drm: sti: add cursor plane

stih407 SoC have a dedicated hardware cursor plane,
this patch enable it.
The hardware have a color look up table, fix it to
be able to use ARGB8888.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
diff --git a/drivers/gpu/drm/sti/Makefile b/drivers/gpu/drm/sti/Makefile
index 04ac2ce..d6128f7f 100644
--- a/drivers/gpu/drm/sti/Makefile
+++ b/drivers/gpu/drm/sti/Makefile
@@ -3,6 +3,7 @@
 	sti_mixer.o \
 	sti_gdp.o \
 	sti_vid.o \
+	sti_cursor.o \
 	sti_compositor.o \
 	sti_drm_crtc.o \
 	sti_drm_plane.o
diff --git a/drivers/gpu/drm/sti/sti_compositor.c b/drivers/gpu/drm/sti/sti_compositor.c
index bbf8462..b9415b3 100644
--- a/drivers/gpu/drm/sti/sti_compositor.c
+++ b/drivers/gpu/drm/sti/sti_compositor.c
@@ -24,8 +24,9 @@
  * stiH407 compositor properties
  */
 struct sti_compositor_data stih407_compositor_data = {
-	.nb_subdev = 7,
+	.nb_subdev = 8,
 	.subdev_desc = {
+			{STI_CURSOR_SUBDEV, (int)STI_CURSOR, 0x000},
 			{STI_GPD_SUBDEV, (int)STI_GDP_0, 0x100},
 			{STI_GPD_SUBDEV, (int)STI_GDP_1, 0x200},
 			{STI_GPD_SUBDEV, (int)STI_GDP_2, 0x300},
@@ -68,11 +69,11 @@
 			break;
 		case STI_GPD_SUBDEV:
 		case STI_VID_SUBDEV:
+		case STI_CURSOR_SUBDEV:
 			compo->layer[layer_id++] =
 			    sti_layer_create(compo->dev, desc[i].id,
 					     compo->regs + desc[i].offset);
 			break;
-			/* case STI_CURSOR_SUBDEV : TODO */
 		default:
 			DRM_ERROR("Unknow subdev compoment type\n");
 			return 1;
@@ -125,11 +126,12 @@
 			}
 
 			/* The first planes are reserved for primary planes*/
-			if (crtc < compo->nb_mixers) {
+			if (crtc < compo->nb_mixers && primary) {
 				sti_drm_crtc_init(drm_dev, compo->mixer[crtc],
 						primary, cursor);
 				crtc++;
 				cursor = NULL;
+				primary = NULL;
 			}
 		}
 	}
diff --git a/drivers/gpu/drm/sti/sti_cursor.c b/drivers/gpu/drm/sti/sti_cursor.c
new file mode 100644
index 0000000..010eaee
--- /dev/null
+++ b/drivers/gpu/drm/sti/sti_cursor.c
@@ -0,0 +1,242 @@
+/*
+ * Copyright (C) STMicroelectronics SA 2014
+ * Authors: Vincent Abriou <vincent.abriou@st.com>
+ *          Fabien Dessenne <fabien.dessenne@st.com>
+ *          for STMicroelectronics.
+ * License terms:  GNU General Public License (GPL), version 2
+ */
+#include <drm/drmP.h>
+
+#include "sti_cursor.h"
+#include "sti_layer.h"
+#include "sti_vtg.h"
+
+/* Registers */
+#define CUR_CTL             0x00
+#define CUR_VPO             0x0C
+#define CUR_PML             0x14
+#define CUR_PMP             0x18
+#define CUR_SIZE            0x1C
+#define CUR_CML             0x20
+#define CUR_AWS             0x28
+#define CUR_AWE             0x2C
+
+#define CUR_CTL_CLUT_UPDATE BIT(1)
+
+#define STI_CURS_MIN_SIZE   1
+#define STI_CURS_MAX_SIZE   128
+
+/*
+ * pixmap dma buffer stucture
+ *
+ * @paddr:  physical address
+ * @size:   buffer size
+ * @base:   virtual address
+ */
+struct dma_pixmap {
+	dma_addr_t paddr;
+	size_t size;
+	void *base;
+};
+
+/**
+ * STI Cursor structure
+ *
+ * @layer:      layer structure
+ * @width:      cursor width
+ * @height:     cursor height
+ * @clut:       color look up table
+ * @clut_paddr: color look up table physical address
+ * @pixmap:     pixmap dma buffer (clut8-format cursor)
+ */
+struct sti_cursor {
+	struct sti_layer layer;
+	unsigned int width;
+	unsigned int height;
+	unsigned short *clut;
+	dma_addr_t clut_paddr;
+	struct dma_pixmap pixmap;
+};
+
+static const uint32_t cursor_supported_formats[] = {
+	DRM_FORMAT_ARGB8888,
+};
+
+#define to_sti_cursor(x) container_of(x, struct sti_cursor, layer)
+
+static const uint32_t *sti_cursor_get_formats(struct sti_layer *layer)
+{
+	return cursor_supported_formats;
+}
+
+static unsigned int sti_cursor_get_nb_formats(struct sti_layer *layer)
+{
+	return ARRAY_SIZE(cursor_supported_formats);
+}
+
+static void sti_cursor_argb8888_to_clut8(struct sti_layer *layer)
+{
+	struct sti_cursor *cursor = to_sti_cursor(layer);
+	u32 *src = layer->vaddr;
+	u8  *dst = cursor->pixmap.base;
+	unsigned int i, j;
+	u32 a, r, g, b;
+
+	for (i = 0; i < cursor->height; i++) {
+		for (j = 0; j < cursor->width; j++) {
+			/* Pick the 2 higher bits of each component */
+			a = (*src >> 30) & 3;
+			r = (*src >> 22) & 3;
+			g = (*src >> 14) & 3;
+			b = (*src >> 6) & 3;
+			*dst = a << 6 | r << 4 | g << 2 | b;
+			src++;
+			dst++;
+		}
+	}
+}
+
+static int sti_cursor_prepare_layer(struct sti_layer *layer, bool first_prepare)
+{
+	struct sti_cursor *cursor = to_sti_cursor(layer);
+	struct drm_display_mode *mode = layer->mode;
+	u32 y, x;
+	u32 val;
+
+	DRM_DEBUG_DRIVER("\n");
+
+	dev_dbg(layer->dev, "%s %s\n", __func__, sti_layer_to_str(layer));
+
+	if (layer->src_w < STI_CURS_MIN_SIZE ||
+	    layer->src_h < STI_CURS_MIN_SIZE ||
+	    layer->src_w > STI_CURS_MAX_SIZE ||
+	    layer->src_h > STI_CURS_MAX_SIZE) {
+		DRM_ERROR("Invalid cursor size (%dx%d)\n",
+				layer->src_w, layer->src_h);
+		return -EINVAL;
+	}
+
+	/* If the cursor size has changed, re-allocated the pixmap */
+	if (!cursor->pixmap.base ||
+	    (cursor->width != layer->src_w) ||
+	    (cursor->height != layer->src_h)) {
+		cursor->width = layer->src_w;
+		cursor->height = layer->src_h;
+
+		if (cursor->pixmap.base)
+			dma_free_writecombine(layer->dev,
+					      cursor->pixmap.size,
+					      cursor->pixmap.base,
+					      cursor->pixmap.paddr);
+
+		cursor->pixmap.size = cursor->width * cursor->height;
+
+		cursor->pixmap.base = dma_alloc_writecombine(layer->dev,
+							cursor->pixmap.size,
+							&cursor->pixmap.paddr,
+							GFP_KERNEL | GFP_DMA);
+		if (!cursor->pixmap.base) {
+			DRM_ERROR("Failed to allocate memory for pixmap\n");
+			return -ENOMEM;
+		}
+	}
+
+	/* Convert ARGB8888 to CLUT8 */
+	sti_cursor_argb8888_to_clut8(layer);
+
+	/* AWS and AWE depend on the mode */
+	y = sti_vtg_get_line_number(*mode, 0);
+	x = sti_vtg_get_pixel_number(*mode, 0);
+	val = y << 16 | x;
+	writel(val, layer->regs + CUR_AWS);
+	y = sti_vtg_get_line_number(*mode, mode->vdisplay - 1);
+	x = sti_vtg_get_pixel_number(*mode, mode->hdisplay - 1);
+	val = y << 16 | x;
+	writel(val, layer->regs + CUR_AWE);
+
+	if (first_prepare) {
+		/* Set and fetch CLUT */
+		writel(cursor->clut_paddr, layer->regs + CUR_CML);
+		writel(CUR_CTL_CLUT_UPDATE, layer->regs + CUR_CTL);
+	}
+
+	return 0;
+}
+
+static int sti_cursor_commit_layer(struct sti_layer *layer)
+{
+	struct sti_cursor *cursor = to_sti_cursor(layer);
+	struct drm_display_mode *mode = layer->mode;
+	u32 ydo, xdo;
+
+	dev_dbg(layer->dev, "%s %s\n", __func__, sti_layer_to_str(layer));
+
+	/* Set memory location, size, and position */
+	writel(cursor->pixmap.paddr, layer->regs + CUR_PML);
+	writel(cursor->width, layer->regs + CUR_PMP);
+	writel(cursor->height << 16 | cursor->width, layer->regs + CUR_SIZE);
+
+	ydo = sti_vtg_get_line_number(*mode, layer->dst_y);
+	xdo = sti_vtg_get_pixel_number(*mode, layer->dst_y);
+	writel((ydo << 16) | xdo, layer->regs + CUR_VPO);
+
+	return 0;
+}
+
+static int sti_cursor_disable_layer(struct sti_layer *layer)
+{
+	return 0;
+}
+
+static void sti_cursor_init(struct sti_layer *layer)
+{
+	struct sti_cursor *cursor = to_sti_cursor(layer);
+	unsigned short *base = cursor->clut;
+	unsigned int a, r, g, b;
+
+	/* Assign CLUT values, ARGB444 format */
+	for (a = 0; a < 4; a++)
+		for (r = 0; r < 4; r++)
+			for (g = 0; g < 4; g++)
+				for (b = 0; b < 4; b++)
+					*base++ = (a * 5) << 12 |
+						  (r * 5) << 8 |
+						  (g * 5) << 4 |
+						  (b * 5);
+}
+
+static const struct sti_layer_funcs cursor_ops = {
+	.get_formats = sti_cursor_get_formats,
+	.get_nb_formats = sti_cursor_get_nb_formats,
+	.init = sti_cursor_init,
+	.prepare = sti_cursor_prepare_layer,
+	.commit = sti_cursor_commit_layer,
+	.disable = sti_cursor_disable_layer,
+};
+
+struct sti_layer *sti_cursor_create(struct device *dev)
+{
+	struct sti_cursor *cursor;
+
+	cursor = devm_kzalloc(dev, sizeof(*cursor), GFP_KERNEL);
+	if (!cursor) {
+		DRM_ERROR("Failed to allocate memory for cursor\n");
+		return NULL;
+	}
+
+	/* Allocate clut buffer */
+	cursor->clut = dma_alloc_writecombine(dev,
+			0x100 * sizeof(unsigned short),
+			&cursor->clut_paddr,
+			GFP_KERNEL | GFP_DMA);
+
+	if (!cursor->clut) {
+		DRM_ERROR("Failed to allocate memory for cursor clut\n");
+		devm_kfree(dev, cursor);
+		return NULL;
+	}
+
+	cursor->layer.ops = &cursor_ops;
+
+	return (struct sti_layer *)cursor;
+}
diff --git a/drivers/gpu/drm/sti/sti_cursor.h b/drivers/gpu/drm/sti/sti_cursor.h
new file mode 100644
index 0000000..3c98274
--- /dev/null
+++ b/drivers/gpu/drm/sti/sti_cursor.h
@@ -0,0 +1,12 @@
+/*
+ * Copyright (C) STMicroelectronics SA 2013
+ * Authors: Vincent Abriou <vincent.abriou@st.com> for STMicroelectronics.
+ * License terms:  GNU General Public License (GPL), version 2
+ */
+
+#ifndef _STI_CURSOR_H_
+#define _STI_CURSOR_H_
+
+struct sti_layer *sti_cursor_create(struct device *dev);
+
+#endif
diff --git a/drivers/gpu/drm/sti/sti_layer.c b/drivers/gpu/drm/sti/sti_layer.c
index 06a587c..5051b4c 100644
--- a/drivers/gpu/drm/sti/sti_layer.c
+++ b/drivers/gpu/drm/sti/sti_layer.c
@@ -11,6 +11,7 @@
 #include <drm/drm_fb_cma_helper.h>
 
 #include "sti_compositor.h"
+#include "sti_cursor.h"
 #include "sti_gdp.h"
 #include "sti_layer.h"
 #include "sti_vid.h"
@@ -50,6 +51,9 @@
 	case STI_VID:
 		layer = sti_vid_create(dev);
 		break;
+	case STI_CUR:
+		layer = sti_cursor_create(dev);
+		break;
 	}
 
 	if (!layer) {
@@ -100,6 +104,7 @@
 	layer->src_w = src_w;
 	layer->src_h = src_h;
 	layer->format = fb->pixel_format;
+	layer->vaddr = cma_obj->vaddr;
 	layer->paddr = cma_obj->paddr;
 	for (i = 0; i < 4; i++) {
 		layer->pitches[i] = fb->pitches[i];
diff --git a/drivers/gpu/drm/sti/sti_layer.h b/drivers/gpu/drm/sti/sti_layer.h
index 198c377..68bfdca 100644
--- a/drivers/gpu/drm/sti/sti_layer.h
+++ b/drivers/gpu/drm/sti/sti_layer.h
@@ -82,6 +82,7 @@
  * @format:             format
  * @pitches:            pitch of 'planes' (eg: Y, U, V)
  * @offsets:            offset of 'planes'
+ * @vaddr:              virtual address of the input buffer
  * @paddr:              physical address of the input buffer
  */
 struct sti_layer {
@@ -102,6 +103,7 @@
 	uint32_t format;
 	unsigned int pitches[4];
 	unsigned int offsets[4];
+	void *vaddr;
 	dma_addr_t paddr;
 };
 
diff --git a/drivers/gpu/drm/sti/sti_mixer.c b/drivers/gpu/drm/sti/sti_mixer.c
index 6bcdf3f..9a4ce74 100644
--- a/drivers/gpu/drm/sti/sti_mixer.c
+++ b/drivers/gpu/drm/sti/sti_mixer.c
@@ -45,6 +45,7 @@
 #define GAM_CTL_GDP1_MASK  BIT(4)
 #define GAM_CTL_GDP2_MASK  BIT(5)
 #define GAM_CTL_GDP3_MASK  BIT(6)
+#define GAM_CTL_CURSOR_MASK BIT(9)
 
 const char *sti_mixer_to_str(struct sti_mixer *mixer)
 {
@@ -127,6 +128,9 @@
 	case STI_VID_1:
 		layer_id = GAM_DEPTH_VID1_ID;
 		break;
+	case STI_CURSOR:
+		/* no need to set depth for cursor */
+		return 0;
 	default:
 		DRM_ERROR("Unknown layer %d\n", layer->desc);
 		return 1;
@@ -188,6 +192,8 @@
 		return GAM_CTL_VID0_MASK;
 	case STI_VID_1:
 		return GAM_CTL_VID1_MASK;
+	case STI_CURSOR:
+		return GAM_CTL_CURSOR_MASK;
 	default:
 		return 0;
 	}