static void rk3399_vpu_set_src_img_ctrl(struct rockchip_vpu_dev *vpu,
struct rockchip_vpu_ctx *ctx)
{
struct v4l2_pix_format_mplane *pix_fmt = &ctx->src_fmt;
u32 reg;
/*
* The pix fmt width/height are already macroblock aligned
* by .vidioc_s_fmt_vid_cap_mplane() callback
*/
reg = VEPU_REG_IN_IMG_CTRL_ROW_LEN(pix_fmt->width);
vepu_write_relaxed(vpu, reg, VEPU_REG_INPUT_LUMA_INFO);
reg = VEPU_REG_IN_IMG_CTRL_OVRFLR_D4(0) |
VEPU_REG_IN_IMG_CTRL_OVRFLB(0);
/*
* This register controls the input crop, as the offset
* from the right/bottom within the last macroblock. The offset from the
* right must be divided by 4 and so the crop must be aligned to 4 pixels
* horizontally.
*/
vepu_write_relaxed(vpu, reg, VEPU_REG_ENC_OVER_FILL_STRM_OFFSET);
reg = VEPU_REG_IN_IMG_CTRL_FMT(ctx->vpu_src_fmt->enc_fmt);
vepu_write_relaxed(vpu, reg, VEPU_REG_ENC_CTRL1);
}
static void rk3399_vpu_jpeg_enc_set_buffers(struct rockchip_vpu_dev *vpu,
struct rockchip_vpu_ctx *ctx,
struct vb2_buffer *src_buf)
{
struct v4l2_pix_format_mplane *pix_fmt = &ctx->src_fmt;
dma_addr_t src[3];
WARN_ON(pix_fmt->num_planes > 3);
vepu_write_relaxed(vpu, ctx->jpeg_enc.bounce_buffer.dma,
VEPU_REG_ADDR_OUTPUT_STREAM);
vepu_write_relaxed(vpu, ctx->jpeg_enc.bounce_buffer.size,
VEPU_REG_STR_BUF_LIMIT);
if (pix_fmt->num_planes == 1) {
src[0] = vb2_dma_contig_plane_dma_addr(src_buf, 0);
vepu_write_relaxed(vpu, src[0], VEPU_REG_ADDR_IN_PLANE_0);
} else if (pix_fmt->num_planes == 2) {
src[0] = vb2_dma_contig_plane_dma_addr(src_buf, 0);
src[1] = vb2_dma_contig_plane_dma_addr(src_buf, 1);
vepu_write_relaxed(vpu, src[0], VEPU_REG_ADDR_IN_PLANE_0);
vepu_write_relaxed(vpu, src[1], VEPU_REG_ADDR_IN_PLANE_1);
} else {
src[0] = vb2_dma_contig_plane_dma_addr(src_buf, 0);
src[1] = vb2_dma_contig_plane_dma_addr(src_buf, 1);
src[2] = vb2_dma_contig_plane_dma_addr(src_buf, 2);
vepu_write_relaxed(vpu, src[0], VEPU_REG_ADDR_IN_PLANE_0);
vepu_write_relaxed(vpu, src[1], VEPU_REG_ADDR_IN_PLANE_1);
vepu_write_relaxed(vpu, src[2], VEPU_REG_ADDR_IN_PLANE_2);
}
}
void rk3399_vpu_jpeg_enc_run(struct rockchip_vpu_ctx *ctx)
{
struct rockchip_vpu_dev *vpu = ctx->dev;
struct vb2_v4l2_buffer *src_buf, *dst_buf;
struct rockchip_vpu_jpeg_ctx jpeg_ctx;
u32 reg;
src_buf = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx);
dst_buf = v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx);
memset(&jpeg_ctx, 0, sizeof(jpeg_ctx));
jpeg_ctx.buffer = vb2_plane_vaddr(&dst_buf->vb2_buf, 0);
jpeg_ctx.width = ctx->dst_fmt.width;
jpeg_ctx.height = ctx->dst_fmt.height;
jpeg_ctx.quality = ctx->jpeg_quality;
rockchip_vpu_jpeg_header_assemble(&jpeg_ctx);
/* Switch to JPEG encoder mode before writing registers */
vepu_write_relaxed(vpu, VEPU_REG_ENCODE_FORMAT_JPEG,
VEPU_REG_ENCODE_START);
rk3399_vpu_set_src_img_ctrl(vpu, ctx);
rk3399_vpu_jpeg_enc_set_buffers(vpu, ctx, &src_buf->vb2_buf);
rk3399_vpu_jpeg_enc_set_qtable(vpu,
rockchip_vpu_jpeg_get_qtable(&jpeg_ctx, 0),
rockchip_vpu_jpeg_get_qtable(&jpeg_ctx, 1));
reg = VEPU_REG_OUTPUT_SWAP32
| VEPU_REG_OUTPUT_SWAP16
| VEPU_REG_OUTPUT_SWAP8
| VEPU_REG_INPUT_SWAP8
| VEPU_REG_INPUT_SWAP16
| VEPU_REG_INPUT_SWAP32;
/* Make sure that all registers are written at this point. */
vepu_write(vpu, reg, VEPU_REG_DATA_ENDIAN);
reg = VEPU_REG_AXI_CTRL_BURST_LEN(16);
vepu_write_relaxed(vpu, reg, VEPU_REG_AXI_CTRL);
reg = VEPU_REG_MB_WIDTH(JPEG_MB_WIDTH(ctx->src_fmt.width))
| VEPU_REG_MB_HEIGHT(JPEG_MB_HEIGHT(ctx->src_fmt.height))
| VEPU_REG_FRAME_TYPE_INTRA
| VEPU_REG_ENCODE_FORMAT_JPEG
| VEPU_REG_ENCODE_ENABLE;
/* Kick the watchdog and start encoding */
schedule_delayed_work(&vpu->watchdog_work, msecs_to_jiffies(2000));
vepu_write(vpu, reg, VEPU_REG_ENCODE_START);
}
static void
rk3399_vpu_jpeg_enc_set_qtable(struct rockchip_vpu_dev *vpu,
unsigned char *luma_qtable,
unsigned char *chroma_qtable)
{
u32 reg, i;
for (i = 0; i < VEPU_JPEG_QUANT_TABLE_COUNT; i++) {
reg = get_unaligned_be32(&luma_qtable[i]);
vepu_write_relaxed(vpu, reg, VEPU_REG_JPEG_LUMA_QUAT(i));
reg = get_unaligned_be32(&chroma_qtable[i]);
vepu_write_relaxed(vpu, reg, VEPU_REG_JPEG_CHROMA_QUAT(i));
}
}
static void rk3288_vpu_set_src_img_ctrl(struct rockchip_vpu_dev *vpu,
struct rockchip_vpu_ctx *ctx)
{
struct v4l2_pix_format_mplane *pix_fmt = &ctx->src_fmt;
u32 reg;
reg = VEPU_REG_IN_IMG_CTRL_ROW_LEN(pix_fmt->width)
| VEPU_REG_IN_IMG_CTRL_OVRFLR_D4(0)
| VEPU_REG_IN_IMG_CTRL_OVRFLB_D4(0)
| VEPU_REG_IN_IMG_CTRL_FMT(ctx->vpu_src_fmt->enc_fmt);
vepu_write_relaxed(vpu, reg, VEPU_REG_IN_IMG_CTRL);
}
void rk3288_vpu_vp8e_run(struct rk3288_vpu_ctx *ctx)
{
struct rk3288_vpu_dev *vpu = ctx->dev;
u32 reg;
/* The hardware expects the control buffer to be zeroed. */
memset(ctx->hw.vp8e.ctrl_buf.cpu, 0,
sizeof(struct rk3288_vpu_vp8e_ctrl_buf));
/*
* Program the hardware.
*/
rk3288_vpu_power_on(vpu);
vepu_write_relaxed(vpu, VEPU_REG_ENC_CTRL_ENC_MODE_VP8,
VEPU_REG_ENC_CTRL);
rk3288_vpu_vp8e_set_params(vpu, ctx);
rk3288_vpu_vp8e_set_buffers(vpu, ctx);
/* Make sure that all registers are written at this point. */
wmb();
/* Set the watchdog. */
schedule_delayed_work(&vpu->watchdog_work, msecs_to_jiffies(2000));
/* Start the hardware. */
reg = VEPU_REG_AXI_CTRL_OUTPUT_SWAP16
| VEPU_REG_AXI_CTRL_INPUT_SWAP16
| VEPU_REG_AXI_CTRL_BURST_LEN(16)
| VEPU_REG_AXI_CTRL_GATE_BIT
| VEPU_REG_AXI_CTRL_OUTPUT_SWAP32
| VEPU_REG_AXI_CTRL_INPUT_SWAP32
| VEPU_REG_AXI_CTRL_OUTPUT_SWAP8
| VEPU_REG_AXI_CTRL_INPUT_SWAP8;
vepu_write(vpu, reg, VEPU_REG_AXI_CTRL);
vepu_write(vpu, 0, VEPU_REG_INTERRUPT);
reg = VEPU_REG_ENC_CTRL_NAL_MODE_BIT
| VEPU_REG_ENC_CTRL_WIDTH(MB_WIDTH(ctx->src_fmt.width))
| VEPU_REG_ENC_CTRL_HEIGHT(MB_HEIGHT(ctx->src_fmt.height))
| VEPU_REG_ENC_CTRL_ENC_MODE_VP8
| VEPU_REG_ENC_CTRL_EN_BIT;
if (ctx->run.dst->b.v4l2_buf.flags & V4L2_BUF_FLAG_KEYFRAME)
reg |= VEPU_REG_ENC_CTRL_KEYFRAME_BIT;
vepu_write(vpu, reg, VEPU_REG_ENC_CTRL);
}
static void rk3288_vpu_vp8e_set_params(struct rk3288_vpu_dev *vpu,
struct rk3288_vpu_ctx *ctx)
{
const struct rk3288_vp8e_reg_params *params = ctx->run.vp8e.reg_params;
int i;
vepu_write_relaxed(vpu, params->enc_ctrl0, VEPU_REG_ENC_CTRL0);
vepu_write_relaxed(vpu, params->enc_ctrl1, VEPU_REG_ENC_CTRL1);
vepu_write_relaxed(vpu, params->enc_ctrl2, VEPU_REG_ENC_CTRL2);
vepu_write_relaxed(vpu, params->enc_ctrl3, VEPU_REG_ENC_CTRL3);
vepu_write_relaxed(vpu, params->enc_ctrl5, VEPU_REG_ENC_CTRL5);
vepu_write_relaxed(vpu, params->enc_ctrl4, VEPU_REG_ENC_CTRL4);
vepu_write_relaxed(vpu, params->str_hdr_rem_msb,
VEPU_REG_STR_HDR_REM_MSB);
vepu_write_relaxed(vpu, params->str_hdr_rem_lsb,
VEPU_REG_STR_HDR_REM_LSB);
vepu_write_relaxed(vpu, params->mad_ctrl, VEPU_REG_MAD_CTRL);
for (i = 0; i < ARRAY_SIZE(params->qp_val); ++i)
vepu_write_relaxed(vpu, params->qp_val[i],
VEPU_REG_VP8_QP_VAL(i));
vepu_write_relaxed(vpu, params->bool_enc, VEPU_REG_VP8_BOOL_ENC);
vepu_write_relaxed(vpu, params->vp8_ctrl0, VEPU_REG_VP8_CTRL0);
vepu_write_relaxed(vpu, params->rlc_ctrl, VEPU_REG_RLC_CTRL);
vepu_write_relaxed(vpu, params->mb_ctrl, VEPU_REG_MB_CTRL);
for (i = 0; i < ARRAY_SIZE(params->rgb_yuv_coeff); ++i)
vepu_write_relaxed(vpu, params->rgb_yuv_coeff[i],
VEPU_REG_RGB_YUV_COEFF(i));
vepu_write_relaxed(vpu, params->rgb_mask_msb,
VEPU_REG_RGB_MASK_MSB);
vepu_write_relaxed(vpu, params->intra_area_ctrl,
VEPU_REG_INTRA_AREA_CTRL);
vepu_write_relaxed(vpu, params->cir_intra_ctrl,
VEPU_REG_CIR_INTRA_CTRL);
vepu_write_relaxed(vpu, params->first_roi_area,
VEPU_REG_FIRST_ROI_AREA);
vepu_write_relaxed(vpu, params->second_roi_area,
VEPU_REG_SECOND_ROI_AREA);
vepu_write_relaxed(vpu, params->mvc_ctrl,
VEPU_REG_MVC_CTRL);
for (i = 0; i < ARRAY_SIZE(params->intra_penalty); ++i)
vepu_write_relaxed(vpu, params->intra_penalty[i],
VEPU_REG_VP8_INTRA_PENALTY(i));
for (i = 0; i < ARRAY_SIZE(params->seg_qp); ++i)
vepu_write_relaxed(vpu, params->seg_qp[i],
VEPU_REG_VP8_SEG_QP(i));
for (i = 0; i < ARRAY_SIZE(params->dmv_4p_1p_penalty); ++i)
vepu_write_relaxed(vpu, params->dmv_4p_1p_penalty[i],
VEPU_REG_DMV_4P_1P_PENALTY(i));
for (i = 0; i < ARRAY_SIZE(params->dmv_qpel_penalty); ++i)
vepu_write_relaxed(vpu, params->dmv_qpel_penalty[i],
VEPU_REG_DMV_QPEL_PENALTY(i));
vepu_write_relaxed(vpu, params->vp8_ctrl1, VEPU_REG_VP8_CTRL1);
vepu_write_relaxed(vpu, params->bit_cost_golden,
VEPU_REG_VP8_BIT_COST_GOLDEN);
for (i = 0; i < ARRAY_SIZE(params->loop_flt_delta); ++i)
vepu_write_relaxed(vpu, params->loop_flt_delta[i],
VEPU_REG_VP8_LOOP_FLT_DELTA(i));
}
static void rk3288_vpu_vp8e_set_buffers(struct rk3288_vpu_dev *vpu,
struct rk3288_vpu_ctx *ctx)
{
const struct rk3288_vp8e_reg_params *params = ctx->run.vp8e.reg_params;
dma_addr_t ref_buf_dma, rec_buf_dma;
dma_addr_t stream_dma;
size_t rounded_size;
dma_addr_t dst_dma;
u32 start_offset;
size_t dst_size;
rounded_size = ref_luma_size(ctx->src_fmt.width,
ctx->src_fmt.height);
ref_buf_dma = rec_buf_dma = ctx->hw.vp8e.ext_buf.dma;
if (ctx->hw.vp8e.ref_rec_ptr)
ref_buf_dma += rounded_size * 3 / 2;
else
rec_buf_dma += rounded_size * 3 / 2;
ctx->hw.vp8e.ref_rec_ptr ^= 1;
dst_dma = vb2_dma_contig_plane_dma_addr(&ctx->run.dst->b, 0);
dst_size = vb2_plane_size(&ctx->run.dst->b, 0);
/*
* stream addr-->|
* align 64bits->|<-start offset->|
* |<---------header size-------->|<---dst buf---
*/
start_offset = (params->rlc_ctrl & VEPU_REG_RLC_CTRL_STR_OFFS_MASK)
>> VEPU_REG_RLC_CTRL_STR_OFFS_SHIFT;
stream_dma = dst_dma + params->hdr_len;
/**
* Userspace will pass 8 bytes aligned size(round_down) to us,
* so we need to plus start offset to get real header size.
*
* |<-aligned size->|<-start offset->|
* |<----------header size---------->|
*/
ctx->run.dst->vp8e.hdr_size = params->hdr_len + (start_offset >> 3);
if (params->enc_ctrl & VEPU_REG_ENC_CTRL_KEYFRAME_BIT)
ctx->run.dst->b.v4l2_buf.flags |= V4L2_BUF_FLAG_KEYFRAME;
else
ctx->run.dst->b.v4l2_buf.flags &= ~V4L2_BUF_FLAG_KEYFRAME;
/*
* We assume here that 1/10 of the buffer is enough for headers.
* DCT partition will be placed in remaining 9/10 of the buffer.
*/
ctx->run.dst->vp8e.dct_offset = round_up(dst_size / 10, 8);
/* Destination buffer. */
vepu_write_relaxed(vpu, stream_dma, VEPU_REG_ADDR_OUTPUT_STREAM);
vepu_write_relaxed(vpu, dst_dma + ctx->run.dst->vp8e.dct_offset,
VEPU_REG_ADDR_VP8_DCT_PART(0));
vepu_write_relaxed(vpu, dst_size - ctx->run.dst->vp8e.dct_offset,
VEPU_REG_STR_BUF_LIMIT);
/* Auxilliary buffers. */
vepu_write_relaxed(vpu, ctx->hw.vp8e.ctrl_buf.dma,
VEPU_REG_ADDR_OUTPUT_CTRL);
vepu_write_relaxed(vpu, ctx->hw.vp8e.mv_buf.dma,
VEPU_REG_ADDR_MV_OUT);
vepu_write_relaxed(vpu, ctx->run.priv_dst.dma,
VEPU_REG_ADDR_VP8_PROB_CNT);
vepu_write_relaxed(vpu, ctx->run.priv_src.dma + VP8_CABAC_CTX_OFFSET,
VEPU_REG_ADDR_CABAC_TBL);
vepu_write_relaxed(vpu, ctx->run.priv_src.dma
+ VP8_CABAC_CTX_OFFSET + VP8_CABAC_CTX_SIZE,
VEPU_REG_ADDR_VP8_SEG_MAP);
/* Reference buffers. */
vepu_write_relaxed(vpu, ref_buf_dma,
VEPU_REG_ADDR_REF_LUMA);
vepu_write_relaxed(vpu, ref_buf_dma + rounded_size,
VEPU_REG_ADDR_REF_CHROMA);
/* Reconstruction buffers. */
vepu_write_relaxed(vpu, rec_buf_dma,
VEPU_REG_ADDR_REC_LUMA);
vepu_write_relaxed(vpu, rec_buf_dma + rounded_size,
VEPU_REG_ADDR_REC_CHROMA);
/* Source buffer. */
vepu_write_relaxed(vpu, vb2_dma_contig_plane_dma_addr(&ctx->run.src->b,
PLANE_Y), VEPU_REG_ADDR_IN_LUMA);
vepu_write_relaxed(vpu, vb2_dma_contig_plane_dma_addr(&ctx->run.src->b,
PLANE_CB), VEPU_REG_ADDR_IN_CB);
vepu_write_relaxed(vpu, vb2_dma_contig_plane_dma_addr(&ctx->run.src->b,
PLANE_CR), VEPU_REG_ADDR_IN_CR);
/* Source parameters. */
vepu_write_relaxed(vpu, enc_in_img_ctrl(ctx), VEPU_REG_IN_IMG_CTRL);
}