]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/commitdiff
drm: sti: add GDP layer
authorBenjamin Gaignard <benjamin.gaignard@linaro.org>
Wed, 30 Jul 2014 16:48:35 +0000 (18:48 +0200)
committerBenjamin Gaignard <benjamin.gaignard@linaro.org>
Wed, 30 Jul 2014 17:26:47 +0000 (19:26 +0200)
Generic Display Pipeline are one of the compositor input sub-devices.
GDP are dedicated to graphic input like RGB plans.
GDP is part of Compositor hardware block which will be introduce later.

A sti_layer structure is used to abstract GDP calls from Compositor.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Reviewed-by: Rob Clark <robdclark@gmail.com>
drivers/gpu/drm/sti/Makefile
drivers/gpu/drm/sti/sti_gdp.c [new file with mode: 0644]
drivers/gpu/drm/sti/sti_gdp.h [new file with mode: 0644]
drivers/gpu/drm/sti/sti_layer.h [new file with mode: 0644]

index 5f6d96d27f3b240e046c9fbc5169a1b40164aa87..7de757f7eb38bf8e7d2f5029fc0fdd25f2e51a20 100644 (file)
@@ -1,3 +1,6 @@
+sticompositor-y := \
+       sti_gdp.o
+
 stihdmi-y := sti_hdmi.o \
        sti_hdmi_tx3g0c55phy.o \
        sti_hdmi_tx3g4c28phy.o \
@@ -7,4 +10,5 @@ obj-$(CONFIG_DRM_STI) = \
        sti_vtac.o \
        stihdmi.o \
        sti_hda.o \
-       sti_tvout.o
\ No newline at end of file
+       sti_tvout.o \
+       sticompositor.o
diff --git a/drivers/gpu/drm/sti/sti_gdp.c b/drivers/gpu/drm/sti/sti_gdp.c
new file mode 100644 (file)
index 0000000..7f55767
--- /dev/null
@@ -0,0 +1,516 @@
+/*
+ * Copyright (C) STMicroelectronics SA 2014
+ * Authors: Benjamin Gaignard <benjamin.gaignard@st.com>
+ *          Fabien Dessenne <fabien.dessenne@st.com>
+ *          for STMicroelectronics.
+ * License terms:  GNU General Public License (GPL), version 2
+ */
+
+#include <linux/clk.h>
+#include <linux/dma-mapping.h>
+
+#include "sti_gdp.h"
+#include "sti_layer.h"
+#include "sti_vtg.h"
+
+#define ENA_COLOR_FILL  BIT(8)
+#define WAIT_NEXT_VSYNC BIT(31)
+
+/* GDP color formats */
+#define GDP_RGB565      0x00
+#define GDP_RGB888      0x01
+#define GDP_RGB888_32   0x02
+#define GDP_ARGB8565    0x04
+#define GDP_ARGB8888    0x05
+#define GDP_ARGB1555    0x06
+#define GDP_ARGB4444    0x07
+#define GDP_CLUT8       0x0B
+#define GDP_YCBR888     0x10
+#define GDP_YCBR422R    0x12
+#define GDP_AYCBR8888   0x15
+
+#define GAM_GDP_CTL_OFFSET      0x00
+#define GAM_GDP_AGC_OFFSET      0x04
+#define GAM_GDP_VPO_OFFSET      0x0C
+#define GAM_GDP_VPS_OFFSET      0x10
+#define GAM_GDP_PML_OFFSET      0x14
+#define GAM_GDP_PMP_OFFSET      0x18
+#define GAM_GDP_SIZE_OFFSET     0x1C
+#define GAM_GDP_NVN_OFFSET      0x24
+#define GAM_GDP_KEY1_OFFSET     0x28
+#define GAM_GDP_KEY2_OFFSET     0x2C
+#define GAM_GDP_PPT_OFFSET      0x34
+#define GAM_GDP_CML_OFFSET      0x3C
+#define GAM_GDP_MST_OFFSET      0x68
+
+#define GAM_GDP_ALPHARANGE_255  BIT(5)
+#define GAM_GDP_AGC_FULL_RANGE  0x00808080
+#define GAM_GDP_PPT_IGNORE      (BIT(1) | BIT(0))
+#define GAM_GDP_SIZE_MAX        0x7FF
+
+#define GDP_NODE_NB_BANK       2
+#define GDP_NODE_PER_FIELD     2
+
+struct sti_gdp_node {
+       u32 gam_gdp_ctl;
+       u32 gam_gdp_agc;
+       u32 reserved1;
+       u32 gam_gdp_vpo;
+       u32 gam_gdp_vps;
+       u32 gam_gdp_pml;
+       u32 gam_gdp_pmp;
+       u32 gam_gdp_size;
+       u32 reserved2;
+       u32 gam_gdp_nvn;
+       u32 gam_gdp_key1;
+       u32 gam_gdp_key2;
+       u32 reserved3;
+       u32 gam_gdp_ppt;
+       u32 reserved4;
+       u32 gam_gdp_cml;
+};
+
+struct sti_gdp_node_list {
+       struct sti_gdp_node *top_field;
+       struct sti_gdp_node *btm_field;
+};
+
+/**
+ * STI GDP structure
+ *
+ * @layer:             layer structure
+ * @clk_pix:            pixel clock for the current gdp
+ * @vtg_field_nb:       callback for VTG FIELD (top or bottom) notification
+ * @is_curr_top:        true if the current node processed is the top field
+ * @node_list:         array of node list
+ */
+struct sti_gdp {
+       struct sti_layer layer;
+       struct clk *clk_pix;
+       struct notifier_block vtg_field_nb;
+       bool is_curr_top;
+       struct sti_gdp_node_list node_list[GDP_NODE_NB_BANK];
+};
+
+#define to_sti_gdp(x) container_of(x, struct sti_gdp, layer)
+
+static const uint32_t gdp_supported_formats[] = {
+       DRM_FORMAT_XRGB8888,
+       DRM_FORMAT_ARGB8888,
+       DRM_FORMAT_ARGB4444,
+       DRM_FORMAT_ARGB1555,
+       DRM_FORMAT_RGB565,
+       DRM_FORMAT_RGB888,
+       DRM_FORMAT_AYUV,
+       DRM_FORMAT_YUV444,
+       DRM_FORMAT_VYUY,
+       DRM_FORMAT_C8,
+};
+
+static const uint32_t *sti_gdp_get_formats(struct sti_layer *layer)
+{
+       return gdp_supported_formats;
+}
+
+static unsigned int sti_gdp_get_nb_formats(struct sti_layer *layer)
+{
+       return ARRAY_SIZE(gdp_supported_formats);
+}
+
+static int sti_gdp_fourcc2format(int fourcc)
+{
+       switch (fourcc) {
+       case DRM_FORMAT_XRGB8888:
+               return GDP_RGB888_32;
+       case DRM_FORMAT_ARGB8888:
+               return GDP_ARGB8888;
+       case DRM_FORMAT_ARGB4444:
+               return GDP_ARGB4444;
+       case DRM_FORMAT_ARGB1555:
+               return GDP_ARGB1555;
+       case DRM_FORMAT_RGB565:
+               return GDP_RGB565;
+       case DRM_FORMAT_RGB888:
+               return GDP_RGB888;
+       case DRM_FORMAT_AYUV:
+               return GDP_AYCBR8888;
+       case DRM_FORMAT_YUV444:
+               return GDP_YCBR888;
+       case DRM_FORMAT_VYUY:
+               return GDP_YCBR422R;
+       case DRM_FORMAT_C8:
+               return GDP_CLUT8;
+       }
+       return -1;
+}
+
+static int sti_gdp_get_alpharange(int format)
+{
+       switch (format) {
+       case GDP_ARGB8565:
+       case GDP_ARGB8888:
+       case GDP_AYCBR8888:
+               return GAM_GDP_ALPHARANGE_255;
+       }
+       return 0;
+}
+
+/**
+ * sti_gdp_get_free_nodes
+ * @layer: gdp layer
+ *
+ * Look for a GDP node list that is not currently read by the HW.
+ *
+ * RETURNS:
+ * Pointer to the free GDP node list
+ */
+static struct sti_gdp_node_list *sti_gdp_get_free_nodes(struct sti_layer *layer)
+{
+       int hw_nvn;
+       void *virt_nvn;
+       struct sti_gdp *gdp = to_sti_gdp(layer);
+       unsigned int i;
+
+       hw_nvn = readl(layer->regs + GAM_GDP_NVN_OFFSET);
+       if (!hw_nvn)
+               goto end;
+
+       virt_nvn = dma_to_virt(layer->dev, (dma_addr_t) hw_nvn);
+
+       for (i = 0; i < GDP_NODE_NB_BANK; i++)
+               if ((virt_nvn != gdp->node_list[i].btm_field) &&
+                   (virt_nvn != gdp->node_list[i].top_field))
+                       return &gdp->node_list[i];
+
+end:
+       return &gdp->node_list[0];
+}
+
+/**
+ * sti_gdp_get_current_nodes
+ * @layer: GDP layer
+ *
+ * Look for GDP nodes that are currently read by the HW.
+ *
+ * RETURNS:
+ * Pointer to the current GDP node list
+ */
+static
+struct sti_gdp_node_list *sti_gdp_get_current_nodes(struct sti_layer *layer)
+{
+       int hw_nvn;
+       void *virt_nvn;
+       struct sti_gdp *gdp = to_sti_gdp(layer);
+       unsigned int i;
+
+       hw_nvn = readl(layer->regs + GAM_GDP_NVN_OFFSET);
+       if (!hw_nvn)
+               goto end;
+
+       virt_nvn = dma_to_virt(layer->dev, (dma_addr_t) hw_nvn);
+
+       for (i = 0; i < GDP_NODE_NB_BANK; i++)
+               if ((virt_nvn == gdp->node_list[i].btm_field) ||
+                               (virt_nvn == gdp->node_list[i].top_field))
+                       return &gdp->node_list[i];
+
+end:
+       return NULL;
+}
+
+/**
+ * sti_gdp_prepare_layer
+ * @lay: gdp layer
+ * @first_prepare: true if it is the first time this function is called
+ *
+ * Update the free GDP node list according to the layer properties.
+ *
+ * RETURNS:
+ * 0 on success.
+ */
+static int sti_gdp_prepare_layer(struct sti_layer *layer, bool first_prepare)
+{
+       struct sti_gdp_node_list *list;
+       struct sti_gdp_node *top_field, *btm_field;
+       struct drm_display_mode *mode = layer->mode;
+       struct device *dev = layer->dev;
+       struct sti_gdp *gdp = to_sti_gdp(layer);
+       int format;
+       unsigned int depth, bpp;
+       int rate = mode->clock * 1000;
+       int res;
+       u32 ydo, xdo, yds, xds;
+
+       list = sti_gdp_get_free_nodes(layer);
+       top_field = list->top_field;
+       btm_field = list->btm_field;
+
+       /* Build the top field from layer params */
+       top_field->gam_gdp_agc = GAM_GDP_AGC_FULL_RANGE;
+       top_field->gam_gdp_ctl = WAIT_NEXT_VSYNC;
+       format = sti_gdp_fourcc2format(layer->format);
+       if (format == -1) {
+               DRM_ERROR("Format not supported by GDP %.4s\n",
+                         (char *)&layer->format);
+               return 1;
+       }
+       top_field->gam_gdp_ctl |= format;
+       top_field->gam_gdp_ctl |= sti_gdp_get_alpharange(format);
+       top_field->gam_gdp_ppt &= ~GAM_GDP_PPT_IGNORE;
+
+       /* pixel memory location */
+       drm_fb_get_bpp_depth(layer->format, &depth, &bpp);
+       top_field->gam_gdp_pml = (u32) layer->paddr + layer->offsets[0];
+       top_field->gam_gdp_pml += layer->src_x * (bpp >> 3);
+       top_field->gam_gdp_pml += layer->src_y * layer->pitches[0];
+
+       /* input parameters */
+       top_field->gam_gdp_pmp = layer->pitches[0];
+       top_field->gam_gdp_size =
+           clamp_val(layer->src_h, 0, GAM_GDP_SIZE_MAX) << 16 |
+           clamp_val(layer->src_w, 0, GAM_GDP_SIZE_MAX);
+
+       /* output parameters */
+       ydo = sti_vtg_get_line_number(*mode, layer->dst_y);
+       yds = sti_vtg_get_line_number(*mode, layer->dst_y + layer->dst_h - 1);
+       xdo = sti_vtg_get_pixel_number(*mode, layer->dst_x);
+       xds = sti_vtg_get_pixel_number(*mode, layer->dst_x + layer->dst_w - 1);
+       top_field->gam_gdp_vpo = (ydo << 16) | xdo;
+       top_field->gam_gdp_vps = (yds << 16) | xds;
+
+       /* Same content and chained together */
+       memcpy(btm_field, top_field, sizeof(*btm_field));
+       top_field->gam_gdp_nvn = virt_to_dma(dev, btm_field);
+       btm_field->gam_gdp_nvn = virt_to_dma(dev, top_field);
+
+       /* Interlaced mode */
+       if (layer->mode->flags & DRM_MODE_FLAG_INTERLACE)
+               btm_field->gam_gdp_pml = top_field->gam_gdp_pml +
+                   layer->pitches[0];
+
+       if (first_prepare) {
+               /* Set and enable gdp clock */
+               if (gdp->clk_pix) {
+                       res = clk_set_rate(gdp->clk_pix, rate);
+                       if (res < 0) {
+                               DRM_ERROR("Cannot set rate (%dHz) for gdp\n",
+                                               rate);
+                               return 1;
+                       }
+
+                       if (clk_prepare_enable(gdp->clk_pix)) {
+                               DRM_ERROR("Failed to prepare/enable gdp\n");
+                               return 1;
+                       }
+               }
+       }
+
+       return 0;
+}
+
+/**
+ * sti_gdp_commit_layer
+ * @lay: gdp layer
+ *
+ * Update the NVN field of the 'right' field of the current GDP node (being
+ * used by the HW) with the address of the updated ('free') top field GDP node.
+ * - In interlaced mode the 'right' field is the bottom field as we update
+ *   frames starting from their top field
+ * - In progressive mode, we update both bottom and top fields which are
+ *   equal nodes.
+ * At the next VSYNC, the updated node list will be used by the HW.
+ *
+ * RETURNS:
+ * 0 on success.
+ */
+static int sti_gdp_commit_layer(struct sti_layer *layer)
+{
+       struct sti_gdp_node_list *updated_list = sti_gdp_get_free_nodes(layer);
+       struct sti_gdp_node *updated_top_node = updated_list->top_field;
+       struct sti_gdp_node *updated_btm_node = updated_list->btm_field;
+       struct sti_gdp *gdp = to_sti_gdp(layer);
+       u32 dma_updated_top = virt_to_dma(layer->dev, updated_top_node);
+       u32 dma_updated_btm = virt_to_dma(layer->dev, updated_btm_node);
+       struct sti_gdp_node_list *curr_list = sti_gdp_get_current_nodes(layer);
+
+       dev_dbg(layer->dev, "Current NVN:0x%X\n",
+               readl(layer->regs + GAM_GDP_NVN_OFFSET));
+       dev_dbg(layer->dev, "Posted buff: %lx current buff: %x\n",
+               (unsigned long)layer->paddr,
+               readl(layer->regs + GAM_GDP_PML_OFFSET));
+
+       if (curr_list == NULL) {
+               /* First update or invalid node should directly write in the
+                * hw register */
+               writel(gdp->is_curr_top == true ?
+                               dma_updated_btm : dma_updated_top,
+                               layer->regs + GAM_GDP_NVN_OFFSET);
+               return 0;
+       }
+
+       if (layer->mode->flags & DRM_MODE_FLAG_INTERLACE) {
+               if (gdp->is_curr_top == true) {
+                       /* Do not update in the middle of the frame, but
+                        * postpone the update after the bottom field has
+                        * been displayed */
+                       curr_list->btm_field->gam_gdp_nvn = dma_updated_top;
+               } else {
+                       /* Direct update to avoid one frame delay */
+                       writel(dma_updated_top,
+                               layer->regs + GAM_GDP_NVN_OFFSET);
+               }
+       } else {
+               /* Direct update for progressive to avoid one frame delay */
+               writel(dma_updated_top, layer->regs + GAM_GDP_NVN_OFFSET);
+       }
+
+       return 0;
+}
+
+/**
+ * sti_gdp_disable_layer
+ * @lay: gdp layer
+ *
+ * Disable a GDP.
+ *
+ * RETURNS:
+ * 0 on success.
+ */
+static int sti_gdp_disable_layer(struct sti_layer *layer)
+{
+       unsigned int i;
+       struct sti_gdp *gdp = to_sti_gdp(layer);
+
+       /* Set the nodes as 'to be ignored on mixer' */
+       for (i = 0; i < GDP_NODE_NB_BANK; i++) {
+               gdp->node_list[i].top_field->gam_gdp_ppt |= GAM_GDP_PPT_IGNORE;
+               gdp->node_list[i].btm_field->gam_gdp_ppt |= GAM_GDP_PPT_IGNORE;
+       }
+
+       if (gdp->clk_pix)
+               clk_disable_unprepare(gdp->clk_pix);
+
+       return 0;
+}
+
+/**
+ * sti_gdp_field_cb
+ * @nb: notifier block
+ * @event: event message
+ * @data: private data
+ *
+ * Handle VTG top field and bottom field event.
+ *
+ * RETURNS:
+ * 0 on success.
+ */
+int sti_gdp_field_cb(struct notifier_block *nb,
+               unsigned long event, void *data)
+{
+       struct sti_gdp *gdp = container_of(nb, struct sti_gdp, vtg_field_nb);
+
+       switch (event) {
+       case VTG_TOP_FIELD_EVENT:
+               gdp->is_curr_top = true;
+               break;
+       case VTG_BOTTOM_FIELD_EVENT:
+               gdp->is_curr_top = false;
+               break;
+       default:
+               DRM_ERROR("unsupported event: %lu\n", event);
+               break;
+       }
+
+       return 0;
+}
+
+static void sti_gdp_init(struct sti_layer *layer)
+{
+       struct sti_gdp *gdp = to_sti_gdp(layer);
+       struct device_node *np = layer->dev->of_node;
+       dma_addr_t dma;
+       void *base;
+       unsigned int i, size;
+
+       /* Allocate all the nodes within a single memory page */
+       size = sizeof(struct sti_gdp_node) *
+           GDP_NODE_PER_FIELD * GDP_NODE_NB_BANK;
+
+       base = dma_alloc_writecombine(layer->dev,
+                       size, &dma, GFP_KERNEL | GFP_DMA);
+       if (!base) {
+               DRM_ERROR("Failed to allocate memory for GDP node\n");
+               return;
+       }
+       memset(base, 0, size);
+
+       for (i = 0; i < GDP_NODE_NB_BANK; i++) {
+               if (virt_to_dma(layer->dev, base) & 0xF) {
+                       DRM_ERROR("Mem alignment failed\n");
+                       return;
+               }
+               gdp->node_list[i].top_field = base;
+               DRM_DEBUG_DRIVER("node[%d].top_field=%p\n", i, base);
+               base += sizeof(struct sti_gdp_node);
+
+               if (virt_to_dma(layer->dev, base) & 0xF) {
+                       DRM_ERROR("Mem alignment failed\n");
+                       return;
+               }
+               gdp->node_list[i].btm_field = base;
+               DRM_DEBUG_DRIVER("node[%d].btm_field=%p\n", i, base);
+               base += sizeof(struct sti_gdp_node);
+       }
+
+       if (of_device_is_compatible(np, "st,stih407-compositor")) {
+               /* GDP of STiH407 chip have its own pixel clock */
+               char *clk_name;
+
+               switch (layer->desc) {
+               case STI_GDP_0:
+                       clk_name = "pix_gdp1";
+                       break;
+               case STI_GDP_1:
+                       clk_name = "pix_gdp2";
+                       break;
+               case STI_GDP_2:
+                       clk_name = "pix_gdp3";
+                       break;
+               case STI_GDP_3:
+                       clk_name = "pix_gdp4";
+                       break;
+               default:
+                       DRM_ERROR("GDP id not recognized\n");
+                       return;
+               }
+
+               gdp->clk_pix = devm_clk_get(layer->dev, clk_name);
+               if (IS_ERR(gdp->clk_pix))
+                       DRM_ERROR("Cannot get %s clock\n", clk_name);
+       }
+}
+
+static const struct sti_layer_funcs gdp_ops = {
+       .get_formats = sti_gdp_get_formats,
+       .get_nb_formats = sti_gdp_get_nb_formats,
+       .init = sti_gdp_init,
+       .prepare = sti_gdp_prepare_layer,
+       .commit = sti_gdp_commit_layer,
+       .disable = sti_gdp_disable_layer,
+};
+
+struct sti_layer *sti_gdp_create(struct device *dev, int id)
+{
+       struct sti_gdp *gdp;
+
+       gdp = devm_kzalloc(dev, sizeof(*gdp), GFP_KERNEL);
+       if (!gdp) {
+               DRM_ERROR("Failed to allocate memory for GDP\n");
+               return NULL;
+       }
+
+       gdp->layer.ops = &gdp_ops;
+       gdp->vtg_field_nb.notifier_call = sti_gdp_field_cb;
+
+       return (struct sti_layer *)gdp;
+}
diff --git a/drivers/gpu/drm/sti/sti_gdp.h b/drivers/gpu/drm/sti/sti_gdp.h
new file mode 100644 (file)
index 0000000..1dab682
--- /dev/null
@@ -0,0 +1,16 @@
+/*
+ * Copyright (C) STMicroelectronics SA 2014
+ * Authors: Benjamin Gaignard <benjamin.gaignard@st.com>
+ *          Fabien Dessenne <fabien.dessenne@st.com>
+ *          for STMicroelectronics.
+ * License terms:  GNU General Public License (GPL), version 2
+ */
+
+#ifndef _STI_GDP_H_
+#define _STI_GDP_H_
+
+#include <linux/types.h>
+
+struct sti_layer *sti_gdp_create(struct device *dev, int id);
+
+#endif
diff --git a/drivers/gpu/drm/sti/sti_layer.h b/drivers/gpu/drm/sti/sti_layer.h
new file mode 100644 (file)
index 0000000..198c377
--- /dev/null
@@ -0,0 +1,123 @@
+/*
+ * Copyright (C) STMicroelectronics SA 2014
+ * Authors: Benjamin Gaignard <benjamin.gaignard@st.com>
+ *          Fabien Dessenne <fabien.dessenne@st.com>
+ *          for STMicroelectronics.
+ * License terms:  GNU General Public License (GPL), version 2
+ */
+
+#ifndef _STI_LAYER_H_
+#define _STI_LAYER_H_
+
+#include <drm/drmP.h>
+
+#define to_sti_layer(x) container_of(x, struct sti_layer, plane)
+
+#define STI_LAYER_TYPE_SHIFT 8
+#define STI_LAYER_TYPE_MASK (~((1<<STI_LAYER_TYPE_SHIFT)-1))
+
+struct sti_layer;
+
+enum sti_layer_type {
+       STI_GDP = 1 << STI_LAYER_TYPE_SHIFT,
+       STI_VID = 2 << STI_LAYER_TYPE_SHIFT,
+       STI_CUR = 3 << STI_LAYER_TYPE_SHIFT,
+       STI_BCK = 4 << STI_LAYER_TYPE_SHIFT
+};
+
+enum sti_layer_id_of_type {
+       STI_ID_0 = 0,
+       STI_ID_1 = 1,
+       STI_ID_2 = 2,
+       STI_ID_3 = 3
+};
+
+enum sti_layer_desc {
+       STI_GDP_0       = STI_GDP | STI_ID_0,
+       STI_GDP_1       = STI_GDP | STI_ID_1,
+       STI_GDP_2       = STI_GDP | STI_ID_2,
+       STI_GDP_3       = STI_GDP | STI_ID_3,
+       STI_VID_0       = STI_VID | STI_ID_0,
+       STI_VID_1       = STI_VID | STI_ID_1,
+       STI_CURSOR      = STI_CUR,
+       STI_BACK        = STI_BCK
+};
+
+/**
+ * STI layer functions structure
+ *
+ * @get_formats:       get layer supported formats
+ * @get_nb_formats:    get number of format supported
+ * @init:               initialize the layer
+ * @prepare:           prepare layer before rendering
+ * @commit:            set layer for rendering
+ * @disable:           disable layer
+ */
+struct sti_layer_funcs {
+       const uint32_t* (*get_formats)(struct sti_layer *layer);
+       unsigned int (*get_nb_formats)(struct sti_layer *layer);
+       void (*init)(struct sti_layer *layer);
+       int (*prepare)(struct sti_layer *layer, bool first_prepare);
+       int (*commit)(struct sti_layer *layer);
+       int (*disable)(struct sti_layer *layer);
+};
+
+/**
+ * STI layer structure
+ *
+ * @plane:              drm plane it is bound to (if any)
+ * @fb:                 drm fb it is bound to
+ * @mode:               display mode
+ * @desc:               layer type & id
+ * @device:            driver device
+ * @regs:              layer registers
+ * @ops:                layer functions
+ * @zorder:             layer z-order
+ * @mixer_id:           id of the mixer used to display the layer
+ * @enabled:            to know if the layer is active or not
+ * @src_x src_y:        coordinates of the input (fb) area
+ * @src_w src_h:        size of the input (fb) area
+ * @dst_x dst_y:        coordinates of the output (crtc) area
+ * @dst_w dst_h:        size of the output (crtc) area
+ * @format:             format
+ * @pitches:            pitch of 'planes' (eg: Y, U, V)
+ * @offsets:            offset of 'planes'
+ * @paddr:              physical address of the input buffer
+ */
+struct sti_layer {
+       struct drm_plane plane;
+       struct drm_framebuffer *fb;
+       struct drm_display_mode *mode;
+       enum sti_layer_desc desc;
+       struct device *dev;
+       void __iomem *regs;
+       const struct sti_layer_funcs *ops;
+       int zorder;
+       int mixer_id;
+       bool enabled;
+       int src_x, src_y;
+       int src_w, src_h;
+       int dst_x, dst_y;
+       int dst_w, dst_h;
+       uint32_t format;
+       unsigned int pitches[4];
+       unsigned int offsets[4];
+       dma_addr_t paddr;
+};
+
+struct sti_layer *sti_layer_create(struct device *dev, int desc,
+                       void __iomem *baseaddr);
+int sti_layer_prepare(struct sti_layer *layer, struct drm_framebuffer *fb,
+                       struct drm_display_mode *mode,
+                       int mixer_id,
+                       int dest_x, int dest_y,
+                       int dest_w, int dest_h,
+                       int src_x, int src_y,
+                       int src_w, int src_h);
+int sti_layer_commit(struct sti_layer *layer);
+int sti_layer_disable(struct sti_layer *layer);
+const uint32_t *sti_layer_get_formats(struct sti_layer *layer);
+unsigned int sti_layer_get_nb_formats(struct sti_layer *layer);
+const char *sti_layer_to_str(struct sti_layer *layer);
+
+#endif