is for scale factors 0.5x to 4.0x.
+ * There shouldn't be any reason to recalculate these, EVER.
+ */
+static const struct isprsz_coef filter_coefs = {
+	/* For 8-phase 4-tap horizontal filter: */
+	{
+		0x0000, 0x0100, 0x0000, 0x0000,
+		0x03FA, 0x00F6, 0x0010, 0x0000,
+		0x03F9, 0x00DB, 0x002C, 0x0000,
+		0x03FB, 0x00B3, 0x0053, 0x03FF,
+		0x03FD, 0x0082, 0x0084, 0x03FD,
+		0x03FF, 0x0053, 0x00B3, 0x03FB,
+		0x0000, 0x002C, 0x00DB, 0x03F9,
+		0x0000, 0x0010, 0x00F6, 0x03FA
+	},
+	/* For 8-phase 4-tap vertical filter: */
+	{
+		0x0000, 0x0100, 0x0000, 0x0000,
+		0x03FA, 0x00F6, 0x0010, 0x0000,
+		0x03F9, 0x00DB, 0x002C, 0x0000,
+		0x03FB, 0x00B3, 0x0053, 0x03FF,
+		0x03FD, 0x0082, 0x0084, 0x03FD,
+		0x03FF, 0x0053, 0x00B3, 0x03FB,
+		0x0000, 0x002C, 0x00DB, 0x03F9,
+		0x0000, 0x0010, 0x00F6, 0x03FA
+	},
+	/* For 4-phase 7-tap horizontal filter: */
+	#define DUMMY 0
+	{
+		0x0004, 0x0023, 0x005A, 0x0058, 0x0023, 0x0004, 0x0000, DUMMY,
+		0x0002, 0x0018, 0x004d, 0x0060, 0x0031, 0x0008, 0x0000, DUMMY,
+		0x0001, 0x000f, 0x003f, 0x0062, 0x003f, 0x000f, 0x0001, DUMMY,
+		0x0000, 0x0008, 0x0031, 0x0060, 0x004d, 0x0018, 0x0002, DUMMY
+	},
+	/* For 4-phase 7-tap vertical filter: */
+	{
+		0x0004, 0x0023, 0x005A, 0x0058, 0x0023, 0x0004, 0x0000, DUMMY,
+		0x0002, 0x0018, 0x004d, 0x0060, 0x0031, 0x0008, 0x0000, DUMMY,
+		0x0001, 0x000f, 0x003f, 0x0062, 0x003f, 0x000f, 0x0001, DUMMY,
+		0x0000, 0x0008, 0x0031, 0x0060, 0x004d, 0x0018, 0x0002, DUMMY
+	}
+	/*
+	 * The dummy padding is required in 7-tap mode because of how the
+	 * registers are arranged physically.
+	 */
+	#undef DUMMY
+};
+
+/*
+ * __resizer_get_format - helper function for getting resizer format
+ * @res   : pointer to resizer private structure
+ * @pad   : pad number
+ * @fh    : V4L2 subdev file handle
+ * @which : wanted subdev format
+ * return zero
+ */
+static struct v4l2_mbus_framefmt *
+__resizer_get_format(struct isp_res_device *res, struct v4l2_subdev_fh *fh,
+		     unsigned int pad, enum v4l2_subdev_format_whence which)
+{
+	if (which == V4L2_SUBDEV_FORMAT_TRY)
+		return v4l2_subdev_get_try_format(fh, pad);
+	else
+		return &res->formats[pad];
+}
+
+/*
+ * __resizer_get_crop - helper function for getting resizer crop rectangle
+ * @res   : pointer to resizer private structure
+ * @fh    : V4L2 subdev file handle
+ * @which : wanted subdev crop rectangle
+ */
+static struct v4l2_rect *
+__resizer_get_crop(struct isp_res_device *res, struct v4l2_subdev_fh *fh,
+		   enum v4l2_subdev_format_whence which)
+{
+	if (which == V4L2_SUBDEV_FORMAT_TRY)
+		return v4l2_subdev_get_try_crop(fh, RESZ_PAD_SINK);
+	else
+		return &res->crop.request;
+}
+
+/*
+ * resizer_set_filters - Set resizer filters
+ * @res: Device context.
+ * @h_coeff: horizontal coefficient
+ * @v_coeff: vertical coefficient
+ * Return none
+ */
+static void resizer_set_filters(struct isp_res_device *res, const u16 *h_coeff,
+				const u16 *v_coeff)
+{
+	struct isp_device *isp = to_isp_device(res);
+	u32 startaddr_h, startaddr_v, tmp_h, tmp_v;
+	int i;
+
+	startaddr_h = ISPRSZ_HFILT10;
+	startaddr_v = ISPRSZ_VFILT10;
+
+	for (i = 0; i < COEFF_CNT; i += 2) {
+		tmp_h = h_coeff[i] |
+			(h_coeff[i + 1] << ISPRSZ_HFILT_COEF1_SHIFT);
+		tmp_v = v_coeff[i] |
+			(v_coeff[i + 1] << ISPRSZ_VFILT_COEF1_SHIFT);
+		isp_reg_writel(isp, tmp_h, OMAP3_ISP_IOMEM_RESZ, startaddr_h);
+		isp_reg_writel(isp, tmp_v, OMAP3_ISP_IOMEM_RESZ, startaddr_v);
+		startaddr_h += 4;
+		startaddr_v += 4;
+	}
+}
+
+/*
+ * resizer_set_bilinear - Chrominance horizontal algorithm select
+ * @res: Device context.
+ * @type: Filtering interpolation type.
+ *
+ * Filtering that is same as luminance processing is
+ * intended only for downsampling, and bilinear interpolation
+ * is intended only for upsampling.
+ */
+static void resizer_set_bilinear(struct isp_res_device *res,
+				 enum resizer_chroma_algo type)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	if (type == RSZ_BILINEAR)
+		isp_reg_set(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT,
+			    ISPRSZ_CNT_CBILIN);
+	else
+		isp_reg_clr(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT,
+			    ISPRSZ_CNT_CBILIN);
+}
+
+/*
+ * resizer_set_ycpos - Luminance and chrominance order
+ * @res: Device context.
+ * @order: order type.
+ */
+static void resizer_set_ycpos(struct isp_res_device *res,
+			      enum v4l2_mbus_pixelcode pixelcode)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	switch (pixelcode) {
+	case V4L2_MBUS_FMT_YUYV8_1X16:
+		isp_reg_set(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT,
+			    ISPRSZ_CNT_YCPOS);
+		break;
+	case V4L2_MBUS_FMT_UYVY8_1X16:
+		isp_reg_clr(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT,
+			    ISPRSZ_CNT_YCPOS);
+		break;
+	default:
+		return;
+	}
+}
+
+/*
+ * resizer_set_phase - Setup horizontal and vertical starting phase
+ * @res: Device context.
+ * @h_phase: horizontal phase parameters.
+ * @v_phase: vertical phase parameters.
+ *
+ * Horizontal and vertical phase range is 0 to 7
+ */
+static void resizer_set_phase(struct isp_res_device *res, u32 h_phase,
+			      u32 v_phase)
+{
+	struct isp_device *isp = to_isp_device(res);
+	u32 rgval = 0;
+
+	rgval = isp_reg_readl(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT) &
+	      ~(ISPRSZ_CNT_HSTPH_MASK | ISPRSZ_CNT_VSTPH_MASK);
+	rgval |= (h_phase << ISPRSZ_CNT_HSTPH_SHIFT) & ISPRSZ_CNT_HSTPH_MASK;
+	rgval |= (v_phase << ISPRSZ_CNT_VSTPH_SHIFT) & ISPRSZ_CNT_VSTPH_MASK;
+
+	isp_reg_writel(isp, rgval, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT);
+}
+
+/*
+ * resizer_set_luma - Setup luminance enhancer parameters
+ * @res: Device context.
+ * @luma: Structure for luminance enhancer parameters.
+ *
+ * Algorithm select:
+ *  0x0: Disable
+ *  0x1: [-1  2 -1]/2 high-pass filter
+ *  0x2: [-1 -2  6 -2 -1]/4 high-pass filter
+ *
+ * Maximum gain:
+ *  The data is coded in U4Q4 representation.
+ *
+ * Slope:
+ *  The data is coded in U4Q4 representation.
+ *
+ * Coring offset:
+ *  The data is coded in U8Q0 representation.
+ *
+ * The new luminance value is computed as:
+ *  Y += HPF(Y) x max(GAIN, (HPF(Y) - CORE) x SLOP + 8) >> 4.
+ */
+static void resizer_set_luma(struct isp_res_device *res,
+			     struct resizer_luma_yenh *luma)
+{
+	struct isp_device *isp = to_isp_device(res);
+	u32 rgval = 0;
+
+	rgval  = (luma->algo << ISPRSZ_YENH_ALGO_SHIFT)
+		  & ISPRSZ_YENH_ALGO_MASK;
+	rgval |= (luma->gain << ISPRSZ_YENH_GAIN_SHIFT)
+		  & ISPRSZ_YENH_GAIN_MASK;
+	rgval |= (luma->slope << ISPRSZ_YENH_SLOP_SHIFT)
+		  & ISPRSZ_YENH_SLOP_MASK;
+	rgval |= (luma->core << ISPRSZ_YENH_CORE_SHIFT)
+		  & ISPRSZ_YENH_CORE_MASK;
+
+	isp_reg_writel(isp, rgval, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_YENH);
+}
+
+/*
+ * resizer_set_source - Input source select
+ * @res: Device context.
+ * @source: Input source type
+ *
+ * If this field is set to RESIZER_INPUT_VP, the resizer input is fed from
+ * Preview/CCDC engine, otherwise from memory.
+ */
+static void resizer_set_source(struct isp_res_device *res,
+			       enum resizer_input_entity source)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	if (source == RESIZER_INPUT_MEMORY)
+		isp_reg_set(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT,
+			    ISPRSZ_CNT_INPSRC);
+	else
+		isp_reg_clr(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT,
+			    ISPRSZ_CNT_INPSRC);
+}
+
+/*
+ * resizer_set_ratio - Setup horizontal and vertical resizing value
+ * @res: Device context.
+ * @ratio: Structure for ratio parameters.
+ *
+ * Resizing range from 64 to 1024
+ */
+static void resizer_set_ratio(struct isp_res_device *res,
+			      const struct resizer_ratio *ratio)
+{
+	struct isp_device *isp = to_isp_device(res);
+	const u16 *h_filter, *v_filter;
+	u32 rgval = 0;
+
+	rgval = isp_reg_readl(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT) &
+			      ~(ISPRSZ_CNT_HRSZ_MASK | ISPRSZ_CNT_VRSZ_MASK);
+	rgval |= ((ratio->horz - 1) << ISPRSZ_CNT_HRSZ_SHIFT)
+		  & ISPRSZ_CNT_HRSZ_MASK;
+	rgval |= ((ratio->vert - 1) << ISPRSZ_CNT_VRSZ_SHIFT)
+		  & ISPRSZ_CNT_VRSZ_MASK;
+	isp_reg_writel(isp, rgval, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT);
+
+	/* prepare horizontal filter coefficients */
+	if (ratio->horz > MID_RESIZE_VALUE)
+		h_filter = &filter_coefs.h_filter_coef_7tap[0];
+	else
+		h_filter = &filter_coefs.h_filter_coef_4tap[0];
+
+	/* prepare vertical filter coefficients */
+	if (ratio->vert > MID_RESIZE_VALUE)
+		v_filter = &filter_coefs.v_filter_coef_7tap[0];
+	else
+		v_filter = &filter_coefs.v_filter_coef_4tap[0];
+
+	resizer_set_filters(res, h_filter, v_filter);
+}
+
+/*
+ * resizer_set_dst_size - Setup the output height and width
+ * @res: Device context.
+ * @width: Output width.
+ * @height: Output height.
+ *
+ * Width :
+ *  The value must be EVEN.
+ *
+ * Height:
+ *  The number of bytes written to SDRAM must be
+ *  a multiple of 16-bytes if the vertical resizing factor
+ *  is greater than 1x (upsizing)
+ */
+static void resizer_set_output_size(struct isp_res_device *res,
+				    u32 width, u32 height)
+{
+	struct isp_device *isp = to_isp_device(res);
+	u32 rgval = 0;
+
+	dev_dbg(isp->dev, "Output size[w/h]: %dx%d\n", width, height);
+	rgval  = (width << ISPRSZ_OUT_SIZE_HORZ_SHIFT)
+		 & ISPRSZ_OUT_SIZE_HORZ_MASK;
+	rgval |= (height << ISPRSZ_OUT_SIZE_VERT_SHIFT)
+		 & ISPRSZ_OUT_SIZE_VERT_MASK;
+	isp_reg_writel(isp, rgval, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_OUT_SIZE);
+}
+
+/*
+ * resizer_set_output_offset - Setup memory offset for the output lines.
+ * @res: Device context.
+ * @offset: Memory offset.
+ *
+ * The 5 LSBs are forced to be zeros by the hardware to align on a 32-byte
+ * boundary; the 5 LSBs are read-only. For optimal use of SDRAM bandwidth,
+ * the SDRAM line offset must be set on a 256-byte boundary
+ */
+static void resizer_set_output_offset(struct isp_res_device *res, u32 offset)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	isp_reg_writel(isp, offset, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_SDR_OUTOFF);
+}
+
+/*
+ * resizer_set_start - Setup vertical and horizontal start position
+ * @res: Device context.
+ * @left: Horizontal start position.
+ * @top: Vertical start position.
+ *
+ * Vertical start line:
+ *  This field makes sense only when the resizer obtains its input
+ *  from the preview engine/CCDC
+ *
+ * Horizontal start pixel:
+ *  Pixels are coded on 16 bits for YUV and 8 bits for color separate data.
+ *  When the resizer gets its input from SDRAM, this field must be set
+ *  to <= 15 for YUV 16-bit data and <= 31 for 8-bit color separate data
+ */
+static void resizer_set_start(struct isp_res_device *res, u32 left, u32 top)
+{
+	struct isp_device *isp = to_isp_device(res);
+	u32 rgval = 0;
+
+	rgval = (left << ISPRSZ_IN_START_HORZ_ST_SHIFT)
+		& ISPRSZ_IN_START_HORZ_ST_MASK;
+	rgval |= (top << ISPRSZ_IN_START_VERT_ST_SHIFT)
+		 & ISPRSZ_IN_START_VERT_ST_MASK;
+
+	isp_reg_writel(isp, rgval, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_IN_START);
+}
+
+/*
+ * resizer_set_input_size - Setup the input size
+ * @res: Device context.
+ * @width: The range is 0 to 4095 pixels
+ * @height: The range is 0 to 4095 lines
+ */
+static void resizer_set_input_size(struct isp_res_device *res,
+				   u32 width, u32 height)
+{
+	struct isp_device *isp = to_isp_device(res);
+	u32 rgval = 0;
+
+	dev_dbg(isp->dev, "Input size[w/h]: %dx%d\n", width, height);
+
+	rgval = (width << ISPRSZ_IN_SIZE_HORZ_SHIFT)
+		& ISPRSZ_IN_SIZE_HORZ_MASK;
+	rgval |= (height << ISPRSZ_IN_SIZE_VERT_SHIFT)
+		 & ISPRSZ_IN_SIZE_VERT_MASK;
+
+	isp_reg_writel(isp, rgval, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_IN_SIZE);
+}
+
+/*
+ * resizer_set_src_offs - Setup the memory offset for the input lines
+ * @res: Device context.
+ * @offset: Memory offset.
+ *
+ * The 5 LSBs are forced to be zeros by the hardware to align on a 32-byte
+ * boundary; the 5 LSBs are read-only. This field must be programmed to be
+ * 0x0 if the resizer input is from preview engine/CCDC.
+ */
+static void resizer_set_input_offset(struct isp_res_device *res, u32 offset)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	isp_reg_writel(isp, offset, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_SDR_INOFF);
+}
+
+/*
+ * resizer_set_intype - Input type select
+ * @res: Device context.
+ * @type: Pixel format type.
+ */
+static void resizer_set_intype(struct isp_res_device *res,
+			       enum resizer_colors_type type)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	if (type == RSZ_COLOR8)
+		isp_reg_set(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT,
+			    ISPRSZ_CNT_INPTYP);
+	else
+		isp_reg_clr(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_CNT,
+			    ISPRSZ_CNT_INPTYP);
+}
+
+/*
+ * __resizer_set_inaddr - Helper function for set input address
+ * @res : pointer to resizer private data structure
+ * @addr: input address
+ * return none
+ */
+static void __resizer_set_inaddr(struct isp_res_device *res, u32 addr)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	isp_reg_writel(isp, addr, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_SDR_INADD);
+}
+
+/*
+ * The data rate at the horizontal resizer output must not exceed half the
+ * functional clock or 100 MP/s, whichever is lower. According to the TRM
+ * there's no similar requirement for the vertical resizer output. However
+ * experience showed that vertical upscaling by 4 leads to SBL overflows (with
+ * data rates at the resizer output exceeding 300 MP/s). Limiting the resizer
+ * output data rate to the functional clock or 200 MP/s, whichever is lower,
+ * seems to get rid of SBL overflows.
+ *
+ * The maximum data rate at the output of the horizontal resizer can thus be
+ * computed with
+ *
+ * max intermediate rate <= L3 clock * input height / output height
+ * max intermediate rate <= L3 clock / 2
+ *
+ * The maximum data rate at the resizer input is then
+ *
+ * max input rate <= max intermediate rate * input width / output width
+ *
+ * where the input width and height are the resizer input crop rectangle size.
+ * The TRM doesn't clearly explain if that's a maximum instant data rate or a
+ * maximum average data rate.
+ */
+void omap3isp_resizer_max_rate(struct isp_res_device *res,
+			       unsigned int *max_rate)
+{
+	struct isp_pipeline *pipe = to_isp_pipeline(&res->subdev.entity);
+	const struct v4l2_mbus_framefmt *ofmt = &res->formats[RESZ_PAD_SOURCE];
+	unsigned long limit = min(pipe->l3_ick, 200000000UL);
+	unsigned long clock;
+
+	clock = div_u64((u64)limit * res->crop.active.height, ofmt->height);
+	clock = min(clock, limit / 2);
+	*max_rate = div_u64((u64)clock * res->crop.active.width, ofmt->width);
+}
+
+/*
+ * When the resizer processes images from memory, the driver must slow down read
+ * requests on the input to at least comply with the internal data rate
+ * requirements. If the application real-time requirements can cope with slower
+ * processing, the resizer can be slowed down even more to put less pressure on
+ * the overall system.
+ *
+ * When the resizer processes images on the fly (either from the CCDC or the
+ * preview module), the same data rate requirements apply but they can't be
+ * enforced at the resizer level. The image input module (sensor, CCP2 or
+ * preview module) must not provide image data faster than the resizer can
+ * process.
+ *
+ * For live image pipelines, the data rate is set by the frame format, size and
+ * rate. The sensor output frame rate must not exceed the maximum resizer data
+ * rate.
+ *
+ * The resizer slows down read requests by inserting wait cycles in the SBL
+ * requests. The maximum number of 256-byte requests per second can be computed
+ * as (the data rate is multiplied by 2 to convert from pixels per second to
+ * bytes per second)
+ *
+ * request per second = data rate * 2 / 256
+ * cycles per request = cycles per second / requests per second
+ *
+ * The number of cycles per second is controlled by the L3 clock, leading to
+ *
+ * cycles per request = L3 frequency / 2 * 256 / data rate
+ */
+static void resizer_adjust_bandwidth(struct isp_res_device *res)
+{
+	struct isp_pipeline *pipe = to_isp_pipeline(&res->subdev.entity);
+	struct isp_device *isp = to_isp_device(res);
+	unsigned long l3_ick = pipe->l3_ick;
+	struct v4l2_fract *timeperframe;
+	unsigned int cycles_per_frame;
+	unsigned int requests_per_frame;
+	unsigned int cycles_per_request;
+	unsigned int granularity;
+	unsigned int minimum;
+	unsigned int maximum;
+	unsigned int value;
+
+	if (res->input != RESIZER_INPUT_MEMORY) {
+		isp_reg_clr(isp, OMAP3_ISP_IOMEM_SBL, ISPSBL_SDR_REQ_EXP,
+			    ISPSBL_SDR_REQ_RSZ_EXP_MASK);
+		return;
+	}
+
+	switch (isp->revision) {
+	case ISP_REVISION_1_0:
+	case ISP_REVISION_2_0:
+	default:
+		granularity = 1024;
+		break;
+
+	case ISP_REVISION_15_0:
+		granularity = 32;
+		break;
+	}
+
+	/* Compute the minimum number of cycles per request, based on the
+	 * pipeline maximum data rate. This is an absolute lower bound if we
+	 * don't want SBL overflows, so round the value up.
+	 */
+	cycles_per_request = div_u64((u64)l3_ick / 2 * 256 + pipe->max_rate - 1,
+				     pipe->max_rate);
+	minimum = DIV_ROUND_UP(cycles_per_request, granularity);
+
+	/* Compute the maximum number of cycles per request, based on the
+	 * requested frame rate. This is a soft upper bound to achieve a frame
+	 * rate equal or higher than the requested value, so round the value
+	 * down.
+	 */
+	timeperframe = &pipe->max_timeperframe;
+
+	requests_per_frame = DIV_ROUND_UP(res->crop.active.width * 2, 256)
+			   * res->crop.active.height;
+	cycles_per_frame = div_u64((u64)l3_ick * timeperframe->numerator,
+				   timeperframe->denominator);
+	cycles_per_request = cycles_per_frame / requests_per_frame;
+
+	maximum = cycles_per_request / granularity;
+
+	value = max(minimum, maximum);
+
+	dev_dbg(isp->dev, "%s: cycles per request = %u\n", __func__, value);
+	isp_reg_clr_set(isp, OMAP3_ISP_IOMEM_SBL, ISPSBL_SDR_REQ_EXP,
+			ISPSBL_SDR_REQ_RSZ_EXP_MASK,
+			value << ISPSBL_SDR_REQ_RSZ_EXP_SHIFT);
+}
+
+/*
+ * omap3isp_resizer_busy - Checks if ISP resizer is busy.
+ *
+ * Returns busy field from ISPRSZ_PCR register.
+ */
+int omap3isp_resizer_busy(struct isp_res_device *res)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	return isp_reg_readl(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_PCR) &
+			     ISPRSZ_PCR_BUSY;
+}
+
+/*
+ * resizer_set_inaddr - Sets the memory address of the input frame.
+ * @addr: 32bit memory address aligned on 32byte boundary.
+ */
+static void resizer_set_inaddr(struct isp_res_device *res, u32 addr)
+{
+	res->addr_base = addr;
+
+	/* This will handle crop settings in stream off state */
+	if (res->crop_offset)
+		addr += res->crop_offset & ~0x1f;
+
+	__resizer_set_inaddr(res, addr);
+}
+
+/*
+ * Configures the memory address to which the output frame is written.
+ * @addr: 32bit memory address aligned on 32byte boundary.
+ * Note: For SBL efficiency reasons the address should be on a 256-byte
+ * boundary.
+ */
+static void resizer_set_outaddr(struct isp_res_device *res, u32 addr)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	/*
+	 * Set output address. This needs to be in its own function
+	 * because it changes often.
+	 */
+	isp_reg_writel(isp, addr << ISPRSZ_SDR_OUTADD_ADDR_SHIFT,
+		       OMAP3_ISP_IOMEM_RESZ, ISPRSZ_SDR_OUTADD);
+}
+
+/*
+ * resizer_print_status - Prints the values of the resizer module registers.
+ */
+#define RSZ_PRINT_REGISTER(isp, name)\
+	dev_dbg(isp->dev, "###RSZ " #name "=0x%08x\n", \
+		isp_reg_readl(isp, OMAP3_ISP_IOMEM_RESZ, ISPRSZ_##name))
+
+static void resizer_print_status(struct isp_res_device *res)
+{
+	struct isp_device *isp = to_isp_device(res);
+
+	dev_dbg(isp->dev, "-------------Resizer Register dump----------\n");
+
+	RSZ_PRINT_REGISTER(isp, PCR);
+	RSZ_PRINT_REGISTER(isp, CNT);
+	RSZ_PRINT_REGISTER(isp, OUT_SIZE);
+	RSZ_PRINT_REGISTER(isp, IN_START);
+	RSZ_PRINT_REGISTER(isp, IN_SIZE);
+	RSZ_PRINT_REGISTER(isp, SDR_INADD);
+	RSZ_PRINT_REGISTER(isp, SDR_INOFF);
+	RSZ_PRINT_REGISTER(isp, SDR_OUTADD);
+	RSZ_PRINT_REGISTER(isp, SDR_OUTOFF);
+	RSZ_PRINT_REGISTER(isp, YENH);
+
+	dev_dbg(isp->dev, "--------------------------------------------\n");
+}
+
+/*
+ * resizer_calc_ratios - Helper function for calculating resizer ratios
+ * @res: pointer to resizer private data structure
+ * @input: input frame size
+ * @output: output frame size
+ * @ratio : return calculated ratios
+ * return none
+ *
+ * The resizer uses a polyphase sample rate converter. The upsampling filter
+ * has a fixed number of phases that depend on the resizing ratio. As the ratio
+ * computation depends on the number of phases, we need to compute a first
+ * approximation and then refine it.
+ *
+ * The input/output/ratio relationship is given by the OMAP34xx TRM:
+ *
+ * - 8-phase, 4-tap mode (RSZ = 64 ~ 512)
+ *	iw = (32 * sph + (ow - 1) * hrsz + 16) >> 8 + 7
+ *	ih = (32 * spv + (oh - 1) * vrsz + 16) >> 8 + 4
+ * - 4-phase, 7-tap mode (RSZ = 513 ~ 1024)
+ *	iw = (64 * sph + (ow - 1) * hrsz + 32) >> 8 + 7
+ *	ih = (64 * spv + (oh - 1) * vrsz + 32) >> 8 + 7
+ *
+ * iw and ih are the input width and height after cropping. Those equations need
+ * to be satisfied exactly for the resizer to work correctly.
+ *
+ * The equations can't be easily reverted, as the >> 8 operation is not linear.
+ * In addition, not all input sizes can be achieved for a given output size. To
+ * get the highest input size lower than or equal to the requested input size,
+ * we need to compute the highest resizing ratio that satisfies the following
+ * inequality (taking the 4-tap mode width equation as an example)
+ *
+ *	iw >= (32 * sph + (ow - 1) * hrsz + 16) >> 8 - 7
+ *
+ * (where iw is the requested input width) which can be rewritten as
+ *
+ *	  iw - 7            >= (32 * sph + (ow - 1) * hrsz + 16) >> 8
+ *	 (iw - 7) << 8      >=  32 * sph + (ow - 1) * hrsz + 16 - b
+ *	((iw - 7) << 8) + b >=  32 * sph + (ow - 1) * hrsz + 16
+ *
+ * where b is the value of the 8 least significant bits of the right hand side
+ * expression of the last inequality. The highest resizing ratio value will be
+ * achieved when b is equal to its maximum value of 255. That resizing ratio
+ * value will still satisfy the original inequality, as b will disappear when
+ * the expression will be shifted right by 8.
+ *
+ * The reverted equations thus become
+ *
+ * - 8-phase, 4-tap mode
+ *	hrsz = ((iw - 7) * 256 + 255 - 16 - 32 * sph) / (ow - 1)
+ *	vrsz = ((ih - 4) * 256 + 255 - 16 - 32 * spv) / (oh - 1)
+ * - 4-phase, 7-tap mode
+ *	hrsz = ((iw - 7) * 256 + 255 - 32 - 64 * sph) / (ow - 1)
+ *	vrsz = ((ih - 7) * 256 + 255 - 32 - 64 * spv) / (oh - 1)
+ *
+ * The ratios are integer values, and are rounded down to ensure that the
+ * cropped input size is not bigger than the uncropped input size.
+ *
+ * As the number of phases/taps, used to select the correct equations to compute
+ * the ratio, depends on the ratio, we start with the 4-tap mode equations to
+ * compute an approximation of the ratio, and switch to the 7-tap mode equations
+ * if the approximation is higher than the ratio threshold.
+ *
+ * As the 7-tap mode equations will return a ratio smaller than or equal to the
+ * 4-tap mode equations, the resulting ratio could become lower than or equal to
+ * the ratio threshold. This 'equations loop' isn't an issue as long as the
+ * correct equations are used to compute the final input size. Starting with the
+ * 4-tap mode equations ensure that, in case of values resulting in a 'ratio
+ * loop', the smallest of the ratio values will be used, never exceeding the
+ * requested input size.
+ *
+ * We first clamp the output size according to the hardware capability to avoid
+ * auto-cropping the input more than required to satisfy the TRM equations. The
+ * minimum output size is achieved with a scaling factor of 1024. It is thus
+ * computed using the 7-tap equations.
+ *
+ *	min ow = ((iw - 7) * 256 - 32 - 64 * sph) / 1024 + 1
+ *	min oh = ((ih - 7) * 256 - 32 - 64 * spv) / 1024 + 1
+ *
+ * Similarly, the maximum output size is achieved with a scaling factor of 64
+ * and computed using the 4-tap equations.
+ *
+ *	max ow = ((iw - 7) * 256 + 255 - 16 - 32 * sph) / 64 + 1
+ *	max oh = ((ih - 4) * 256 + 255 - 16 - 32 * spv) / 64 + 1
+ *
+ * The additional +255 term compensates for the round down operation performed
+ * by the TRM equations when shifting the value right by 8 bits.
+ *
+ * We then compute and clamp the ratios (x1/4 ~ x4). Clamping the output size to
+ * the maximum value guarantees that the ratio value will never be smaller than
+ * the minimum, but it could still slightly exceed the maximum. Clamping the
+ * ratio will thus result in a resizing factor slightly larger than the
+ * requested value.
+ *
+ * To accommodate that, and make sure the TRM equations are satisfied exactly, we
+ * compute the input crop rectangle as the last step.
+ *
+ * As if the situation wasn't complex enough, the maximum output width depends
+ * on the vertical resizing ratio.  Fortunately, the output height doesn't
+ * depend on the horizontal resizing ratio. We can then start by computing the
+ * output height and the vertical ratio, and then move to computing the output
+ * width and the horizontal ratio.
+ */
+static void resizer_calc_ratios(struct isp_res_device *res,
+				struct v4l2_rect *input,
+				struct v4l2_mbus_framefmt *output,
+				struct resizer_ratio *ratio)
+{
+	struct isp_device *isp = to_isp_device(res);
+	const unsigned int spv = DEFAULT_PHASE;
+	const unsigned int sph = DEFAULT_PHASE;
+	unsigned int upscaled_width;
+	unsigned int upscaled_height;
+	unsigned int min_width;
+	unsigned int min_height;
+	unsigned int max_width;
+	unsigned int max_height;
+	unsigned int width_alignment;
+	unsigned int width;
+	unsigned int height;
+
+	/*
+	 * Clamp the output height based on the hardware capabilities and
+	 * compute the vertical resizing ratio.
+	 */
+	min_height = ((input->height - 7) * 256 - 32 - 64 * spv) / 1024 + 1;
+	min_height = max_t(unsigned int, min_height, MIN_OUT_HEIGHT);
+	max_height = ((input->height - 4) * 256 + 255 - 16 - 32 * spv) / 64 + 1;
+	max_height = min_t(unsigned int, max_height, MAX_OUT_HEIGHT);
+	output->height = clamp(output->height, min_height, max_height);
+
+	ratio->vert = ((input->height - 4) * 256 + 255 - 16 - 32 * spv)
+		    / (output->height - 1);
+	if (ratio->vert > MID_RESIZE_VALUE)
+		ratio->vert = ((input->height - 7) * 256 + 255 - 32 - 64 * spv)
+			    / (output->height - 1);
+	ratio->vert = clamp_t(unsigned int, ratio->vert,
+			      MIN_RESIZE_VALUE, MAX_RESIZE_VALUE);
+
+	if (ratio->vert <= MID_RESIZE_VALUE) {
+		upscaled_height = (output->height - 1) * ratio->vert
+				+ 32 * spv + 16;
+		height = (upscaled_height >> 8) + 4;
+	} else {
+		upscaled_height = (output->height - 1) * ratio->vert
+				+ 64 * spv + 32;
+		height = (upscaled_height >> 8) + 7;
+	}
+
+	/*
+	 * Compute the minimum and maximum output widths based on the hardware
+	 * capabilities. The maximum depends on the vertical resizing ratio.
+	 */
+	min_width = ((input->width - 7) * 256 - 32 - 64 * sph) / 1024 + 1;
+	min_width = max_t(unsigned int, min_width, MIN_OUT_WIDTH);
+
+	if (ratio->vert <= MID_RESIZE_VALUE) {
+		switch (isp->revision) {
+		case ISP_REVISION_1_0:
+			max_width = MAX_4TAP_OUT_WIDTH_ES1;
+			break;
+
+		case ISP_REVISION_2_0:
+		default:
+			max_width = MAX_4TAP_OUT_WIDTH_ES2;
+			break;
+
+		case ISP_REVISION_15_0:
+			max_width = MAX_4TAP_OUT_WIDTH_3630;
+			break;
+		}
+	} else {
+		switch (isp->revision) {
+		case ISP_REVISION_1_0:
+			max_width = MAX_7TAP_OUT_WIDTH_ES1;
+			break;
+
+		case ISP_REVISION_2_0:
+		default:
+			max_width = MAX_7TAP_OUT_WIDTH_ES2;
+			break;
+
+		case ISP_REVISION_15_0:
+			max_width = MAX_7TAP_OUT_WIDTH_3630;
+			break;
+		}
+	}
+	max_width = min(((input->width - 7) * 256 + 255 - 16 - 32 * sph) / 64
+			+ 1, max_width);
+
+	/*
+	 * The output width must be even, and must be a multiple of 16 bytes
+	 * when upscaling vertically. Clamp the output width to the valid range.
+	 * Take the alignment into account (the maximum width in 7-tap mode on
+	 * ES2 isn't a multiple of 8) and align the result up to make sure it
+	 * won't be smaller than the minimum.
+	 */
+	width_alignment = ratio->vert < 256 ? 8 : 2;
+	output->width = clamp(output->width, min_width,
+			      max_width & ~(width_alignment - 1));
+	output->width = ALIGN(output->width, width_alignment);
+
+	ratio->horz = ((input->width - 7) * 256 + 255 - 16 - 32 * sph)
+		    / (output->width - 1);
+	if (ratio->horz > MID_RESIZE_VALUE)
+		ratio->horz = ((input->width - 7) * 256 + 255 - 32 - 64 * sph)
+			    / (output->width - 1);
+	ratio->horz = clamp_t(unsigned int, ratio->horz,
+			      MIN_RESIZE_VALUE, MAX_RESIZE_VALUE);
+
+	if (ratio->horz <= MID_RESIZE_VALUE) {
+		upscaled_width = (output->width - 1) * ratio->horz
+			       + 32 * sph + 16;
+		width = (upscaled_width >> 8) + 7;
+	} else {
+		upscaled_width = (output->width - 1) * ratio->horz
+			       + 64 * sph + 32;
+		width = (upscaled_width >> 8) + 7;
+	}
+
+	/* Center the new crop rectangle. */
+	input->left += (input->width - width) / 2;
+	input->top += (input->height - height) / 2;
+	input->width = width;
+	input->height = height;
+}
+
+/*
+ * resizer_set_crop_params - Setup hardware with cropping parameters
+ * @res : resizer private structure
+ * @crop_rect : current crop rectangle
+ * @ratio : resizer ratios
+ * return none
+ */
+static void resizer_set_crop_params(struct isp_res_device *res,
+				    const struct v4l2_mbus_framefmt *input,
+				    const struct v4l2_mbus_framefmt *output)
+{
+	resizer_set_ratio(res, &res->ratio);
+
+	/* Set chrominance horizontal algorithm */
+	if (res->ratio.horz >= RESIZE_DIVISOR)
+		resizer_set_bilinear(res, RSZ_THE_SAME);
+	else
+		resizer_set_bilinear(res, RSZ_BILINEAR);
+
+	resizer_adjust_bandwidth(res);
+
+	if (res->input == RESIZER_INPUT_MEMORY) {
+		/* Calculate additional offset for crop */
+		res->crop_offset = (res->crop.active.top * input->width +
+				    res->crop.active.left) * 2;
+		/*
+		 * Write lowest 4 bits of horizontal pixel offset (in pixels),
+		 * vertical start must be 0.
+		 */
+		resizer_set_start(res, (res->crop_offset / 2) & 0xf, 0);
+
+		/*
+		 * Set start (read) address for cropping, in bytes.
+		 * Lowest 5 bits must be zero.
+		 */
+		__resizer_set_inaddr(res,
+				res->addr_base + (res->crop_offset & ~0x1f));
+	} else {
+		/*
+		 * Set vertical start line and horizontal starting pixel.
+		 * If the input is from CCDC/PREV, horizontal start field is
+		 * in bytes (twice number of pixels).
+		 */
+		resizer_set_start(res, res->crop.active.left * 2,
+				  res->crop.active.top);
+		/* Input address and offset must be 0 for preview/ccdc input */
+		__resizer_set_inaddr(res, 0);
+		resizer_set_input_offset(res, 0);
+	}
+
+	/* Set the input size */
+	resizer_set_input_size(res, res->crop.active.width,
+			       res->crop.active.height);
+}
+
+static void resizer_configure(struct isp_res_device *res)
+{
+	struct v4l2_mbus_framefmt *informat, *outformat;
+	struct resizer_luma_yenh luma = {0, 0, 0, 0};
+
+	resizer_set_source(res, res->input);
+
+	informat = &res->formats[RESZ_PAD_SINK];
+	outformat = &res->formats[RESZ_PAD_SOURCE];
+
+	/* RESZ_PAD_SINK */
+	if (res->input == RESIZER_INPUT_VP)
+		resizer_set_input_offset(res, 0);
+	else
+		resizer_set_input_offset(res, ALIGN(informat->width, 0x10) * 2);
+
+	/* YUV422 interleaved, default phase, no luma enhancement */
+	resizer_set_intype(res, RSZ_YUV422);
+	resizer_set_ycpos(res, informat->code);
+	resizer_set_phase(res, DEFAULT_PHASE, DEFAULT_PHASE);
+	resizer_set_luma(res, &luma);
+