drm/mediatek: Add AFBC support to Mediatek DRM driver

Tested on MT8195 and confirmed both correct video output and improved DRAM
bandwidth performance.

Signed-off-by: Justin Green <greenjustin@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://patchwork.kernel.org/project/linux-mediatek/patch/20221116193335.36320-1-greenjustin@google.com/
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>

Author: Justin Green

drivers/gpu/drm/mediatek/mtk_disp_ovl.c

@@ -29,17 +29,22 @@
 #define DISP_REG_OVL_DATAPATH_CON		0x0024
 #define OVL_LAYER_SMI_ID_EN				BIT(0)
 #define OVL_BGCLR_SEL_IN				BIT(2)
+#define OVL_LAYER_AFBC_EN(n)				BIT(4+n)
 #define DISP_REG_OVL_ROI_BGCLR			0x0028
 #define DISP_REG_OVL_SRC_CON			0x002c
 #define DISP_REG_OVL_CON(n)			(0x0030 + 0x20 * (n))
 #define DISP_REG_OVL_SRC_SIZE(n)		(0x0038 + 0x20 * (n))
 #define DISP_REG_OVL_OFFSET(n)			(0x003c + 0x20 * (n))
+#define DISP_REG_OVL_PITCH_MSB(n)		(0x0040 + 0x20 * (n))
+#define OVL_PITCH_MSB_2ND_SUBBUF			BIT(16)
 #define DISP_REG_OVL_PITCH(n)			(0x0044 + 0x20 * (n))
 #define DISP_REG_OVL_RDMA_CTRL(n)		(0x00c0 + 0x20 * (n))
 #define DISP_REG_OVL_RDMA_GMC(n)		(0x00c8 + 0x20 * (n))
 #define DISP_REG_OVL_ADDR_MT2701		0x0040
 #define DISP_REG_OVL_ADDR_MT8173		0x0f40
 #define DISP_REG_OVL_ADDR(ovl, n)		((ovl)->data->addr + 0x20 * (n))
+#define DISP_REG_OVL_HDR_ADDR(ovl, n)		((ovl)->data->addr + 0x20 * (n) + 0x04)
+#define DISP_REG_OVL_HDR_PITCH(ovl, n)		((ovl)->data->addr + 0x20 * (n) + 0x08)
 
 #define GMC_THRESHOLD_BITS	16
 #define GMC_THRESHOLD_HIGH	((1 << GMC_THRESHOLD_BITS) / 4)
@@ -67,6 +72,7 @@ struct mtk_disp_ovl_data {
 	unsigned int layer_nr;
 	bool fmt_rgb565_is_0;
 	bool smi_id_en;
+	bool supports_afbc;
 };
 
 /*
@@ -172,7 +178,14 @@ void mtk_ovl_stop(struct device *dev)
 		reg = reg & ~OVL_LAYER_SMI_ID_EN;
 		writel_relaxed(reg, ovl->regs + DISP_REG_OVL_DATAPATH_CON);
 	}
+}
 
+static void mtk_ovl_set_afbc(struct mtk_disp_ovl *ovl, struct cmdq_pkt *cmdq_pkt,
+			     int idx, bool enabled)
+{
+	mtk_ddp_write_mask(cmdq_pkt, enabled ? OVL_LAYER_AFBC_EN(idx) : 0,
+			   &ovl->cmdq_reg, ovl->regs,
+			   DISP_REG_OVL_DATAPATH_CON, OVL_LAYER_AFBC_EN(idx));
 }
 
 void mtk_ovl_config(struct device *dev, unsigned int w,
@@ -310,11 +323,23 @@ void mtk_ovl_layer_config(struct device *dev, unsigned int idx,
 	struct mtk_disp_ovl *ovl = dev_get_drvdata(dev);
 	struct mtk_plane_pending_state *pending = &state->pending;
 	unsigned int addr = pending->addr;
-	unsigned int pitch = pending->pitch & 0xffff;
+	unsigned int hdr_addr = pending->hdr_addr;
+	unsigned int pitch = pending->pitch;
+	unsigned int hdr_pitch = pending->hdr_pitch;
 	unsigned int fmt = pending->format;
 	unsigned int offset = (pending->y << 16) | pending->x;
 	unsigned int src_size = (pending->height << 16) | pending->width;
 	unsigned int con;
+	bool is_afbc = pending->modifier != DRM_FORMAT_MOD_LINEAR;
+	union overlay_pitch {
+		struct split_pitch {
+			u16 lsb;
+			u16 msb;
+		} split_pitch;
+		u32 pitch;
+	} overlay_pitch;
+
+	overlay_pitch.pitch = pitch;
 
 	if (!pending->enable) {
 		mtk_ovl_layer_off(dev, idx, cmdq_pkt);
@@ -335,9 +360,12 @@ void mtk_ovl_layer_config(struct device *dev, unsigned int idx,
 		addr += pending->pitch - 1;
 	}
 
+	if (ovl->data->supports_afbc)
+		mtk_ovl_set_afbc(ovl, cmdq_pkt, idx, is_afbc);
+
 	mtk_ddp_write_relaxed(cmdq_pkt, con, &ovl->cmdq_reg, ovl->regs,
 			      DISP_REG_OVL_CON(idx));
-	mtk_ddp_write_relaxed(cmdq_pkt, pitch, &ovl->cmdq_reg, ovl->regs,
+	mtk_ddp_write_relaxed(cmdq_pkt, overlay_pitch.split_pitch.lsb, &ovl->cmdq_reg, ovl->regs,
 			      DISP_REG_OVL_PITCH(idx));
 	mtk_ddp_write_relaxed(cmdq_pkt, src_size, &ovl->cmdq_reg, ovl->regs,
 			      DISP_REG_OVL_SRC_SIZE(idx));
@@ -346,6 +374,20 @@ void mtk_ovl_layer_config(struct device *dev, unsigned int idx,
 	mtk_ddp_write_relaxed(cmdq_pkt, addr, &ovl->cmdq_reg, ovl->regs,
 			      DISP_REG_OVL_ADDR(ovl, idx));
 
+	if (is_afbc) {
+		mtk_ddp_write_relaxed(cmdq_pkt, hdr_addr, &ovl->cmdq_reg, ovl->regs,
+				      DISP_REG_OVL_HDR_ADDR(ovl, idx));
+		mtk_ddp_write_relaxed(cmdq_pkt,
+				      OVL_PITCH_MSB_2ND_SUBBUF | overlay_pitch.split_pitch.msb,
+				      &ovl->cmdq_reg, ovl->regs, DISP_REG_OVL_PITCH_MSB(idx));
+		mtk_ddp_write_relaxed(cmdq_pkt, hdr_pitch, &ovl->cmdq_reg, ovl->regs,
+				      DISP_REG_OVL_HDR_PITCH(ovl, idx));
+	} else {
+		mtk_ddp_write_relaxed(cmdq_pkt,
+				      overlay_pitch.split_pitch.msb,
+				      &ovl->cmdq_reg, ovl->regs, DISP_REG_OVL_PITCH_MSB(idx));
+	}
+
 	mtk_ovl_layer_on(dev, idx, cmdq_pkt);
 }
 

drivers/gpu/drm/mediatek/mtk_drm_plane.c

@@ -11,6 +11,7 @@
 #include <drm/drm_fourcc.h>
 #include <drm/drm_framebuffer.h>
 #include <drm/drm_gem_atomic_helper.h>
+#include <linux/align.h>
 
 #include "mtk_drm_crtc.h"
 #include "mtk_drm_ddp_comp.h"
@@ -32,6 +33,14 @@ static const u32 formats[] = {
 	DRM_FORMAT_YUYV,
 };
 
+static const u64 modifiers[] = {
+	DRM_FORMAT_MOD_LINEAR,
+	DRM_FORMAT_MOD_ARM_AFBC(AFBC_FORMAT_MOD_BLOCK_SIZE_32x8 |
+				AFBC_FORMAT_MOD_SPLIT |
+				AFBC_FORMAT_MOD_SPARSE),
+	DRM_FORMAT_MOD_INVALID,
+};
+
 static void mtk_plane_reset(struct drm_plane *plane)
 {
 	struct mtk_plane_state *state;
@@ -51,6 +60,7 @@ static void mtk_plane_reset(struct drm_plane *plane)
 
 	state->base.plane = plane;
 	state->pending.format = DRM_FORMAT_RGB565;
+	state->pending.modifier = DRM_FORMAT_MOD_LINEAR;
 }
 
 static struct drm_plane_state *mtk_plane_duplicate_state(struct drm_plane *plane)
@@ -71,6 +81,32 @@ static struct drm_plane_state *mtk_plane_duplicate_state(struct drm_plane *plane
 	return &state->base;
 }
 
+static bool mtk_plane_format_mod_supported(struct drm_plane *plane,
+					   uint32_t format,
+					   uint64_t modifier)
+{
+	if (modifier == DRM_FORMAT_MOD_LINEAR)
+		return true;
+
+	if (modifier != DRM_FORMAT_MOD_ARM_AFBC(
+				AFBC_FORMAT_MOD_BLOCK_SIZE_32x8 |
+				AFBC_FORMAT_MOD_SPLIT |
+				AFBC_FORMAT_MOD_SPARSE))
+		return false;
+
+	if (format != DRM_FORMAT_XRGB8888 &&
+	    format != DRM_FORMAT_ARGB8888 &&
+	    format != DRM_FORMAT_BGRX8888 &&
+	    format != DRM_FORMAT_BGRA8888 &&
+	    format != DRM_FORMAT_ABGR8888 &&
+	    format != DRM_FORMAT_XBGR8888 &&
+	    format != DRM_FORMAT_RGB888 &&
+	    format != DRM_FORMAT_BGR888)
+		return false;
+
+	return true;
+}
+
 static void mtk_drm_plane_destroy_state(struct drm_plane *plane,
 					struct drm_plane_state *state)
 {
@@ -119,21 +155,52 @@ static void mtk_plane_update_new_state(struct drm_plane_state *new_state,
 	struct drm_gem_object *gem;
 	struct mtk_drm_gem_obj *mtk_gem;
 	unsigned int pitch, format;
+	u64 modifier;
 	dma_addr_t addr;
+	dma_addr_t hdr_addr = 0;
+	unsigned int hdr_pitch = 0;
 
 	gem = fb->obj[0];
 	mtk_gem = to_mtk_gem_obj(gem);
 	addr = mtk_gem->dma_addr;
 	pitch = fb->pitches[0];
 	format = fb->format->format;
+	modifier = fb->modifier;
 
-	addr += (new_state->src.x1 >> 16) * fb->format->cpp[0];
-	addr += (new_state->src.y1 >> 16) * pitch;
+	if (modifier == DRM_FORMAT_MOD_LINEAR) {
+		addr += (new_state->src.x1 >> 16) * fb->format->cpp[0];
+		addr += (new_state->src.y1 >> 16) * pitch;
+	} else {
+		int width_in_blocks = ALIGN(fb->width, AFBC_DATA_BLOCK_WIDTH)
+				      / AFBC_DATA_BLOCK_WIDTH;
+		int height_in_blocks = ALIGN(fb->height, AFBC_DATA_BLOCK_HEIGHT)
+				       / AFBC_DATA_BLOCK_HEIGHT;
+		int x_offset_in_blocks = (new_state->src.x1 >> 16) / AFBC_DATA_BLOCK_WIDTH;
+		int y_offset_in_blocks = (new_state->src.y1 >> 16) / AFBC_DATA_BLOCK_HEIGHT;
+		int hdr_size;
+
+		hdr_pitch = width_in_blocks * AFBC_HEADER_BLOCK_SIZE;
+		pitch = width_in_blocks * AFBC_DATA_BLOCK_WIDTH *
+			AFBC_DATA_BLOCK_HEIGHT * fb->format->cpp[0];
+
+		hdr_size = ALIGN(hdr_pitch * height_in_blocks, AFBC_HEADER_ALIGNMENT);
+
+		hdr_addr = addr + hdr_pitch * y_offset_in_blocks +
+			   AFBC_HEADER_BLOCK_SIZE * x_offset_in_blocks;
+		/* The data plane is offset by 1 additional block. */
+		addr = addr + hdr_size +
+		       pitch * y_offset_in_blocks +
+		       AFBC_DATA_BLOCK_WIDTH * AFBC_DATA_BLOCK_HEIGHT *
+		       fb->format->cpp[0] * (x_offset_in_blocks + 1);
+	}
 
 	mtk_plane_state->pending.enable = true;
 	mtk_plane_state->pending.pitch = pitch;
+	mtk_plane_state->pending.hdr_pitch = hdr_pitch;
 	mtk_plane_state->pending.format = format;
+	mtk_plane_state->pending.modifier = modifier;
 	mtk_plane_state->pending.addr = addr;
+	mtk_plane_state->pending.hdr_addr = hdr_addr;
 	mtk_plane_state->pending.x = new_state->dst.x1;
 	mtk_plane_state->pending.y = new_state->dst.y1;
 	mtk_plane_state->pending.width = drm_rect_width(&new_state->dst);
@@ -172,6 +239,7 @@ static const struct drm_plane_funcs mtk_plane_funcs = {
 	.reset = mtk_plane_reset,
 	.atomic_duplicate_state = mtk_plane_duplicate_state,
 	.atomic_destroy_state = mtk_drm_plane_destroy_state,
+	.format_mod_supported = mtk_plane_format_mod_supported,
 };
 
 static int mtk_plane_atomic_check(struct drm_plane *plane,
@@ -253,7 +321,7 @@ int mtk_plane_init(struct drm_device *dev, struct drm_plane *plane,
 
 	err = drm_universal_plane_init(dev, plane, possible_crtcs,
 				       &mtk_plane_funcs, formats,
-				       ARRAY_SIZE(formats), NULL, type, NULL);
+				       ARRAY_SIZE(formats), modifiers, type, NULL);
 	if (err) {
 		DRM_ERROR("failed to initialize plane\n");
 		return err;

drivers/gpu/drm/mediatek/mtk_drm_plane.h

@@ -10,12 +10,20 @@
 #include <drm/drm_crtc.h>
 #include <linux/types.h>
 
+#define AFBC_DATA_BLOCK_WIDTH 32
+#define AFBC_DATA_BLOCK_HEIGHT 8
+#define AFBC_HEADER_BLOCK_SIZE 16
+#define AFBC_HEADER_ALIGNMENT 1024
+
 struct mtk_plane_pending_state {
 	bool				config;
 	bool				enable;
 	dma_addr_t			addr;
+	dma_addr_t			hdr_addr;
 	unsigned int			pitch;
+	unsigned int			hdr_pitch;
 	unsigned int			format;
+	unsigned long long		modifier;
 	unsigned int			x;
 	unsigned int			y;
 	unsigned int			width;